feat: Passwordless cross-device authentication

- Arkitektur: docs/auth/passwordless-architecture.md
- Backend: iom/quixzoom-auth-service/ (FastAPI + Redis)
- Webb: quixzoom-market-pages/se/login/ (QR-kod + polling)
- App: iom/quixzoom-app/src/features/auth/ (push + deep links)

Flöde: QR-kod → app-godkännande → webb-inloggad
This commit is contained in:
Bernt
2026-07-07 07:11:50 +00:00
parent 4aa984ad74
commit 6989a98d75
61843 changed files with 5491611 additions and 872231 deletions
+4
View File
@@ -0,0 +1,4 @@
import { DOCUMENT_MODE } from './html.js';
import type { DoctypeToken } from './token.js';
export declare function isConforming(token: DoctypeToken): boolean;
export declare function getDocumentMode(token: DoctypeToken): DOCUMENT_MODE;
+118
View File
@@ -0,0 +1,118 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.isConforming = isConforming;
exports.getDocumentMode = getDocumentMode;
const html_js_1 = require("./html.js");
//Const
const VALID_DOCTYPE_NAME = 'html';
const VALID_SYSTEM_ID = 'about:legacy-compat';
const QUIRKS_MODE_SYSTEM_ID = 'http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd';
const QUIRKS_MODE_PUBLIC_ID_PREFIXES = [
'+//silmaril//dtd html pro v0r11 19970101//',
'-//as//dtd html 3.0 aswedit + extensions//',
'-//advasoft ltd//dtd html 3.0 aswedit + extensions//',
'-//ietf//dtd html 2.0 level 1//',
'-//ietf//dtd html 2.0 level 2//',
'-//ietf//dtd html 2.0 strict level 1//',
'-//ietf//dtd html 2.0 strict level 2//',
'-//ietf//dtd html 2.0 strict//',
'-//ietf//dtd html 2.0//',
'-//ietf//dtd html 2.1e//',
'-//ietf//dtd html 3.0//',
'-//ietf//dtd html 3.2 final//',
'-//ietf//dtd html 3.2//',
'-//ietf//dtd html 3//',
'-//ietf//dtd html level 0//',
'-//ietf//dtd html level 1//',
'-//ietf//dtd html level 2//',
'-//ietf//dtd html level 3//',
'-//ietf//dtd html strict level 0//',
'-//ietf//dtd html strict level 1//',
'-//ietf//dtd html strict level 2//',
'-//ietf//dtd html strict level 3//',
'-//ietf//dtd html strict//',
'-//ietf//dtd html//',
'-//metrius//dtd metrius presentational//',
'-//microsoft//dtd internet explorer 2.0 html strict//',
'-//microsoft//dtd internet explorer 2.0 html//',
'-//microsoft//dtd internet explorer 2.0 tables//',
'-//microsoft//dtd internet explorer 3.0 html strict//',
'-//microsoft//dtd internet explorer 3.0 html//',
'-//microsoft//dtd internet explorer 3.0 tables//',
'-//netscape comm. corp.//dtd html//',
'-//netscape comm. corp.//dtd strict html//',
"-//o'reilly and associates//dtd html 2.0//",
"-//o'reilly and associates//dtd html extended 1.0//",
"-//o'reilly and associates//dtd html extended relaxed 1.0//",
'-//sq//dtd html 2.0 hotmetal + extensions//',
'-//softquad software//dtd hotmetal pro 6.0::19990601::extensions to html 4.0//',
'-//softquad//dtd hotmetal pro 4.0::19971010::extensions to html 4.0//',
'-//spyglass//dtd html 2.0 extended//',
'-//sun microsystems corp.//dtd hotjava html//',
'-//sun microsystems corp.//dtd hotjava strict html//',
'-//w3c//dtd html 3 1995-03-24//',
'-//w3c//dtd html 3.2 draft//',
'-//w3c//dtd html 3.2 final//',
'-//w3c//dtd html 3.2//',
'-//w3c//dtd html 3.2s draft//',
'-//w3c//dtd html 4.0 frameset//',
'-//w3c//dtd html 4.0 transitional//',
'-//w3c//dtd html experimental 19960712//',
'-//w3c//dtd html experimental 970421//',
'-//w3c//dtd w3 html//',
'-//w3o//dtd w3 html 3.0//',
'-//webtechs//dtd mozilla html 2.0//',
'-//webtechs//dtd mozilla html//',
];
const QUIRKS_MODE_NO_SYSTEM_ID_PUBLIC_ID_PREFIXES = [
...QUIRKS_MODE_PUBLIC_ID_PREFIXES,
'-//w3c//dtd html 4.01 frameset//',
'-//w3c//dtd html 4.01 transitional//',
];
const QUIRKS_MODE_PUBLIC_IDS = new Set([
'-//w3o//dtd w3 html strict 3.0//en//',
'-/w3c/dtd html 4.0 transitional/en',
'html',
]);
const LIMITED_QUIRKS_PUBLIC_ID_PREFIXES = ['-//w3c//dtd xhtml 1.0 frameset//', '-//w3c//dtd xhtml 1.0 transitional//'];
const LIMITED_QUIRKS_WITH_SYSTEM_ID_PUBLIC_ID_PREFIXES = [
...LIMITED_QUIRKS_PUBLIC_ID_PREFIXES,
'-//w3c//dtd html 4.01 frameset//',
'-//w3c//dtd html 4.01 transitional//',
];
//Utils
function hasPrefix(publicId, prefixes) {
return prefixes.some((prefix) => publicId.startsWith(prefix));
}
//API
function isConforming(token) {
return (token.name === VALID_DOCTYPE_NAME &&
token.publicId === null &&
(token.systemId === null || token.systemId === VALID_SYSTEM_ID));
}
function getDocumentMode(token) {
if (token.name !== VALID_DOCTYPE_NAME) {
return html_js_1.DOCUMENT_MODE.QUIRKS;
}
const { systemId } = token;
if (systemId && systemId.toLowerCase() === QUIRKS_MODE_SYSTEM_ID) {
return html_js_1.DOCUMENT_MODE.QUIRKS;
}
let { publicId } = token;
if (publicId !== null) {
publicId = publicId.toLowerCase();
if (QUIRKS_MODE_PUBLIC_IDS.has(publicId)) {
return html_js_1.DOCUMENT_MODE.QUIRKS;
}
let prefixes = systemId === null ? QUIRKS_MODE_NO_SYSTEM_ID_PUBLIC_ID_PREFIXES : QUIRKS_MODE_PUBLIC_ID_PREFIXES;
if (hasPrefix(publicId, prefixes)) {
return html_js_1.DOCUMENT_MODE.QUIRKS;
}
prefixes =
systemId === null ? LIMITED_QUIRKS_PUBLIC_ID_PREFIXES : LIMITED_QUIRKS_WITH_SYSTEM_ID_PUBLIC_ID_PREFIXES;
if (hasPrefix(publicId, prefixes)) {
return html_js_1.DOCUMENT_MODE.LIMITED_QUIRKS;
}
}
return html_js_1.DOCUMENT_MODE.NO_QUIRKS;
}
+67
View File
@@ -0,0 +1,67 @@
import type { Location } from './token.js';
export interface ParserError extends Location {
code: ERR;
}
export type ParserErrorHandler = (error: ParserError) => void;
export declare enum ERR {
controlCharacterInInputStream = "control-character-in-input-stream",
noncharacterInInputStream = "noncharacter-in-input-stream",
surrogateInInputStream = "surrogate-in-input-stream",
nonVoidHtmlElementStartTagWithTrailingSolidus = "non-void-html-element-start-tag-with-trailing-solidus",
endTagWithAttributes = "end-tag-with-attributes",
endTagWithTrailingSolidus = "end-tag-with-trailing-solidus",
unexpectedSolidusInTag = "unexpected-solidus-in-tag",
unexpectedNullCharacter = "unexpected-null-character",
unexpectedQuestionMarkInsteadOfTagName = "unexpected-question-mark-instead-of-tag-name",
invalidFirstCharacterOfTagName = "invalid-first-character-of-tag-name",
unexpectedEqualsSignBeforeAttributeName = "unexpected-equals-sign-before-attribute-name",
missingEndTagName = "missing-end-tag-name",
unexpectedCharacterInAttributeName = "unexpected-character-in-attribute-name",
unknownNamedCharacterReference = "unknown-named-character-reference",
missingSemicolonAfterCharacterReference = "missing-semicolon-after-character-reference",
unexpectedCharacterAfterDoctypeSystemIdentifier = "unexpected-character-after-doctype-system-identifier",
unexpectedCharacterInUnquotedAttributeValue = "unexpected-character-in-unquoted-attribute-value",
eofBeforeTagName = "eof-before-tag-name",
eofInTag = "eof-in-tag",
missingAttributeValue = "missing-attribute-value",
missingWhitespaceBetweenAttributes = "missing-whitespace-between-attributes",
missingWhitespaceAfterDoctypePublicKeyword = "missing-whitespace-after-doctype-public-keyword",
missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers = "missing-whitespace-between-doctype-public-and-system-identifiers",
missingWhitespaceAfterDoctypeSystemKeyword = "missing-whitespace-after-doctype-system-keyword",
missingQuoteBeforeDoctypePublicIdentifier = "missing-quote-before-doctype-public-identifier",
missingQuoteBeforeDoctypeSystemIdentifier = "missing-quote-before-doctype-system-identifier",
missingDoctypePublicIdentifier = "missing-doctype-public-identifier",
missingDoctypeSystemIdentifier = "missing-doctype-system-identifier",
abruptDoctypePublicIdentifier = "abrupt-doctype-public-identifier",
abruptDoctypeSystemIdentifier = "abrupt-doctype-system-identifier",
cdataInHtmlContent = "cdata-in-html-content",
incorrectlyOpenedComment = "incorrectly-opened-comment",
eofInScriptHtmlCommentLikeText = "eof-in-script-html-comment-like-text",
eofInDoctype = "eof-in-doctype",
nestedComment = "nested-comment",
abruptClosingOfEmptyComment = "abrupt-closing-of-empty-comment",
eofInComment = "eof-in-comment",
incorrectlyClosedComment = "incorrectly-closed-comment",
eofInCdata = "eof-in-cdata",
absenceOfDigitsInNumericCharacterReference = "absence-of-digits-in-numeric-character-reference",
nullCharacterReference = "null-character-reference",
surrogateCharacterReference = "surrogate-character-reference",
characterReferenceOutsideUnicodeRange = "character-reference-outside-unicode-range",
controlCharacterReference = "control-character-reference",
noncharacterCharacterReference = "noncharacter-character-reference",
missingWhitespaceBeforeDoctypeName = "missing-whitespace-before-doctype-name",
missingDoctypeName = "missing-doctype-name",
invalidCharacterSequenceAfterDoctypeName = "invalid-character-sequence-after-doctype-name",
duplicateAttribute = "duplicate-attribute",
nonConformingDoctype = "non-conforming-doctype",
missingDoctype = "missing-doctype",
misplacedDoctype = "misplaced-doctype",
endTagWithoutMatchingOpenElement = "end-tag-without-matching-open-element",
closingOfElementWithOpenChildElements = "closing-of-element-with-open-child-elements",
disallowedContentInNoscriptInHead = "disallowed-content-in-noscript-in-head",
openElementsLeftAfterEof = "open-elements-left-after-eof",
abandonedHeadElementChild = "abandoned-head-element-child",
misplacedStartTagForHeadElement = "misplaced-start-tag-for-head-element",
nestedNoscriptInHead = "nested-noscript-in-head",
eofInElementThatCanContainOnlyText = "eof-in-element-that-can-contain-only-text"
}
+66
View File
@@ -0,0 +1,66 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.ERR = void 0;
var ERR;
(function (ERR) {
ERR["controlCharacterInInputStream"] = "control-character-in-input-stream";
ERR["noncharacterInInputStream"] = "noncharacter-in-input-stream";
ERR["surrogateInInputStream"] = "surrogate-in-input-stream";
ERR["nonVoidHtmlElementStartTagWithTrailingSolidus"] = "non-void-html-element-start-tag-with-trailing-solidus";
ERR["endTagWithAttributes"] = "end-tag-with-attributes";
ERR["endTagWithTrailingSolidus"] = "end-tag-with-trailing-solidus";
ERR["unexpectedSolidusInTag"] = "unexpected-solidus-in-tag";
ERR["unexpectedNullCharacter"] = "unexpected-null-character";
ERR["unexpectedQuestionMarkInsteadOfTagName"] = "unexpected-question-mark-instead-of-tag-name";
ERR["invalidFirstCharacterOfTagName"] = "invalid-first-character-of-tag-name";
ERR["unexpectedEqualsSignBeforeAttributeName"] = "unexpected-equals-sign-before-attribute-name";
ERR["missingEndTagName"] = "missing-end-tag-name";
ERR["unexpectedCharacterInAttributeName"] = "unexpected-character-in-attribute-name";
ERR["unknownNamedCharacterReference"] = "unknown-named-character-reference";
ERR["missingSemicolonAfterCharacterReference"] = "missing-semicolon-after-character-reference";
ERR["unexpectedCharacterAfterDoctypeSystemIdentifier"] = "unexpected-character-after-doctype-system-identifier";
ERR["unexpectedCharacterInUnquotedAttributeValue"] = "unexpected-character-in-unquoted-attribute-value";
ERR["eofBeforeTagName"] = "eof-before-tag-name";
ERR["eofInTag"] = "eof-in-tag";
ERR["missingAttributeValue"] = "missing-attribute-value";
ERR["missingWhitespaceBetweenAttributes"] = "missing-whitespace-between-attributes";
ERR["missingWhitespaceAfterDoctypePublicKeyword"] = "missing-whitespace-after-doctype-public-keyword";
ERR["missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers"] = "missing-whitespace-between-doctype-public-and-system-identifiers";
ERR["missingWhitespaceAfterDoctypeSystemKeyword"] = "missing-whitespace-after-doctype-system-keyword";
ERR["missingQuoteBeforeDoctypePublicIdentifier"] = "missing-quote-before-doctype-public-identifier";
ERR["missingQuoteBeforeDoctypeSystemIdentifier"] = "missing-quote-before-doctype-system-identifier";
ERR["missingDoctypePublicIdentifier"] = "missing-doctype-public-identifier";
ERR["missingDoctypeSystemIdentifier"] = "missing-doctype-system-identifier";
ERR["abruptDoctypePublicIdentifier"] = "abrupt-doctype-public-identifier";
ERR["abruptDoctypeSystemIdentifier"] = "abrupt-doctype-system-identifier";
ERR["cdataInHtmlContent"] = "cdata-in-html-content";
ERR["incorrectlyOpenedComment"] = "incorrectly-opened-comment";
ERR["eofInScriptHtmlCommentLikeText"] = "eof-in-script-html-comment-like-text";
ERR["eofInDoctype"] = "eof-in-doctype";
ERR["nestedComment"] = "nested-comment";
ERR["abruptClosingOfEmptyComment"] = "abrupt-closing-of-empty-comment";
ERR["eofInComment"] = "eof-in-comment";
ERR["incorrectlyClosedComment"] = "incorrectly-closed-comment";
ERR["eofInCdata"] = "eof-in-cdata";
ERR["absenceOfDigitsInNumericCharacterReference"] = "absence-of-digits-in-numeric-character-reference";
ERR["nullCharacterReference"] = "null-character-reference";
ERR["surrogateCharacterReference"] = "surrogate-character-reference";
ERR["characterReferenceOutsideUnicodeRange"] = "character-reference-outside-unicode-range";
ERR["controlCharacterReference"] = "control-character-reference";
ERR["noncharacterCharacterReference"] = "noncharacter-character-reference";
ERR["missingWhitespaceBeforeDoctypeName"] = "missing-whitespace-before-doctype-name";
ERR["missingDoctypeName"] = "missing-doctype-name";
ERR["invalidCharacterSequenceAfterDoctypeName"] = "invalid-character-sequence-after-doctype-name";
ERR["duplicateAttribute"] = "duplicate-attribute";
ERR["nonConformingDoctype"] = "non-conforming-doctype";
ERR["missingDoctype"] = "missing-doctype";
ERR["misplacedDoctype"] = "misplaced-doctype";
ERR["endTagWithoutMatchingOpenElement"] = "end-tag-without-matching-open-element";
ERR["closingOfElementWithOpenChildElements"] = "closing-of-element-with-open-child-elements";
ERR["disallowedContentInNoscriptInHead"] = "disallowed-content-in-noscript-in-head";
ERR["openElementsLeftAfterEof"] = "open-elements-left-after-eof";
ERR["abandonedHeadElementChild"] = "abandoned-head-element-child";
ERR["misplacedStartTagForHeadElement"] = "misplaced-start-tag-for-head-element";
ERR["nestedNoscriptInHead"] = "nested-noscript-in-head";
ERR["eofInElementThatCanContainOnlyText"] = "eof-in-element-that-can-contain-only-text";
})(ERR || (exports.ERR = ERR = {}));
+9
View File
@@ -0,0 +1,9 @@
import { TAG_ID as $, NS } from './html.js';
import type { TagToken, Attribute } from './token.js';
export declare const SVG_TAG_NAMES_ADJUSTMENT_MAP: Map<string, string>;
export declare function causesExit(startTagToken: TagToken): boolean;
export declare function adjustTokenMathMLAttrs(token: TagToken): void;
export declare function adjustTokenSVGAttrs(token: TagToken): void;
export declare function adjustTokenXMLAttrs(token: TagToken): void;
export declare function adjustTokenSVGTagName(token: TagToken): void;
export declare function isIntegrationPoint(tn: $, ns: NS, attrs: Attribute[], foreignNS?: NS): boolean;
+237
View File
@@ -0,0 +1,237 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.SVG_TAG_NAMES_ADJUSTMENT_MAP = void 0;
exports.causesExit = causesExit;
exports.adjustTokenMathMLAttrs = adjustTokenMathMLAttrs;
exports.adjustTokenSVGAttrs = adjustTokenSVGAttrs;
exports.adjustTokenXMLAttrs = adjustTokenXMLAttrs;
exports.adjustTokenSVGTagName = adjustTokenSVGTagName;
exports.isIntegrationPoint = isIntegrationPoint;
const html_js_1 = require("./html.js");
//MIME types
const MIME_TYPES = {
TEXT_HTML: 'text/html',
APPLICATION_XML: 'application/xhtml+xml',
};
//Attributes
const DEFINITION_URL_ATTR = 'definitionurl';
const ADJUSTED_DEFINITION_URL_ATTR = 'definitionURL';
const SVG_ATTRS_ADJUSTMENT_MAP = new Map([
'attributeName',
'attributeType',
'baseFrequency',
'baseProfile',
'calcMode',
'clipPathUnits',
'diffuseConstant',
'edgeMode',
'filterUnits',
'glyphRef',
'gradientTransform',
'gradientUnits',
'kernelMatrix',
'kernelUnitLength',
'keyPoints',
'keySplines',
'keyTimes',
'lengthAdjust',
'limitingConeAngle',
'markerHeight',
'markerUnits',
'markerWidth',
'maskContentUnits',
'maskUnits',
'numOctaves',
'pathLength',
'patternContentUnits',
'patternTransform',
'patternUnits',
'pointsAtX',
'pointsAtY',
'pointsAtZ',
'preserveAlpha',
'preserveAspectRatio',
'primitiveUnits',
'refX',
'refY',
'repeatCount',
'repeatDur',
'requiredExtensions',
'requiredFeatures',
'specularConstant',
'specularExponent',
'spreadMethod',
'startOffset',
'stdDeviation',
'stitchTiles',
'surfaceScale',
'systemLanguage',
'tableValues',
'targetX',
'targetY',
'textLength',
'viewBox',
'viewTarget',
'xChannelSelector',
'yChannelSelector',
'zoomAndPan',
].map((attr) => [attr.toLowerCase(), attr]));
const XML_ATTRS_ADJUSTMENT_MAP = new Map([
['xlink:actuate', { prefix: 'xlink', name: 'actuate', namespace: html_js_1.NS.XLINK }],
['xlink:arcrole', { prefix: 'xlink', name: 'arcrole', namespace: html_js_1.NS.XLINK }],
['xlink:href', { prefix: 'xlink', name: 'href', namespace: html_js_1.NS.XLINK }],
['xlink:role', { prefix: 'xlink', name: 'role', namespace: html_js_1.NS.XLINK }],
['xlink:show', { prefix: 'xlink', name: 'show', namespace: html_js_1.NS.XLINK }],
['xlink:title', { prefix: 'xlink', name: 'title', namespace: html_js_1.NS.XLINK }],
['xlink:type', { prefix: 'xlink', name: 'type', namespace: html_js_1.NS.XLINK }],
['xml:lang', { prefix: 'xml', name: 'lang', namespace: html_js_1.NS.XML }],
['xml:space', { prefix: 'xml', name: 'space', namespace: html_js_1.NS.XML }],
['xmlns', { prefix: '', name: 'xmlns', namespace: html_js_1.NS.XMLNS }],
['xmlns:xlink', { prefix: 'xmlns', name: 'xlink', namespace: html_js_1.NS.XMLNS }],
]);
//SVG tag names adjustment map
exports.SVG_TAG_NAMES_ADJUSTMENT_MAP = new Map([
'altGlyph',
'altGlyphDef',
'altGlyphItem',
'animateColor',
'animateMotion',
'animateTransform',
'clipPath',
'feBlend',
'feColorMatrix',
'feComponentTransfer',
'feComposite',
'feConvolveMatrix',
'feDiffuseLighting',
'feDisplacementMap',
'feDistantLight',
'feFlood',
'feFuncA',
'feFuncB',
'feFuncG',
'feFuncR',
'feGaussianBlur',
'feImage',
'feMerge',
'feMergeNode',
'feMorphology',
'feOffset',
'fePointLight',
'feSpecularLighting',
'feSpotLight',
'feTile',
'feTurbulence',
'foreignObject',
'glyphRef',
'linearGradient',
'radialGradient',
'textPath',
].map((tn) => [tn.toLowerCase(), tn]));
//Tags that causes exit from foreign content
const EXITS_FOREIGN_CONTENT = new Set([
html_js_1.TAG_ID.B,
html_js_1.TAG_ID.BIG,
html_js_1.TAG_ID.BLOCKQUOTE,
html_js_1.TAG_ID.BODY,
html_js_1.TAG_ID.BR,
html_js_1.TAG_ID.CENTER,
html_js_1.TAG_ID.CODE,
html_js_1.TAG_ID.DD,
html_js_1.TAG_ID.DIV,
html_js_1.TAG_ID.DL,
html_js_1.TAG_ID.DT,
html_js_1.TAG_ID.EM,
html_js_1.TAG_ID.EMBED,
html_js_1.TAG_ID.H1,
html_js_1.TAG_ID.H2,
html_js_1.TAG_ID.H3,
html_js_1.TAG_ID.H4,
html_js_1.TAG_ID.H5,
html_js_1.TAG_ID.H6,
html_js_1.TAG_ID.HEAD,
html_js_1.TAG_ID.HR,
html_js_1.TAG_ID.I,
html_js_1.TAG_ID.IMG,
html_js_1.TAG_ID.LI,
html_js_1.TAG_ID.LISTING,
html_js_1.TAG_ID.MENU,
html_js_1.TAG_ID.META,
html_js_1.TAG_ID.NOBR,
html_js_1.TAG_ID.OL,
html_js_1.TAG_ID.P,
html_js_1.TAG_ID.PRE,
html_js_1.TAG_ID.RUBY,
html_js_1.TAG_ID.S,
html_js_1.TAG_ID.SMALL,
html_js_1.TAG_ID.SPAN,
html_js_1.TAG_ID.STRONG,
html_js_1.TAG_ID.STRIKE,
html_js_1.TAG_ID.SUB,
html_js_1.TAG_ID.SUP,
html_js_1.TAG_ID.TABLE,
html_js_1.TAG_ID.TT,
html_js_1.TAG_ID.U,
html_js_1.TAG_ID.UL,
html_js_1.TAG_ID.VAR,
]);
//Check exit from foreign content
function causesExit(startTagToken) {
const tn = startTagToken.tagID;
const isFontWithAttrs = tn === html_js_1.TAG_ID.FONT &&
startTagToken.attrs.some(({ name }) => name === html_js_1.ATTRS.COLOR || name === html_js_1.ATTRS.SIZE || name === html_js_1.ATTRS.FACE);
return isFontWithAttrs || EXITS_FOREIGN_CONTENT.has(tn);
}
//Token adjustments
function adjustTokenMathMLAttrs(token) {
for (let i = 0; i < token.attrs.length; i++) {
if (token.attrs[i].name === DEFINITION_URL_ATTR) {
token.attrs[i].name = ADJUSTED_DEFINITION_URL_ATTR;
break;
}
}
}
function adjustTokenSVGAttrs(token) {
for (let i = 0; i < token.attrs.length; i++) {
const adjustedAttrName = SVG_ATTRS_ADJUSTMENT_MAP.get(token.attrs[i].name);
if (adjustedAttrName != null) {
token.attrs[i].name = adjustedAttrName;
}
}
}
function adjustTokenXMLAttrs(token) {
for (let i = 0; i < token.attrs.length; i++) {
const adjustedAttrEntry = XML_ATTRS_ADJUSTMENT_MAP.get(token.attrs[i].name);
if (adjustedAttrEntry) {
token.attrs[i].prefix = adjustedAttrEntry.prefix;
token.attrs[i].name = adjustedAttrEntry.name;
token.attrs[i].namespace = adjustedAttrEntry.namespace;
}
}
}
function adjustTokenSVGTagName(token) {
const adjustedTagName = exports.SVG_TAG_NAMES_ADJUSTMENT_MAP.get(token.tagName);
if (adjustedTagName != null) {
token.tagName = adjustedTagName;
token.tagID = (0, html_js_1.getTagID)(token.tagName);
}
}
//Integration points
function isMathMLTextIntegrationPoint(tn, ns) {
return ns === html_js_1.NS.MATHML && (tn === html_js_1.TAG_ID.MI || tn === html_js_1.TAG_ID.MO || tn === html_js_1.TAG_ID.MN || tn === html_js_1.TAG_ID.MS || tn === html_js_1.TAG_ID.MTEXT);
}
function isHtmlIntegrationPoint(tn, ns, attrs) {
if (ns === html_js_1.NS.MATHML && tn === html_js_1.TAG_ID.ANNOTATION_XML) {
for (let i = 0; i < attrs.length; i++) {
if (attrs[i].name === html_js_1.ATTRS.ENCODING) {
const value = attrs[i].value.toLowerCase();
return value === MIME_TYPES.TEXT_HTML || value === MIME_TYPES.APPLICATION_XML;
}
}
}
return ns === html_js_1.NS.SVG && (tn === html_js_1.TAG_ID.FOREIGN_OBJECT || tn === html_js_1.TAG_ID.DESC || tn === html_js_1.TAG_ID.TITLE);
}
function isIntegrationPoint(tn, ns, attrs, foreignNS) {
return (((!foreignNS || foreignNS === html_js_1.NS.HTML) && isHtmlIntegrationPoint(tn, ns, attrs)) ||
((!foreignNS || foreignNS === html_js_1.NS.MATHML) && isMathMLTextIntegrationPoint(tn, ns)));
}
+289
View File
@@ -0,0 +1,289 @@
/** All valid namespaces in HTML. */
export declare enum NS {
HTML = "http://www.w3.org/1999/xhtml",
MATHML = "http://www.w3.org/1998/Math/MathML",
SVG = "http://www.w3.org/2000/svg",
XLINK = "http://www.w3.org/1999/xlink",
XML = "http://www.w3.org/XML/1998/namespace",
XMLNS = "http://www.w3.org/2000/xmlns/"
}
export declare enum ATTRS {
TYPE = "type",
ACTION = "action",
ENCODING = "encoding",
PROMPT = "prompt",
NAME = "name",
COLOR = "color",
FACE = "face",
SIZE = "size"
}
/**
* The mode of the document.
*
* @see {@link https://dom.spec.whatwg.org/#concept-document-limited-quirks}
*/
export declare enum DOCUMENT_MODE {
NO_QUIRKS = "no-quirks",
QUIRKS = "quirks",
LIMITED_QUIRKS = "limited-quirks"
}
export declare enum TAG_NAMES {
A = "a",
ADDRESS = "address",
ANNOTATION_XML = "annotation-xml",
APPLET = "applet",
AREA = "area",
ARTICLE = "article",
ASIDE = "aside",
B = "b",
BASE = "base",
BASEFONT = "basefont",
BGSOUND = "bgsound",
BIG = "big",
BLOCKQUOTE = "blockquote",
BODY = "body",
BR = "br",
BUTTON = "button",
CAPTION = "caption",
CENTER = "center",
CODE = "code",
COL = "col",
COLGROUP = "colgroup",
DD = "dd",
DESC = "desc",
DETAILS = "details",
DIALOG = "dialog",
DIR = "dir",
DIV = "div",
DL = "dl",
DT = "dt",
EM = "em",
EMBED = "embed",
FIELDSET = "fieldset",
FIGCAPTION = "figcaption",
FIGURE = "figure",
FONT = "font",
FOOTER = "footer",
FOREIGN_OBJECT = "foreignObject",
FORM = "form",
FRAME = "frame",
FRAMESET = "frameset",
H1 = "h1",
H2 = "h2",
H3 = "h3",
H4 = "h4",
H5 = "h5",
H6 = "h6",
HEAD = "head",
HEADER = "header",
HGROUP = "hgroup",
HR = "hr",
HTML = "html",
I = "i",
IMG = "img",
IMAGE = "image",
INPUT = "input",
IFRAME = "iframe",
KEYGEN = "keygen",
LABEL = "label",
LI = "li",
LINK = "link",
LISTING = "listing",
MAIN = "main",
MALIGNMARK = "malignmark",
MARQUEE = "marquee",
MATH = "math",
MENU = "menu",
META = "meta",
MGLYPH = "mglyph",
MI = "mi",
MO = "mo",
MN = "mn",
MS = "ms",
MTEXT = "mtext",
NAV = "nav",
NOBR = "nobr",
NOFRAMES = "noframes",
NOEMBED = "noembed",
NOSCRIPT = "noscript",
OBJECT = "object",
OL = "ol",
OPTGROUP = "optgroup",
OPTION = "option",
P = "p",
PARAM = "param",
PLAINTEXT = "plaintext",
PRE = "pre",
RB = "rb",
RP = "rp",
RT = "rt",
RTC = "rtc",
RUBY = "ruby",
S = "s",
SCRIPT = "script",
SEARCH = "search",
SECTION = "section",
SELECT = "select",
SOURCE = "source",
SMALL = "small",
SPAN = "span",
STRIKE = "strike",
STRONG = "strong",
STYLE = "style",
SUB = "sub",
SUMMARY = "summary",
SUP = "sup",
TABLE = "table",
TBODY = "tbody",
TEMPLATE = "template",
TEXTAREA = "textarea",
TFOOT = "tfoot",
TD = "td",
TH = "th",
THEAD = "thead",
TITLE = "title",
TR = "tr",
TRACK = "track",
TT = "tt",
U = "u",
UL = "ul",
SVG = "svg",
VAR = "var",
WBR = "wbr",
XMP = "xmp"
}
/**
* Tag IDs are numeric IDs for known tag names.
*
* We use tag IDs to improve the performance of tag name comparisons.
*/
export declare enum TAG_ID {
UNKNOWN = 0,
A = 1,
ADDRESS = 2,
ANNOTATION_XML = 3,
APPLET = 4,
AREA = 5,
ARTICLE = 6,
ASIDE = 7,
B = 8,
BASE = 9,
BASEFONT = 10,
BGSOUND = 11,
BIG = 12,
BLOCKQUOTE = 13,
BODY = 14,
BR = 15,
BUTTON = 16,
CAPTION = 17,
CENTER = 18,
CODE = 19,
COL = 20,
COLGROUP = 21,
DD = 22,
DESC = 23,
DETAILS = 24,
DIALOG = 25,
DIR = 26,
DIV = 27,
DL = 28,
DT = 29,
EM = 30,
EMBED = 31,
FIELDSET = 32,
FIGCAPTION = 33,
FIGURE = 34,
FONT = 35,
FOOTER = 36,
FOREIGN_OBJECT = 37,
FORM = 38,
FRAME = 39,
FRAMESET = 40,
H1 = 41,
H2 = 42,
H3 = 43,
H4 = 44,
H5 = 45,
H6 = 46,
HEAD = 47,
HEADER = 48,
HGROUP = 49,
HR = 50,
HTML = 51,
I = 52,
IMG = 53,
IMAGE = 54,
INPUT = 55,
IFRAME = 56,
KEYGEN = 57,
LABEL = 58,
LI = 59,
LINK = 60,
LISTING = 61,
MAIN = 62,
MALIGNMARK = 63,
MARQUEE = 64,
MATH = 65,
MENU = 66,
META = 67,
MGLYPH = 68,
MI = 69,
MO = 70,
MN = 71,
MS = 72,
MTEXT = 73,
NAV = 74,
NOBR = 75,
NOFRAMES = 76,
NOEMBED = 77,
NOSCRIPT = 78,
OBJECT = 79,
OL = 80,
OPTGROUP = 81,
OPTION = 82,
P = 83,
PARAM = 84,
PLAINTEXT = 85,
PRE = 86,
RB = 87,
RP = 88,
RT = 89,
RTC = 90,
RUBY = 91,
S = 92,
SCRIPT = 93,
SEARCH = 94,
SECTION = 95,
SELECT = 96,
SOURCE = 97,
SMALL = 98,
SPAN = 99,
STRIKE = 100,
STRONG = 101,
STYLE = 102,
SUB = 103,
SUMMARY = 104,
SUP = 105,
TABLE = 106,
TBODY = 107,
TEMPLATE = 108,
TEXTAREA = 109,
TFOOT = 110,
TD = 111,
TH = 112,
THEAD = 113,
TITLE = 114,
TR = 115,
TRACK = 116,
TT = 117,
U = 118,
UL = 119,
SVG = 120,
VAR = 121,
WBR = 122,
XMP = 123
}
export declare function getTagID(tagName: string): TAG_ID;
export declare const SPECIAL_ELEMENTS: Record<NS, Set<TAG_ID>>;
export declare const NUMBERED_HEADERS: Set<TAG_ID>;
export declare function hasUnescapedText(tn: string, scriptingEnabled: boolean): boolean;
+528
View File
@@ -0,0 +1,528 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.NUMBERED_HEADERS = exports.SPECIAL_ELEMENTS = exports.TAG_ID = exports.TAG_NAMES = exports.DOCUMENT_MODE = exports.ATTRS = exports.NS = void 0;
exports.getTagID = getTagID;
exports.hasUnescapedText = hasUnescapedText;
/** All valid namespaces in HTML. */
var NS;
(function (NS) {
NS["HTML"] = "http://www.w3.org/1999/xhtml";
NS["MATHML"] = "http://www.w3.org/1998/Math/MathML";
NS["SVG"] = "http://www.w3.org/2000/svg";
NS["XLINK"] = "http://www.w3.org/1999/xlink";
NS["XML"] = "http://www.w3.org/XML/1998/namespace";
NS["XMLNS"] = "http://www.w3.org/2000/xmlns/";
})(NS || (exports.NS = NS = {}));
var ATTRS;
(function (ATTRS) {
ATTRS["TYPE"] = "type";
ATTRS["ACTION"] = "action";
ATTRS["ENCODING"] = "encoding";
ATTRS["PROMPT"] = "prompt";
ATTRS["NAME"] = "name";
ATTRS["COLOR"] = "color";
ATTRS["FACE"] = "face";
ATTRS["SIZE"] = "size";
})(ATTRS || (exports.ATTRS = ATTRS = {}));
/**
* The mode of the document.
*
* @see {@link https://dom.spec.whatwg.org/#concept-document-limited-quirks}
*/
var DOCUMENT_MODE;
(function (DOCUMENT_MODE) {
DOCUMENT_MODE["NO_QUIRKS"] = "no-quirks";
DOCUMENT_MODE["QUIRKS"] = "quirks";
DOCUMENT_MODE["LIMITED_QUIRKS"] = "limited-quirks";
})(DOCUMENT_MODE || (exports.DOCUMENT_MODE = DOCUMENT_MODE = {}));
var TAG_NAMES;
(function (TAG_NAMES) {
TAG_NAMES["A"] = "a";
TAG_NAMES["ADDRESS"] = "address";
TAG_NAMES["ANNOTATION_XML"] = "annotation-xml";
TAG_NAMES["APPLET"] = "applet";
TAG_NAMES["AREA"] = "area";
TAG_NAMES["ARTICLE"] = "article";
TAG_NAMES["ASIDE"] = "aside";
TAG_NAMES["B"] = "b";
TAG_NAMES["BASE"] = "base";
TAG_NAMES["BASEFONT"] = "basefont";
TAG_NAMES["BGSOUND"] = "bgsound";
TAG_NAMES["BIG"] = "big";
TAG_NAMES["BLOCKQUOTE"] = "blockquote";
TAG_NAMES["BODY"] = "body";
TAG_NAMES["BR"] = "br";
TAG_NAMES["BUTTON"] = "button";
TAG_NAMES["CAPTION"] = "caption";
TAG_NAMES["CENTER"] = "center";
TAG_NAMES["CODE"] = "code";
TAG_NAMES["COL"] = "col";
TAG_NAMES["COLGROUP"] = "colgroup";
TAG_NAMES["DD"] = "dd";
TAG_NAMES["DESC"] = "desc";
TAG_NAMES["DETAILS"] = "details";
TAG_NAMES["DIALOG"] = "dialog";
TAG_NAMES["DIR"] = "dir";
TAG_NAMES["DIV"] = "div";
TAG_NAMES["DL"] = "dl";
TAG_NAMES["DT"] = "dt";
TAG_NAMES["EM"] = "em";
TAG_NAMES["EMBED"] = "embed";
TAG_NAMES["FIELDSET"] = "fieldset";
TAG_NAMES["FIGCAPTION"] = "figcaption";
TAG_NAMES["FIGURE"] = "figure";
TAG_NAMES["FONT"] = "font";
TAG_NAMES["FOOTER"] = "footer";
TAG_NAMES["FOREIGN_OBJECT"] = "foreignObject";
TAG_NAMES["FORM"] = "form";
TAG_NAMES["FRAME"] = "frame";
TAG_NAMES["FRAMESET"] = "frameset";
TAG_NAMES["H1"] = "h1";
TAG_NAMES["H2"] = "h2";
TAG_NAMES["H3"] = "h3";
TAG_NAMES["H4"] = "h4";
TAG_NAMES["H5"] = "h5";
TAG_NAMES["H6"] = "h6";
TAG_NAMES["HEAD"] = "head";
TAG_NAMES["HEADER"] = "header";
TAG_NAMES["HGROUP"] = "hgroup";
TAG_NAMES["HR"] = "hr";
TAG_NAMES["HTML"] = "html";
TAG_NAMES["I"] = "i";
TAG_NAMES["IMG"] = "img";
TAG_NAMES["IMAGE"] = "image";
TAG_NAMES["INPUT"] = "input";
TAG_NAMES["IFRAME"] = "iframe";
TAG_NAMES["KEYGEN"] = "keygen";
TAG_NAMES["LABEL"] = "label";
TAG_NAMES["LI"] = "li";
TAG_NAMES["LINK"] = "link";
TAG_NAMES["LISTING"] = "listing";
TAG_NAMES["MAIN"] = "main";
TAG_NAMES["MALIGNMARK"] = "malignmark";
TAG_NAMES["MARQUEE"] = "marquee";
TAG_NAMES["MATH"] = "math";
TAG_NAMES["MENU"] = "menu";
TAG_NAMES["META"] = "meta";
TAG_NAMES["MGLYPH"] = "mglyph";
TAG_NAMES["MI"] = "mi";
TAG_NAMES["MO"] = "mo";
TAG_NAMES["MN"] = "mn";
TAG_NAMES["MS"] = "ms";
TAG_NAMES["MTEXT"] = "mtext";
TAG_NAMES["NAV"] = "nav";
TAG_NAMES["NOBR"] = "nobr";
TAG_NAMES["NOFRAMES"] = "noframes";
TAG_NAMES["NOEMBED"] = "noembed";
TAG_NAMES["NOSCRIPT"] = "noscript";
TAG_NAMES["OBJECT"] = "object";
TAG_NAMES["OL"] = "ol";
TAG_NAMES["OPTGROUP"] = "optgroup";
TAG_NAMES["OPTION"] = "option";
TAG_NAMES["P"] = "p";
TAG_NAMES["PARAM"] = "param";
TAG_NAMES["PLAINTEXT"] = "plaintext";
TAG_NAMES["PRE"] = "pre";
TAG_NAMES["RB"] = "rb";
TAG_NAMES["RP"] = "rp";
TAG_NAMES["RT"] = "rt";
TAG_NAMES["RTC"] = "rtc";
TAG_NAMES["RUBY"] = "ruby";
TAG_NAMES["S"] = "s";
TAG_NAMES["SCRIPT"] = "script";
TAG_NAMES["SEARCH"] = "search";
TAG_NAMES["SECTION"] = "section";
TAG_NAMES["SELECT"] = "select";
TAG_NAMES["SOURCE"] = "source";
TAG_NAMES["SMALL"] = "small";
TAG_NAMES["SPAN"] = "span";
TAG_NAMES["STRIKE"] = "strike";
TAG_NAMES["STRONG"] = "strong";
TAG_NAMES["STYLE"] = "style";
TAG_NAMES["SUB"] = "sub";
TAG_NAMES["SUMMARY"] = "summary";
TAG_NAMES["SUP"] = "sup";
TAG_NAMES["TABLE"] = "table";
TAG_NAMES["TBODY"] = "tbody";
TAG_NAMES["TEMPLATE"] = "template";
TAG_NAMES["TEXTAREA"] = "textarea";
TAG_NAMES["TFOOT"] = "tfoot";
TAG_NAMES["TD"] = "td";
TAG_NAMES["TH"] = "th";
TAG_NAMES["THEAD"] = "thead";
TAG_NAMES["TITLE"] = "title";
TAG_NAMES["TR"] = "tr";
TAG_NAMES["TRACK"] = "track";
TAG_NAMES["TT"] = "tt";
TAG_NAMES["U"] = "u";
TAG_NAMES["UL"] = "ul";
TAG_NAMES["SVG"] = "svg";
TAG_NAMES["VAR"] = "var";
TAG_NAMES["WBR"] = "wbr";
TAG_NAMES["XMP"] = "xmp";
})(TAG_NAMES || (exports.TAG_NAMES = TAG_NAMES = {}));
/**
* Tag IDs are numeric IDs for known tag names.
*
* We use tag IDs to improve the performance of tag name comparisons.
*/
var TAG_ID;
(function (TAG_ID) {
TAG_ID[TAG_ID["UNKNOWN"] = 0] = "UNKNOWN";
TAG_ID[TAG_ID["A"] = 1] = "A";
TAG_ID[TAG_ID["ADDRESS"] = 2] = "ADDRESS";
TAG_ID[TAG_ID["ANNOTATION_XML"] = 3] = "ANNOTATION_XML";
TAG_ID[TAG_ID["APPLET"] = 4] = "APPLET";
TAG_ID[TAG_ID["AREA"] = 5] = "AREA";
TAG_ID[TAG_ID["ARTICLE"] = 6] = "ARTICLE";
TAG_ID[TAG_ID["ASIDE"] = 7] = "ASIDE";
TAG_ID[TAG_ID["B"] = 8] = "B";
TAG_ID[TAG_ID["BASE"] = 9] = "BASE";
TAG_ID[TAG_ID["BASEFONT"] = 10] = "BASEFONT";
TAG_ID[TAG_ID["BGSOUND"] = 11] = "BGSOUND";
TAG_ID[TAG_ID["BIG"] = 12] = "BIG";
TAG_ID[TAG_ID["BLOCKQUOTE"] = 13] = "BLOCKQUOTE";
TAG_ID[TAG_ID["BODY"] = 14] = "BODY";
TAG_ID[TAG_ID["BR"] = 15] = "BR";
TAG_ID[TAG_ID["BUTTON"] = 16] = "BUTTON";
TAG_ID[TAG_ID["CAPTION"] = 17] = "CAPTION";
TAG_ID[TAG_ID["CENTER"] = 18] = "CENTER";
TAG_ID[TAG_ID["CODE"] = 19] = "CODE";
TAG_ID[TAG_ID["COL"] = 20] = "COL";
TAG_ID[TAG_ID["COLGROUP"] = 21] = "COLGROUP";
TAG_ID[TAG_ID["DD"] = 22] = "DD";
TAG_ID[TAG_ID["DESC"] = 23] = "DESC";
TAG_ID[TAG_ID["DETAILS"] = 24] = "DETAILS";
TAG_ID[TAG_ID["DIALOG"] = 25] = "DIALOG";
TAG_ID[TAG_ID["DIR"] = 26] = "DIR";
TAG_ID[TAG_ID["DIV"] = 27] = "DIV";
TAG_ID[TAG_ID["DL"] = 28] = "DL";
TAG_ID[TAG_ID["DT"] = 29] = "DT";
TAG_ID[TAG_ID["EM"] = 30] = "EM";
TAG_ID[TAG_ID["EMBED"] = 31] = "EMBED";
TAG_ID[TAG_ID["FIELDSET"] = 32] = "FIELDSET";
TAG_ID[TAG_ID["FIGCAPTION"] = 33] = "FIGCAPTION";
TAG_ID[TAG_ID["FIGURE"] = 34] = "FIGURE";
TAG_ID[TAG_ID["FONT"] = 35] = "FONT";
TAG_ID[TAG_ID["FOOTER"] = 36] = "FOOTER";
TAG_ID[TAG_ID["FOREIGN_OBJECT"] = 37] = "FOREIGN_OBJECT";
TAG_ID[TAG_ID["FORM"] = 38] = "FORM";
TAG_ID[TAG_ID["FRAME"] = 39] = "FRAME";
TAG_ID[TAG_ID["FRAMESET"] = 40] = "FRAMESET";
TAG_ID[TAG_ID["H1"] = 41] = "H1";
TAG_ID[TAG_ID["H2"] = 42] = "H2";
TAG_ID[TAG_ID["H3"] = 43] = "H3";
TAG_ID[TAG_ID["H4"] = 44] = "H4";
TAG_ID[TAG_ID["H5"] = 45] = "H5";
TAG_ID[TAG_ID["H6"] = 46] = "H6";
TAG_ID[TAG_ID["HEAD"] = 47] = "HEAD";
TAG_ID[TAG_ID["HEADER"] = 48] = "HEADER";
TAG_ID[TAG_ID["HGROUP"] = 49] = "HGROUP";
TAG_ID[TAG_ID["HR"] = 50] = "HR";
TAG_ID[TAG_ID["HTML"] = 51] = "HTML";
TAG_ID[TAG_ID["I"] = 52] = "I";
TAG_ID[TAG_ID["IMG"] = 53] = "IMG";
TAG_ID[TAG_ID["IMAGE"] = 54] = "IMAGE";
TAG_ID[TAG_ID["INPUT"] = 55] = "INPUT";
TAG_ID[TAG_ID["IFRAME"] = 56] = "IFRAME";
TAG_ID[TAG_ID["KEYGEN"] = 57] = "KEYGEN";
TAG_ID[TAG_ID["LABEL"] = 58] = "LABEL";
TAG_ID[TAG_ID["LI"] = 59] = "LI";
TAG_ID[TAG_ID["LINK"] = 60] = "LINK";
TAG_ID[TAG_ID["LISTING"] = 61] = "LISTING";
TAG_ID[TAG_ID["MAIN"] = 62] = "MAIN";
TAG_ID[TAG_ID["MALIGNMARK"] = 63] = "MALIGNMARK";
TAG_ID[TAG_ID["MARQUEE"] = 64] = "MARQUEE";
TAG_ID[TAG_ID["MATH"] = 65] = "MATH";
TAG_ID[TAG_ID["MENU"] = 66] = "MENU";
TAG_ID[TAG_ID["META"] = 67] = "META";
TAG_ID[TAG_ID["MGLYPH"] = 68] = "MGLYPH";
TAG_ID[TAG_ID["MI"] = 69] = "MI";
TAG_ID[TAG_ID["MO"] = 70] = "MO";
TAG_ID[TAG_ID["MN"] = 71] = "MN";
TAG_ID[TAG_ID["MS"] = 72] = "MS";
TAG_ID[TAG_ID["MTEXT"] = 73] = "MTEXT";
TAG_ID[TAG_ID["NAV"] = 74] = "NAV";
TAG_ID[TAG_ID["NOBR"] = 75] = "NOBR";
TAG_ID[TAG_ID["NOFRAMES"] = 76] = "NOFRAMES";
TAG_ID[TAG_ID["NOEMBED"] = 77] = "NOEMBED";
TAG_ID[TAG_ID["NOSCRIPT"] = 78] = "NOSCRIPT";
TAG_ID[TAG_ID["OBJECT"] = 79] = "OBJECT";
TAG_ID[TAG_ID["OL"] = 80] = "OL";
TAG_ID[TAG_ID["OPTGROUP"] = 81] = "OPTGROUP";
TAG_ID[TAG_ID["OPTION"] = 82] = "OPTION";
TAG_ID[TAG_ID["P"] = 83] = "P";
TAG_ID[TAG_ID["PARAM"] = 84] = "PARAM";
TAG_ID[TAG_ID["PLAINTEXT"] = 85] = "PLAINTEXT";
TAG_ID[TAG_ID["PRE"] = 86] = "PRE";
TAG_ID[TAG_ID["RB"] = 87] = "RB";
TAG_ID[TAG_ID["RP"] = 88] = "RP";
TAG_ID[TAG_ID["RT"] = 89] = "RT";
TAG_ID[TAG_ID["RTC"] = 90] = "RTC";
TAG_ID[TAG_ID["RUBY"] = 91] = "RUBY";
TAG_ID[TAG_ID["S"] = 92] = "S";
TAG_ID[TAG_ID["SCRIPT"] = 93] = "SCRIPT";
TAG_ID[TAG_ID["SEARCH"] = 94] = "SEARCH";
TAG_ID[TAG_ID["SECTION"] = 95] = "SECTION";
TAG_ID[TAG_ID["SELECT"] = 96] = "SELECT";
TAG_ID[TAG_ID["SOURCE"] = 97] = "SOURCE";
TAG_ID[TAG_ID["SMALL"] = 98] = "SMALL";
TAG_ID[TAG_ID["SPAN"] = 99] = "SPAN";
TAG_ID[TAG_ID["STRIKE"] = 100] = "STRIKE";
TAG_ID[TAG_ID["STRONG"] = 101] = "STRONG";
TAG_ID[TAG_ID["STYLE"] = 102] = "STYLE";
TAG_ID[TAG_ID["SUB"] = 103] = "SUB";
TAG_ID[TAG_ID["SUMMARY"] = 104] = "SUMMARY";
TAG_ID[TAG_ID["SUP"] = 105] = "SUP";
TAG_ID[TAG_ID["TABLE"] = 106] = "TABLE";
TAG_ID[TAG_ID["TBODY"] = 107] = "TBODY";
TAG_ID[TAG_ID["TEMPLATE"] = 108] = "TEMPLATE";
TAG_ID[TAG_ID["TEXTAREA"] = 109] = "TEXTAREA";
TAG_ID[TAG_ID["TFOOT"] = 110] = "TFOOT";
TAG_ID[TAG_ID["TD"] = 111] = "TD";
TAG_ID[TAG_ID["TH"] = 112] = "TH";
TAG_ID[TAG_ID["THEAD"] = 113] = "THEAD";
TAG_ID[TAG_ID["TITLE"] = 114] = "TITLE";
TAG_ID[TAG_ID["TR"] = 115] = "TR";
TAG_ID[TAG_ID["TRACK"] = 116] = "TRACK";
TAG_ID[TAG_ID["TT"] = 117] = "TT";
TAG_ID[TAG_ID["U"] = 118] = "U";
TAG_ID[TAG_ID["UL"] = 119] = "UL";
TAG_ID[TAG_ID["SVG"] = 120] = "SVG";
TAG_ID[TAG_ID["VAR"] = 121] = "VAR";
TAG_ID[TAG_ID["WBR"] = 122] = "WBR";
TAG_ID[TAG_ID["XMP"] = 123] = "XMP";
})(TAG_ID || (exports.TAG_ID = TAG_ID = {}));
const TAG_NAME_TO_ID = new Map([
[TAG_NAMES.A, TAG_ID.A],
[TAG_NAMES.ADDRESS, TAG_ID.ADDRESS],
[TAG_NAMES.ANNOTATION_XML, TAG_ID.ANNOTATION_XML],
[TAG_NAMES.APPLET, TAG_ID.APPLET],
[TAG_NAMES.AREA, TAG_ID.AREA],
[TAG_NAMES.ARTICLE, TAG_ID.ARTICLE],
[TAG_NAMES.ASIDE, TAG_ID.ASIDE],
[TAG_NAMES.B, TAG_ID.B],
[TAG_NAMES.BASE, TAG_ID.BASE],
[TAG_NAMES.BASEFONT, TAG_ID.BASEFONT],
[TAG_NAMES.BGSOUND, TAG_ID.BGSOUND],
[TAG_NAMES.BIG, TAG_ID.BIG],
[TAG_NAMES.BLOCKQUOTE, TAG_ID.BLOCKQUOTE],
[TAG_NAMES.BODY, TAG_ID.BODY],
[TAG_NAMES.BR, TAG_ID.BR],
[TAG_NAMES.BUTTON, TAG_ID.BUTTON],
[TAG_NAMES.CAPTION, TAG_ID.CAPTION],
[TAG_NAMES.CENTER, TAG_ID.CENTER],
[TAG_NAMES.CODE, TAG_ID.CODE],
[TAG_NAMES.COL, TAG_ID.COL],
[TAG_NAMES.COLGROUP, TAG_ID.COLGROUP],
[TAG_NAMES.DD, TAG_ID.DD],
[TAG_NAMES.DESC, TAG_ID.DESC],
[TAG_NAMES.DETAILS, TAG_ID.DETAILS],
[TAG_NAMES.DIALOG, TAG_ID.DIALOG],
[TAG_NAMES.DIR, TAG_ID.DIR],
[TAG_NAMES.DIV, TAG_ID.DIV],
[TAG_NAMES.DL, TAG_ID.DL],
[TAG_NAMES.DT, TAG_ID.DT],
[TAG_NAMES.EM, TAG_ID.EM],
[TAG_NAMES.EMBED, TAG_ID.EMBED],
[TAG_NAMES.FIELDSET, TAG_ID.FIELDSET],
[TAG_NAMES.FIGCAPTION, TAG_ID.FIGCAPTION],
[TAG_NAMES.FIGURE, TAG_ID.FIGURE],
[TAG_NAMES.FONT, TAG_ID.FONT],
[TAG_NAMES.FOOTER, TAG_ID.FOOTER],
[TAG_NAMES.FOREIGN_OBJECT, TAG_ID.FOREIGN_OBJECT],
[TAG_NAMES.FORM, TAG_ID.FORM],
[TAG_NAMES.FRAME, TAG_ID.FRAME],
[TAG_NAMES.FRAMESET, TAG_ID.FRAMESET],
[TAG_NAMES.H1, TAG_ID.H1],
[TAG_NAMES.H2, TAG_ID.H2],
[TAG_NAMES.H3, TAG_ID.H3],
[TAG_NAMES.H4, TAG_ID.H4],
[TAG_NAMES.H5, TAG_ID.H5],
[TAG_NAMES.H6, TAG_ID.H6],
[TAG_NAMES.HEAD, TAG_ID.HEAD],
[TAG_NAMES.HEADER, TAG_ID.HEADER],
[TAG_NAMES.HGROUP, TAG_ID.HGROUP],
[TAG_NAMES.HR, TAG_ID.HR],
[TAG_NAMES.HTML, TAG_ID.HTML],
[TAG_NAMES.I, TAG_ID.I],
[TAG_NAMES.IMG, TAG_ID.IMG],
[TAG_NAMES.IMAGE, TAG_ID.IMAGE],
[TAG_NAMES.INPUT, TAG_ID.INPUT],
[TAG_NAMES.IFRAME, TAG_ID.IFRAME],
[TAG_NAMES.KEYGEN, TAG_ID.KEYGEN],
[TAG_NAMES.LABEL, TAG_ID.LABEL],
[TAG_NAMES.LI, TAG_ID.LI],
[TAG_NAMES.LINK, TAG_ID.LINK],
[TAG_NAMES.LISTING, TAG_ID.LISTING],
[TAG_NAMES.MAIN, TAG_ID.MAIN],
[TAG_NAMES.MALIGNMARK, TAG_ID.MALIGNMARK],
[TAG_NAMES.MARQUEE, TAG_ID.MARQUEE],
[TAG_NAMES.MATH, TAG_ID.MATH],
[TAG_NAMES.MENU, TAG_ID.MENU],
[TAG_NAMES.META, TAG_ID.META],
[TAG_NAMES.MGLYPH, TAG_ID.MGLYPH],
[TAG_NAMES.MI, TAG_ID.MI],
[TAG_NAMES.MO, TAG_ID.MO],
[TAG_NAMES.MN, TAG_ID.MN],
[TAG_NAMES.MS, TAG_ID.MS],
[TAG_NAMES.MTEXT, TAG_ID.MTEXT],
[TAG_NAMES.NAV, TAG_ID.NAV],
[TAG_NAMES.NOBR, TAG_ID.NOBR],
[TAG_NAMES.NOFRAMES, TAG_ID.NOFRAMES],
[TAG_NAMES.NOEMBED, TAG_ID.NOEMBED],
[TAG_NAMES.NOSCRIPT, TAG_ID.NOSCRIPT],
[TAG_NAMES.OBJECT, TAG_ID.OBJECT],
[TAG_NAMES.OL, TAG_ID.OL],
[TAG_NAMES.OPTGROUP, TAG_ID.OPTGROUP],
[TAG_NAMES.OPTION, TAG_ID.OPTION],
[TAG_NAMES.P, TAG_ID.P],
[TAG_NAMES.PARAM, TAG_ID.PARAM],
[TAG_NAMES.PLAINTEXT, TAG_ID.PLAINTEXT],
[TAG_NAMES.PRE, TAG_ID.PRE],
[TAG_NAMES.RB, TAG_ID.RB],
[TAG_NAMES.RP, TAG_ID.RP],
[TAG_NAMES.RT, TAG_ID.RT],
[TAG_NAMES.RTC, TAG_ID.RTC],
[TAG_NAMES.RUBY, TAG_ID.RUBY],
[TAG_NAMES.S, TAG_ID.S],
[TAG_NAMES.SCRIPT, TAG_ID.SCRIPT],
[TAG_NAMES.SEARCH, TAG_ID.SEARCH],
[TAG_NAMES.SECTION, TAG_ID.SECTION],
[TAG_NAMES.SELECT, TAG_ID.SELECT],
[TAG_NAMES.SOURCE, TAG_ID.SOURCE],
[TAG_NAMES.SMALL, TAG_ID.SMALL],
[TAG_NAMES.SPAN, TAG_ID.SPAN],
[TAG_NAMES.STRIKE, TAG_ID.STRIKE],
[TAG_NAMES.STRONG, TAG_ID.STRONG],
[TAG_NAMES.STYLE, TAG_ID.STYLE],
[TAG_NAMES.SUB, TAG_ID.SUB],
[TAG_NAMES.SUMMARY, TAG_ID.SUMMARY],
[TAG_NAMES.SUP, TAG_ID.SUP],
[TAG_NAMES.TABLE, TAG_ID.TABLE],
[TAG_NAMES.TBODY, TAG_ID.TBODY],
[TAG_NAMES.TEMPLATE, TAG_ID.TEMPLATE],
[TAG_NAMES.TEXTAREA, TAG_ID.TEXTAREA],
[TAG_NAMES.TFOOT, TAG_ID.TFOOT],
[TAG_NAMES.TD, TAG_ID.TD],
[TAG_NAMES.TH, TAG_ID.TH],
[TAG_NAMES.THEAD, TAG_ID.THEAD],
[TAG_NAMES.TITLE, TAG_ID.TITLE],
[TAG_NAMES.TR, TAG_ID.TR],
[TAG_NAMES.TRACK, TAG_ID.TRACK],
[TAG_NAMES.TT, TAG_ID.TT],
[TAG_NAMES.U, TAG_ID.U],
[TAG_NAMES.UL, TAG_ID.UL],
[TAG_NAMES.SVG, TAG_ID.SVG],
[TAG_NAMES.VAR, TAG_ID.VAR],
[TAG_NAMES.WBR, TAG_ID.WBR],
[TAG_NAMES.XMP, TAG_ID.XMP],
]);
function getTagID(tagName) {
var _a;
return (_a = TAG_NAME_TO_ID.get(tagName)) !== null && _a !== void 0 ? _a : TAG_ID.UNKNOWN;
}
const $ = TAG_ID;
exports.SPECIAL_ELEMENTS = {
[NS.HTML]: new Set([
$.ADDRESS,
$.APPLET,
$.AREA,
$.ARTICLE,
$.ASIDE,
$.BASE,
$.BASEFONT,
$.BGSOUND,
$.BLOCKQUOTE,
$.BODY,
$.BR,
$.BUTTON,
$.CAPTION,
$.CENTER,
$.COL,
$.COLGROUP,
$.DD,
$.DETAILS,
$.DIR,
$.DIV,
$.DL,
$.DT,
$.EMBED,
$.FIELDSET,
$.FIGCAPTION,
$.FIGURE,
$.FOOTER,
$.FORM,
$.FRAME,
$.FRAMESET,
$.H1,
$.H2,
$.H3,
$.H4,
$.H5,
$.H6,
$.HEAD,
$.HEADER,
$.HGROUP,
$.HR,
$.HTML,
$.IFRAME,
$.IMG,
$.INPUT,
$.LI,
$.LINK,
$.LISTING,
$.MAIN,
$.MARQUEE,
$.MENU,
$.META,
$.NAV,
$.NOEMBED,
$.NOFRAMES,
$.NOSCRIPT,
$.OBJECT,
$.OL,
$.P,
$.PARAM,
$.PLAINTEXT,
$.PRE,
$.SCRIPT,
$.SECTION,
$.SELECT,
$.SOURCE,
$.STYLE,
$.SUMMARY,
$.TABLE,
$.TBODY,
$.TD,
$.TEMPLATE,
$.TEXTAREA,
$.TFOOT,
$.TH,
$.THEAD,
$.TITLE,
$.TR,
$.TRACK,
$.UL,
$.WBR,
$.XMP,
]),
[NS.MATHML]: new Set([$.MI, $.MO, $.MN, $.MS, $.MTEXT, $.ANNOTATION_XML]),
[NS.SVG]: new Set([$.TITLE, $.FOREIGN_OBJECT, $.DESC]),
[NS.XLINK]: new Set(),
[NS.XML]: new Set(),
[NS.XMLNS]: new Set(),
};
exports.NUMBERED_HEADERS = new Set([$.H1, $.H2, $.H3, $.H4, $.H5, $.H6]);
const UNESCAPED_TEXT = new Set([
TAG_NAMES.STYLE,
TAG_NAMES.SCRIPT,
TAG_NAMES.XMP,
TAG_NAMES.IFRAME,
TAG_NAMES.NOEMBED,
TAG_NAMES.NOFRAMES,
TAG_NAMES.PLAINTEXT,
]);
function hasUnescapedText(tn, scriptingEnabled) {
return UNESCAPED_TEXT.has(tn) || (scriptingEnabled && tn === TAG_NAMES.NOSCRIPT);
}
+84
View File
@@ -0,0 +1,84 @@
import type { TAG_ID } from './html.js';
export declare enum TokenType {
CHARACTER = 0,
NULL_CHARACTER = 1,
WHITESPACE_CHARACTER = 2,
START_TAG = 3,
END_TAG = 4,
COMMENT = 5,
DOCTYPE = 6,
EOF = 7,
HIBERNATION = 8
}
export interface Location {
/** One-based line index of the first character. */
startLine: number;
/** One-based column index of the first character. */
startCol: number;
/** Zero-based first character index. */
startOffset: number;
/** One-based line index of the last character. */
endLine: number;
/** One-based column index of the last character. Points directly *after* the last character. */
endCol: number;
/** Zero-based last character index. Points directly *after* the last character. */
endOffset: number;
}
export interface LocationWithAttributes extends Location {
/** Start tag attributes' location info. */
attrs?: Record<string, Location>;
}
export interface ElementLocation extends LocationWithAttributes {
/** Element's start tag location info. */
startTag?: Location;
/**
* Element's end tag location info.
* This property is undefined, if the element has no closing tag.
*/
endTag?: Location;
}
interface TokenBase {
readonly type: TokenType;
location: Location | null;
}
export interface DoctypeToken extends TokenBase {
readonly type: TokenType.DOCTYPE;
name: string | null;
forceQuirks: boolean;
publicId: string | null;
systemId: string | null;
}
export interface Attribute {
/** The name of the attribute. */
name: string;
/** The namespace of the attribute. */
namespace?: string;
/** The namespace-related prefix of the attribute. */
prefix?: string;
/** The value of the attribute. */
value: string;
}
export interface TagToken extends TokenBase {
readonly type: TokenType.START_TAG | TokenType.END_TAG;
tagName: string;
/** Used to cache the ID of the tag name. */
tagID: TAG_ID;
selfClosing: boolean;
ackSelfClosing: boolean;
attrs: Attribute[];
location: LocationWithAttributes | null;
}
export declare function getTokenAttr(token: TagToken, attrName: string): string | null;
export interface CommentToken extends TokenBase {
readonly type: TokenType.COMMENT;
data: string;
}
export interface EOFToken extends TokenBase {
readonly type: TokenType.EOF;
}
export interface CharacterToken extends TokenBase {
type: TokenType.CHARACTER | TokenType.NULL_CHARACTER | TokenType.WHITESPACE_CHARACTER;
chars: string;
}
export type Token = DoctypeToken | TagToken | CommentToken | EOFToken | CharacterToken;
export {};
+24
View File
@@ -0,0 +1,24 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.TokenType = void 0;
exports.getTokenAttr = getTokenAttr;
var TokenType;
(function (TokenType) {
TokenType[TokenType["CHARACTER"] = 0] = "CHARACTER";
TokenType[TokenType["NULL_CHARACTER"] = 1] = "NULL_CHARACTER";
TokenType[TokenType["WHITESPACE_CHARACTER"] = 2] = "WHITESPACE_CHARACTER";
TokenType[TokenType["START_TAG"] = 3] = "START_TAG";
TokenType[TokenType["END_TAG"] = 4] = "END_TAG";
TokenType[TokenType["COMMENT"] = 5] = "COMMENT";
TokenType[TokenType["DOCTYPE"] = 6] = "DOCTYPE";
TokenType[TokenType["EOF"] = 7] = "EOF";
TokenType[TokenType["HIBERNATION"] = 8] = "HIBERNATION";
})(TokenType || (exports.TokenType = TokenType = {}));
function getTokenAttr(token, attrName) {
for (let i = token.attrs.length - 1; i >= 0; i--) {
if (token.attrs[i].name === attrName) {
return token.attrs[i].value;
}
}
return null;
}
+42
View File
@@ -0,0 +1,42 @@
export declare const REPLACEMENT_CHARACTER = "\uFFFD";
export declare enum CODE_POINTS {
EOF = -1,
NULL = 0,
TABULATION = 9,
CARRIAGE_RETURN = 13,
LINE_FEED = 10,
FORM_FEED = 12,
SPACE = 32,
EXCLAMATION_MARK = 33,
QUOTATION_MARK = 34,
AMPERSAND = 38,
APOSTROPHE = 39,
HYPHEN_MINUS = 45,
SOLIDUS = 47,
DIGIT_0 = 48,
DIGIT_9 = 57,
SEMICOLON = 59,
LESS_THAN_SIGN = 60,
EQUALS_SIGN = 61,
GREATER_THAN_SIGN = 62,
QUESTION_MARK = 63,
LATIN_CAPITAL_A = 65,
LATIN_CAPITAL_Z = 90,
RIGHT_SQUARE_BRACKET = 93,
GRAVE_ACCENT = 96,
LATIN_SMALL_A = 97,
LATIN_SMALL_Z = 122
}
export declare const SEQUENCES: {
DASH_DASH: string;
CDATA_START: string;
DOCTYPE: string;
SCRIPT: string;
PUBLIC: string;
SYSTEM: string;
};
export declare function isSurrogate(cp: number): boolean;
export declare function isSurrogatePair(cp: number): boolean;
export declare function getSurrogatePairCodePoint(cp1: number, cp2: number): number;
export declare function isControlCodePoint(cp: number): boolean;
export declare function isUndefinedCodePoint(cp: number): boolean;
+70
View File
@@ -0,0 +1,70 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.SEQUENCES = exports.CODE_POINTS = exports.REPLACEMENT_CHARACTER = void 0;
exports.isSurrogate = isSurrogate;
exports.isSurrogatePair = isSurrogatePair;
exports.getSurrogatePairCodePoint = getSurrogatePairCodePoint;
exports.isControlCodePoint = isControlCodePoint;
exports.isUndefinedCodePoint = isUndefinedCodePoint;
const UNDEFINED_CODE_POINTS = new Set([
65534, 65535, 131070, 131071, 196606, 196607, 262142, 262143, 327678, 327679, 393214,
393215, 458750, 458751, 524286, 524287, 589822, 589823, 655358, 655359, 720894,
720895, 786430, 786431, 851966, 851967, 917502, 917503, 983038, 983039, 1048574,
1048575, 1114110, 1114111,
]);
exports.REPLACEMENT_CHARACTER = '\uFFFD';
var CODE_POINTS;
(function (CODE_POINTS) {
CODE_POINTS[CODE_POINTS["EOF"] = -1] = "EOF";
CODE_POINTS[CODE_POINTS["NULL"] = 0] = "NULL";
CODE_POINTS[CODE_POINTS["TABULATION"] = 9] = "TABULATION";
CODE_POINTS[CODE_POINTS["CARRIAGE_RETURN"] = 13] = "CARRIAGE_RETURN";
CODE_POINTS[CODE_POINTS["LINE_FEED"] = 10] = "LINE_FEED";
CODE_POINTS[CODE_POINTS["FORM_FEED"] = 12] = "FORM_FEED";
CODE_POINTS[CODE_POINTS["SPACE"] = 32] = "SPACE";
CODE_POINTS[CODE_POINTS["EXCLAMATION_MARK"] = 33] = "EXCLAMATION_MARK";
CODE_POINTS[CODE_POINTS["QUOTATION_MARK"] = 34] = "QUOTATION_MARK";
CODE_POINTS[CODE_POINTS["AMPERSAND"] = 38] = "AMPERSAND";
CODE_POINTS[CODE_POINTS["APOSTROPHE"] = 39] = "APOSTROPHE";
CODE_POINTS[CODE_POINTS["HYPHEN_MINUS"] = 45] = "HYPHEN_MINUS";
CODE_POINTS[CODE_POINTS["SOLIDUS"] = 47] = "SOLIDUS";
CODE_POINTS[CODE_POINTS["DIGIT_0"] = 48] = "DIGIT_0";
CODE_POINTS[CODE_POINTS["DIGIT_9"] = 57] = "DIGIT_9";
CODE_POINTS[CODE_POINTS["SEMICOLON"] = 59] = "SEMICOLON";
CODE_POINTS[CODE_POINTS["LESS_THAN_SIGN"] = 60] = "LESS_THAN_SIGN";
CODE_POINTS[CODE_POINTS["EQUALS_SIGN"] = 61] = "EQUALS_SIGN";
CODE_POINTS[CODE_POINTS["GREATER_THAN_SIGN"] = 62] = "GREATER_THAN_SIGN";
CODE_POINTS[CODE_POINTS["QUESTION_MARK"] = 63] = "QUESTION_MARK";
CODE_POINTS[CODE_POINTS["LATIN_CAPITAL_A"] = 65] = "LATIN_CAPITAL_A";
CODE_POINTS[CODE_POINTS["LATIN_CAPITAL_Z"] = 90] = "LATIN_CAPITAL_Z";
CODE_POINTS[CODE_POINTS["RIGHT_SQUARE_BRACKET"] = 93] = "RIGHT_SQUARE_BRACKET";
CODE_POINTS[CODE_POINTS["GRAVE_ACCENT"] = 96] = "GRAVE_ACCENT";
CODE_POINTS[CODE_POINTS["LATIN_SMALL_A"] = 97] = "LATIN_SMALL_A";
CODE_POINTS[CODE_POINTS["LATIN_SMALL_Z"] = 122] = "LATIN_SMALL_Z";
})(CODE_POINTS || (exports.CODE_POINTS = CODE_POINTS = {}));
exports.SEQUENCES = {
DASH_DASH: '--',
CDATA_START: '[CDATA[',
DOCTYPE: 'doctype',
SCRIPT: 'script',
PUBLIC: 'public',
SYSTEM: 'system',
};
//Surrogates
function isSurrogate(cp) {
return cp >= 55296 && cp <= 57343;
}
function isSurrogatePair(cp) {
return cp >= 56320 && cp <= 57343;
}
function getSurrogatePairCodePoint(cp1, cp2) {
return (cp1 - 55296) * 1024 + 9216 + cp2;
}
//NOTE: excluding NULL and ASCII whitespace
function isControlCodePoint(cp) {
return ((cp !== 0x20 && cp !== 0x0a && cp !== 0x0d && cp !== 0x09 && cp !== 0x0c && cp >= 0x01 && cp <= 0x1f) ||
(cp >= 0x7f && cp <= 0x9f));
}
function isUndefinedCodePoint(cp) {
return (cp >= 64976 && cp <= 65007) || UNDEFINED_CODE_POINTS.has(cp);
}
+71
View File
@@ -0,0 +1,71 @@
import { type ParserOptions } from './parser/index.js';
import type { DefaultTreeAdapterMap } from './tree-adapters/default.js';
import type { TreeAdapterTypeMap } from './tree-adapters/interface.js';
export { type DefaultTreeAdapterMap, defaultTreeAdapter } from './tree-adapters/default.js';
import type * as DefaultTreeAdapter from './tree-adapters/default.js';
export declare namespace DefaultTreeAdapterTypes {
type Document = DefaultTreeAdapter.Document;
type DocumentFragment = DefaultTreeAdapter.DocumentFragment;
type Element = DefaultTreeAdapter.Element;
type CommentNode = DefaultTreeAdapter.CommentNode;
type TextNode = DefaultTreeAdapter.TextNode;
type Template = DefaultTreeAdapter.Template;
type DocumentType = DefaultTreeAdapter.DocumentType;
type ParentNode = DefaultTreeAdapter.ParentNode;
type ChildNode = DefaultTreeAdapter.ChildNode;
type Node = DefaultTreeAdapter.Node;
type DefaultTreeAdapterMap = DefaultTreeAdapter.DefaultTreeAdapterMap;
}
export type { TreeAdapter, TreeAdapterTypeMap } from './tree-adapters/interface.js';
export { type ParserOptions, /** @internal */ Parser } from './parser/index.js';
export { serialize, serializeOuter, type SerializerOptions } from './serializer/index.js';
export { ERR as ErrorCodes, type ParserError, type ParserErrorHandler } from './common/error-codes.js';
/** @internal */
export * as foreignContent from './common/foreign-content.js';
export * as html from './common/html.js';
export * as Token from './common/token.js';
/** @internal */
export { Tokenizer, type TokenizerOptions, TokenizerMode, type TokenHandler } from './tokenizer/index.js';
/**
* Parses an HTML string.
*
* @param html Input HTML string.
* @param options Parsing options.
* @returns Document
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
*
* console.log(document.childNodes[1].tagName); //> 'html'
*```
*/
export declare function parse<T extends TreeAdapterTypeMap = DefaultTreeAdapterMap>(html: string, options?: ParserOptions<T>): T['document'];
/**
* Parses an HTML fragment.
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const documentFragment = parse5.parseFragment('<table></table>');
*
* console.log(documentFragment.childNodes[0].tagName); //> 'table'
*
* // Parses the html fragment in the context of the parsed <table> element.
* const trFragment = parse5.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
*
* console.log(trFragment.childNodes[0].childNodes[0].tagName); //> 'td'
* ```
*
* @param fragmentContext Parsing context element. If specified, given fragment will be parsed as if it was set to the context element's `innerHTML` property.
* @param html Input HTML fragment string.
* @param options Parsing options.
* @returns DocumentFragment
*/
export declare function parseFragment<T extends TreeAdapterTypeMap = DefaultTreeAdapterMap>(fragmentContext: T['parentNode'] | null, html: string, options: ParserOptions<T>): T['documentFragment'];
export declare function parseFragment<T extends TreeAdapterTypeMap = DefaultTreeAdapterMap>(html: string, options?: ParserOptions<T>): T['documentFragment'];
+54
View File
@@ -0,0 +1,54 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.TokenizerMode = exports.Tokenizer = exports.Token = exports.html = exports.foreignContent = exports.ErrorCodes = exports.serializeOuter = exports.serialize = exports.Parser = exports.defaultTreeAdapter = void 0;
exports.parse = parse;
exports.parseFragment = parseFragment;
const index_js_1 = require("./parser/index.js");
var default_js_1 = require("./tree-adapters/default.js");
Object.defineProperty(exports, "defaultTreeAdapter", { enumerable: true, get: function () { return default_js_1.defaultTreeAdapter; } });
var index_js_2 = require("./parser/index.js");
Object.defineProperty(exports, "Parser", { enumerable: true, get: function () { return index_js_2.Parser; } });
var index_js_3 = require("./serializer/index.js");
Object.defineProperty(exports, "serialize", { enumerable: true, get: function () { return index_js_3.serialize; } });
Object.defineProperty(exports, "serializeOuter", { enumerable: true, get: function () { return index_js_3.serializeOuter; } });
var error_codes_js_1 = require("./common/error-codes.js");
Object.defineProperty(exports, "ErrorCodes", { enumerable: true, get: function () { return error_codes_js_1.ERR; } });
/** @internal */
exports.foreignContent = require("./common/foreign-content.js");
exports.html = require("./common/html.js");
exports.Token = require("./common/token.js");
/** @internal */
var index_js_4 = require("./tokenizer/index.js");
Object.defineProperty(exports, "Tokenizer", { enumerable: true, get: function () { return index_js_4.Tokenizer; } });
Object.defineProperty(exports, "TokenizerMode", { enumerable: true, get: function () { return index_js_4.TokenizerMode; } });
// Shorthands
/**
* Parses an HTML string.
*
* @param html Input HTML string.
* @param options Parsing options.
* @returns Document
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
*
* console.log(document.childNodes[1].tagName); //> 'html'
*```
*/
function parse(html, options) {
return index_js_1.Parser.parse(html, options);
}
function parseFragment(fragmentContext, html, options) {
if (typeof fragmentContext === 'string') {
options = html;
html = fragmentContext;
fragmentContext = null;
}
const parser = index_js_1.Parser.getFragmentParser(fragmentContext, options);
parser.tokenizer.write(html, true);
return parser.getFragment();
}
+1
View File
@@ -0,0 +1 @@
{"type":"commonjs"}
@@ -0,0 +1,36 @@
import type { TagToken } from '../common/token.js';
import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
export declare enum EntryType {
Marker = 0,
Element = 1
}
interface MarkerEntry {
type: EntryType.Marker;
}
export interface ElementEntry<T extends TreeAdapterTypeMap> {
type: EntryType.Element;
element: T['element'];
token: TagToken;
}
export type Entry<T extends TreeAdapterTypeMap> = MarkerEntry | ElementEntry<T>;
export declare class FormattingElementList<T extends TreeAdapterTypeMap> {
private treeAdapter;
entries: Entry<T>[];
bookmark: Entry<T> | null;
constructor(treeAdapter: TreeAdapter<T>);
private _getNoahArkConditionCandidates;
private _ensureNoahArkCondition;
insertMarker(): void;
pushElement(element: T['element'], token: TagToken): void;
insertElementAfterBookmark(element: T['element'], token: TagToken): void;
removeEntry(entry: Entry<T>): void;
/**
* Clears the list of formatting elements up to the last marker.
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
*/
clearToLastMarker(): void;
getElementEntryInScopeWithTagName(tagName: string): ElementEntry<T> | null;
getElementEntry(element: T['element']): ElementEntry<T> | undefined;
}
export {};
+114
View File
@@ -0,0 +1,114 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.FormattingElementList = exports.EntryType = void 0;
//Const
const NOAH_ARK_CAPACITY = 3;
var EntryType;
(function (EntryType) {
EntryType[EntryType["Marker"] = 0] = "Marker";
EntryType[EntryType["Element"] = 1] = "Element";
})(EntryType || (exports.EntryType = EntryType = {}));
const MARKER = { type: EntryType.Marker };
//List of formatting elements
class FormattingElementList {
constructor(treeAdapter) {
this.treeAdapter = treeAdapter;
this.entries = [];
this.bookmark = null;
}
//Noah Ark's condition
//OPTIMIZATION: at first we try to find possible candidates for exclusion using
//lightweight heuristics without thorough attributes check.
_getNoahArkConditionCandidates(newElement, neAttrs) {
const candidates = [];
const neAttrsLength = neAttrs.length;
const neTagName = this.treeAdapter.getTagName(newElement);
const neNamespaceURI = this.treeAdapter.getNamespaceURI(newElement);
for (let i = 0; i < this.entries.length; i++) {
const entry = this.entries[i];
if (entry.type === EntryType.Marker) {
break;
}
const { element } = entry;
if (this.treeAdapter.getTagName(element) === neTagName &&
this.treeAdapter.getNamespaceURI(element) === neNamespaceURI) {
const elementAttrs = this.treeAdapter.getAttrList(element);
if (elementAttrs.length === neAttrsLength) {
candidates.push({ idx: i, attrs: elementAttrs });
}
}
}
return candidates;
}
_ensureNoahArkCondition(newElement) {
if (this.entries.length < NOAH_ARK_CAPACITY)
return;
const neAttrs = this.treeAdapter.getAttrList(newElement);
const candidates = this._getNoahArkConditionCandidates(newElement, neAttrs);
if (candidates.length < NOAH_ARK_CAPACITY)
return;
//NOTE: build attrs map for the new element, so we can perform fast lookups
const neAttrsMap = new Map(neAttrs.map((neAttr) => [neAttr.name, neAttr.value]));
let validCandidates = 0;
//NOTE: remove bottommost candidates, until Noah's Ark condition will not be met
for (let i = 0; i < candidates.length; i++) {
const candidate = candidates[i];
// We know that `candidate.attrs.length === neAttrs.length`
if (candidate.attrs.every((cAttr) => neAttrsMap.get(cAttr.name) === cAttr.value)) {
validCandidates += 1;
if (validCandidates >= NOAH_ARK_CAPACITY) {
this.entries.splice(candidate.idx, 1);
}
}
}
}
//Mutations
insertMarker() {
this.entries.unshift(MARKER);
}
pushElement(element, token) {
this._ensureNoahArkCondition(element);
this.entries.unshift({
type: EntryType.Element,
element,
token,
});
}
insertElementAfterBookmark(element, token) {
const bookmarkIdx = this.entries.indexOf(this.bookmark);
this.entries.splice(bookmarkIdx, 0, {
type: EntryType.Element,
element,
token,
});
}
removeEntry(entry) {
const entryIndex = this.entries.indexOf(entry);
if (entryIndex !== -1) {
this.entries.splice(entryIndex, 1);
}
}
/**
* Clears the list of formatting elements up to the last marker.
*
* @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
*/
clearToLastMarker() {
const markerIdx = this.entries.indexOf(MARKER);
if (markerIdx === -1) {
this.entries.length = 0;
}
else {
this.entries.splice(0, markerIdx + 1);
}
}
//Search
getElementEntryInScopeWithTagName(tagName) {
const entry = this.entries.find((entry) => entry.type === EntryType.Marker || this.treeAdapter.getTagName(entry.element) === tagName);
return entry && entry.type === EntryType.Element ? entry : null;
}
getElementEntry(element) {
return this.entries.find((entry) => entry.type === EntryType.Element && entry.element === element);
}
}
exports.FormattingElementList = FormattingElementList;
+221
View File
@@ -0,0 +1,221 @@
import { Tokenizer, TokenizerMode, type TokenHandler } from '../tokenizer/index.js';
import { OpenElementStack, type StackHandler } from './open-element-stack.js';
import { FormattingElementList } from './formatting-element-list.js';
import { ERR, type ParserErrorHandler } from '../common/error-codes.js';
import { TAG_ID as $, NS } from '../common/html.js';
import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
import { type Token, type CommentToken, type CharacterToken, type TagToken, type DoctypeToken, type EOFToken, type LocationWithAttributes } from '../common/token.js';
declare enum InsertionMode {
INITIAL = 0,
BEFORE_HTML = 1,
BEFORE_HEAD = 2,
IN_HEAD = 3,
IN_HEAD_NO_SCRIPT = 4,
AFTER_HEAD = 5,
IN_BODY = 6,
TEXT = 7,
IN_TABLE = 8,
IN_TABLE_TEXT = 9,
IN_CAPTION = 10,
IN_COLUMN_GROUP = 11,
IN_TABLE_BODY = 12,
IN_ROW = 13,
IN_CELL = 14,
IN_SELECT = 15,
IN_SELECT_IN_TABLE = 16,
IN_TEMPLATE = 17,
AFTER_BODY = 18,
IN_FRAMESET = 19,
AFTER_FRAMESET = 20,
AFTER_AFTER_BODY = 21,
AFTER_AFTER_FRAMESET = 22
}
export interface ParserOptions<T extends TreeAdapterTypeMap> {
/**
* The [scripting flag](https://html.spec.whatwg.org/multipage/parsing.html#scripting-flag). If set
* to `true`, `noscript` element content will be parsed as text.
*
* @default `true`
*/
scriptingEnabled?: boolean;
/**
* Enables source code location information. When enabled, each node (except the root node)
* will have a `sourceCodeLocation` property. If the node is not an empty element, `sourceCodeLocation` will
* be a {@link ElementLocation} object, otherwise it will be {@link Location}.
* If the element was implicitly created by the parser (as part of
* [tree correction](https://html.spec.whatwg.org/multipage/syntax.html#an-introduction-to-error-handling-and-strange-cases-in-the-parser)),
* its `sourceCodeLocation` property will be `undefined`.
*
* @default `false`
*/
sourceCodeLocationInfo?: boolean;
/**
* Specifies the resulting tree format.
*
* @default `treeAdapters.default`
*/
treeAdapter?: TreeAdapter<T>;
/**
* Callback for parse errors.
*
* @default `null`
*/
onParseError?: ParserErrorHandler | null;
}
export declare class Parser<T extends TreeAdapterTypeMap> implements TokenHandler, StackHandler<T> {
/** @internal */
fragmentContext: T['element'] | null;
/** @internal */
scriptHandler: null | ((pendingScript: T['element']) => void);
treeAdapter: TreeAdapter<T>;
/** @internal */
onParseError: ParserErrorHandler | null;
protected currentToken: Token | null;
options: Required<ParserOptions<T>>;
document: T['document'];
constructor(options?: ParserOptions<T>, document?: T['document'],
/** @internal */
fragmentContext?: T['element'] | null,
/** @internal */
scriptHandler?: null | ((pendingScript: T['element']) => void));
static parse<T extends TreeAdapterTypeMap>(html: string, options?: ParserOptions<T>): T['document'];
static getFragmentParser<T extends TreeAdapterTypeMap>(fragmentContext?: T['parentNode'] | null, options?: ParserOptions<T>): Parser<T>;
getFragment(): T['documentFragment'];
tokenizer: Tokenizer;
stopped: boolean;
/** @internal */
insertionMode: InsertionMode;
/** @internal */
originalInsertionMode: InsertionMode;
/** @internal */
fragmentContextID: $;
/** @internal */
headElement: null | T['element'];
/** @internal */
formElement: null | T['element'];
/** @internal */
openElements: OpenElementStack<T>;
/** @internal */
activeFormattingElements: FormattingElementList<T>;
/** Indicates that the current node is not an element in the HTML namespace */
protected currentNotInHTML: boolean;
/**
* The template insertion mode stack is maintained from the left.
* Ie. the topmost element will always have index 0.
*
* @internal
*/
tmplInsertionModeStack: InsertionMode[];
/** @internal */
pendingCharacterTokens: CharacterToken[];
/** @internal */
hasNonWhitespacePendingCharacterToken: boolean;
/** @internal */
framesetOk: boolean;
/** @internal */
skipNextNewLine: boolean;
/** @internal */
fosterParentingEnabled: boolean;
/** @internal */
_err(token: Token, code: ERR, beforeToken?: boolean): void;
/** @internal */
onItemPush(node: T['parentNode'], tid: number, isTop: boolean): void;
/** @internal */
onItemPop(node: T['parentNode'], isTop: boolean): void;
protected _setContextModes(current: T['parentNode'] | undefined, tid: number | undefined): void;
/** @protected */
_switchToTextParsing(currentToken: TagToken, nextTokenizerState: (typeof TokenizerMode)[keyof typeof TokenizerMode]): void;
switchToPlaintextParsing(): void;
/** @protected */
_getAdjustedCurrentElement(): T['element'];
/** @protected */
_findFormInFragmentContext(): void;
protected _initTokenizerForFragmentParsing(): void;
/** @protected */
_setDocumentType(token: DoctypeToken): void;
/** @protected */
_attachElementToTree(element: T['element'], location: LocationWithAttributes | null): void;
/**
* For self-closing tags. Add an element to the tree, but skip adding it
* to the stack.
*/
/** @protected */
_appendElement(token: TagToken, namespaceURI: NS): void;
/** @protected */
_insertElement(token: TagToken, namespaceURI: NS): void;
/** @protected */
_insertFakeElement(tagName: string, tagID: $): void;
/** @protected */
_insertTemplate(token: TagToken): void;
/** @protected */
_insertFakeRootElement(): void;
/** @protected */
_appendCommentNode(token: CommentToken, parent: T['parentNode']): void;
/** @protected */
_insertCharacters(token: CharacterToken): void;
/** @protected */
_adoptNodes(donor: T['parentNode'], recipient: T['parentNode']): void;
/** @protected */
_setEndLocation(element: T['element'], closingToken: Token): void;
protected shouldProcessStartTagTokenInForeignContent(token: TagToken): boolean;
/** @protected */
_processToken(token: Token): void;
/** @protected */
_isIntegrationPoint(tid: $, element: T['element'], foreignNS?: NS): boolean;
/** @protected */
_reconstructActiveFormattingElements(): void;
/** @protected */
_closeTableCell(): void;
/** @protected */
_closePElement(): void;
/** @protected */
_resetInsertionMode(): void;
/** @protected */
_resetInsertionModeForSelect(selectIdx: number): void;
/** @protected */
_isElementCausesFosterParenting(tn: $): boolean;
/** @protected */
_shouldFosterParentOnInsertion(): boolean;
/** @protected */
_findFosterParentingLocation(): {
parent: T['parentNode'];
beforeElement: T['element'] | null;
};
/** @protected */
_fosterParentElement(element: T['element']): void;
/** @protected */
_isSpecialElement(element: T['element'], id: $): boolean;
/** @internal */
onCharacter(token: CharacterToken): void;
/** @internal */
onNullCharacter(token: CharacterToken): void;
/** @internal */
onComment(token: CommentToken): void;
/** @internal */
onDoctype(token: DoctypeToken): void;
/** @internal */
onStartTag(token: TagToken): void;
/**
* Processes a given start tag.
*
* `onStartTag` checks if a self-closing tag was recognized. When a token
* is moved inbetween multiple insertion modes, this check for self-closing
* could lead to false positives. To avoid this, `_processStartTag` is used
* for nested calls.
*
* @param token The token to process.
* @protected
*/
_processStartTag(token: TagToken): void;
/** @protected */
_startTagOutsideForeignContent(token: TagToken): void;
/** @internal */
onEndTag(token: TagToken): void;
/** @protected */
_endTagOutsideForeignContent(token: TagToken): void;
/** @internal */
onEof(token: EOFToken): void;
/** @internal */
onWhitespaceCharacter(token: CharacterToken): void;
}
export {};
+3240
View File
@@ -0,0 +1,3240 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.Parser = void 0;
const index_js_1 = require("../tokenizer/index.js");
const open_element_stack_js_1 = require("./open-element-stack.js");
const formatting_element_list_js_1 = require("./formatting-element-list.js");
const default_js_1 = require("../tree-adapters/default.js");
const doctype = require("../common/doctype.js");
const foreignContent = require("../common/foreign-content.js");
const error_codes_js_1 = require("../common/error-codes.js");
const unicode = require("../common/unicode.js");
const html_js_1 = require("../common/html.js");
const token_js_1 = require("../common/token.js");
//Misc constants
const HIDDEN_INPUT_TYPE = 'hidden';
//Adoption agency loops iteration count
const AA_OUTER_LOOP_ITER = 8;
const AA_INNER_LOOP_ITER = 3;
//Insertion modes
var InsertionMode;
(function (InsertionMode) {
InsertionMode[InsertionMode["INITIAL"] = 0] = "INITIAL";
InsertionMode[InsertionMode["BEFORE_HTML"] = 1] = "BEFORE_HTML";
InsertionMode[InsertionMode["BEFORE_HEAD"] = 2] = "BEFORE_HEAD";
InsertionMode[InsertionMode["IN_HEAD"] = 3] = "IN_HEAD";
InsertionMode[InsertionMode["IN_HEAD_NO_SCRIPT"] = 4] = "IN_HEAD_NO_SCRIPT";
InsertionMode[InsertionMode["AFTER_HEAD"] = 5] = "AFTER_HEAD";
InsertionMode[InsertionMode["IN_BODY"] = 6] = "IN_BODY";
InsertionMode[InsertionMode["TEXT"] = 7] = "TEXT";
InsertionMode[InsertionMode["IN_TABLE"] = 8] = "IN_TABLE";
InsertionMode[InsertionMode["IN_TABLE_TEXT"] = 9] = "IN_TABLE_TEXT";
InsertionMode[InsertionMode["IN_CAPTION"] = 10] = "IN_CAPTION";
InsertionMode[InsertionMode["IN_COLUMN_GROUP"] = 11] = "IN_COLUMN_GROUP";
InsertionMode[InsertionMode["IN_TABLE_BODY"] = 12] = "IN_TABLE_BODY";
InsertionMode[InsertionMode["IN_ROW"] = 13] = "IN_ROW";
InsertionMode[InsertionMode["IN_CELL"] = 14] = "IN_CELL";
InsertionMode[InsertionMode["IN_SELECT"] = 15] = "IN_SELECT";
InsertionMode[InsertionMode["IN_SELECT_IN_TABLE"] = 16] = "IN_SELECT_IN_TABLE";
InsertionMode[InsertionMode["IN_TEMPLATE"] = 17] = "IN_TEMPLATE";
InsertionMode[InsertionMode["AFTER_BODY"] = 18] = "AFTER_BODY";
InsertionMode[InsertionMode["IN_FRAMESET"] = 19] = "IN_FRAMESET";
InsertionMode[InsertionMode["AFTER_FRAMESET"] = 20] = "AFTER_FRAMESET";
InsertionMode[InsertionMode["AFTER_AFTER_BODY"] = 21] = "AFTER_AFTER_BODY";
InsertionMode[InsertionMode["AFTER_AFTER_FRAMESET"] = 22] = "AFTER_AFTER_FRAMESET";
})(InsertionMode || (InsertionMode = {}));
const BASE_LOC = {
startLine: -1,
startCol: -1,
startOffset: -1,
endLine: -1,
endCol: -1,
endOffset: -1,
};
const TABLE_STRUCTURE_TAGS = new Set([html_js_1.TAG_ID.TABLE, html_js_1.TAG_ID.TBODY, html_js_1.TAG_ID.TFOOT, html_js_1.TAG_ID.THEAD, html_js_1.TAG_ID.TR]);
const defaultParserOptions = {
scriptingEnabled: true,
sourceCodeLocationInfo: false,
treeAdapter: default_js_1.defaultTreeAdapter,
onParseError: null,
};
//Parser
class Parser {
constructor(options, document,
/** @internal */
fragmentContext = null,
/** @internal */
scriptHandler = null) {
this.fragmentContext = fragmentContext;
this.scriptHandler = scriptHandler;
this.currentToken = null;
this.stopped = false;
/** @internal */
this.insertionMode = InsertionMode.INITIAL;
/** @internal */
this.originalInsertionMode = InsertionMode.INITIAL;
/** @internal */
this.headElement = null;
/** @internal */
this.formElement = null;
/** Indicates that the current node is not an element in the HTML namespace */
this.currentNotInHTML = false;
/**
* The template insertion mode stack is maintained from the left.
* Ie. the topmost element will always have index 0.
*
* @internal
*/
this.tmplInsertionModeStack = [];
/** @internal */
this.pendingCharacterTokens = [];
/** @internal */
this.hasNonWhitespacePendingCharacterToken = false;
/** @internal */
this.framesetOk = true;
/** @internal */
this.skipNextNewLine = false;
/** @internal */
this.fosterParentingEnabled = false;
this.options = Object.assign(Object.assign({}, defaultParserOptions), options);
this.treeAdapter = this.options.treeAdapter;
this.onParseError = this.options.onParseError;
// Always enable location info if we report parse errors.
if (this.onParseError) {
this.options.sourceCodeLocationInfo = true;
}
this.document = document !== null && document !== void 0 ? document : this.treeAdapter.createDocument();
this.tokenizer = new index_js_1.Tokenizer(this.options, this);
this.activeFormattingElements = new formatting_element_list_js_1.FormattingElementList(this.treeAdapter);
this.fragmentContextID = fragmentContext ? (0, html_js_1.getTagID)(this.treeAdapter.getTagName(fragmentContext)) : html_js_1.TAG_ID.UNKNOWN;
this._setContextModes(fragmentContext !== null && fragmentContext !== void 0 ? fragmentContext : this.document, this.fragmentContextID);
this.openElements = new open_element_stack_js_1.OpenElementStack(this.document, this.treeAdapter, this);
}
// API
static parse(html, options) {
const parser = new this(options);
parser.tokenizer.write(html, true);
return parser.document;
}
static getFragmentParser(fragmentContext, options) {
const opts = Object.assign(Object.assign({}, defaultParserOptions), options);
//NOTE: use a <template> element as the fragment context if no context element was provided,
//so we will parse in a "forgiving" manner
fragmentContext !== null && fragmentContext !== void 0 ? fragmentContext : (fragmentContext = opts.treeAdapter.createElement(html_js_1.TAG_NAMES.TEMPLATE, html_js_1.NS.HTML, []));
//NOTE: create a fake element which will be used as the `document` for fragment parsing.
//This is important for jsdom, where a new `document` cannot be created. This led to
//fragment parsing messing with the main `document`.
const documentMock = opts.treeAdapter.createElement('documentmock', html_js_1.NS.HTML, []);
const parser = new this(opts, documentMock, fragmentContext);
if (parser.fragmentContextID === html_js_1.TAG_ID.TEMPLATE) {
parser.tmplInsertionModeStack.unshift(InsertionMode.IN_TEMPLATE);
}
parser._initTokenizerForFragmentParsing();
parser._insertFakeRootElement();
parser._resetInsertionMode();
parser._findFormInFragmentContext();
return parser;
}
getFragment() {
const rootElement = this.treeAdapter.getFirstChild(this.document);
const fragment = this.treeAdapter.createDocumentFragment();
this._adoptNodes(rootElement, fragment);
return fragment;
}
//Errors
/** @internal */
_err(token, code, beforeToken) {
var _a;
if (!this.onParseError)
return;
const loc = (_a = token.location) !== null && _a !== void 0 ? _a : BASE_LOC;
const err = {
code,
startLine: loc.startLine,
startCol: loc.startCol,
startOffset: loc.startOffset,
endLine: beforeToken ? loc.startLine : loc.endLine,
endCol: beforeToken ? loc.startCol : loc.endCol,
endOffset: beforeToken ? loc.startOffset : loc.endOffset,
};
this.onParseError(err);
}
//Stack events
/** @internal */
onItemPush(node, tid, isTop) {
var _a, _b;
(_b = (_a = this.treeAdapter).onItemPush) === null || _b === void 0 ? void 0 : _b.call(_a, node);
if (isTop && this.openElements.stackTop > 0)
this._setContextModes(node, tid);
}
/** @internal */
onItemPop(node, isTop) {
var _a, _b;
if (this.options.sourceCodeLocationInfo) {
this._setEndLocation(node, this.currentToken);
}
(_b = (_a = this.treeAdapter).onItemPop) === null || _b === void 0 ? void 0 : _b.call(_a, node, this.openElements.current);
if (isTop) {
let current;
let currentTagId;
if (this.openElements.stackTop === 0 && this.fragmentContext) {
current = this.fragmentContext;
currentTagId = this.fragmentContextID;
}
else {
({ current, currentTagId } = this.openElements);
}
this._setContextModes(current, currentTagId);
}
}
_setContextModes(current, tid) {
const isHTML = current === this.document || (current && this.treeAdapter.getNamespaceURI(current) === html_js_1.NS.HTML);
this.currentNotInHTML = !isHTML;
this.tokenizer.inForeignNode =
!isHTML && current !== undefined && tid !== undefined && !this._isIntegrationPoint(tid, current);
}
/** @protected */
_switchToTextParsing(currentToken, nextTokenizerState) {
this._insertElement(currentToken, html_js_1.NS.HTML);
this.tokenizer.state = nextTokenizerState;
this.originalInsertionMode = this.insertionMode;
this.insertionMode = InsertionMode.TEXT;
}
switchToPlaintextParsing() {
this.insertionMode = InsertionMode.TEXT;
this.originalInsertionMode = InsertionMode.IN_BODY;
this.tokenizer.state = index_js_1.TokenizerMode.PLAINTEXT;
}
//Fragment parsing
/** @protected */
_getAdjustedCurrentElement() {
return this.openElements.stackTop === 0 && this.fragmentContext
? this.fragmentContext
: this.openElements.current;
}
/** @protected */
_findFormInFragmentContext() {
let node = this.fragmentContext;
while (node) {
if (this.treeAdapter.getTagName(node) === html_js_1.TAG_NAMES.FORM) {
this.formElement = node;
break;
}
node = this.treeAdapter.getParentNode(node);
}
}
_initTokenizerForFragmentParsing() {
if (!this.fragmentContext || this.treeAdapter.getNamespaceURI(this.fragmentContext) !== html_js_1.NS.HTML) {
return;
}
switch (this.fragmentContextID) {
case html_js_1.TAG_ID.TITLE:
case html_js_1.TAG_ID.TEXTAREA: {
this.tokenizer.state = index_js_1.TokenizerMode.RCDATA;
break;
}
case html_js_1.TAG_ID.STYLE:
case html_js_1.TAG_ID.XMP:
case html_js_1.TAG_ID.IFRAME:
case html_js_1.TAG_ID.NOEMBED:
case html_js_1.TAG_ID.NOFRAMES:
case html_js_1.TAG_ID.NOSCRIPT: {
this.tokenizer.state = index_js_1.TokenizerMode.RAWTEXT;
break;
}
case html_js_1.TAG_ID.SCRIPT: {
this.tokenizer.state = index_js_1.TokenizerMode.SCRIPT_DATA;
break;
}
case html_js_1.TAG_ID.PLAINTEXT: {
this.tokenizer.state = index_js_1.TokenizerMode.PLAINTEXT;
break;
}
default:
// Do nothing
}
}
//Tree mutation
/** @protected */
_setDocumentType(token) {
const name = token.name || '';
const publicId = token.publicId || '';
const systemId = token.systemId || '';
this.treeAdapter.setDocumentType(this.document, name, publicId, systemId);
if (token.location) {
const documentChildren = this.treeAdapter.getChildNodes(this.document);
const docTypeNode = documentChildren.find((node) => this.treeAdapter.isDocumentTypeNode(node));
if (docTypeNode) {
this.treeAdapter.setNodeSourceCodeLocation(docTypeNode, token.location);
}
}
}
/** @protected */
_attachElementToTree(element, location) {
if (this.options.sourceCodeLocationInfo) {
const loc = location && Object.assign(Object.assign({}, location), { startTag: location });
this.treeAdapter.setNodeSourceCodeLocation(element, loc);
}
if (this._shouldFosterParentOnInsertion()) {
this._fosterParentElement(element);
}
else {
const parent = this.openElements.currentTmplContentOrNode;
this.treeAdapter.appendChild(parent !== null && parent !== void 0 ? parent : this.document, element);
}
}
/**
* For self-closing tags. Add an element to the tree, but skip adding it
* to the stack.
*/
/** @protected */
_appendElement(token, namespaceURI) {
const element = this.treeAdapter.createElement(token.tagName, namespaceURI, token.attrs);
this._attachElementToTree(element, token.location);
}
/** @protected */
_insertElement(token, namespaceURI) {
const element = this.treeAdapter.createElement(token.tagName, namespaceURI, token.attrs);
this._attachElementToTree(element, token.location);
this.openElements.push(element, token.tagID);
}
/** @protected */
_insertFakeElement(tagName, tagID) {
const element = this.treeAdapter.createElement(tagName, html_js_1.NS.HTML, []);
this._attachElementToTree(element, null);
this.openElements.push(element, tagID);
}
/** @protected */
_insertTemplate(token) {
const tmpl = this.treeAdapter.createElement(token.tagName, html_js_1.NS.HTML, token.attrs);
const content = this.treeAdapter.createDocumentFragment();
this.treeAdapter.setTemplateContent(tmpl, content);
this._attachElementToTree(tmpl, token.location);
this.openElements.push(tmpl, token.tagID);
if (this.options.sourceCodeLocationInfo)
this.treeAdapter.setNodeSourceCodeLocation(content, null);
}
/** @protected */
_insertFakeRootElement() {
const element = this.treeAdapter.createElement(html_js_1.TAG_NAMES.HTML, html_js_1.NS.HTML, []);
if (this.options.sourceCodeLocationInfo)
this.treeAdapter.setNodeSourceCodeLocation(element, null);
this.treeAdapter.appendChild(this.openElements.current, element);
this.openElements.push(element, html_js_1.TAG_ID.HTML);
}
/** @protected */
_appendCommentNode(token, parent) {
const commentNode = this.treeAdapter.createCommentNode(token.data);
this.treeAdapter.appendChild(parent, commentNode);
if (this.options.sourceCodeLocationInfo) {
this.treeAdapter.setNodeSourceCodeLocation(commentNode, token.location);
}
}
/** @protected */
_insertCharacters(token) {
let parent;
let beforeElement;
if (this._shouldFosterParentOnInsertion()) {
({ parent, beforeElement } = this._findFosterParentingLocation());
if (beforeElement) {
this.treeAdapter.insertTextBefore(parent, token.chars, beforeElement);
}
else {
this.treeAdapter.insertText(parent, token.chars);
}
}
else {
parent = this.openElements.currentTmplContentOrNode;
this.treeAdapter.insertText(parent, token.chars);
}
if (!token.location)
return;
const siblings = this.treeAdapter.getChildNodes(parent);
const textNodeIdx = beforeElement ? siblings.lastIndexOf(beforeElement) : siblings.length;
const textNode = siblings[textNodeIdx - 1];
//NOTE: if we have a location assigned by another token, then just update the end position
const tnLoc = this.treeAdapter.getNodeSourceCodeLocation(textNode);
if (tnLoc) {
const { endLine, endCol, endOffset } = token.location;
this.treeAdapter.updateNodeSourceCodeLocation(textNode, { endLine, endCol, endOffset });
}
else if (this.options.sourceCodeLocationInfo) {
this.treeAdapter.setNodeSourceCodeLocation(textNode, token.location);
}
}
/** @protected */
_adoptNodes(donor, recipient) {
for (let child = this.treeAdapter.getFirstChild(donor); child; child = this.treeAdapter.getFirstChild(donor)) {
this.treeAdapter.detachNode(child);
this.treeAdapter.appendChild(recipient, child);
}
}
/** @protected */
_setEndLocation(element, closingToken) {
if (this.treeAdapter.getNodeSourceCodeLocation(element) && closingToken.location) {
const ctLoc = closingToken.location;
const tn = this.treeAdapter.getTagName(element);
const endLoc =
// NOTE: For cases like <p> <p> </p> - First 'p' closes without a closing
// tag and for cases like <td> <p> </td> - 'p' closes without a closing tag.
closingToken.type === token_js_1.TokenType.END_TAG && tn === closingToken.tagName
? {
endTag: Object.assign({}, ctLoc),
endLine: ctLoc.endLine,
endCol: ctLoc.endCol,
endOffset: ctLoc.endOffset,
}
: {
endLine: ctLoc.startLine,
endCol: ctLoc.startCol,
endOffset: ctLoc.startOffset,
};
this.treeAdapter.updateNodeSourceCodeLocation(element, endLoc);
}
}
//Token processing
shouldProcessStartTagTokenInForeignContent(token) {
// Check that neither current === document, or ns === NS.HTML
if (!this.currentNotInHTML)
return false;
let current;
let currentTagId;
if (this.openElements.stackTop === 0 && this.fragmentContext) {
current = this.fragmentContext;
currentTagId = this.fragmentContextID;
}
else {
({ current, currentTagId } = this.openElements);
}
if (token.tagID === html_js_1.TAG_ID.SVG &&
this.treeAdapter.getTagName(current) === html_js_1.TAG_NAMES.ANNOTATION_XML &&
this.treeAdapter.getNamespaceURI(current) === html_js_1.NS.MATHML) {
return false;
}
return (
// Check that `current` is not an integration point for HTML or MathML elements.
this.tokenizer.inForeignNode ||
// If it _is_ an integration point, then we might have to check that it is not an HTML
// integration point.
((token.tagID === html_js_1.TAG_ID.MGLYPH || token.tagID === html_js_1.TAG_ID.MALIGNMARK) &&
currentTagId !== undefined &&
!this._isIntegrationPoint(currentTagId, current, html_js_1.NS.HTML)));
}
/** @protected */
_processToken(token) {
switch (token.type) {
case token_js_1.TokenType.CHARACTER: {
this.onCharacter(token);
break;
}
case token_js_1.TokenType.NULL_CHARACTER: {
this.onNullCharacter(token);
break;
}
case token_js_1.TokenType.COMMENT: {
this.onComment(token);
break;
}
case token_js_1.TokenType.DOCTYPE: {
this.onDoctype(token);
break;
}
case token_js_1.TokenType.START_TAG: {
this._processStartTag(token);
break;
}
case token_js_1.TokenType.END_TAG: {
this.onEndTag(token);
break;
}
case token_js_1.TokenType.EOF: {
this.onEof(token);
break;
}
case token_js_1.TokenType.WHITESPACE_CHARACTER: {
this.onWhitespaceCharacter(token);
break;
}
}
}
//Integration points
/** @protected */
_isIntegrationPoint(tid, element, foreignNS) {
const ns = this.treeAdapter.getNamespaceURI(element);
const attrs = this.treeAdapter.getAttrList(element);
return foreignContent.isIntegrationPoint(tid, ns, attrs, foreignNS);
}
//Active formatting elements reconstruction
/** @protected */
_reconstructActiveFormattingElements() {
const listLength = this.activeFormattingElements.entries.length;
if (listLength) {
const endIndex = this.activeFormattingElements.entries.findIndex((entry) => entry.type === formatting_element_list_js_1.EntryType.Marker || this.openElements.contains(entry.element));
const unopenIdx = endIndex === -1 ? listLength - 1 : endIndex - 1;
for (let i = unopenIdx; i >= 0; i--) {
const entry = this.activeFormattingElements.entries[i];
this._insertElement(entry.token, this.treeAdapter.getNamespaceURI(entry.element));
entry.element = this.openElements.current;
}
}
}
//Close elements
/** @protected */
_closeTableCell() {
this.openElements.generateImpliedEndTags();
this.openElements.popUntilTableCellPopped();
this.activeFormattingElements.clearToLastMarker();
this.insertionMode = InsertionMode.IN_ROW;
}
/** @protected */
_closePElement() {
this.openElements.generateImpliedEndTagsWithExclusion(html_js_1.TAG_ID.P);
this.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.P);
}
//Insertion modes
/** @protected */
_resetInsertionMode() {
for (let i = this.openElements.stackTop; i >= 0; i--) {
//Insertion mode reset map
switch (i === 0 && this.fragmentContext ? this.fragmentContextID : this.openElements.tagIDs[i]) {
case html_js_1.TAG_ID.TR: {
this.insertionMode = InsertionMode.IN_ROW;
return;
}
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TFOOT: {
this.insertionMode = InsertionMode.IN_TABLE_BODY;
return;
}
case html_js_1.TAG_ID.CAPTION: {
this.insertionMode = InsertionMode.IN_CAPTION;
return;
}
case html_js_1.TAG_ID.COLGROUP: {
this.insertionMode = InsertionMode.IN_COLUMN_GROUP;
return;
}
case html_js_1.TAG_ID.TABLE: {
this.insertionMode = InsertionMode.IN_TABLE;
return;
}
case html_js_1.TAG_ID.BODY: {
this.insertionMode = InsertionMode.IN_BODY;
return;
}
case html_js_1.TAG_ID.FRAMESET: {
this.insertionMode = InsertionMode.IN_FRAMESET;
return;
}
case html_js_1.TAG_ID.SELECT: {
this._resetInsertionModeForSelect(i);
return;
}
case html_js_1.TAG_ID.TEMPLATE: {
this.insertionMode = this.tmplInsertionModeStack[0];
return;
}
case html_js_1.TAG_ID.HTML: {
this.insertionMode = this.headElement ? InsertionMode.AFTER_HEAD : InsertionMode.BEFORE_HEAD;
return;
}
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH: {
if (i > 0) {
this.insertionMode = InsertionMode.IN_CELL;
return;
}
break;
}
case html_js_1.TAG_ID.HEAD: {
if (i > 0) {
this.insertionMode = InsertionMode.IN_HEAD;
return;
}
break;
}
}
}
this.insertionMode = InsertionMode.IN_BODY;
}
/** @protected */
_resetInsertionModeForSelect(selectIdx) {
if (selectIdx > 0) {
for (let i = selectIdx - 1; i > 0; i--) {
const tn = this.openElements.tagIDs[i];
if (tn === html_js_1.TAG_ID.TEMPLATE) {
break;
}
else if (tn === html_js_1.TAG_ID.TABLE) {
this.insertionMode = InsertionMode.IN_SELECT_IN_TABLE;
return;
}
}
}
this.insertionMode = InsertionMode.IN_SELECT;
}
//Foster parenting
/** @protected */
_isElementCausesFosterParenting(tn) {
return TABLE_STRUCTURE_TAGS.has(tn);
}
/** @protected */
_shouldFosterParentOnInsertion() {
return (this.fosterParentingEnabled &&
this.openElements.currentTagId !== undefined &&
this._isElementCausesFosterParenting(this.openElements.currentTagId));
}
/** @protected */
_findFosterParentingLocation() {
for (let i = this.openElements.stackTop; i >= 0; i--) {
const openElement = this.openElements.items[i];
switch (this.openElements.tagIDs[i]) {
case html_js_1.TAG_ID.TEMPLATE: {
if (this.treeAdapter.getNamespaceURI(openElement) === html_js_1.NS.HTML) {
return { parent: this.treeAdapter.getTemplateContent(openElement), beforeElement: null };
}
break;
}
case html_js_1.TAG_ID.TABLE: {
const parent = this.treeAdapter.getParentNode(openElement);
if (parent) {
return { parent, beforeElement: openElement };
}
return { parent: this.openElements.items[i - 1], beforeElement: null };
}
default:
// Do nothing
}
}
return { parent: this.openElements.items[0], beforeElement: null };
}
/** @protected */
_fosterParentElement(element) {
const location = this._findFosterParentingLocation();
if (location.beforeElement) {
this.treeAdapter.insertBefore(location.parent, element, location.beforeElement);
}
else {
this.treeAdapter.appendChild(location.parent, element);
}
}
//Special elements
/** @protected */
_isSpecialElement(element, id) {
const ns = this.treeAdapter.getNamespaceURI(element);
return html_js_1.SPECIAL_ELEMENTS[ns].has(id);
}
/** @internal */
onCharacter(token) {
this.skipNextNewLine = false;
if (this.tokenizer.inForeignNode) {
characterInForeignContent(this, token);
return;
}
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
tokenInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HTML: {
tokenBeforeHtml(this, token);
break;
}
case InsertionMode.BEFORE_HEAD: {
tokenBeforeHead(this, token);
break;
}
case InsertionMode.IN_HEAD: {
tokenInHead(this, token);
break;
}
case InsertionMode.IN_HEAD_NO_SCRIPT: {
tokenInHeadNoScript(this, token);
break;
}
case InsertionMode.AFTER_HEAD: {
tokenAfterHead(this, token);
break;
}
case InsertionMode.IN_BODY:
case InsertionMode.IN_CAPTION:
case InsertionMode.IN_CELL:
case InsertionMode.IN_TEMPLATE: {
characterInBody(this, token);
break;
}
case InsertionMode.TEXT:
case InsertionMode.IN_SELECT:
case InsertionMode.IN_SELECT_IN_TABLE: {
this._insertCharacters(token);
break;
}
case InsertionMode.IN_TABLE:
case InsertionMode.IN_TABLE_BODY:
case InsertionMode.IN_ROW: {
characterInTable(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
characterInTableText(this, token);
break;
}
case InsertionMode.IN_COLUMN_GROUP: {
tokenInColumnGroup(this, token);
break;
}
case InsertionMode.AFTER_BODY: {
tokenAfterBody(this, token);
break;
}
case InsertionMode.AFTER_AFTER_BODY: {
tokenAfterAfterBody(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onNullCharacter(token) {
this.skipNextNewLine = false;
if (this.tokenizer.inForeignNode) {
nullCharacterInForeignContent(this, token);
return;
}
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
tokenInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HTML: {
tokenBeforeHtml(this, token);
break;
}
case InsertionMode.BEFORE_HEAD: {
tokenBeforeHead(this, token);
break;
}
case InsertionMode.IN_HEAD: {
tokenInHead(this, token);
break;
}
case InsertionMode.IN_HEAD_NO_SCRIPT: {
tokenInHeadNoScript(this, token);
break;
}
case InsertionMode.AFTER_HEAD: {
tokenAfterHead(this, token);
break;
}
case InsertionMode.TEXT: {
this._insertCharacters(token);
break;
}
case InsertionMode.IN_TABLE:
case InsertionMode.IN_TABLE_BODY:
case InsertionMode.IN_ROW: {
characterInTable(this, token);
break;
}
case InsertionMode.IN_COLUMN_GROUP: {
tokenInColumnGroup(this, token);
break;
}
case InsertionMode.AFTER_BODY: {
tokenAfterBody(this, token);
break;
}
case InsertionMode.AFTER_AFTER_BODY: {
tokenAfterAfterBody(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onComment(token) {
this.skipNextNewLine = false;
if (this.currentNotInHTML) {
appendComment(this, token);
return;
}
switch (this.insertionMode) {
case InsertionMode.INITIAL:
case InsertionMode.BEFORE_HTML:
case InsertionMode.BEFORE_HEAD:
case InsertionMode.IN_HEAD:
case InsertionMode.IN_HEAD_NO_SCRIPT:
case InsertionMode.AFTER_HEAD:
case InsertionMode.IN_BODY:
case InsertionMode.IN_TABLE:
case InsertionMode.IN_CAPTION:
case InsertionMode.IN_COLUMN_GROUP:
case InsertionMode.IN_TABLE_BODY:
case InsertionMode.IN_ROW:
case InsertionMode.IN_CELL:
case InsertionMode.IN_SELECT:
case InsertionMode.IN_SELECT_IN_TABLE:
case InsertionMode.IN_TEMPLATE:
case InsertionMode.IN_FRAMESET:
case InsertionMode.AFTER_FRAMESET: {
appendComment(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
tokenInTableText(this, token);
break;
}
case InsertionMode.AFTER_BODY: {
appendCommentToRootHtmlElement(this, token);
break;
}
case InsertionMode.AFTER_AFTER_BODY:
case InsertionMode.AFTER_AFTER_FRAMESET: {
appendCommentToDocument(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onDoctype(token) {
this.skipNextNewLine = false;
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
doctypeInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HEAD:
case InsertionMode.IN_HEAD:
case InsertionMode.IN_HEAD_NO_SCRIPT:
case InsertionMode.AFTER_HEAD: {
this._err(token, error_codes_js_1.ERR.misplacedDoctype);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
tokenInTableText(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onStartTag(token) {
this.skipNextNewLine = false;
this.currentToken = token;
this._processStartTag(token);
if (token.selfClosing && !token.ackSelfClosing) {
this._err(token, error_codes_js_1.ERR.nonVoidHtmlElementStartTagWithTrailingSolidus);
}
}
/**
* Processes a given start tag.
*
* `onStartTag` checks if a self-closing tag was recognized. When a token
* is moved inbetween multiple insertion modes, this check for self-closing
* could lead to false positives. To avoid this, `_processStartTag` is used
* for nested calls.
*
* @param token The token to process.
* @protected
*/
_processStartTag(token) {
if (this.shouldProcessStartTagTokenInForeignContent(token)) {
startTagInForeignContent(this, token);
}
else {
this._startTagOutsideForeignContent(token);
}
}
/** @protected */
_startTagOutsideForeignContent(token) {
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
tokenInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HTML: {
startTagBeforeHtml(this, token);
break;
}
case InsertionMode.BEFORE_HEAD: {
startTagBeforeHead(this, token);
break;
}
case InsertionMode.IN_HEAD: {
startTagInHead(this, token);
break;
}
case InsertionMode.IN_HEAD_NO_SCRIPT: {
startTagInHeadNoScript(this, token);
break;
}
case InsertionMode.AFTER_HEAD: {
startTagAfterHead(this, token);
break;
}
case InsertionMode.IN_BODY: {
startTagInBody(this, token);
break;
}
case InsertionMode.IN_TABLE: {
startTagInTable(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
tokenInTableText(this, token);
break;
}
case InsertionMode.IN_CAPTION: {
startTagInCaption(this, token);
break;
}
case InsertionMode.IN_COLUMN_GROUP: {
startTagInColumnGroup(this, token);
break;
}
case InsertionMode.IN_TABLE_BODY: {
startTagInTableBody(this, token);
break;
}
case InsertionMode.IN_ROW: {
startTagInRow(this, token);
break;
}
case InsertionMode.IN_CELL: {
startTagInCell(this, token);
break;
}
case InsertionMode.IN_SELECT: {
startTagInSelect(this, token);
break;
}
case InsertionMode.IN_SELECT_IN_TABLE: {
startTagInSelectInTable(this, token);
break;
}
case InsertionMode.IN_TEMPLATE: {
startTagInTemplate(this, token);
break;
}
case InsertionMode.AFTER_BODY: {
startTagAfterBody(this, token);
break;
}
case InsertionMode.IN_FRAMESET: {
startTagInFrameset(this, token);
break;
}
case InsertionMode.AFTER_FRAMESET: {
startTagAfterFrameset(this, token);
break;
}
case InsertionMode.AFTER_AFTER_BODY: {
startTagAfterAfterBody(this, token);
break;
}
case InsertionMode.AFTER_AFTER_FRAMESET: {
startTagAfterAfterFrameset(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onEndTag(token) {
this.skipNextNewLine = false;
this.currentToken = token;
if (this.currentNotInHTML) {
endTagInForeignContent(this, token);
}
else {
this._endTagOutsideForeignContent(token);
}
}
/** @protected */
_endTagOutsideForeignContent(token) {
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
tokenInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HTML: {
endTagBeforeHtml(this, token);
break;
}
case InsertionMode.BEFORE_HEAD: {
endTagBeforeHead(this, token);
break;
}
case InsertionMode.IN_HEAD: {
endTagInHead(this, token);
break;
}
case InsertionMode.IN_HEAD_NO_SCRIPT: {
endTagInHeadNoScript(this, token);
break;
}
case InsertionMode.AFTER_HEAD: {
endTagAfterHead(this, token);
break;
}
case InsertionMode.IN_BODY: {
endTagInBody(this, token);
break;
}
case InsertionMode.TEXT: {
endTagInText(this, token);
break;
}
case InsertionMode.IN_TABLE: {
endTagInTable(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
tokenInTableText(this, token);
break;
}
case InsertionMode.IN_CAPTION: {
endTagInCaption(this, token);
break;
}
case InsertionMode.IN_COLUMN_GROUP: {
endTagInColumnGroup(this, token);
break;
}
case InsertionMode.IN_TABLE_BODY: {
endTagInTableBody(this, token);
break;
}
case InsertionMode.IN_ROW: {
endTagInRow(this, token);
break;
}
case InsertionMode.IN_CELL: {
endTagInCell(this, token);
break;
}
case InsertionMode.IN_SELECT: {
endTagInSelect(this, token);
break;
}
case InsertionMode.IN_SELECT_IN_TABLE: {
endTagInSelectInTable(this, token);
break;
}
case InsertionMode.IN_TEMPLATE: {
endTagInTemplate(this, token);
break;
}
case InsertionMode.AFTER_BODY: {
endTagAfterBody(this, token);
break;
}
case InsertionMode.IN_FRAMESET: {
endTagInFrameset(this, token);
break;
}
case InsertionMode.AFTER_FRAMESET: {
endTagAfterFrameset(this, token);
break;
}
case InsertionMode.AFTER_AFTER_BODY: {
tokenAfterAfterBody(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onEof(token) {
switch (this.insertionMode) {
case InsertionMode.INITIAL: {
tokenInInitialMode(this, token);
break;
}
case InsertionMode.BEFORE_HTML: {
tokenBeforeHtml(this, token);
break;
}
case InsertionMode.BEFORE_HEAD: {
tokenBeforeHead(this, token);
break;
}
case InsertionMode.IN_HEAD: {
tokenInHead(this, token);
break;
}
case InsertionMode.IN_HEAD_NO_SCRIPT: {
tokenInHeadNoScript(this, token);
break;
}
case InsertionMode.AFTER_HEAD: {
tokenAfterHead(this, token);
break;
}
case InsertionMode.IN_BODY:
case InsertionMode.IN_TABLE:
case InsertionMode.IN_CAPTION:
case InsertionMode.IN_COLUMN_GROUP:
case InsertionMode.IN_TABLE_BODY:
case InsertionMode.IN_ROW:
case InsertionMode.IN_CELL:
case InsertionMode.IN_SELECT:
case InsertionMode.IN_SELECT_IN_TABLE: {
eofInBody(this, token);
break;
}
case InsertionMode.TEXT: {
eofInText(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
tokenInTableText(this, token);
break;
}
case InsertionMode.IN_TEMPLATE: {
eofInTemplate(this, token);
break;
}
case InsertionMode.AFTER_BODY:
case InsertionMode.IN_FRAMESET:
case InsertionMode.AFTER_FRAMESET:
case InsertionMode.AFTER_AFTER_BODY:
case InsertionMode.AFTER_AFTER_FRAMESET: {
stopParsing(this, token);
break;
}
default:
// Do nothing
}
}
/** @internal */
onWhitespaceCharacter(token) {
if (this.skipNextNewLine) {
this.skipNextNewLine = false;
if (token.chars.charCodeAt(0) === unicode.CODE_POINTS.LINE_FEED) {
if (token.chars.length === 1) {
return;
}
token.chars = token.chars.substr(1);
}
}
if (this.tokenizer.inForeignNode) {
this._insertCharacters(token);
return;
}
switch (this.insertionMode) {
case InsertionMode.IN_HEAD:
case InsertionMode.IN_HEAD_NO_SCRIPT:
case InsertionMode.AFTER_HEAD:
case InsertionMode.TEXT:
case InsertionMode.IN_COLUMN_GROUP:
case InsertionMode.IN_SELECT:
case InsertionMode.IN_SELECT_IN_TABLE:
case InsertionMode.IN_FRAMESET:
case InsertionMode.AFTER_FRAMESET: {
this._insertCharacters(token);
break;
}
case InsertionMode.IN_BODY:
case InsertionMode.IN_CAPTION:
case InsertionMode.IN_CELL:
case InsertionMode.IN_TEMPLATE:
case InsertionMode.AFTER_BODY:
case InsertionMode.AFTER_AFTER_BODY:
case InsertionMode.AFTER_AFTER_FRAMESET: {
whitespaceCharacterInBody(this, token);
break;
}
case InsertionMode.IN_TABLE:
case InsertionMode.IN_TABLE_BODY:
case InsertionMode.IN_ROW: {
characterInTable(this, token);
break;
}
case InsertionMode.IN_TABLE_TEXT: {
whitespaceCharacterInTableText(this, token);
break;
}
default:
// Do nothing
}
}
}
exports.Parser = Parser;
//Adoption agency algorithm
//(see: http://www.whatwg.org/specs/web-apps/current-work/multipage/tree-construction.html#adoptionAgency)
//------------------------------------------------------------------
//Steps 5-8 of the algorithm
function aaObtainFormattingElementEntry(p, token) {
let formattingElementEntry = p.activeFormattingElements.getElementEntryInScopeWithTagName(token.tagName);
if (formattingElementEntry) {
if (!p.openElements.contains(formattingElementEntry.element)) {
p.activeFormattingElements.removeEntry(formattingElementEntry);
formattingElementEntry = null;
}
else if (!p.openElements.hasInScope(token.tagID)) {
formattingElementEntry = null;
}
}
else {
genericEndTagInBody(p, token);
}
return formattingElementEntry;
}
//Steps 9 and 10 of the algorithm
function aaObtainFurthestBlock(p, formattingElementEntry) {
let furthestBlock = null;
let idx = p.openElements.stackTop;
for (; idx >= 0; idx--) {
const element = p.openElements.items[idx];
if (element === formattingElementEntry.element) {
break;
}
if (p._isSpecialElement(element, p.openElements.tagIDs[idx])) {
furthestBlock = element;
}
}
if (!furthestBlock) {
p.openElements.shortenToLength(Math.max(idx, 0));
p.activeFormattingElements.removeEntry(formattingElementEntry);
}
return furthestBlock;
}
//Step 13 of the algorithm
function aaInnerLoop(p, furthestBlock, formattingElement) {
let lastElement = furthestBlock;
let nextElement = p.openElements.getCommonAncestor(furthestBlock);
for (let i = 0, element = nextElement; element !== formattingElement; i++, element = nextElement) {
//NOTE: store the next element for the next loop iteration (it may be deleted from the stack by step 9.5)
nextElement = p.openElements.getCommonAncestor(element);
const elementEntry = p.activeFormattingElements.getElementEntry(element);
const counterOverflow = elementEntry && i >= AA_INNER_LOOP_ITER;
const shouldRemoveFromOpenElements = !elementEntry || counterOverflow;
if (shouldRemoveFromOpenElements) {
if (counterOverflow) {
p.activeFormattingElements.removeEntry(elementEntry);
}
p.openElements.remove(element);
}
else {
element = aaRecreateElementFromEntry(p, elementEntry);
if (lastElement === furthestBlock) {
p.activeFormattingElements.bookmark = elementEntry;
}
p.treeAdapter.detachNode(lastElement);
p.treeAdapter.appendChild(element, lastElement);
lastElement = element;
}
}
return lastElement;
}
//Step 13.7 of the algorithm
function aaRecreateElementFromEntry(p, elementEntry) {
const ns = p.treeAdapter.getNamespaceURI(elementEntry.element);
const newElement = p.treeAdapter.createElement(elementEntry.token.tagName, ns, elementEntry.token.attrs);
p.openElements.replace(elementEntry.element, newElement);
elementEntry.element = newElement;
return newElement;
}
//Step 14 of the algorithm
function aaInsertLastNodeInCommonAncestor(p, commonAncestor, lastElement) {
const tn = p.treeAdapter.getTagName(commonAncestor);
const tid = (0, html_js_1.getTagID)(tn);
if (p._isElementCausesFosterParenting(tid)) {
p._fosterParentElement(lastElement);
}
else {
const ns = p.treeAdapter.getNamespaceURI(commonAncestor);
if (tid === html_js_1.TAG_ID.TEMPLATE && ns === html_js_1.NS.HTML) {
commonAncestor = p.treeAdapter.getTemplateContent(commonAncestor);
}
p.treeAdapter.appendChild(commonAncestor, lastElement);
}
}
//Steps 15-19 of the algorithm
function aaReplaceFormattingElement(p, furthestBlock, formattingElementEntry) {
const ns = p.treeAdapter.getNamespaceURI(formattingElementEntry.element);
const { token } = formattingElementEntry;
const newElement = p.treeAdapter.createElement(token.tagName, ns, token.attrs);
p._adoptNodes(furthestBlock, newElement);
p.treeAdapter.appendChild(furthestBlock, newElement);
p.activeFormattingElements.insertElementAfterBookmark(newElement, token);
p.activeFormattingElements.removeEntry(formattingElementEntry);
p.openElements.remove(formattingElementEntry.element);
p.openElements.insertAfter(furthestBlock, newElement, token.tagID);
}
//Algorithm entry point
function callAdoptionAgency(p, token) {
for (let i = 0; i < AA_OUTER_LOOP_ITER; i++) {
const formattingElementEntry = aaObtainFormattingElementEntry(p, token);
if (!formattingElementEntry) {
break;
}
const furthestBlock = aaObtainFurthestBlock(p, formattingElementEntry);
if (!furthestBlock) {
break;
}
p.activeFormattingElements.bookmark = formattingElementEntry;
const lastElement = aaInnerLoop(p, furthestBlock, formattingElementEntry.element);
const commonAncestor = p.openElements.getCommonAncestor(formattingElementEntry.element);
p.treeAdapter.detachNode(lastElement);
if (commonAncestor)
aaInsertLastNodeInCommonAncestor(p, commonAncestor, lastElement);
aaReplaceFormattingElement(p, furthestBlock, formattingElementEntry);
}
}
//Generic token handlers
//------------------------------------------------------------------
function appendComment(p, token) {
p._appendCommentNode(token, p.openElements.currentTmplContentOrNode);
}
function appendCommentToRootHtmlElement(p, token) {
p._appendCommentNode(token, p.openElements.items[0]);
}
function appendCommentToDocument(p, token) {
p._appendCommentNode(token, p.document);
}
function stopParsing(p, token) {
p.stopped = true;
// NOTE: Set end locations for elements that remain on the open element stack.
if (token.location) {
// NOTE: If we are not in a fragment, `html` and `body` will stay on the stack.
// This is a problem, as we might overwrite their end position here.
const target = p.fragmentContext ? 0 : 2;
for (let i = p.openElements.stackTop; i >= target; i--) {
p._setEndLocation(p.openElements.items[i], token);
}
// Handle `html` and `body`
if (!p.fragmentContext && p.openElements.stackTop >= 0) {
const htmlElement = p.openElements.items[0];
const htmlLocation = p.treeAdapter.getNodeSourceCodeLocation(htmlElement);
if (htmlLocation && !htmlLocation.endTag) {
p._setEndLocation(htmlElement, token);
if (p.openElements.stackTop >= 1) {
const bodyElement = p.openElements.items[1];
const bodyLocation = p.treeAdapter.getNodeSourceCodeLocation(bodyElement);
if (bodyLocation && !bodyLocation.endTag) {
p._setEndLocation(bodyElement, token);
}
}
}
}
}
}
// The "initial" insertion mode
//------------------------------------------------------------------
function doctypeInInitialMode(p, token) {
p._setDocumentType(token);
const mode = token.forceQuirks ? html_js_1.DOCUMENT_MODE.QUIRKS : doctype.getDocumentMode(token);
if (!doctype.isConforming(token)) {
p._err(token, error_codes_js_1.ERR.nonConformingDoctype);
}
p.treeAdapter.setDocumentMode(p.document, mode);
p.insertionMode = InsertionMode.BEFORE_HTML;
}
function tokenInInitialMode(p, token) {
p._err(token, error_codes_js_1.ERR.missingDoctype, true);
p.treeAdapter.setDocumentMode(p.document, html_js_1.DOCUMENT_MODE.QUIRKS);
p.insertionMode = InsertionMode.BEFORE_HTML;
p._processToken(token);
}
// The "before html" insertion mode
//------------------------------------------------------------------
function startTagBeforeHtml(p, token) {
if (token.tagID === html_js_1.TAG_ID.HTML) {
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.BEFORE_HEAD;
}
else {
tokenBeforeHtml(p, token);
}
}
function endTagBeforeHtml(p, token) {
const tn = token.tagID;
if (tn === html_js_1.TAG_ID.HTML || tn === html_js_1.TAG_ID.HEAD || tn === html_js_1.TAG_ID.BODY || tn === html_js_1.TAG_ID.BR) {
tokenBeforeHtml(p, token);
}
}
function tokenBeforeHtml(p, token) {
p._insertFakeRootElement();
p.insertionMode = InsertionMode.BEFORE_HEAD;
p._processToken(token);
}
// The "before head" insertion mode
//------------------------------------------------------------------
function startTagBeforeHead(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.HEAD: {
p._insertElement(token, html_js_1.NS.HTML);
p.headElement = p.openElements.current;
p.insertionMode = InsertionMode.IN_HEAD;
break;
}
default: {
tokenBeforeHead(p, token);
}
}
}
function endTagBeforeHead(p, token) {
const tn = token.tagID;
if (tn === html_js_1.TAG_ID.HEAD || tn === html_js_1.TAG_ID.BODY || tn === html_js_1.TAG_ID.HTML || tn === html_js_1.TAG_ID.BR) {
tokenBeforeHead(p, token);
}
else {
p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
}
}
function tokenBeforeHead(p, token) {
p._insertFakeElement(html_js_1.TAG_NAMES.HEAD, html_js_1.TAG_ID.HEAD);
p.headElement = p.openElements.current;
p.insertionMode = InsertionMode.IN_HEAD;
p._processToken(token);
}
// The "in head" insertion mode
//------------------------------------------------------------------
function startTagInHead(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BASE:
case html_js_1.TAG_ID.BASEFONT:
case html_js_1.TAG_ID.BGSOUND:
case html_js_1.TAG_ID.LINK:
case html_js_1.TAG_ID.META: {
p._appendElement(token, html_js_1.NS.HTML);
token.ackSelfClosing = true;
break;
}
case html_js_1.TAG_ID.TITLE: {
p._switchToTextParsing(token, index_js_1.TokenizerMode.RCDATA);
break;
}
case html_js_1.TAG_ID.NOSCRIPT: {
if (p.options.scriptingEnabled) {
p._switchToTextParsing(token, index_js_1.TokenizerMode.RAWTEXT);
}
else {
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_HEAD_NO_SCRIPT;
}
break;
}
case html_js_1.TAG_ID.NOFRAMES:
case html_js_1.TAG_ID.STYLE: {
p._switchToTextParsing(token, index_js_1.TokenizerMode.RAWTEXT);
break;
}
case html_js_1.TAG_ID.SCRIPT: {
p._switchToTextParsing(token, index_js_1.TokenizerMode.SCRIPT_DATA);
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
p._insertTemplate(token);
p.activeFormattingElements.insertMarker();
p.framesetOk = false;
p.insertionMode = InsertionMode.IN_TEMPLATE;
p.tmplInsertionModeStack.unshift(InsertionMode.IN_TEMPLATE);
break;
}
case html_js_1.TAG_ID.HEAD: {
p._err(token, error_codes_js_1.ERR.misplacedStartTagForHeadElement);
break;
}
default: {
tokenInHead(p, token);
}
}
}
function endTagInHead(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HEAD: {
p.openElements.pop();
p.insertionMode = InsertionMode.AFTER_HEAD;
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.BR:
case html_js_1.TAG_ID.HTML: {
tokenInHead(p, token);
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
default: {
p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
}
}
}
function templateEndTagInHead(p, token) {
if (p.openElements.tmplCount > 0) {
p.openElements.generateImpliedEndTagsThoroughly();
if (p.openElements.currentTagId !== html_js_1.TAG_ID.TEMPLATE) {
p._err(token, error_codes_js_1.ERR.closingOfElementWithOpenChildElements);
}
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TEMPLATE);
p.activeFormattingElements.clearToLastMarker();
p.tmplInsertionModeStack.shift();
p._resetInsertionMode();
}
else {
p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
}
}
function tokenInHead(p, token) {
p.openElements.pop();
p.insertionMode = InsertionMode.AFTER_HEAD;
p._processToken(token);
}
// The "in head no script" insertion mode
//------------------------------------------------------------------
function startTagInHeadNoScript(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BASEFONT:
case html_js_1.TAG_ID.BGSOUND:
case html_js_1.TAG_ID.HEAD:
case html_js_1.TAG_ID.LINK:
case html_js_1.TAG_ID.META:
case html_js_1.TAG_ID.NOFRAMES:
case html_js_1.TAG_ID.STYLE: {
startTagInHead(p, token);
break;
}
case html_js_1.TAG_ID.NOSCRIPT: {
p._err(token, error_codes_js_1.ERR.nestedNoscriptInHead);
break;
}
default: {
tokenInHeadNoScript(p, token);
}
}
}
function endTagInHeadNoScript(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.NOSCRIPT: {
p.openElements.pop();
p.insertionMode = InsertionMode.IN_HEAD;
break;
}
case html_js_1.TAG_ID.BR: {
tokenInHeadNoScript(p, token);
break;
}
default: {
p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
}
}
}
function tokenInHeadNoScript(p, token) {
const errCode = token.type === token_js_1.TokenType.EOF ? error_codes_js_1.ERR.openElementsLeftAfterEof : error_codes_js_1.ERR.disallowedContentInNoscriptInHead;
p._err(token, errCode);
p.openElements.pop();
p.insertionMode = InsertionMode.IN_HEAD;
p._processToken(token);
}
// The "after head" insertion mode
//------------------------------------------------------------------
function startTagAfterHead(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BODY: {
p._insertElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
p.insertionMode = InsertionMode.IN_BODY;
break;
}
case html_js_1.TAG_ID.FRAMESET: {
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_FRAMESET;
break;
}
case html_js_1.TAG_ID.BASE:
case html_js_1.TAG_ID.BASEFONT:
case html_js_1.TAG_ID.BGSOUND:
case html_js_1.TAG_ID.LINK:
case html_js_1.TAG_ID.META:
case html_js_1.TAG_ID.NOFRAMES:
case html_js_1.TAG_ID.SCRIPT:
case html_js_1.TAG_ID.STYLE:
case html_js_1.TAG_ID.TEMPLATE:
case html_js_1.TAG_ID.TITLE: {
p._err(token, error_codes_js_1.ERR.abandonedHeadElementChild);
p.openElements.push(p.headElement, html_js_1.TAG_ID.HEAD);
startTagInHead(p, token);
p.openElements.remove(p.headElement);
break;
}
case html_js_1.TAG_ID.HEAD: {
p._err(token, error_codes_js_1.ERR.misplacedStartTagForHeadElement);
break;
}
default: {
tokenAfterHead(p, token);
}
}
}
function endTagAfterHead(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.HTML:
case html_js_1.TAG_ID.BR: {
tokenAfterHead(p, token);
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
default: {
p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
}
}
}
function tokenAfterHead(p, token) {
p._insertFakeElement(html_js_1.TAG_NAMES.BODY, html_js_1.TAG_ID.BODY);
p.insertionMode = InsertionMode.IN_BODY;
modeInBody(p, token);
}
// The "in body" insertion mode
//------------------------------------------------------------------
function modeInBody(p, token) {
switch (token.type) {
case token_js_1.TokenType.CHARACTER: {
characterInBody(p, token);
break;
}
case token_js_1.TokenType.WHITESPACE_CHARACTER: {
whitespaceCharacterInBody(p, token);
break;
}
case token_js_1.TokenType.COMMENT: {
appendComment(p, token);
break;
}
case token_js_1.TokenType.START_TAG: {
startTagInBody(p, token);
break;
}
case token_js_1.TokenType.END_TAG: {
endTagInBody(p, token);
break;
}
case token_js_1.TokenType.EOF: {
eofInBody(p, token);
break;
}
default:
// Do nothing
}
}
function whitespaceCharacterInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertCharacters(token);
}
function characterInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertCharacters(token);
p.framesetOk = false;
}
function htmlStartTagInBody(p, token) {
if (p.openElements.tmplCount === 0) {
p.treeAdapter.adoptAttributes(p.openElements.items[0], token.attrs);
}
}
function bodyStartTagInBody(p, token) {
const bodyElement = p.openElements.tryPeekProperlyNestedBodyElement();
if (bodyElement && p.openElements.tmplCount === 0) {
p.framesetOk = false;
p.treeAdapter.adoptAttributes(bodyElement, token.attrs);
}
}
function framesetStartTagInBody(p, token) {
const bodyElement = p.openElements.tryPeekProperlyNestedBodyElement();
if (p.framesetOk && bodyElement) {
p.treeAdapter.detachNode(bodyElement);
p.openElements.popAllUpToHtmlElement();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_FRAMESET;
}
}
function addressStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
}
function numberedHeaderStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
if (p.openElements.currentTagId !== undefined && html_js_1.NUMBERED_HEADERS.has(p.openElements.currentTagId)) {
p.openElements.pop();
}
p._insertElement(token, html_js_1.NS.HTML);
}
function preStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
//NOTE: If the next token is a U+000A LINE FEED (LF) character token, then ignore that token and move
//on to the next one. (Newlines at the start of pre blocks are ignored as an authoring convenience.)
p.skipNextNewLine = true;
p.framesetOk = false;
}
function formStartTagInBody(p, token) {
const inTemplate = p.openElements.tmplCount > 0;
if (!p.formElement || inTemplate) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
if (!inTemplate) {
p.formElement = p.openElements.current;
}
}
}
function listItemStartTagInBody(p, token) {
p.framesetOk = false;
const tn = token.tagID;
for (let i = p.openElements.stackTop; i >= 0; i--) {
const elementId = p.openElements.tagIDs[i];
if ((tn === html_js_1.TAG_ID.LI && elementId === html_js_1.TAG_ID.LI) ||
((tn === html_js_1.TAG_ID.DD || tn === html_js_1.TAG_ID.DT) && (elementId === html_js_1.TAG_ID.DD || elementId === html_js_1.TAG_ID.DT))) {
p.openElements.generateImpliedEndTagsWithExclusion(elementId);
p.openElements.popUntilTagNamePopped(elementId);
break;
}
if (elementId !== html_js_1.TAG_ID.ADDRESS &&
elementId !== html_js_1.TAG_ID.DIV &&
elementId !== html_js_1.TAG_ID.P &&
p._isSpecialElement(p.openElements.items[i], elementId)) {
break;
}
}
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
}
function plaintextStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
p.tokenizer.state = index_js_1.TokenizerMode.PLAINTEXT;
}
function buttonStartTagInBody(p, token) {
if (p.openElements.hasInScope(html_js_1.TAG_ID.BUTTON)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.BUTTON);
}
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
}
function aStartTagInBody(p, token) {
const activeElementEntry = p.activeFormattingElements.getElementEntryInScopeWithTagName(html_js_1.TAG_NAMES.A);
if (activeElementEntry) {
callAdoptionAgency(p, token);
p.openElements.remove(activeElementEntry.element);
p.activeFormattingElements.removeEntry(activeElementEntry);
}
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
p.activeFormattingElements.pushElement(p.openElements.current, token);
}
function bStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
p.activeFormattingElements.pushElement(p.openElements.current, token);
}
function nobrStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
if (p.openElements.hasInScope(html_js_1.TAG_ID.NOBR)) {
callAdoptionAgency(p, token);
p._reconstructActiveFormattingElements();
}
p._insertElement(token, html_js_1.NS.HTML);
p.activeFormattingElements.pushElement(p.openElements.current, token);
}
function appletStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
p.activeFormattingElements.insertMarker();
p.framesetOk = false;
}
function tableStartTagInBody(p, token) {
if (p.treeAdapter.getDocumentMode(p.document) !== html_js_1.DOCUMENT_MODE.QUIRKS && p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._insertElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
p.insertionMode = InsertionMode.IN_TABLE;
}
function areaStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._appendElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
token.ackSelfClosing = true;
}
function isHiddenInput(token) {
const inputType = (0, token_js_1.getTokenAttr)(token, html_js_1.ATTRS.TYPE);
return inputType != null && inputType.toLowerCase() === HIDDEN_INPUT_TYPE;
}
function inputStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._appendElement(token, html_js_1.NS.HTML);
if (!isHiddenInput(token)) {
p.framesetOk = false;
}
token.ackSelfClosing = true;
}
function paramStartTagInBody(p, token) {
p._appendElement(token, html_js_1.NS.HTML);
token.ackSelfClosing = true;
}
function hrStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._appendElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
token.ackSelfClosing = true;
}
function imageStartTagInBody(p, token) {
token.tagName = html_js_1.TAG_NAMES.IMG;
token.tagID = html_js_1.TAG_ID.IMG;
areaStartTagInBody(p, token);
}
function textareaStartTagInBody(p, token) {
p._insertElement(token, html_js_1.NS.HTML);
//NOTE: If the next token is a U+000A LINE FEED (LF) character token, then ignore that token and move
//on to the next one. (Newlines at the start of textarea elements are ignored as an authoring convenience.)
p.skipNextNewLine = true;
p.tokenizer.state = index_js_1.TokenizerMode.RCDATA;
p.originalInsertionMode = p.insertionMode;
p.framesetOk = false;
p.insertionMode = InsertionMode.TEXT;
}
function xmpStartTagInBody(p, token) {
if (p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._closePElement();
}
p._reconstructActiveFormattingElements();
p.framesetOk = false;
p._switchToTextParsing(token, index_js_1.TokenizerMode.RAWTEXT);
}
function iframeStartTagInBody(p, token) {
p.framesetOk = false;
p._switchToTextParsing(token, index_js_1.TokenizerMode.RAWTEXT);
}
//NOTE: here we assume that we always act as a user agent with enabled plugins/frames, so we parse
//<noembed>/<noframes> as rawtext.
function rawTextStartTagInBody(p, token) {
p._switchToTextParsing(token, index_js_1.TokenizerMode.RAWTEXT);
}
function selectStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
p.framesetOk = false;
p.insertionMode =
p.insertionMode === InsertionMode.IN_TABLE ||
p.insertionMode === InsertionMode.IN_CAPTION ||
p.insertionMode === InsertionMode.IN_TABLE_BODY ||
p.insertionMode === InsertionMode.IN_ROW ||
p.insertionMode === InsertionMode.IN_CELL
? InsertionMode.IN_SELECT_IN_TABLE
: InsertionMode.IN_SELECT;
}
function optgroupStartTagInBody(p, token) {
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTION) {
p.openElements.pop();
}
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
}
function rbStartTagInBody(p, token) {
if (p.openElements.hasInScope(html_js_1.TAG_ID.RUBY)) {
p.openElements.generateImpliedEndTags();
}
p._insertElement(token, html_js_1.NS.HTML);
}
function rtStartTagInBody(p, token) {
if (p.openElements.hasInScope(html_js_1.TAG_ID.RUBY)) {
p.openElements.generateImpliedEndTagsWithExclusion(html_js_1.TAG_ID.RTC);
}
p._insertElement(token, html_js_1.NS.HTML);
}
function mathStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
foreignContent.adjustTokenMathMLAttrs(token);
foreignContent.adjustTokenXMLAttrs(token);
if (token.selfClosing) {
p._appendElement(token, html_js_1.NS.MATHML);
}
else {
p._insertElement(token, html_js_1.NS.MATHML);
}
token.ackSelfClosing = true;
}
function svgStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
foreignContent.adjustTokenSVGAttrs(token);
foreignContent.adjustTokenXMLAttrs(token);
if (token.selfClosing) {
p._appendElement(token, html_js_1.NS.SVG);
}
else {
p._insertElement(token, html_js_1.NS.SVG);
}
token.ackSelfClosing = true;
}
function genericStartTagInBody(p, token) {
p._reconstructActiveFormattingElements();
p._insertElement(token, html_js_1.NS.HTML);
}
function startTagInBody(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.I:
case html_js_1.TAG_ID.S:
case html_js_1.TAG_ID.B:
case html_js_1.TAG_ID.U:
case html_js_1.TAG_ID.EM:
case html_js_1.TAG_ID.TT:
case html_js_1.TAG_ID.BIG:
case html_js_1.TAG_ID.CODE:
case html_js_1.TAG_ID.FONT:
case html_js_1.TAG_ID.SMALL:
case html_js_1.TAG_ID.STRIKE:
case html_js_1.TAG_ID.STRONG: {
bStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.A: {
aStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.H1:
case html_js_1.TAG_ID.H2:
case html_js_1.TAG_ID.H3:
case html_js_1.TAG_ID.H4:
case html_js_1.TAG_ID.H5:
case html_js_1.TAG_ID.H6: {
numberedHeaderStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.P:
case html_js_1.TAG_ID.DL:
case html_js_1.TAG_ID.OL:
case html_js_1.TAG_ID.UL:
case html_js_1.TAG_ID.DIV:
case html_js_1.TAG_ID.DIR:
case html_js_1.TAG_ID.NAV:
case html_js_1.TAG_ID.MAIN:
case html_js_1.TAG_ID.MENU:
case html_js_1.TAG_ID.ASIDE:
case html_js_1.TAG_ID.CENTER:
case html_js_1.TAG_ID.FIGURE:
case html_js_1.TAG_ID.FOOTER:
case html_js_1.TAG_ID.HEADER:
case html_js_1.TAG_ID.HGROUP:
case html_js_1.TAG_ID.DIALOG:
case html_js_1.TAG_ID.DETAILS:
case html_js_1.TAG_ID.ADDRESS:
case html_js_1.TAG_ID.ARTICLE:
case html_js_1.TAG_ID.SEARCH:
case html_js_1.TAG_ID.SECTION:
case html_js_1.TAG_ID.SUMMARY:
case html_js_1.TAG_ID.FIELDSET:
case html_js_1.TAG_ID.BLOCKQUOTE:
case html_js_1.TAG_ID.FIGCAPTION: {
addressStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.LI:
case html_js_1.TAG_ID.DD:
case html_js_1.TAG_ID.DT: {
listItemStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BR:
case html_js_1.TAG_ID.IMG:
case html_js_1.TAG_ID.WBR:
case html_js_1.TAG_ID.AREA:
case html_js_1.TAG_ID.EMBED:
case html_js_1.TAG_ID.KEYGEN: {
areaStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.HR: {
hrStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.RB:
case html_js_1.TAG_ID.RTC: {
rbStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.RT:
case html_js_1.TAG_ID.RP: {
rtStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.PRE:
case html_js_1.TAG_ID.LISTING: {
preStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.XMP: {
xmpStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.SVG: {
svgStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.HTML: {
htmlStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BASE:
case html_js_1.TAG_ID.LINK:
case html_js_1.TAG_ID.META:
case html_js_1.TAG_ID.STYLE:
case html_js_1.TAG_ID.TITLE:
case html_js_1.TAG_ID.SCRIPT:
case html_js_1.TAG_ID.BGSOUND:
case html_js_1.TAG_ID.BASEFONT:
case html_js_1.TAG_ID.TEMPLATE: {
startTagInHead(p, token);
break;
}
case html_js_1.TAG_ID.BODY: {
bodyStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.FORM: {
formStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.NOBR: {
nobrStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.MATH: {
mathStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.TABLE: {
tableStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.INPUT: {
inputStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.PARAM:
case html_js_1.TAG_ID.TRACK:
case html_js_1.TAG_ID.SOURCE: {
paramStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.IMAGE: {
imageStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.BUTTON: {
buttonStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.APPLET:
case html_js_1.TAG_ID.OBJECT:
case html_js_1.TAG_ID.MARQUEE: {
appletStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.IFRAME: {
iframeStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.SELECT: {
selectStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.OPTION:
case html_js_1.TAG_ID.OPTGROUP: {
optgroupStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.NOEMBED:
case html_js_1.TAG_ID.NOFRAMES: {
rawTextStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.FRAMESET: {
framesetStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.TEXTAREA: {
textareaStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.NOSCRIPT: {
if (p.options.scriptingEnabled) {
rawTextStartTagInBody(p, token);
}
else {
genericStartTagInBody(p, token);
}
break;
}
case html_js_1.TAG_ID.PLAINTEXT: {
plaintextStartTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TR:
case html_js_1.TAG_ID.HEAD:
case html_js_1.TAG_ID.FRAME:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COLGROUP: {
// Ignore token
break;
}
default: {
genericStartTagInBody(p, token);
}
}
}
function bodyEndTagInBody(p, token) {
if (p.openElements.hasInScope(html_js_1.TAG_ID.BODY)) {
p.insertionMode = InsertionMode.AFTER_BODY;
//NOTE: <body> is never popped from the stack, so we need to updated
//the end location explicitly.
if (p.options.sourceCodeLocationInfo) {
const bodyElement = p.openElements.tryPeekProperlyNestedBodyElement();
if (bodyElement) {
p._setEndLocation(bodyElement, token);
}
}
}
}
function htmlEndTagInBody(p, token) {
if (p.openElements.hasInScope(html_js_1.TAG_ID.BODY)) {
p.insertionMode = InsertionMode.AFTER_BODY;
endTagAfterBody(p, token);
}
}
function addressEndTagInBody(p, token) {
const tn = token.tagID;
if (p.openElements.hasInScope(tn)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(tn);
}
}
function formEndTagInBody(p) {
const inTemplate = p.openElements.tmplCount > 0;
const { formElement } = p;
if (!inTemplate) {
p.formElement = null;
}
if ((formElement || inTemplate) && p.openElements.hasInScope(html_js_1.TAG_ID.FORM)) {
p.openElements.generateImpliedEndTags();
if (inTemplate) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.FORM);
}
else if (formElement) {
p.openElements.remove(formElement);
}
}
}
function pEndTagInBody(p) {
if (!p.openElements.hasInButtonScope(html_js_1.TAG_ID.P)) {
p._insertFakeElement(html_js_1.TAG_NAMES.P, html_js_1.TAG_ID.P);
}
p._closePElement();
}
function liEndTagInBody(p) {
if (p.openElements.hasInListItemScope(html_js_1.TAG_ID.LI)) {
p.openElements.generateImpliedEndTagsWithExclusion(html_js_1.TAG_ID.LI);
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.LI);
}
}
function ddEndTagInBody(p, token) {
const tn = token.tagID;
if (p.openElements.hasInScope(tn)) {
p.openElements.generateImpliedEndTagsWithExclusion(tn);
p.openElements.popUntilTagNamePopped(tn);
}
}
function numberedHeaderEndTagInBody(p) {
if (p.openElements.hasNumberedHeaderInScope()) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilNumberedHeaderPopped();
}
}
function appletEndTagInBody(p, token) {
const tn = token.tagID;
if (p.openElements.hasInScope(tn)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(tn);
p.activeFormattingElements.clearToLastMarker();
}
}
function brEndTagInBody(p) {
p._reconstructActiveFormattingElements();
p._insertFakeElement(html_js_1.TAG_NAMES.BR, html_js_1.TAG_ID.BR);
p.openElements.pop();
p.framesetOk = false;
}
function genericEndTagInBody(p, token) {
const tn = token.tagName;
const tid = token.tagID;
for (let i = p.openElements.stackTop; i > 0; i--) {
const element = p.openElements.items[i];
const elementId = p.openElements.tagIDs[i];
// Compare the tag name here, as the tag might not be a known tag with an ID.
if (tid === elementId && (tid !== html_js_1.TAG_ID.UNKNOWN || p.treeAdapter.getTagName(element) === tn)) {
p.openElements.generateImpliedEndTagsWithExclusion(tid);
if (p.openElements.stackTop >= i)
p.openElements.shortenToLength(i);
break;
}
if (p._isSpecialElement(element, elementId)) {
break;
}
}
}
function endTagInBody(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.A:
case html_js_1.TAG_ID.B:
case html_js_1.TAG_ID.I:
case html_js_1.TAG_ID.S:
case html_js_1.TAG_ID.U:
case html_js_1.TAG_ID.EM:
case html_js_1.TAG_ID.TT:
case html_js_1.TAG_ID.BIG:
case html_js_1.TAG_ID.CODE:
case html_js_1.TAG_ID.FONT:
case html_js_1.TAG_ID.NOBR:
case html_js_1.TAG_ID.SMALL:
case html_js_1.TAG_ID.STRIKE:
case html_js_1.TAG_ID.STRONG: {
callAdoptionAgency(p, token);
break;
}
case html_js_1.TAG_ID.P: {
pEndTagInBody(p);
break;
}
case html_js_1.TAG_ID.DL:
case html_js_1.TAG_ID.UL:
case html_js_1.TAG_ID.OL:
case html_js_1.TAG_ID.DIR:
case html_js_1.TAG_ID.DIV:
case html_js_1.TAG_ID.NAV:
case html_js_1.TAG_ID.PRE:
case html_js_1.TAG_ID.MAIN:
case html_js_1.TAG_ID.MENU:
case html_js_1.TAG_ID.ASIDE:
case html_js_1.TAG_ID.BUTTON:
case html_js_1.TAG_ID.CENTER:
case html_js_1.TAG_ID.FIGURE:
case html_js_1.TAG_ID.FOOTER:
case html_js_1.TAG_ID.HEADER:
case html_js_1.TAG_ID.HGROUP:
case html_js_1.TAG_ID.DIALOG:
case html_js_1.TAG_ID.ADDRESS:
case html_js_1.TAG_ID.ARTICLE:
case html_js_1.TAG_ID.DETAILS:
case html_js_1.TAG_ID.SEARCH:
case html_js_1.TAG_ID.SECTION:
case html_js_1.TAG_ID.SUMMARY:
case html_js_1.TAG_ID.LISTING:
case html_js_1.TAG_ID.FIELDSET:
case html_js_1.TAG_ID.BLOCKQUOTE:
case html_js_1.TAG_ID.FIGCAPTION: {
addressEndTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.LI: {
liEndTagInBody(p);
break;
}
case html_js_1.TAG_ID.DD:
case html_js_1.TAG_ID.DT: {
ddEndTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.H1:
case html_js_1.TAG_ID.H2:
case html_js_1.TAG_ID.H3:
case html_js_1.TAG_ID.H4:
case html_js_1.TAG_ID.H5:
case html_js_1.TAG_ID.H6: {
numberedHeaderEndTagInBody(p);
break;
}
case html_js_1.TAG_ID.BR: {
brEndTagInBody(p);
break;
}
case html_js_1.TAG_ID.BODY: {
bodyEndTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.HTML: {
htmlEndTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.FORM: {
formEndTagInBody(p);
break;
}
case html_js_1.TAG_ID.APPLET:
case html_js_1.TAG_ID.OBJECT:
case html_js_1.TAG_ID.MARQUEE: {
appletEndTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
default: {
genericEndTagInBody(p, token);
}
}
}
function eofInBody(p, token) {
if (p.tmplInsertionModeStack.length > 0) {
eofInTemplate(p, token);
}
else {
stopParsing(p, token);
}
}
// The "text" insertion mode
//------------------------------------------------------------------
function endTagInText(p, token) {
var _a;
if (token.tagID === html_js_1.TAG_ID.SCRIPT) {
(_a = p.scriptHandler) === null || _a === void 0 ? void 0 : _a.call(p, p.openElements.current);
}
p.openElements.pop();
p.insertionMode = p.originalInsertionMode;
}
function eofInText(p, token) {
p._err(token, error_codes_js_1.ERR.eofInElementThatCanContainOnlyText);
p.openElements.pop();
p.insertionMode = p.originalInsertionMode;
p.onEof(token);
}
// The "in table" insertion mode
//------------------------------------------------------------------
function characterInTable(p, token) {
if (p.openElements.currentTagId !== undefined && TABLE_STRUCTURE_TAGS.has(p.openElements.currentTagId)) {
p.pendingCharacterTokens.length = 0;
p.hasNonWhitespacePendingCharacterToken = false;
p.originalInsertionMode = p.insertionMode;
p.insertionMode = InsertionMode.IN_TABLE_TEXT;
switch (token.type) {
case token_js_1.TokenType.CHARACTER: {
characterInTableText(p, token);
break;
}
case token_js_1.TokenType.WHITESPACE_CHARACTER: {
whitespaceCharacterInTableText(p, token);
break;
}
// Ignore null
}
}
else {
tokenInTable(p, token);
}
}
function captionStartTagInTable(p, token) {
p.openElements.clearBackToTableContext();
p.activeFormattingElements.insertMarker();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_CAPTION;
}
function colgroupStartTagInTable(p, token) {
p.openElements.clearBackToTableContext();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_COLUMN_GROUP;
}
function colStartTagInTable(p, token) {
p.openElements.clearBackToTableContext();
p._insertFakeElement(html_js_1.TAG_NAMES.COLGROUP, html_js_1.TAG_ID.COLGROUP);
p.insertionMode = InsertionMode.IN_COLUMN_GROUP;
startTagInColumnGroup(p, token);
}
function tbodyStartTagInTable(p, token) {
p.openElements.clearBackToTableContext();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_TABLE_BODY;
}
function tdStartTagInTable(p, token) {
p.openElements.clearBackToTableContext();
p._insertFakeElement(html_js_1.TAG_NAMES.TBODY, html_js_1.TAG_ID.TBODY);
p.insertionMode = InsertionMode.IN_TABLE_BODY;
startTagInTableBody(p, token);
}
function tableStartTagInTable(p, token) {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TABLE)) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TABLE);
p._resetInsertionMode();
p._processStartTag(token);
}
}
function inputStartTagInTable(p, token) {
if (isHiddenInput(token)) {
p._appendElement(token, html_js_1.NS.HTML);
}
else {
tokenInTable(p, token);
}
token.ackSelfClosing = true;
}
function formStartTagInTable(p, token) {
if (!p.formElement && p.openElements.tmplCount === 0) {
p._insertElement(token, html_js_1.NS.HTML);
p.formElement = p.openElements.current;
p.openElements.pop();
}
}
function startTagInTable(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.TR: {
tdStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.STYLE:
case html_js_1.TAG_ID.SCRIPT:
case html_js_1.TAG_ID.TEMPLATE: {
startTagInHead(p, token);
break;
}
case html_js_1.TAG_ID.COL: {
colStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.FORM: {
formStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.TABLE: {
tableStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD: {
tbodyStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.INPUT: {
inputStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.CAPTION: {
captionStartTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.COLGROUP: {
colgroupStartTagInTable(p, token);
break;
}
default: {
tokenInTable(p, token);
}
}
}
function endTagInTable(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.TABLE: {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TABLE)) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TABLE);
p._resetInsertionMode();
}
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.HTML:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TR: {
// Ignore token
break;
}
default: {
tokenInTable(p, token);
}
}
}
function tokenInTable(p, token) {
const savedFosterParentingState = p.fosterParentingEnabled;
p.fosterParentingEnabled = true;
// Process token in `In Body` mode
modeInBody(p, token);
p.fosterParentingEnabled = savedFosterParentingState;
}
// The "in table text" insertion mode
//------------------------------------------------------------------
function whitespaceCharacterInTableText(p, token) {
p.pendingCharacterTokens.push(token);
}
function characterInTableText(p, token) {
p.pendingCharacterTokens.push(token);
p.hasNonWhitespacePendingCharacterToken = true;
}
function tokenInTableText(p, token) {
let i = 0;
if (p.hasNonWhitespacePendingCharacterToken) {
for (; i < p.pendingCharacterTokens.length; i++) {
tokenInTable(p, p.pendingCharacterTokens[i]);
}
}
else {
for (; i < p.pendingCharacterTokens.length; i++) {
p._insertCharacters(p.pendingCharacterTokens[i]);
}
}
p.insertionMode = p.originalInsertionMode;
p._processToken(token);
}
// The "in caption" insertion mode
//------------------------------------------------------------------
const TABLE_VOID_ELEMENTS = new Set([html_js_1.TAG_ID.CAPTION, html_js_1.TAG_ID.COL, html_js_1.TAG_ID.COLGROUP, html_js_1.TAG_ID.TBODY, html_js_1.TAG_ID.TD, html_js_1.TAG_ID.TFOOT, html_js_1.TAG_ID.TH, html_js_1.TAG_ID.THEAD, html_js_1.TAG_ID.TR]);
function startTagInCaption(p, token) {
const tn = token.tagID;
if (TABLE_VOID_ELEMENTS.has(tn)) {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.CAPTION)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.CAPTION);
p.activeFormattingElements.clearToLastMarker();
p.insertionMode = InsertionMode.IN_TABLE;
startTagInTable(p, token);
}
}
else {
startTagInBody(p, token);
}
}
function endTagInCaption(p, token) {
const tn = token.tagID;
switch (tn) {
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.TABLE: {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.CAPTION)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.CAPTION);
p.activeFormattingElements.clearToLastMarker();
p.insertionMode = InsertionMode.IN_TABLE;
if (tn === html_js_1.TAG_ID.TABLE) {
endTagInTable(p, token);
}
}
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.HTML:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TR: {
// Ignore token
break;
}
default: {
endTagInBody(p, token);
}
}
}
// The "in column group" insertion mode
//------------------------------------------------------------------
function startTagInColumnGroup(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.COL: {
p._appendElement(token, html_js_1.NS.HTML);
token.ackSelfClosing = true;
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
startTagInHead(p, token);
break;
}
default: {
tokenInColumnGroup(p, token);
}
}
}
function endTagInColumnGroup(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.COLGROUP: {
if (p.openElements.currentTagId === html_js_1.TAG_ID.COLGROUP) {
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE;
}
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
case html_js_1.TAG_ID.COL: {
// Ignore token
break;
}
default: {
tokenInColumnGroup(p, token);
}
}
}
function tokenInColumnGroup(p, token) {
if (p.openElements.currentTagId === html_js_1.TAG_ID.COLGROUP) {
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE;
p._processToken(token);
}
}
// The "in table body" insertion mode
//------------------------------------------------------------------
function startTagInTableBody(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.TR: {
p.openElements.clearBackToTableBodyContext();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_ROW;
break;
}
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.TD: {
p.openElements.clearBackToTableBodyContext();
p._insertFakeElement(html_js_1.TAG_NAMES.TR, html_js_1.TAG_ID.TR);
p.insertionMode = InsertionMode.IN_ROW;
startTagInRow(p, token);
break;
}
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD: {
if (p.openElements.hasTableBodyContextInTableScope()) {
p.openElements.clearBackToTableBodyContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE;
startTagInTable(p, token);
}
break;
}
default: {
startTagInTable(p, token);
}
}
}
function endTagInTableBody(p, token) {
const tn = token.tagID;
switch (token.tagID) {
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD: {
if (p.openElements.hasInTableScope(tn)) {
p.openElements.clearBackToTableBodyContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE;
}
break;
}
case html_js_1.TAG_ID.TABLE: {
if (p.openElements.hasTableBodyContextInTableScope()) {
p.openElements.clearBackToTableBodyContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE;
endTagInTable(p, token);
}
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.HTML:
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.TR: {
// Ignore token
break;
}
default: {
endTagInTable(p, token);
}
}
}
// The "in row" insertion mode
//------------------------------------------------------------------
function startTagInRow(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.TH:
case html_js_1.TAG_ID.TD: {
p.openElements.clearBackToTableRowContext();
p._insertElement(token, html_js_1.NS.HTML);
p.insertionMode = InsertionMode.IN_CELL;
p.activeFormattingElements.insertMarker();
break;
}
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TR: {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TR)) {
p.openElements.clearBackToTableRowContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE_BODY;
startTagInTableBody(p, token);
}
break;
}
default: {
startTagInTable(p, token);
}
}
}
function endTagInRow(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.TR: {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TR)) {
p.openElements.clearBackToTableRowContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE_BODY;
}
break;
}
case html_js_1.TAG_ID.TABLE: {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TR)) {
p.openElements.clearBackToTableRowContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE_BODY;
endTagInTableBody(p, token);
}
break;
}
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD: {
if (p.openElements.hasInTableScope(token.tagID) || p.openElements.hasInTableScope(html_js_1.TAG_ID.TR)) {
p.openElements.clearBackToTableRowContext();
p.openElements.pop();
p.insertionMode = InsertionMode.IN_TABLE_BODY;
endTagInTableBody(p, token);
}
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.HTML:
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH: {
// Ignore end tag
break;
}
default: {
endTagInTable(p, token);
}
}
}
// The "in cell" insertion mode
//------------------------------------------------------------------
function startTagInCell(p, token) {
const tn = token.tagID;
if (TABLE_VOID_ELEMENTS.has(tn)) {
if (p.openElements.hasInTableScope(html_js_1.TAG_ID.TD) || p.openElements.hasInTableScope(html_js_1.TAG_ID.TH)) {
p._closeTableCell();
startTagInRow(p, token);
}
}
else {
startTagInBody(p, token);
}
}
function endTagInCell(p, token) {
const tn = token.tagID;
switch (tn) {
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH: {
if (p.openElements.hasInTableScope(tn)) {
p.openElements.generateImpliedEndTags();
p.openElements.popUntilTagNamePopped(tn);
p.activeFormattingElements.clearToLastMarker();
p.insertionMode = InsertionMode.IN_ROW;
}
break;
}
case html_js_1.TAG_ID.TABLE:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TR: {
if (p.openElements.hasInTableScope(tn)) {
p._closeTableCell();
endTagInRow(p, token);
}
break;
}
case html_js_1.TAG_ID.BODY:
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COL:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.HTML: {
// Ignore token
break;
}
default: {
endTagInBody(p, token);
}
}
}
// The "in select" insertion mode
//------------------------------------------------------------------
function startTagInSelect(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.OPTION: {
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTION) {
p.openElements.pop();
}
p._insertElement(token, html_js_1.NS.HTML);
break;
}
case html_js_1.TAG_ID.OPTGROUP: {
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTION) {
p.openElements.pop();
}
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTGROUP) {
p.openElements.pop();
}
p._insertElement(token, html_js_1.NS.HTML);
break;
}
case html_js_1.TAG_ID.HR: {
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTION) {
p.openElements.pop();
}
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTGROUP) {
p.openElements.pop();
}
p._appendElement(token, html_js_1.NS.HTML);
token.ackSelfClosing = true;
break;
}
case html_js_1.TAG_ID.INPUT:
case html_js_1.TAG_ID.KEYGEN:
case html_js_1.TAG_ID.TEXTAREA:
case html_js_1.TAG_ID.SELECT: {
if (p.openElements.hasInSelectScope(html_js_1.TAG_ID.SELECT)) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.SELECT);
p._resetInsertionMode();
if (token.tagID !== html_js_1.TAG_ID.SELECT) {
p._processStartTag(token);
}
}
break;
}
case html_js_1.TAG_ID.SCRIPT:
case html_js_1.TAG_ID.TEMPLATE: {
startTagInHead(p, token);
break;
}
default:
// Do nothing
}
}
function endTagInSelect(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.OPTGROUP: {
if (p.openElements.stackTop > 0 &&
p.openElements.currentTagId === html_js_1.TAG_ID.OPTION &&
p.openElements.tagIDs[p.openElements.stackTop - 1] === html_js_1.TAG_ID.OPTGROUP) {
p.openElements.pop();
}
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTGROUP) {
p.openElements.pop();
}
break;
}
case html_js_1.TAG_ID.OPTION: {
if (p.openElements.currentTagId === html_js_1.TAG_ID.OPTION) {
p.openElements.pop();
}
break;
}
case html_js_1.TAG_ID.SELECT: {
if (p.openElements.hasInSelectScope(html_js_1.TAG_ID.SELECT)) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.SELECT);
p._resetInsertionMode();
}
break;
}
case html_js_1.TAG_ID.TEMPLATE: {
templateEndTagInHead(p, token);
break;
}
default:
// Do nothing
}
}
// The "in select in table" insertion mode
//------------------------------------------------------------------
function startTagInSelectInTable(p, token) {
const tn = token.tagID;
if (tn === html_js_1.TAG_ID.CAPTION ||
tn === html_js_1.TAG_ID.TABLE ||
tn === html_js_1.TAG_ID.TBODY ||
tn === html_js_1.TAG_ID.TFOOT ||
tn === html_js_1.TAG_ID.THEAD ||
tn === html_js_1.TAG_ID.TR ||
tn === html_js_1.TAG_ID.TD ||
tn === html_js_1.TAG_ID.TH) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.SELECT);
p._resetInsertionMode();
p._processStartTag(token);
}
else {
startTagInSelect(p, token);
}
}
function endTagInSelectInTable(p, token) {
const tn = token.tagID;
if (tn === html_js_1.TAG_ID.CAPTION ||
tn === html_js_1.TAG_ID.TABLE ||
tn === html_js_1.TAG_ID.TBODY ||
tn === html_js_1.TAG_ID.TFOOT ||
tn === html_js_1.TAG_ID.THEAD ||
tn === html_js_1.TAG_ID.TR ||
tn === html_js_1.TAG_ID.TD ||
tn === html_js_1.TAG_ID.TH) {
if (p.openElements.hasInTableScope(tn)) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.SELECT);
p._resetInsertionMode();
p.onEndTag(token);
}
}
else {
endTagInSelect(p, token);
}
}
// The "in template" insertion mode
//------------------------------------------------------------------
function startTagInTemplate(p, token) {
switch (token.tagID) {
// First, handle tags that can start without a mode change
case html_js_1.TAG_ID.BASE:
case html_js_1.TAG_ID.BASEFONT:
case html_js_1.TAG_ID.BGSOUND:
case html_js_1.TAG_ID.LINK:
case html_js_1.TAG_ID.META:
case html_js_1.TAG_ID.NOFRAMES:
case html_js_1.TAG_ID.SCRIPT:
case html_js_1.TAG_ID.STYLE:
case html_js_1.TAG_ID.TEMPLATE:
case html_js_1.TAG_ID.TITLE: {
startTagInHead(p, token);
break;
}
// Re-process the token in the appropriate mode
case html_js_1.TAG_ID.CAPTION:
case html_js_1.TAG_ID.COLGROUP:
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.TFOOT:
case html_js_1.TAG_ID.THEAD: {
p.tmplInsertionModeStack[0] = InsertionMode.IN_TABLE;
p.insertionMode = InsertionMode.IN_TABLE;
startTagInTable(p, token);
break;
}
case html_js_1.TAG_ID.COL: {
p.tmplInsertionModeStack[0] = InsertionMode.IN_COLUMN_GROUP;
p.insertionMode = InsertionMode.IN_COLUMN_GROUP;
startTagInColumnGroup(p, token);
break;
}
case html_js_1.TAG_ID.TR: {
p.tmplInsertionModeStack[0] = InsertionMode.IN_TABLE_BODY;
p.insertionMode = InsertionMode.IN_TABLE_BODY;
startTagInTableBody(p, token);
break;
}
case html_js_1.TAG_ID.TD:
case html_js_1.TAG_ID.TH: {
p.tmplInsertionModeStack[0] = InsertionMode.IN_ROW;
p.insertionMode = InsertionMode.IN_ROW;
startTagInRow(p, token);
break;
}
default: {
p.tmplInsertionModeStack[0] = InsertionMode.IN_BODY;
p.insertionMode = InsertionMode.IN_BODY;
startTagInBody(p, token);
}
}
}
function endTagInTemplate(p, token) {
if (token.tagID === html_js_1.TAG_ID.TEMPLATE) {
templateEndTagInHead(p, token);
}
}
function eofInTemplate(p, token) {
if (p.openElements.tmplCount > 0) {
p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TEMPLATE);
p.activeFormattingElements.clearToLastMarker();
p.tmplInsertionModeStack.shift();
p._resetInsertionMode();
p.onEof(token);
}
else {
stopParsing(p, token);
}
}
// The "after body" insertion mode
//------------------------------------------------------------------
function startTagAfterBody(p, token) {
if (token.tagID === html_js_1.TAG_ID.HTML) {
startTagInBody(p, token);
}
else {
tokenAfterBody(p, token);
}
}
function endTagAfterBody(p, token) {
var _a;
if (token.tagID === html_js_1.TAG_ID.HTML) {
if (!p.fragmentContext) {
p.insertionMode = InsertionMode.AFTER_AFTER_BODY;
}
//NOTE: <html> is never popped from the stack, so we need to updated
//the end location explicitly.
if (p.options.sourceCodeLocationInfo && p.openElements.tagIDs[0] === html_js_1.TAG_ID.HTML) {
p._setEndLocation(p.openElements.items[0], token);
// Update the body element, if it doesn't have an end tag
const bodyElement = p.openElements.items[1];
if (bodyElement && !((_a = p.treeAdapter.getNodeSourceCodeLocation(bodyElement)) === null || _a === void 0 ? void 0 : _a.endTag)) {
p._setEndLocation(bodyElement, token);
}
}
}
else {
tokenAfterBody(p, token);
}
}
function tokenAfterBody(p, token) {
p.insertionMode = InsertionMode.IN_BODY;
modeInBody(p, token);
}
// The "in frameset" insertion mode
//------------------------------------------------------------------
function startTagInFrameset(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.FRAMESET: {
p._insertElement(token, html_js_1.NS.HTML);
break;
}
case html_js_1.TAG_ID.FRAME: {
p._appendElement(token, html_js_1.NS.HTML);
token.ackSelfClosing = true;
break;
}
case html_js_1.TAG_ID.NOFRAMES: {
startTagInHead(p, token);
break;
}
default:
// Do nothing
}
}
function endTagInFrameset(p, token) {
if (token.tagID === html_js_1.TAG_ID.FRAMESET && !p.openElements.isRootHtmlElementCurrent()) {
p.openElements.pop();
if (!p.fragmentContext && p.openElements.currentTagId !== html_js_1.TAG_ID.FRAMESET) {
p.insertionMode = InsertionMode.AFTER_FRAMESET;
}
}
}
// The "after frameset" insertion mode
//------------------------------------------------------------------
function startTagAfterFrameset(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.NOFRAMES: {
startTagInHead(p, token);
break;
}
default:
// Do nothing
}
}
function endTagAfterFrameset(p, token) {
if (token.tagID === html_js_1.TAG_ID.HTML) {
p.insertionMode = InsertionMode.AFTER_AFTER_FRAMESET;
}
}
// The "after after body" insertion mode
//------------------------------------------------------------------
function startTagAfterAfterBody(p, token) {
if (token.tagID === html_js_1.TAG_ID.HTML) {
startTagInBody(p, token);
}
else {
tokenAfterAfterBody(p, token);
}
}
function tokenAfterAfterBody(p, token) {
p.insertionMode = InsertionMode.IN_BODY;
modeInBody(p, token);
}
// The "after after frameset" insertion mode
//------------------------------------------------------------------
function startTagAfterAfterFrameset(p, token) {
switch (token.tagID) {
case html_js_1.TAG_ID.HTML: {
startTagInBody(p, token);
break;
}
case html_js_1.TAG_ID.NOFRAMES: {
startTagInHead(p, token);
break;
}
default:
// Do nothing
}
}
// The rules for parsing tokens in foreign content
//------------------------------------------------------------------
function nullCharacterInForeignContent(p, token) {
token.chars = unicode.REPLACEMENT_CHARACTER;
p._insertCharacters(token);
}
function characterInForeignContent(p, token) {
p._insertCharacters(token);
p.framesetOk = false;
}
function popUntilHtmlOrIntegrationPoint(p) {
while (p.treeAdapter.getNamespaceURI(p.openElements.current) !== html_js_1.NS.HTML &&
p.openElements.currentTagId !== undefined &&
!p._isIntegrationPoint(p.openElements.currentTagId, p.openElements.current)) {
p.openElements.pop();
}
}
function startTagInForeignContent(p, token) {
if (foreignContent.causesExit(token)) {
popUntilHtmlOrIntegrationPoint(p);
p._startTagOutsideForeignContent(token);
}
else {
const current = p._getAdjustedCurrentElement();
const currentNs = p.treeAdapter.getNamespaceURI(current);
if (currentNs === html_js_1.NS.MATHML) {
foreignContent.adjustTokenMathMLAttrs(token);
}
else if (currentNs === html_js_1.NS.SVG) {
foreignContent.adjustTokenSVGTagName(token);
foreignContent.adjustTokenSVGAttrs(token);
}
foreignContent.adjustTokenXMLAttrs(token);
if (token.selfClosing) {
p._appendElement(token, currentNs);
}
else {
p._insertElement(token, currentNs);
}
token.ackSelfClosing = true;
}
}
function endTagInForeignContent(p, token) {
if (token.tagID === html_js_1.TAG_ID.P || token.tagID === html_js_1.TAG_ID.BR) {
popUntilHtmlOrIntegrationPoint(p);
p._endTagOutsideForeignContent(token);
return;
}
for (let i = p.openElements.stackTop; i > 0; i--) {
const element = p.openElements.items[i];
if (p.treeAdapter.getNamespaceURI(element) === html_js_1.NS.HTML) {
p._endTagOutsideForeignContent(token);
break;
}
const tagName = p.treeAdapter.getTagName(element);
if (tagName.toLowerCase() === token.tagName) {
//NOTE: update the token tag name for `_setEndLocation`.
token.tagName = tagName;
p.openElements.shortenToLength(i);
break;
}
}
}
+53
View File
@@ -0,0 +1,53 @@
import { TAG_ID as $ } from '../common/html.js';
import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
export interface StackHandler<T extends TreeAdapterTypeMap> {
onItemPush: (node: T['parentNode'], tid: number, isTop: boolean) => void;
onItemPop: (node: T['parentNode'], isTop: boolean) => void;
}
export declare class OpenElementStack<T extends TreeAdapterTypeMap> {
private treeAdapter;
private handler;
items: T['parentNode'][];
tagIDs: $[];
current: T['parentNode'] | undefined;
stackTop: number;
tmplCount: number;
currentTagId: number | undefined;
get currentTmplContentOrNode(): T['parentNode'];
constructor(document: T['document'], treeAdapter: TreeAdapter<T>, handler: StackHandler<T>);
private _indexOf;
private _isInTemplate;
private _updateCurrentElement;
push(element: T['element'], tagID: $): void;
pop(): void;
replace(oldElement: T['element'], newElement: T['element']): void;
insertAfter(referenceElement: T['element'], newElement: T['element'], newElementID: $): void;
popUntilTagNamePopped(tagName: $): void;
shortenToLength(idx: number): void;
popUntilElementPopped(element: T['element']): void;
private popUntilPopped;
popUntilNumberedHeaderPopped(): void;
popUntilTableCellPopped(): void;
popAllUpToHtmlElement(): void;
private _indexOfTagNames;
private clearBackTo;
clearBackToTableContext(): void;
clearBackToTableBodyContext(): void;
clearBackToTableRowContext(): void;
remove(element: T['element']): void;
tryPeekProperlyNestedBodyElement(): T['element'] | null;
contains(element: T['element']): boolean;
getCommonAncestor(element: T['element']): T['element'] | null;
isRootHtmlElementCurrent(): boolean;
private hasInDynamicScope;
hasInScope(tagName: $): boolean;
hasInListItemScope(tagName: $): boolean;
hasInButtonScope(tagName: $): boolean;
hasNumberedHeaderInScope(): boolean;
hasInTableScope(tagName: $): boolean;
hasTableBodyContextInTableScope(): boolean;
hasInSelectScope(tagName: $): boolean;
generateImpliedEndTags(): void;
generateImpliedEndTagsThoroughly(): void;
generateImpliedEndTagsWithExclusion(exclusionId: $): void;
}
+328
View File
@@ -0,0 +1,328 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.OpenElementStack = void 0;
const html_js_1 = require("../common/html.js");
//Element utils
const IMPLICIT_END_TAG_REQUIRED = new Set([html_js_1.TAG_ID.DD, html_js_1.TAG_ID.DT, html_js_1.TAG_ID.LI, html_js_1.TAG_ID.OPTGROUP, html_js_1.TAG_ID.OPTION, html_js_1.TAG_ID.P, html_js_1.TAG_ID.RB, html_js_1.TAG_ID.RP, html_js_1.TAG_ID.RT, html_js_1.TAG_ID.RTC]);
const IMPLICIT_END_TAG_REQUIRED_THOROUGHLY = new Set([
...IMPLICIT_END_TAG_REQUIRED,
html_js_1.TAG_ID.CAPTION,
html_js_1.TAG_ID.COLGROUP,
html_js_1.TAG_ID.TBODY,
html_js_1.TAG_ID.TD,
html_js_1.TAG_ID.TFOOT,
html_js_1.TAG_ID.TH,
html_js_1.TAG_ID.THEAD,
html_js_1.TAG_ID.TR,
]);
const SCOPING_ELEMENTS_HTML = new Set([
html_js_1.TAG_ID.APPLET,
html_js_1.TAG_ID.CAPTION,
html_js_1.TAG_ID.HTML,
html_js_1.TAG_ID.MARQUEE,
html_js_1.TAG_ID.OBJECT,
html_js_1.TAG_ID.TABLE,
html_js_1.TAG_ID.TD,
html_js_1.TAG_ID.TEMPLATE,
html_js_1.TAG_ID.TH,
]);
const SCOPING_ELEMENTS_HTML_LIST = new Set([...SCOPING_ELEMENTS_HTML, html_js_1.TAG_ID.OL, html_js_1.TAG_ID.UL]);
const SCOPING_ELEMENTS_HTML_BUTTON = new Set([...SCOPING_ELEMENTS_HTML, html_js_1.TAG_ID.BUTTON]);
const SCOPING_ELEMENTS_MATHML = new Set([html_js_1.TAG_ID.ANNOTATION_XML, html_js_1.TAG_ID.MI, html_js_1.TAG_ID.MN, html_js_1.TAG_ID.MO, html_js_1.TAG_ID.MS, html_js_1.TAG_ID.MTEXT]);
const SCOPING_ELEMENTS_SVG = new Set([html_js_1.TAG_ID.DESC, html_js_1.TAG_ID.FOREIGN_OBJECT, html_js_1.TAG_ID.TITLE]);
const TABLE_ROW_CONTEXT = new Set([html_js_1.TAG_ID.TR, html_js_1.TAG_ID.TEMPLATE, html_js_1.TAG_ID.HTML]);
const TABLE_BODY_CONTEXT = new Set([html_js_1.TAG_ID.TBODY, html_js_1.TAG_ID.TFOOT, html_js_1.TAG_ID.THEAD, html_js_1.TAG_ID.TEMPLATE, html_js_1.TAG_ID.HTML]);
const TABLE_CONTEXT = new Set([html_js_1.TAG_ID.TABLE, html_js_1.TAG_ID.TEMPLATE, html_js_1.TAG_ID.HTML]);
const TABLE_CELLS = new Set([html_js_1.TAG_ID.TD, html_js_1.TAG_ID.TH]);
//Stack of open elements
class OpenElementStack {
get currentTmplContentOrNode() {
return this._isInTemplate() ? this.treeAdapter.getTemplateContent(this.current) : this.current;
}
constructor(document, treeAdapter, handler) {
this.treeAdapter = treeAdapter;
this.handler = handler;
this.items = [];
this.tagIDs = [];
this.stackTop = -1;
this.tmplCount = 0;
this.currentTagId = html_js_1.TAG_ID.UNKNOWN;
this.current = document;
}
//Index of element
_indexOf(element) {
return this.items.lastIndexOf(element, this.stackTop);
}
//Update current element
_isInTemplate() {
return this.currentTagId === html_js_1.TAG_ID.TEMPLATE && this.treeAdapter.getNamespaceURI(this.current) === html_js_1.NS.HTML;
}
_updateCurrentElement() {
this.current = this.items[this.stackTop];
this.currentTagId = this.tagIDs[this.stackTop];
}
//Mutations
push(element, tagID) {
this.stackTop++;
this.items[this.stackTop] = element;
this.current = element;
this.tagIDs[this.stackTop] = tagID;
this.currentTagId = tagID;
if (this._isInTemplate()) {
this.tmplCount++;
}
this.handler.onItemPush(element, tagID, true);
}
pop() {
const popped = this.current;
if (this.tmplCount > 0 && this._isInTemplate()) {
this.tmplCount--;
}
this.stackTop--;
this._updateCurrentElement();
this.handler.onItemPop(popped, true);
}
replace(oldElement, newElement) {
const idx = this._indexOf(oldElement);
this.items[idx] = newElement;
if (idx === this.stackTop) {
this.current = newElement;
}
}
insertAfter(referenceElement, newElement, newElementID) {
const insertionIdx = this._indexOf(referenceElement) + 1;
this.items.splice(insertionIdx, 0, newElement);
this.tagIDs.splice(insertionIdx, 0, newElementID);
this.stackTop++;
if (insertionIdx === this.stackTop) {
this._updateCurrentElement();
}
if (this.current && this.currentTagId !== undefined) {
this.handler.onItemPush(this.current, this.currentTagId, insertionIdx === this.stackTop);
}
}
popUntilTagNamePopped(tagName) {
let targetIdx = this.stackTop + 1;
do {
targetIdx = this.tagIDs.lastIndexOf(tagName, targetIdx - 1);
} while (targetIdx > 0 && this.treeAdapter.getNamespaceURI(this.items[targetIdx]) !== html_js_1.NS.HTML);
this.shortenToLength(Math.max(targetIdx, 0));
}
shortenToLength(idx) {
while (this.stackTop >= idx) {
const popped = this.current;
if (this.tmplCount > 0 && this._isInTemplate()) {
this.tmplCount -= 1;
}
this.stackTop--;
this._updateCurrentElement();
this.handler.onItemPop(popped, this.stackTop < idx);
}
}
popUntilElementPopped(element) {
const idx = this._indexOf(element);
this.shortenToLength(Math.max(idx, 0));
}
popUntilPopped(tagNames, targetNS) {
const idx = this._indexOfTagNames(tagNames, targetNS);
this.shortenToLength(Math.max(idx, 0));
}
popUntilNumberedHeaderPopped() {
this.popUntilPopped(html_js_1.NUMBERED_HEADERS, html_js_1.NS.HTML);
}
popUntilTableCellPopped() {
this.popUntilPopped(TABLE_CELLS, html_js_1.NS.HTML);
}
popAllUpToHtmlElement() {
//NOTE: here we assume that the root <html> element is always first in the open element stack, so
//we perform this fast stack clean up.
this.tmplCount = 0;
this.shortenToLength(1);
}
_indexOfTagNames(tagNames, namespace) {
for (let i = this.stackTop; i >= 0; i--) {
if (tagNames.has(this.tagIDs[i]) && this.treeAdapter.getNamespaceURI(this.items[i]) === namespace) {
return i;
}
}
return -1;
}
clearBackTo(tagNames, targetNS) {
const idx = this._indexOfTagNames(tagNames, targetNS);
this.shortenToLength(idx + 1);
}
clearBackToTableContext() {
this.clearBackTo(TABLE_CONTEXT, html_js_1.NS.HTML);
}
clearBackToTableBodyContext() {
this.clearBackTo(TABLE_BODY_CONTEXT, html_js_1.NS.HTML);
}
clearBackToTableRowContext() {
this.clearBackTo(TABLE_ROW_CONTEXT, html_js_1.NS.HTML);
}
remove(element) {
const idx = this._indexOf(element);
if (idx >= 0) {
if (idx === this.stackTop) {
this.pop();
}
else {
this.items.splice(idx, 1);
this.tagIDs.splice(idx, 1);
this.stackTop--;
this._updateCurrentElement();
this.handler.onItemPop(element, false);
}
}
}
//Search
tryPeekProperlyNestedBodyElement() {
//Properly nested <body> element (should be second element in stack).
return this.stackTop >= 1 && this.tagIDs[1] === html_js_1.TAG_ID.BODY ? this.items[1] : null;
}
contains(element) {
return this._indexOf(element) > -1;
}
getCommonAncestor(element) {
const elementIdx = this._indexOf(element) - 1;
return elementIdx >= 0 ? this.items[elementIdx] : null;
}
isRootHtmlElementCurrent() {
return this.stackTop === 0 && this.tagIDs[0] === html_js_1.TAG_ID.HTML;
}
//Element in scope
hasInDynamicScope(tagName, htmlScope) {
for (let i = this.stackTop; i >= 0; i--) {
const tn = this.tagIDs[i];
switch (this.treeAdapter.getNamespaceURI(this.items[i])) {
case html_js_1.NS.HTML: {
if (tn === tagName)
return true;
if (htmlScope.has(tn))
return false;
break;
}
case html_js_1.NS.SVG: {
if (SCOPING_ELEMENTS_SVG.has(tn))
return false;
break;
}
case html_js_1.NS.MATHML: {
if (SCOPING_ELEMENTS_MATHML.has(tn))
return false;
break;
}
}
}
return true;
}
hasInScope(tagName) {
return this.hasInDynamicScope(tagName, SCOPING_ELEMENTS_HTML);
}
hasInListItemScope(tagName) {
return this.hasInDynamicScope(tagName, SCOPING_ELEMENTS_HTML_LIST);
}
hasInButtonScope(tagName) {
return this.hasInDynamicScope(tagName, SCOPING_ELEMENTS_HTML_BUTTON);
}
hasNumberedHeaderInScope() {
for (let i = this.stackTop; i >= 0; i--) {
const tn = this.tagIDs[i];
switch (this.treeAdapter.getNamespaceURI(this.items[i])) {
case html_js_1.NS.HTML: {
if (html_js_1.NUMBERED_HEADERS.has(tn))
return true;
if (SCOPING_ELEMENTS_HTML.has(tn))
return false;
break;
}
case html_js_1.NS.SVG: {
if (SCOPING_ELEMENTS_SVG.has(tn))
return false;
break;
}
case html_js_1.NS.MATHML: {
if (SCOPING_ELEMENTS_MATHML.has(tn))
return false;
break;
}
}
}
return true;
}
hasInTableScope(tagName) {
for (let i = this.stackTop; i >= 0; i--) {
if (this.treeAdapter.getNamespaceURI(this.items[i]) !== html_js_1.NS.HTML) {
continue;
}
switch (this.tagIDs[i]) {
case tagName: {
return true;
}
case html_js_1.TAG_ID.TABLE:
case html_js_1.TAG_ID.HTML: {
return false;
}
}
}
return true;
}
hasTableBodyContextInTableScope() {
for (let i = this.stackTop; i >= 0; i--) {
if (this.treeAdapter.getNamespaceURI(this.items[i]) !== html_js_1.NS.HTML) {
continue;
}
switch (this.tagIDs[i]) {
case html_js_1.TAG_ID.TBODY:
case html_js_1.TAG_ID.THEAD:
case html_js_1.TAG_ID.TFOOT: {
return true;
}
case html_js_1.TAG_ID.TABLE:
case html_js_1.TAG_ID.HTML: {
return false;
}
}
}
return true;
}
hasInSelectScope(tagName) {
for (let i = this.stackTop; i >= 0; i--) {
if (this.treeAdapter.getNamespaceURI(this.items[i]) !== html_js_1.NS.HTML) {
continue;
}
switch (this.tagIDs[i]) {
case tagName: {
return true;
}
case html_js_1.TAG_ID.OPTION:
case html_js_1.TAG_ID.OPTGROUP: {
break;
}
default: {
return false;
}
}
}
return true;
}
//Implied end tags
generateImpliedEndTags() {
while (this.currentTagId !== undefined && IMPLICIT_END_TAG_REQUIRED.has(this.currentTagId)) {
this.pop();
}
}
generateImpliedEndTagsThoroughly() {
while (this.currentTagId !== undefined && IMPLICIT_END_TAG_REQUIRED_THOROUGHLY.has(this.currentTagId)) {
this.pop();
}
}
generateImpliedEndTagsWithExclusion(exclusionId) {
while (this.currentTagId !== undefined &&
this.currentTagId !== exclusionId &&
IMPLICIT_END_TAG_REQUIRED_THOROUGHLY.has(this.currentTagId)) {
this.pop();
}
}
}
exports.OpenElementStack = OpenElementStack;
+60
View File
@@ -0,0 +1,60 @@
import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
import { type DefaultTreeAdapterMap } from '../tree-adapters/default.js';
export interface SerializerOptions<T extends TreeAdapterTypeMap> {
/**
* Specifies input tree format.
*
* @default `treeAdapters.default`
*/
treeAdapter?: TreeAdapter<T>;
/**
* The [scripting flag](https://html.spec.whatwg.org/multipage/parsing.html#scripting-flag). If set
* to `true`, `noscript` element content will not be escaped.
*
* @default `true`
*/
scriptingEnabled?: boolean;
}
/**
* Serializes an AST node to an HTML string.
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
*
* // Serializes a document.
* const html = parse5.serialize(document);
*
* // Serializes the <html> element content.
* const str = parse5.serialize(document.childNodes[1]);
*
* console.log(str); //> '<head></head><body>Hi there!</body>'
* ```
*
* @param node Node to serialize.
* @param options Serialization options.
*/
export declare function serialize<T extends TreeAdapterTypeMap = DefaultTreeAdapterMap>(node: T['parentNode'], options?: SerializerOptions<T>): string;
/**
* Serializes an AST element node to an HTML string, including the element node.
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parseFragment('<div>Hello, <b>world</b>!</div>');
*
* // Serializes the <div> element.
* const str = parse5.serializeOuter(document.childNodes[0]);
*
* console.log(str); //> '<div>Hello, <b>world</b>!</div>'
* ```
*
* @param node Node to serialize.
* @param options Serialization options.
*/
export declare function serializeOuter<T extends TreeAdapterTypeMap = DefaultTreeAdapterMap>(node: T['node'], options?: SerializerOptions<T>): string;
+172
View File
@@ -0,0 +1,172 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.serialize = serialize;
exports.serializeOuter = serializeOuter;
const html_js_1 = require("../common/html.js");
const escape_1 = require("entities/escape");
const default_js_1 = require("../tree-adapters/default.js");
// Sets
const VOID_ELEMENTS = new Set([
html_js_1.TAG_NAMES.AREA,
html_js_1.TAG_NAMES.BASE,
html_js_1.TAG_NAMES.BASEFONT,
html_js_1.TAG_NAMES.BGSOUND,
html_js_1.TAG_NAMES.BR,
html_js_1.TAG_NAMES.COL,
html_js_1.TAG_NAMES.EMBED,
html_js_1.TAG_NAMES.FRAME,
html_js_1.TAG_NAMES.HR,
html_js_1.TAG_NAMES.IMG,
html_js_1.TAG_NAMES.INPUT,
html_js_1.TAG_NAMES.KEYGEN,
html_js_1.TAG_NAMES.LINK,
html_js_1.TAG_NAMES.META,
html_js_1.TAG_NAMES.PARAM,
html_js_1.TAG_NAMES.SOURCE,
html_js_1.TAG_NAMES.TRACK,
html_js_1.TAG_NAMES.WBR,
]);
function isVoidElement(node, options) {
return (options.treeAdapter.isElementNode(node) &&
options.treeAdapter.getNamespaceURI(node) === html_js_1.NS.HTML &&
VOID_ELEMENTS.has(options.treeAdapter.getTagName(node)));
}
const defaultOpts = { treeAdapter: default_js_1.defaultTreeAdapter, scriptingEnabled: true };
/**
* Serializes an AST node to an HTML string.
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parse('<!DOCTYPE html><html><head></head><body>Hi there!</body></html>');
*
* // Serializes a document.
* const html = parse5.serialize(document);
*
* // Serializes the <html> element content.
* const str = parse5.serialize(document.childNodes[1]);
*
* console.log(str); //> '<head></head><body>Hi there!</body>'
* ```
*
* @param node Node to serialize.
* @param options Serialization options.
*/
function serialize(node, options) {
const opts = Object.assign(Object.assign({}, defaultOpts), options);
if (isVoidElement(node, opts)) {
return '';
}
return serializeChildNodes(node, opts);
}
/**
* Serializes an AST element node to an HTML string, including the element node.
*
* @example
*
* ```js
* const parse5 = require('parse5');
*
* const document = parse5.parseFragment('<div>Hello, <b>world</b>!</div>');
*
* // Serializes the <div> element.
* const str = parse5.serializeOuter(document.childNodes[0]);
*
* console.log(str); //> '<div>Hello, <b>world</b>!</div>'
* ```
*
* @param node Node to serialize.
* @param options Serialization options.
*/
function serializeOuter(node, options) {
const opts = Object.assign(Object.assign({}, defaultOpts), options);
return serializeNode(node, opts);
}
function serializeChildNodes(parentNode, options) {
let html = '';
// Get container of the child nodes
const container = options.treeAdapter.isElementNode(parentNode) &&
options.treeAdapter.getTagName(parentNode) === html_js_1.TAG_NAMES.TEMPLATE &&
options.treeAdapter.getNamespaceURI(parentNode) === html_js_1.NS.HTML
? options.treeAdapter.getTemplateContent(parentNode)
: parentNode;
const childNodes = options.treeAdapter.getChildNodes(container);
if (childNodes) {
for (const currentNode of childNodes) {
html += serializeNode(currentNode, options);
}
}
return html;
}
function serializeNode(node, options) {
if (options.treeAdapter.isElementNode(node)) {
return serializeElement(node, options);
}
if (options.treeAdapter.isTextNode(node)) {
return serializeTextNode(node, options);
}
if (options.treeAdapter.isCommentNode(node)) {
return serializeCommentNode(node, options);
}
if (options.treeAdapter.isDocumentTypeNode(node)) {
return serializeDocumentTypeNode(node, options);
}
// Return an empty string for unknown nodes
return '';
}
function serializeElement(node, options) {
const tn = options.treeAdapter.getTagName(node);
return `<${tn}${serializeAttributes(node, options)}>${isVoidElement(node, options) ? '' : `${serializeChildNodes(node, options)}</${tn}>`}`;
}
function serializeAttributes(node, { treeAdapter }) {
let html = '';
for (const attr of treeAdapter.getAttrList(node)) {
html += ' ';
if (attr.namespace) {
switch (attr.namespace) {
case html_js_1.NS.XML: {
html += `xml:${attr.name}`;
break;
}
case html_js_1.NS.XMLNS: {
if (attr.name !== 'xmlns') {
html += 'xmlns:';
}
html += attr.name;
break;
}
case html_js_1.NS.XLINK: {
html += `xlink:${attr.name}`;
break;
}
default: {
html += `${attr.prefix}:${attr.name}`;
}
}
}
else {
html += attr.name;
}
html += `="${(0, escape_1.escapeAttribute)(attr.value)}"`;
}
return html;
}
function serializeTextNode(node, options) {
const { treeAdapter } = options;
const content = treeAdapter.getTextNodeContent(node);
const parent = treeAdapter.getParentNode(node);
const parentTn = parent && treeAdapter.isElementNode(parent) && treeAdapter.getTagName(parent);
return parentTn &&
treeAdapter.getNamespaceURI(parent) === html_js_1.NS.HTML &&
(0, html_js_1.hasUnescapedText)(parentTn, options.scriptingEnabled)
? content
: (0, escape_1.escapeText)(content);
}
function serializeCommentNode(node, { treeAdapter }) {
return `<!--${treeAdapter.getCommentNodeContent(node)}-->`;
}
function serializeDocumentTypeNode(node, { treeAdapter }) {
return `<!DOCTYPE ${treeAdapter.getDocumentTypeNodeName(node)}>`;
}
+247
View File
@@ -0,0 +1,247 @@
import { Preprocessor } from './preprocessor.js';
import { type Token, type CharacterToken, type DoctypeToken, type TagToken, type EOFToken, type CommentToken, type Attribute, type Location } from '../common/token.js';
import { EntityDecoder } from 'entities/decode';
import { ERR, type ParserErrorHandler } from '../common/error-codes.js';
declare const enum State {
DATA = 0,
RCDATA = 1,
RAWTEXT = 2,
SCRIPT_DATA = 3,
PLAINTEXT = 4,
TAG_OPEN = 5,
END_TAG_OPEN = 6,
TAG_NAME = 7,
RCDATA_LESS_THAN_SIGN = 8,
RCDATA_END_TAG_OPEN = 9,
RCDATA_END_TAG_NAME = 10,
RAWTEXT_LESS_THAN_SIGN = 11,
RAWTEXT_END_TAG_OPEN = 12,
RAWTEXT_END_TAG_NAME = 13,
SCRIPT_DATA_LESS_THAN_SIGN = 14,
SCRIPT_DATA_END_TAG_OPEN = 15,
SCRIPT_DATA_END_TAG_NAME = 16,
SCRIPT_DATA_ESCAPE_START = 17,
SCRIPT_DATA_ESCAPE_START_DASH = 18,
SCRIPT_DATA_ESCAPED = 19,
SCRIPT_DATA_ESCAPED_DASH = 20,
SCRIPT_DATA_ESCAPED_DASH_DASH = 21,
SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN = 22,
SCRIPT_DATA_ESCAPED_END_TAG_OPEN = 23,
SCRIPT_DATA_ESCAPED_END_TAG_NAME = 24,
SCRIPT_DATA_DOUBLE_ESCAPE_START = 25,
SCRIPT_DATA_DOUBLE_ESCAPED = 26,
SCRIPT_DATA_DOUBLE_ESCAPED_DASH = 27,
SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH = 28,
SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN = 29,
SCRIPT_DATA_DOUBLE_ESCAPE_END = 30,
BEFORE_ATTRIBUTE_NAME = 31,
ATTRIBUTE_NAME = 32,
AFTER_ATTRIBUTE_NAME = 33,
BEFORE_ATTRIBUTE_VALUE = 34,
ATTRIBUTE_VALUE_DOUBLE_QUOTED = 35,
ATTRIBUTE_VALUE_SINGLE_QUOTED = 36,
ATTRIBUTE_VALUE_UNQUOTED = 37,
AFTER_ATTRIBUTE_VALUE_QUOTED = 38,
SELF_CLOSING_START_TAG = 39,
BOGUS_COMMENT = 40,
MARKUP_DECLARATION_OPEN = 41,
COMMENT_START = 42,
COMMENT_START_DASH = 43,
COMMENT = 44,
COMMENT_LESS_THAN_SIGN = 45,
COMMENT_LESS_THAN_SIGN_BANG = 46,
COMMENT_LESS_THAN_SIGN_BANG_DASH = 47,
COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH = 48,
COMMENT_END_DASH = 49,
COMMENT_END = 50,
COMMENT_END_BANG = 51,
DOCTYPE = 52,
BEFORE_DOCTYPE_NAME = 53,
DOCTYPE_NAME = 54,
AFTER_DOCTYPE_NAME = 55,
AFTER_DOCTYPE_PUBLIC_KEYWORD = 56,
BEFORE_DOCTYPE_PUBLIC_IDENTIFIER = 57,
DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED = 58,
DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED = 59,
AFTER_DOCTYPE_PUBLIC_IDENTIFIER = 60,
BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS = 61,
AFTER_DOCTYPE_SYSTEM_KEYWORD = 62,
BEFORE_DOCTYPE_SYSTEM_IDENTIFIER = 63,
DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED = 64,
DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED = 65,
AFTER_DOCTYPE_SYSTEM_IDENTIFIER = 66,
BOGUS_DOCTYPE = 67,
CDATA_SECTION = 68,
CDATA_SECTION_BRACKET = 69,
CDATA_SECTION_END = 70,
CHARACTER_REFERENCE = 71,
AMBIGUOUS_AMPERSAND = 72
}
export declare const TokenizerMode: {
readonly DATA: State.DATA;
readonly RCDATA: State.RCDATA;
readonly RAWTEXT: State.RAWTEXT;
readonly SCRIPT_DATA: State.SCRIPT_DATA;
readonly PLAINTEXT: State.PLAINTEXT;
readonly CDATA_SECTION: State.CDATA_SECTION;
};
export interface TokenizerOptions {
sourceCodeLocationInfo?: boolean;
}
export interface TokenHandler {
onComment(token: CommentToken): void;
onDoctype(token: DoctypeToken): void;
onStartTag(token: TagToken): void;
onEndTag(token: TagToken): void;
onEof(token: EOFToken): void;
onCharacter(token: CharacterToken): void;
onNullCharacter(token: CharacterToken): void;
onWhitespaceCharacter(token: CharacterToken): void;
onParseError?: ParserErrorHandler | null;
}
export declare class Tokenizer {
protected options: TokenizerOptions;
protected handler: TokenHandler;
preprocessor: Preprocessor;
protected paused: boolean;
/** Ensures that the parsing loop isn't run multiple times at once. */
protected inLoop: boolean;
/**
* Indicates that the current adjusted node exists, is not an element in the HTML namespace,
* and that it is not an integration point for either MathML or HTML.
*
* @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
*/
inForeignNode: boolean;
lastStartTagName: string;
active: boolean;
state: State;
protected returnState: State;
/**
* We use `entities`' `EntityDecoder` to parse character references.
*
* All of the following states are handled by the `EntityDecoder`:
*
* - Named character reference state
* - Numeric character reference state
* - Hexademical character reference start state
* - Hexademical character reference state
* - Decimal character reference state
* - Numeric character reference end state
*/
protected entityDecoder: EntityDecoder;
protected entityStartPos: number;
protected consumedAfterSnapshot: number;
protected currentLocation: Location | null;
protected currentCharacterToken: CharacterToken | null;
protected currentToken: Token | null;
protected currentAttr: Attribute;
constructor(options: TokenizerOptions, handler: TokenHandler);
protected _err(code: ERR, cpOffset?: number): void;
protected getCurrentLocation(offset: number): Location | null;
protected _runParsingLoop(): void;
pause(): void;
resume(writeCallback?: () => void): void;
write(chunk: string, isLastChunk: boolean, writeCallback?: () => void): void;
insertHtmlAtCurrentPos(chunk: string): void;
protected _ensureHibernation(): boolean;
protected _consume(): number;
protected _advanceBy(count: number): void;
protected _consumeSequenceIfMatch(pattern: string, caseSensitive: boolean): boolean;
protected _createStartTagToken(): void;
protected _createEndTagToken(): void;
protected _createCommentToken(offset: number): void;
protected _createDoctypeToken(initialName: string | null): void;
protected _createCharacterToken(type: CharacterToken['type'], chars: string): void;
protected _createAttr(attrNameFirstCh: string): void;
protected _leaveAttrName(): void;
protected _leaveAttrValue(): void;
protected prepareToken(ct: Token): void;
protected emitCurrentTagToken(): void;
protected emitCurrentComment(ct: CommentToken): void;
protected emitCurrentDoctype(ct: DoctypeToken): void;
protected _emitCurrentCharacterToken(nextLocation: Location | null): void;
protected _emitEOFToken(): void;
protected _appendCharToCurrentCharacterToken(type: CharacterToken['type'], ch: string): void;
protected _emitCodePoint(cp: number): void;
protected _emitChars(ch: string): void;
protected _startCharacterReference(): void;
protected _isCharacterReferenceInAttribute(): boolean;
protected _flushCodePointConsumedAsCharacterReference(cp: number): void;
protected _callState(cp: number): void;
protected _stateData(cp: number): void;
protected _stateRcdata(cp: number): void;
protected _stateRawtext(cp: number): void;
protected _stateScriptData(cp: number): void;
protected _statePlaintext(cp: number): void;
protected _stateTagOpen(cp: number): void;
protected _stateEndTagOpen(cp: number): void;
protected _stateTagName(cp: number): void;
protected _stateRcdataLessThanSign(cp: number): void;
protected _stateRcdataEndTagOpen(cp: number): void;
protected handleSpecialEndTag(_cp: number): boolean;
protected _stateRcdataEndTagName(cp: number): void;
protected _stateRawtextLessThanSign(cp: number): void;
protected _stateRawtextEndTagOpen(cp: number): void;
protected _stateRawtextEndTagName(cp: number): void;
protected _stateScriptDataLessThanSign(cp: number): void;
protected _stateScriptDataEndTagOpen(cp: number): void;
protected _stateScriptDataEndTagName(cp: number): void;
protected _stateScriptDataEscapeStart(cp: number): void;
protected _stateScriptDataEscapeStartDash(cp: number): void;
protected _stateScriptDataEscaped(cp: number): void;
protected _stateScriptDataEscapedDash(cp: number): void;
protected _stateScriptDataEscapedDashDash(cp: number): void;
protected _stateScriptDataEscapedLessThanSign(cp: number): void;
protected _stateScriptDataEscapedEndTagOpen(cp: number): void;
protected _stateScriptDataEscapedEndTagName(cp: number): void;
protected _stateScriptDataDoubleEscapeStart(cp: number): void;
protected _stateScriptDataDoubleEscaped(cp: number): void;
protected _stateScriptDataDoubleEscapedDash(cp: number): void;
protected _stateScriptDataDoubleEscapedDashDash(cp: number): void;
protected _stateScriptDataDoubleEscapedLessThanSign(cp: number): void;
protected _stateScriptDataDoubleEscapeEnd(cp: number): void;
protected _stateBeforeAttributeName(cp: number): void;
protected _stateAttributeName(cp: number): void;
protected _stateAfterAttributeName(cp: number): void;
protected _stateBeforeAttributeValue(cp: number): void;
protected _stateAttributeValueDoubleQuoted(cp: number): void;
protected _stateAttributeValueSingleQuoted(cp: number): void;
protected _stateAttributeValueUnquoted(cp: number): void;
protected _stateAfterAttributeValueQuoted(cp: number): void;
protected _stateSelfClosingStartTag(cp: number): void;
protected _stateBogusComment(cp: number): void;
protected _stateMarkupDeclarationOpen(cp: number): void;
protected _stateCommentStart(cp: number): void;
protected _stateCommentStartDash(cp: number): void;
protected _stateComment(cp: number): void;
protected _stateCommentLessThanSign(cp: number): void;
protected _stateCommentLessThanSignBang(cp: number): void;
protected _stateCommentLessThanSignBangDash(cp: number): void;
protected _stateCommentLessThanSignBangDashDash(cp: number): void;
protected _stateCommentEndDash(cp: number): void;
protected _stateCommentEnd(cp: number): void;
protected _stateCommentEndBang(cp: number): void;
protected _stateDoctype(cp: number): void;
protected _stateBeforeDoctypeName(cp: number): void;
protected _stateDoctypeName(cp: number): void;
protected _stateAfterDoctypeName(cp: number): void;
protected _stateAfterDoctypePublicKeyword(cp: number): void;
protected _stateBeforeDoctypePublicIdentifier(cp: number): void;
protected _stateDoctypePublicIdentifierDoubleQuoted(cp: number): void;
protected _stateDoctypePublicIdentifierSingleQuoted(cp: number): void;
protected _stateAfterDoctypePublicIdentifier(cp: number): void;
protected _stateBetweenDoctypePublicAndSystemIdentifiers(cp: number): void;
protected _stateAfterDoctypeSystemKeyword(cp: number): void;
protected _stateBeforeDoctypeSystemIdentifier(cp: number): void;
protected _stateDoctypeSystemIdentifierDoubleQuoted(cp: number): void;
protected _stateDoctypeSystemIdentifierSingleQuoted(cp: number): void;
protected _stateAfterDoctypeSystemIdentifier(cp: number): void;
protected _stateBogusDoctype(cp: number): void;
protected _stateCdataSection(cp: number): void;
protected _stateCdataSectionBracket(cp: number): void;
protected _stateCdataSectionEnd(cp: number): void;
protected _stateCharacterReference(): void;
protected _stateAmbiguousAmpersand(cp: number): void;
}
export {};
+2714
View File
@@ -0,0 +1,2714 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.Tokenizer = exports.TokenizerMode = void 0;
const preprocessor_js_1 = require("./preprocessor.js");
const unicode_js_1 = require("../common/unicode.js");
const token_js_1 = require("../common/token.js");
const decode_1 = require("entities/decode");
const error_codes_js_1 = require("../common/error-codes.js");
const html_js_1 = require("../common/html.js");
//States
var State;
(function (State) {
State[State["DATA"] = 0] = "DATA";
State[State["RCDATA"] = 1] = "RCDATA";
State[State["RAWTEXT"] = 2] = "RAWTEXT";
State[State["SCRIPT_DATA"] = 3] = "SCRIPT_DATA";
State[State["PLAINTEXT"] = 4] = "PLAINTEXT";
State[State["TAG_OPEN"] = 5] = "TAG_OPEN";
State[State["END_TAG_OPEN"] = 6] = "END_TAG_OPEN";
State[State["TAG_NAME"] = 7] = "TAG_NAME";
State[State["RCDATA_LESS_THAN_SIGN"] = 8] = "RCDATA_LESS_THAN_SIGN";
State[State["RCDATA_END_TAG_OPEN"] = 9] = "RCDATA_END_TAG_OPEN";
State[State["RCDATA_END_TAG_NAME"] = 10] = "RCDATA_END_TAG_NAME";
State[State["RAWTEXT_LESS_THAN_SIGN"] = 11] = "RAWTEXT_LESS_THAN_SIGN";
State[State["RAWTEXT_END_TAG_OPEN"] = 12] = "RAWTEXT_END_TAG_OPEN";
State[State["RAWTEXT_END_TAG_NAME"] = 13] = "RAWTEXT_END_TAG_NAME";
State[State["SCRIPT_DATA_LESS_THAN_SIGN"] = 14] = "SCRIPT_DATA_LESS_THAN_SIGN";
State[State["SCRIPT_DATA_END_TAG_OPEN"] = 15] = "SCRIPT_DATA_END_TAG_OPEN";
State[State["SCRIPT_DATA_END_TAG_NAME"] = 16] = "SCRIPT_DATA_END_TAG_NAME";
State[State["SCRIPT_DATA_ESCAPE_START"] = 17] = "SCRIPT_DATA_ESCAPE_START";
State[State["SCRIPT_DATA_ESCAPE_START_DASH"] = 18] = "SCRIPT_DATA_ESCAPE_START_DASH";
State[State["SCRIPT_DATA_ESCAPED"] = 19] = "SCRIPT_DATA_ESCAPED";
State[State["SCRIPT_DATA_ESCAPED_DASH"] = 20] = "SCRIPT_DATA_ESCAPED_DASH";
State[State["SCRIPT_DATA_ESCAPED_DASH_DASH"] = 21] = "SCRIPT_DATA_ESCAPED_DASH_DASH";
State[State["SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN"] = 22] = "SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN";
State[State["SCRIPT_DATA_ESCAPED_END_TAG_OPEN"] = 23] = "SCRIPT_DATA_ESCAPED_END_TAG_OPEN";
State[State["SCRIPT_DATA_ESCAPED_END_TAG_NAME"] = 24] = "SCRIPT_DATA_ESCAPED_END_TAG_NAME";
State[State["SCRIPT_DATA_DOUBLE_ESCAPE_START"] = 25] = "SCRIPT_DATA_DOUBLE_ESCAPE_START";
State[State["SCRIPT_DATA_DOUBLE_ESCAPED"] = 26] = "SCRIPT_DATA_DOUBLE_ESCAPED";
State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH"] = 27] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH";
State[State["SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH"] = 28] = "SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH";
State[State["SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN"] = 29] = "SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN";
State[State["SCRIPT_DATA_DOUBLE_ESCAPE_END"] = 30] = "SCRIPT_DATA_DOUBLE_ESCAPE_END";
State[State["BEFORE_ATTRIBUTE_NAME"] = 31] = "BEFORE_ATTRIBUTE_NAME";
State[State["ATTRIBUTE_NAME"] = 32] = "ATTRIBUTE_NAME";
State[State["AFTER_ATTRIBUTE_NAME"] = 33] = "AFTER_ATTRIBUTE_NAME";
State[State["BEFORE_ATTRIBUTE_VALUE"] = 34] = "BEFORE_ATTRIBUTE_VALUE";
State[State["ATTRIBUTE_VALUE_DOUBLE_QUOTED"] = 35] = "ATTRIBUTE_VALUE_DOUBLE_QUOTED";
State[State["ATTRIBUTE_VALUE_SINGLE_QUOTED"] = 36] = "ATTRIBUTE_VALUE_SINGLE_QUOTED";
State[State["ATTRIBUTE_VALUE_UNQUOTED"] = 37] = "ATTRIBUTE_VALUE_UNQUOTED";
State[State["AFTER_ATTRIBUTE_VALUE_QUOTED"] = 38] = "AFTER_ATTRIBUTE_VALUE_QUOTED";
State[State["SELF_CLOSING_START_TAG"] = 39] = "SELF_CLOSING_START_TAG";
State[State["BOGUS_COMMENT"] = 40] = "BOGUS_COMMENT";
State[State["MARKUP_DECLARATION_OPEN"] = 41] = "MARKUP_DECLARATION_OPEN";
State[State["COMMENT_START"] = 42] = "COMMENT_START";
State[State["COMMENT_START_DASH"] = 43] = "COMMENT_START_DASH";
State[State["COMMENT"] = 44] = "COMMENT";
State[State["COMMENT_LESS_THAN_SIGN"] = 45] = "COMMENT_LESS_THAN_SIGN";
State[State["COMMENT_LESS_THAN_SIGN_BANG"] = 46] = "COMMENT_LESS_THAN_SIGN_BANG";
State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH"] = 47] = "COMMENT_LESS_THAN_SIGN_BANG_DASH";
State[State["COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH"] = 48] = "COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH";
State[State["COMMENT_END_DASH"] = 49] = "COMMENT_END_DASH";
State[State["COMMENT_END"] = 50] = "COMMENT_END";
State[State["COMMENT_END_BANG"] = 51] = "COMMENT_END_BANG";
State[State["DOCTYPE"] = 52] = "DOCTYPE";
State[State["BEFORE_DOCTYPE_NAME"] = 53] = "BEFORE_DOCTYPE_NAME";
State[State["DOCTYPE_NAME"] = 54] = "DOCTYPE_NAME";
State[State["AFTER_DOCTYPE_NAME"] = 55] = "AFTER_DOCTYPE_NAME";
State[State["AFTER_DOCTYPE_PUBLIC_KEYWORD"] = 56] = "AFTER_DOCTYPE_PUBLIC_KEYWORD";
State[State["BEFORE_DOCTYPE_PUBLIC_IDENTIFIER"] = 57] = "BEFORE_DOCTYPE_PUBLIC_IDENTIFIER";
State[State["DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED"] = 58] = "DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED";
State[State["DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED"] = 59] = "DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED";
State[State["AFTER_DOCTYPE_PUBLIC_IDENTIFIER"] = 60] = "AFTER_DOCTYPE_PUBLIC_IDENTIFIER";
State[State["BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS"] = 61] = "BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS";
State[State["AFTER_DOCTYPE_SYSTEM_KEYWORD"] = 62] = "AFTER_DOCTYPE_SYSTEM_KEYWORD";
State[State["BEFORE_DOCTYPE_SYSTEM_IDENTIFIER"] = 63] = "BEFORE_DOCTYPE_SYSTEM_IDENTIFIER";
State[State["DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED"] = 64] = "DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED";
State[State["DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED"] = 65] = "DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED";
State[State["AFTER_DOCTYPE_SYSTEM_IDENTIFIER"] = 66] = "AFTER_DOCTYPE_SYSTEM_IDENTIFIER";
State[State["BOGUS_DOCTYPE"] = 67] = "BOGUS_DOCTYPE";
State[State["CDATA_SECTION"] = 68] = "CDATA_SECTION";
State[State["CDATA_SECTION_BRACKET"] = 69] = "CDATA_SECTION_BRACKET";
State[State["CDATA_SECTION_END"] = 70] = "CDATA_SECTION_END";
State[State["CHARACTER_REFERENCE"] = 71] = "CHARACTER_REFERENCE";
State[State["AMBIGUOUS_AMPERSAND"] = 72] = "AMBIGUOUS_AMPERSAND";
})(State || (State = {}));
//Tokenizer initial states for different modes
exports.TokenizerMode = {
DATA: State.DATA,
RCDATA: State.RCDATA,
RAWTEXT: State.RAWTEXT,
SCRIPT_DATA: State.SCRIPT_DATA,
PLAINTEXT: State.PLAINTEXT,
CDATA_SECTION: State.CDATA_SECTION,
};
//Utils
//OPTIMIZATION: these utility functions should not be moved out of this module. V8 Crankshaft will not inline
//this functions if they will be situated in another module due to context switch.
//Always perform inlining check before modifying this functions ('node --trace-inlining').
function isAsciiDigit(cp) {
return cp >= unicode_js_1.CODE_POINTS.DIGIT_0 && cp <= unicode_js_1.CODE_POINTS.DIGIT_9;
}
function isAsciiUpper(cp) {
return cp >= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_CAPITAL_Z;
}
function isAsciiLower(cp) {
return cp >= unicode_js_1.CODE_POINTS.LATIN_SMALL_A && cp <= unicode_js_1.CODE_POINTS.LATIN_SMALL_Z;
}
function isAsciiLetter(cp) {
return isAsciiLower(cp) || isAsciiUpper(cp);
}
function isAsciiAlphaNumeric(cp) {
return isAsciiLetter(cp) || isAsciiDigit(cp);
}
function toAsciiLower(cp) {
return cp + 32;
}
function isWhitespace(cp) {
return cp === unicode_js_1.CODE_POINTS.SPACE || cp === unicode_js_1.CODE_POINTS.LINE_FEED || cp === unicode_js_1.CODE_POINTS.TABULATION || cp === unicode_js_1.CODE_POINTS.FORM_FEED;
}
function isScriptDataDoubleEscapeSequenceEnd(cp) {
return isWhitespace(cp) || cp === unicode_js_1.CODE_POINTS.SOLIDUS || cp === unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN;
}
function getErrorForNumericCharacterReference(code) {
if (code === unicode_js_1.CODE_POINTS.NULL) {
return error_codes_js_1.ERR.nullCharacterReference;
}
else if (code > 1114111) {
return error_codes_js_1.ERR.characterReferenceOutsideUnicodeRange;
}
else if ((0, unicode_js_1.isSurrogate)(code)) {
return error_codes_js_1.ERR.surrogateCharacterReference;
}
else if ((0, unicode_js_1.isUndefinedCodePoint)(code)) {
return error_codes_js_1.ERR.noncharacterCharacterReference;
}
else if ((0, unicode_js_1.isControlCodePoint)(code) || code === unicode_js_1.CODE_POINTS.CARRIAGE_RETURN) {
return error_codes_js_1.ERR.controlCharacterReference;
}
return null;
}
//Tokenizer
class Tokenizer {
constructor(options, handler) {
this.options = options;
this.handler = handler;
this.paused = false;
/** Ensures that the parsing loop isn't run multiple times at once. */
this.inLoop = false;
/**
* Indicates that the current adjusted node exists, is not an element in the HTML namespace,
* and that it is not an integration point for either MathML or HTML.
*
* @see {@link https://html.spec.whatwg.org/multipage/parsing.html#tree-construction}
*/
this.inForeignNode = false;
this.lastStartTagName = '';
this.active = false;
this.state = State.DATA;
this.returnState = State.DATA;
this.entityStartPos = 0;
this.consumedAfterSnapshot = -1;
this.currentCharacterToken = null;
this.currentToken = null;
this.currentAttr = { name: '', value: '' };
this.preprocessor = new preprocessor_js_1.Preprocessor(handler);
this.currentLocation = this.getCurrentLocation(-1);
this.entityDecoder = new decode_1.EntityDecoder(decode_1.htmlDecodeTree, (cp, consumed) => {
// Note: Set `pos` _before_ flushing, as flushing might drop
// the current chunk and invalidate `entityStartPos`.
this.preprocessor.pos = this.entityStartPos + consumed - 1;
this._flushCodePointConsumedAsCharacterReference(cp);
}, handler.onParseError
? {
missingSemicolonAfterCharacterReference: () => {
this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference, 1);
},
absenceOfDigitsInNumericCharacterReference: (consumed) => {
this._err(error_codes_js_1.ERR.absenceOfDigitsInNumericCharacterReference, this.entityStartPos - this.preprocessor.pos + consumed);
},
validateNumericCharacterReference: (code) => {
const error = getErrorForNumericCharacterReference(code);
if (error)
this._err(error, 1);
},
}
: undefined);
}
//Errors
_err(code, cpOffset = 0) {
var _a, _b;
(_b = (_a = this.handler).onParseError) === null || _b === void 0 ? void 0 : _b.call(_a, this.preprocessor.getError(code, cpOffset));
}
// NOTE: `offset` may never run across line boundaries.
getCurrentLocation(offset) {
if (!this.options.sourceCodeLocationInfo) {
return null;
}
return {
startLine: this.preprocessor.line,
startCol: this.preprocessor.col - offset,
startOffset: this.preprocessor.offset - offset,
endLine: -1,
endCol: -1,
endOffset: -1,
};
}
_runParsingLoop() {
if (this.inLoop)
return;
this.inLoop = true;
while (this.active && !this.paused) {
this.consumedAfterSnapshot = 0;
const cp = this._consume();
if (!this._ensureHibernation()) {
this._callState(cp);
}
}
this.inLoop = false;
}
//API
pause() {
this.paused = true;
}
resume(writeCallback) {
if (!this.paused) {
throw new Error('Parser was already resumed');
}
this.paused = false;
// Necessary for synchronous resume.
if (this.inLoop)
return;
this._runParsingLoop();
if (!this.paused) {
writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
}
}
write(chunk, isLastChunk, writeCallback) {
this.active = true;
this.preprocessor.write(chunk, isLastChunk);
this._runParsingLoop();
if (!this.paused) {
writeCallback === null || writeCallback === void 0 ? void 0 : writeCallback();
}
}
insertHtmlAtCurrentPos(chunk) {
this.active = true;
this.preprocessor.insertHtmlAtCurrentPos(chunk);
this._runParsingLoop();
}
//Hibernation
_ensureHibernation() {
if (this.preprocessor.endOfChunkHit) {
this.preprocessor.retreat(this.consumedAfterSnapshot);
this.consumedAfterSnapshot = 0;
this.active = false;
return true;
}
return false;
}
//Consumption
_consume() {
this.consumedAfterSnapshot++;
return this.preprocessor.advance();
}
_advanceBy(count) {
this.consumedAfterSnapshot += count;
for (let i = 0; i < count; i++) {
this.preprocessor.advance();
}
}
_consumeSequenceIfMatch(pattern, caseSensitive) {
if (this.preprocessor.startsWith(pattern, caseSensitive)) {
// We will already have consumed one character before calling this method.
this._advanceBy(pattern.length - 1);
return true;
}
return false;
}
//Token creation
_createStartTagToken() {
this.currentToken = {
type: token_js_1.TokenType.START_TAG,
tagName: '',
tagID: html_js_1.TAG_ID.UNKNOWN,
selfClosing: false,
ackSelfClosing: false,
attrs: [],
location: this.getCurrentLocation(1),
};
}
_createEndTagToken() {
this.currentToken = {
type: token_js_1.TokenType.END_TAG,
tagName: '',
tagID: html_js_1.TAG_ID.UNKNOWN,
selfClosing: false,
ackSelfClosing: false,
attrs: [],
location: this.getCurrentLocation(2),
};
}
_createCommentToken(offset) {
this.currentToken = {
type: token_js_1.TokenType.COMMENT,
data: '',
location: this.getCurrentLocation(offset),
};
}
_createDoctypeToken(initialName) {
this.currentToken = {
type: token_js_1.TokenType.DOCTYPE,
name: initialName,
forceQuirks: false,
publicId: null,
systemId: null,
location: this.currentLocation,
};
}
_createCharacterToken(type, chars) {
this.currentCharacterToken = {
type,
chars,
location: this.currentLocation,
};
}
//Tag attributes
_createAttr(attrNameFirstCh) {
this.currentAttr = {
name: attrNameFirstCh,
value: '',
};
this.currentLocation = this.getCurrentLocation(0);
}
_leaveAttrName() {
var _a;
var _b;
const token = this.currentToken;
if ((0, token_js_1.getTokenAttr)(token, this.currentAttr.name) === null) {
token.attrs.push(this.currentAttr);
if (token.location && this.currentLocation) {
const attrLocations = ((_a = (_b = token.location).attrs) !== null && _a !== void 0 ? _a : (_b.attrs = Object.create(null)));
attrLocations[this.currentAttr.name] = this.currentLocation;
// Set end location
this._leaveAttrValue();
}
}
else {
this._err(error_codes_js_1.ERR.duplicateAttribute);
}
}
_leaveAttrValue() {
if (this.currentLocation) {
this.currentLocation.endLine = this.preprocessor.line;
this.currentLocation.endCol = this.preprocessor.col;
this.currentLocation.endOffset = this.preprocessor.offset;
}
}
//Token emission
prepareToken(ct) {
this._emitCurrentCharacterToken(ct.location);
this.currentToken = null;
if (ct.location) {
ct.location.endLine = this.preprocessor.line;
ct.location.endCol = this.preprocessor.col + 1;
ct.location.endOffset = this.preprocessor.offset + 1;
}
this.currentLocation = this.getCurrentLocation(-1);
}
emitCurrentTagToken() {
const ct = this.currentToken;
this.prepareToken(ct);
ct.tagID = (0, html_js_1.getTagID)(ct.tagName);
if (ct.type === token_js_1.TokenType.START_TAG) {
this.lastStartTagName = ct.tagName;
this.handler.onStartTag(ct);
}
else {
if (ct.attrs.length > 0) {
this._err(error_codes_js_1.ERR.endTagWithAttributes);
}
if (ct.selfClosing) {
this._err(error_codes_js_1.ERR.endTagWithTrailingSolidus);
}
this.handler.onEndTag(ct);
}
this.preprocessor.dropParsedChunk();
}
emitCurrentComment(ct) {
this.prepareToken(ct);
this.handler.onComment(ct);
this.preprocessor.dropParsedChunk();
}
emitCurrentDoctype(ct) {
this.prepareToken(ct);
this.handler.onDoctype(ct);
this.preprocessor.dropParsedChunk();
}
_emitCurrentCharacterToken(nextLocation) {
if (this.currentCharacterToken) {
//NOTE: if we have a pending character token, make it's end location equal to the
//current token's start location.
if (nextLocation && this.currentCharacterToken.location) {
this.currentCharacterToken.location.endLine = nextLocation.startLine;
this.currentCharacterToken.location.endCol = nextLocation.startCol;
this.currentCharacterToken.location.endOffset = nextLocation.startOffset;
}
switch (this.currentCharacterToken.type) {
case token_js_1.TokenType.CHARACTER: {
this.handler.onCharacter(this.currentCharacterToken);
break;
}
case token_js_1.TokenType.NULL_CHARACTER: {
this.handler.onNullCharacter(this.currentCharacterToken);
break;
}
case token_js_1.TokenType.WHITESPACE_CHARACTER: {
this.handler.onWhitespaceCharacter(this.currentCharacterToken);
break;
}
}
this.currentCharacterToken = null;
}
}
_emitEOFToken() {
const location = this.getCurrentLocation(0);
if (location) {
location.endLine = location.startLine;
location.endCol = location.startCol;
location.endOffset = location.startOffset;
}
this._emitCurrentCharacterToken(location);
this.handler.onEof({ type: token_js_1.TokenType.EOF, location });
this.active = false;
}
//Characters emission
//OPTIMIZATION: The specification uses only one type of character token (one token per character).
//This causes a huge memory overhead and a lot of unnecessary parser loops. parse5 uses 3 groups of characters.
//If we have a sequence of characters that belong to the same group, the parser can process it
//as a single solid character token.
//So, there are 3 types of character tokens in parse5:
//1)TokenType.NULL_CHARACTER - \u0000-character sequences (e.g. '\u0000\u0000\u0000')
//2)TokenType.WHITESPACE_CHARACTER - any whitespace/new-line character sequences (e.g. '\n \r\t \f')
//3)TokenType.CHARACTER - any character sequence which don't belong to groups 1 and 2 (e.g. 'abcdef1234@@#$%^')
_appendCharToCurrentCharacterToken(type, ch) {
if (this.currentCharacterToken) {
if (this.currentCharacterToken.type === type) {
this.currentCharacterToken.chars += ch;
return;
}
else {
this.currentLocation = this.getCurrentLocation(0);
this._emitCurrentCharacterToken(this.currentLocation);
this.preprocessor.dropParsedChunk();
}
}
this._createCharacterToken(type, ch);
}
_emitCodePoint(cp) {
const type = isWhitespace(cp)
? token_js_1.TokenType.WHITESPACE_CHARACTER
: cp === unicode_js_1.CODE_POINTS.NULL
? token_js_1.TokenType.NULL_CHARACTER
: token_js_1.TokenType.CHARACTER;
this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
}
//NOTE: used when we emit characters explicitly.
//This is always for non-whitespace and non-null characters, which allows us to avoid additional checks.
_emitChars(ch) {
this._appendCharToCurrentCharacterToken(token_js_1.TokenType.CHARACTER, ch);
}
// Character reference helpers
_startCharacterReference() {
this.returnState = this.state;
this.state = State.CHARACTER_REFERENCE;
this.entityStartPos = this.preprocessor.pos;
this.entityDecoder.startEntity(this._isCharacterReferenceInAttribute() ? decode_1.DecodingMode.Attribute : decode_1.DecodingMode.Legacy);
}
_isCharacterReferenceInAttribute() {
return (this.returnState === State.ATTRIBUTE_VALUE_DOUBLE_QUOTED ||
this.returnState === State.ATTRIBUTE_VALUE_SINGLE_QUOTED ||
this.returnState === State.ATTRIBUTE_VALUE_UNQUOTED);
}
_flushCodePointConsumedAsCharacterReference(cp) {
if (this._isCharacterReferenceInAttribute()) {
this.currentAttr.value += String.fromCodePoint(cp);
}
else {
this._emitCodePoint(cp);
}
}
// Calling states this way turns out to be much faster than any other approach.
_callState(cp) {
switch (this.state) {
case State.DATA: {
this._stateData(cp);
break;
}
case State.RCDATA: {
this._stateRcdata(cp);
break;
}
case State.RAWTEXT: {
this._stateRawtext(cp);
break;
}
case State.SCRIPT_DATA: {
this._stateScriptData(cp);
break;
}
case State.PLAINTEXT: {
this._statePlaintext(cp);
break;
}
case State.TAG_OPEN: {
this._stateTagOpen(cp);
break;
}
case State.END_TAG_OPEN: {
this._stateEndTagOpen(cp);
break;
}
case State.TAG_NAME: {
this._stateTagName(cp);
break;
}
case State.RCDATA_LESS_THAN_SIGN: {
this._stateRcdataLessThanSign(cp);
break;
}
case State.RCDATA_END_TAG_OPEN: {
this._stateRcdataEndTagOpen(cp);
break;
}
case State.RCDATA_END_TAG_NAME: {
this._stateRcdataEndTagName(cp);
break;
}
case State.RAWTEXT_LESS_THAN_SIGN: {
this._stateRawtextLessThanSign(cp);
break;
}
case State.RAWTEXT_END_TAG_OPEN: {
this._stateRawtextEndTagOpen(cp);
break;
}
case State.RAWTEXT_END_TAG_NAME: {
this._stateRawtextEndTagName(cp);
break;
}
case State.SCRIPT_DATA_LESS_THAN_SIGN: {
this._stateScriptDataLessThanSign(cp);
break;
}
case State.SCRIPT_DATA_END_TAG_OPEN: {
this._stateScriptDataEndTagOpen(cp);
break;
}
case State.SCRIPT_DATA_END_TAG_NAME: {
this._stateScriptDataEndTagName(cp);
break;
}
case State.SCRIPT_DATA_ESCAPE_START: {
this._stateScriptDataEscapeStart(cp);
break;
}
case State.SCRIPT_DATA_ESCAPE_START_DASH: {
this._stateScriptDataEscapeStartDash(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED: {
this._stateScriptDataEscaped(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED_DASH: {
this._stateScriptDataEscapedDash(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED_DASH_DASH: {
this._stateScriptDataEscapedDashDash(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN: {
this._stateScriptDataEscapedLessThanSign(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN: {
this._stateScriptDataEscapedEndTagOpen(cp);
break;
}
case State.SCRIPT_DATA_ESCAPED_END_TAG_NAME: {
this._stateScriptDataEscapedEndTagName(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPE_START: {
this._stateScriptDataDoubleEscapeStart(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPED: {
this._stateScriptDataDoubleEscaped(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH: {
this._stateScriptDataDoubleEscapedDash(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH: {
this._stateScriptDataDoubleEscapedDashDash(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN: {
this._stateScriptDataDoubleEscapedLessThanSign(cp);
break;
}
case State.SCRIPT_DATA_DOUBLE_ESCAPE_END: {
this._stateScriptDataDoubleEscapeEnd(cp);
break;
}
case State.BEFORE_ATTRIBUTE_NAME: {
this._stateBeforeAttributeName(cp);
break;
}
case State.ATTRIBUTE_NAME: {
this._stateAttributeName(cp);
break;
}
case State.AFTER_ATTRIBUTE_NAME: {
this._stateAfterAttributeName(cp);
break;
}
case State.BEFORE_ATTRIBUTE_VALUE: {
this._stateBeforeAttributeValue(cp);
break;
}
case State.ATTRIBUTE_VALUE_DOUBLE_QUOTED: {
this._stateAttributeValueDoubleQuoted(cp);
break;
}
case State.ATTRIBUTE_VALUE_SINGLE_QUOTED: {
this._stateAttributeValueSingleQuoted(cp);
break;
}
case State.ATTRIBUTE_VALUE_UNQUOTED: {
this._stateAttributeValueUnquoted(cp);
break;
}
case State.AFTER_ATTRIBUTE_VALUE_QUOTED: {
this._stateAfterAttributeValueQuoted(cp);
break;
}
case State.SELF_CLOSING_START_TAG: {
this._stateSelfClosingStartTag(cp);
break;
}
case State.BOGUS_COMMENT: {
this._stateBogusComment(cp);
break;
}
case State.MARKUP_DECLARATION_OPEN: {
this._stateMarkupDeclarationOpen(cp);
break;
}
case State.COMMENT_START: {
this._stateCommentStart(cp);
break;
}
case State.COMMENT_START_DASH: {
this._stateCommentStartDash(cp);
break;
}
case State.COMMENT: {
this._stateComment(cp);
break;
}
case State.COMMENT_LESS_THAN_SIGN: {
this._stateCommentLessThanSign(cp);
break;
}
case State.COMMENT_LESS_THAN_SIGN_BANG: {
this._stateCommentLessThanSignBang(cp);
break;
}
case State.COMMENT_LESS_THAN_SIGN_BANG_DASH: {
this._stateCommentLessThanSignBangDash(cp);
break;
}
case State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH: {
this._stateCommentLessThanSignBangDashDash(cp);
break;
}
case State.COMMENT_END_DASH: {
this._stateCommentEndDash(cp);
break;
}
case State.COMMENT_END: {
this._stateCommentEnd(cp);
break;
}
case State.COMMENT_END_BANG: {
this._stateCommentEndBang(cp);
break;
}
case State.DOCTYPE: {
this._stateDoctype(cp);
break;
}
case State.BEFORE_DOCTYPE_NAME: {
this._stateBeforeDoctypeName(cp);
break;
}
case State.DOCTYPE_NAME: {
this._stateDoctypeName(cp);
break;
}
case State.AFTER_DOCTYPE_NAME: {
this._stateAfterDoctypeName(cp);
break;
}
case State.AFTER_DOCTYPE_PUBLIC_KEYWORD: {
this._stateAfterDoctypePublicKeyword(cp);
break;
}
case State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER: {
this._stateBeforeDoctypePublicIdentifier(cp);
break;
}
case State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED: {
this._stateDoctypePublicIdentifierDoubleQuoted(cp);
break;
}
case State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED: {
this._stateDoctypePublicIdentifierSingleQuoted(cp);
break;
}
case State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER: {
this._stateAfterDoctypePublicIdentifier(cp);
break;
}
case State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS: {
this._stateBetweenDoctypePublicAndSystemIdentifiers(cp);
break;
}
case State.AFTER_DOCTYPE_SYSTEM_KEYWORD: {
this._stateAfterDoctypeSystemKeyword(cp);
break;
}
case State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER: {
this._stateBeforeDoctypeSystemIdentifier(cp);
break;
}
case State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED: {
this._stateDoctypeSystemIdentifierDoubleQuoted(cp);
break;
}
case State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED: {
this._stateDoctypeSystemIdentifierSingleQuoted(cp);
break;
}
case State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER: {
this._stateAfterDoctypeSystemIdentifier(cp);
break;
}
case State.BOGUS_DOCTYPE: {
this._stateBogusDoctype(cp);
break;
}
case State.CDATA_SECTION: {
this._stateCdataSection(cp);
break;
}
case State.CDATA_SECTION_BRACKET: {
this._stateCdataSectionBracket(cp);
break;
}
case State.CDATA_SECTION_END: {
this._stateCdataSectionEnd(cp);
break;
}
case State.CHARACTER_REFERENCE: {
this._stateCharacterReference();
break;
}
case State.AMBIGUOUS_AMPERSAND: {
this._stateAmbiguousAmpersand(cp);
break;
}
default: {
throw new Error('Unknown state');
}
}
}
// State machine
// Data state
//------------------------------------------------------------------
_stateData(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.TAG_OPEN;
break;
}
case unicode_js_1.CODE_POINTS.AMPERSAND: {
this._startCharacterReference();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitCodePoint(cp);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// RCDATA state
//------------------------------------------------------------------
_stateRcdata(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.AMPERSAND: {
this._startCharacterReference();
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.RCDATA_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// RAWTEXT state
//------------------------------------------------------------------
_stateRawtext(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.RAWTEXT_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// Script data state
//------------------------------------------------------------------
_stateScriptData(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// PLAINTEXT state
//------------------------------------------------------------------
_statePlaintext(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// Tag open state
//------------------------------------------------------------------
_stateTagOpen(cp) {
if (isAsciiLetter(cp)) {
this._createStartTagToken();
this.state = State.TAG_NAME;
this._stateTagName(cp);
}
else
switch (cp) {
case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
this.state = State.MARKUP_DECLARATION_OPEN;
break;
}
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this.state = State.END_TAG_OPEN;
break;
}
case unicode_js_1.CODE_POINTS.QUESTION_MARK: {
this._err(error_codes_js_1.ERR.unexpectedQuestionMarkInsteadOfTagName);
this._createCommentToken(1);
this.state = State.BOGUS_COMMENT;
this._stateBogusComment(cp);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofBeforeTagName);
this._emitChars('<');
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.invalidFirstCharacterOfTagName);
this._emitChars('<');
this.state = State.DATA;
this._stateData(cp);
}
}
}
// End tag open state
//------------------------------------------------------------------
_stateEndTagOpen(cp) {
if (isAsciiLetter(cp)) {
this._createEndTagToken();
this.state = State.TAG_NAME;
this._stateTagName(cp);
}
else
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingEndTagName);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofBeforeTagName);
this._emitChars('</');
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.invalidFirstCharacterOfTagName);
this._createCommentToken(2);
this.state = State.BOGUS_COMMENT;
this._stateBogusComment(cp);
}
}
}
// Tag name state
//------------------------------------------------------------------
_stateTagName(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.BEFORE_ATTRIBUTE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this.state = State.SELF_CLOSING_START_TAG;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.tagName += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
token.tagName += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
}
}
}
// RCDATA less-than sign state
//------------------------------------------------------------------
_stateRcdataLessThanSign(cp) {
if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
this.state = State.RCDATA_END_TAG_OPEN;
}
else {
this._emitChars('<');
this.state = State.RCDATA;
this._stateRcdata(cp);
}
}
// RCDATA end tag open state
//------------------------------------------------------------------
_stateRcdataEndTagOpen(cp) {
if (isAsciiLetter(cp)) {
this.state = State.RCDATA_END_TAG_NAME;
this._stateRcdataEndTagName(cp);
}
else {
this._emitChars('</');
this.state = State.RCDATA;
this._stateRcdata(cp);
}
}
handleSpecialEndTag(_cp) {
if (!this.preprocessor.startsWith(this.lastStartTagName, false)) {
return !this._ensureHibernation();
}
this._createEndTagToken();
const token = this.currentToken;
token.tagName = this.lastStartTagName;
const cp = this.preprocessor.peek(this.lastStartTagName.length);
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this._advanceBy(this.lastStartTagName.length);
this.state = State.BEFORE_ATTRIBUTE_NAME;
return false;
}
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this._advanceBy(this.lastStartTagName.length);
this.state = State.SELF_CLOSING_START_TAG;
return false;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._advanceBy(this.lastStartTagName.length);
this.emitCurrentTagToken();
this.state = State.DATA;
return false;
}
default: {
return !this._ensureHibernation();
}
}
}
// RCDATA end tag name state
//------------------------------------------------------------------
_stateRcdataEndTagName(cp) {
if (this.handleSpecialEndTag(cp)) {
this._emitChars('</');
this.state = State.RCDATA;
this._stateRcdata(cp);
}
}
// RAWTEXT less-than sign state
//------------------------------------------------------------------
_stateRawtextLessThanSign(cp) {
if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
this.state = State.RAWTEXT_END_TAG_OPEN;
}
else {
this._emitChars('<');
this.state = State.RAWTEXT;
this._stateRawtext(cp);
}
}
// RAWTEXT end tag open state
//------------------------------------------------------------------
_stateRawtextEndTagOpen(cp) {
if (isAsciiLetter(cp)) {
this.state = State.RAWTEXT_END_TAG_NAME;
this._stateRawtextEndTagName(cp);
}
else {
this._emitChars('</');
this.state = State.RAWTEXT;
this._stateRawtext(cp);
}
}
// RAWTEXT end tag name state
//------------------------------------------------------------------
_stateRawtextEndTagName(cp) {
if (this.handleSpecialEndTag(cp)) {
this._emitChars('</');
this.state = State.RAWTEXT;
this._stateRawtext(cp);
}
}
// Script data less-than sign state
//------------------------------------------------------------------
_stateScriptDataLessThanSign(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this.state = State.SCRIPT_DATA_END_TAG_OPEN;
break;
}
case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
this.state = State.SCRIPT_DATA_ESCAPE_START;
this._emitChars('<!');
break;
}
default: {
this._emitChars('<');
this.state = State.SCRIPT_DATA;
this._stateScriptData(cp);
}
}
}
// Script data end tag open state
//------------------------------------------------------------------
_stateScriptDataEndTagOpen(cp) {
if (isAsciiLetter(cp)) {
this.state = State.SCRIPT_DATA_END_TAG_NAME;
this._stateScriptDataEndTagName(cp);
}
else {
this._emitChars('</');
this.state = State.SCRIPT_DATA;
this._stateScriptData(cp);
}
}
// Script data end tag name state
//------------------------------------------------------------------
_stateScriptDataEndTagName(cp) {
if (this.handleSpecialEndTag(cp)) {
this._emitChars('</');
this.state = State.SCRIPT_DATA;
this._stateScriptData(cp);
}
}
// Script data escape start state
//------------------------------------------------------------------
_stateScriptDataEscapeStart(cp) {
if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
this.state = State.SCRIPT_DATA_ESCAPE_START_DASH;
this._emitChars('-');
}
else {
this.state = State.SCRIPT_DATA;
this._stateScriptData(cp);
}
}
// Script data escape start dash state
//------------------------------------------------------------------
_stateScriptDataEscapeStartDash(cp) {
if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
this._emitChars('-');
}
else {
this.state = State.SCRIPT_DATA;
this._stateScriptData(cp);
}
}
// Script data escaped state
//------------------------------------------------------------------
_stateScriptDataEscaped(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.SCRIPT_DATA_ESCAPED_DASH;
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// Script data escaped dash state
//------------------------------------------------------------------
_stateScriptDataEscapedDash(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.SCRIPT_DATA_ESCAPED_DASH_DASH;
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.state = State.SCRIPT_DATA_ESCAPED;
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this.state = State.SCRIPT_DATA_ESCAPED;
this._emitCodePoint(cp);
}
}
}
// Script data escaped dash dash state
//------------------------------------------------------------------
_stateScriptDataEscapedDashDash(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.SCRIPT_DATA;
this._emitChars('>');
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.state = State.SCRIPT_DATA_ESCAPED;
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this.state = State.SCRIPT_DATA_ESCAPED;
this._emitCodePoint(cp);
}
}
}
// Script data escaped less-than sign state
//------------------------------------------------------------------
_stateScriptDataEscapedLessThanSign(cp) {
if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_OPEN;
}
else if (isAsciiLetter(cp)) {
this._emitChars('<');
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_START;
this._stateScriptDataDoubleEscapeStart(cp);
}
else {
this._emitChars('<');
this.state = State.SCRIPT_DATA_ESCAPED;
this._stateScriptDataEscaped(cp);
}
}
// Script data escaped end tag open state
//------------------------------------------------------------------
_stateScriptDataEscapedEndTagOpen(cp) {
if (isAsciiLetter(cp)) {
this.state = State.SCRIPT_DATA_ESCAPED_END_TAG_NAME;
this._stateScriptDataEscapedEndTagName(cp);
}
else {
this._emitChars('</');
this.state = State.SCRIPT_DATA_ESCAPED;
this._stateScriptDataEscaped(cp);
}
}
// Script data escaped end tag name state
//------------------------------------------------------------------
_stateScriptDataEscapedEndTagName(cp) {
if (this.handleSpecialEndTag(cp)) {
this._emitChars('</');
this.state = State.SCRIPT_DATA_ESCAPED;
this._stateScriptDataEscaped(cp);
}
}
// Script data double escape start state
//------------------------------------------------------------------
_stateScriptDataDoubleEscapeStart(cp) {
if (this.preprocessor.startsWith(unicode_js_1.SEQUENCES.SCRIPT, false) &&
isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek(unicode_js_1.SEQUENCES.SCRIPT.length))) {
this._emitCodePoint(cp);
for (let i = 0; i < unicode_js_1.SEQUENCES.SCRIPT.length; i++) {
this._emitCodePoint(this._consume());
}
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
}
else if (!this._ensureHibernation()) {
this.state = State.SCRIPT_DATA_ESCAPED;
this._stateScriptDataEscaped(cp);
}
}
// Script data double escaped state
//------------------------------------------------------------------
_stateScriptDataDoubleEscaped(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH;
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
this._emitChars('<');
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// Script data double escaped dash state
//------------------------------------------------------------------
_stateScriptDataDoubleEscapedDash(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH;
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
this._emitChars('<');
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._emitCodePoint(cp);
}
}
}
// Script data double escaped dash dash state
//------------------------------------------------------------------
_stateScriptDataDoubleEscapedDashDash(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this._emitChars('-');
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN;
this._emitChars('<');
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.SCRIPT_DATA;
this._emitChars('>');
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._emitChars(unicode_js_1.REPLACEMENT_CHARACTER);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInScriptHtmlCommentLikeText);
this._emitEOFToken();
break;
}
default: {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._emitCodePoint(cp);
}
}
}
// Script data double escaped less-than sign state
//------------------------------------------------------------------
_stateScriptDataDoubleEscapedLessThanSign(cp) {
if (cp === unicode_js_1.CODE_POINTS.SOLIDUS) {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPE_END;
this._emitChars('/');
}
else {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._stateScriptDataDoubleEscaped(cp);
}
}
// Script data double escape end state
//------------------------------------------------------------------
_stateScriptDataDoubleEscapeEnd(cp) {
if (this.preprocessor.startsWith(unicode_js_1.SEQUENCES.SCRIPT, false) &&
isScriptDataDoubleEscapeSequenceEnd(this.preprocessor.peek(unicode_js_1.SEQUENCES.SCRIPT.length))) {
this._emitCodePoint(cp);
for (let i = 0; i < unicode_js_1.SEQUENCES.SCRIPT.length; i++) {
this._emitCodePoint(this._consume());
}
this.state = State.SCRIPT_DATA_ESCAPED;
}
else if (!this._ensureHibernation()) {
this.state = State.SCRIPT_DATA_DOUBLE_ESCAPED;
this._stateScriptDataDoubleEscaped(cp);
}
}
// Before attribute name state
//------------------------------------------------------------------
_stateBeforeAttributeName(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.SOLIDUS:
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN:
case unicode_js_1.CODE_POINTS.EOF: {
this.state = State.AFTER_ATTRIBUTE_NAME;
this._stateAfterAttributeName(cp);
break;
}
case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
this._err(error_codes_js_1.ERR.unexpectedEqualsSignBeforeAttributeName);
this._createAttr('=');
this.state = State.ATTRIBUTE_NAME;
break;
}
default: {
this._createAttr('');
this.state = State.ATTRIBUTE_NAME;
this._stateAttributeName(cp);
}
}
}
// Attribute name state
//------------------------------------------------------------------
_stateAttributeName(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED:
case unicode_js_1.CODE_POINTS.SOLIDUS:
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN:
case unicode_js_1.CODE_POINTS.EOF: {
this._leaveAttrName();
this.state = State.AFTER_ATTRIBUTE_NAME;
this._stateAfterAttributeName(cp);
break;
}
case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
this._leaveAttrName();
this.state = State.BEFORE_ATTRIBUTE_VALUE;
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK:
case unicode_js_1.CODE_POINTS.APOSTROPHE:
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
this._err(error_codes_js_1.ERR.unexpectedCharacterInAttributeName);
this.currentAttr.name += String.fromCodePoint(cp);
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.currentAttr.name += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
default: {
this.currentAttr.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
}
}
}
// After attribute name state
//------------------------------------------------------------------
_stateAfterAttributeName(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this.state = State.SELF_CLOSING_START_TAG;
break;
}
case unicode_js_1.CODE_POINTS.EQUALS_SIGN: {
this.state = State.BEFORE_ATTRIBUTE_VALUE;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this._createAttr('');
this.state = State.ATTRIBUTE_NAME;
this._stateAttributeName(cp);
}
}
}
// Before attribute value state
//------------------------------------------------------------------
_stateBeforeAttributeValue(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this.state = State.ATTRIBUTE_VALUE_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this.state = State.ATTRIBUTE_VALUE_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingAttributeValue);
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
default: {
this.state = State.ATTRIBUTE_VALUE_UNQUOTED;
this._stateAttributeValueUnquoted(cp);
}
}
}
// Attribute value (double-quoted) state
//------------------------------------------------------------------
_stateAttributeValueDoubleQuoted(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.AMPERSAND: {
this._startCharacterReference();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this.currentAttr.value += String.fromCodePoint(cp);
}
}
}
// Attribute value (single-quoted) state
//------------------------------------------------------------------
_stateAttributeValueSingleQuoted(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this.state = State.AFTER_ATTRIBUTE_VALUE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.AMPERSAND: {
this._startCharacterReference();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this.currentAttr.value += String.fromCodePoint(cp);
}
}
}
// Attribute value (unquoted) state
//------------------------------------------------------------------
_stateAttributeValueUnquoted(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this._leaveAttrValue();
this.state = State.BEFORE_ATTRIBUTE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.AMPERSAND: {
this._startCharacterReference();
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._leaveAttrValue();
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this.currentAttr.value += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK:
case unicode_js_1.CODE_POINTS.APOSTROPHE:
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN:
case unicode_js_1.CODE_POINTS.EQUALS_SIGN:
case unicode_js_1.CODE_POINTS.GRAVE_ACCENT: {
this._err(error_codes_js_1.ERR.unexpectedCharacterInUnquotedAttributeValue);
this.currentAttr.value += String.fromCodePoint(cp);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this.currentAttr.value += String.fromCodePoint(cp);
}
}
}
// After attribute value (quoted) state
//------------------------------------------------------------------
_stateAfterAttributeValueQuoted(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this._leaveAttrValue();
this.state = State.BEFORE_ATTRIBUTE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.SOLIDUS: {
this._leaveAttrValue();
this.state = State.SELF_CLOSING_START_TAG;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._leaveAttrValue();
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingWhitespaceBetweenAttributes);
this.state = State.BEFORE_ATTRIBUTE_NAME;
this._stateBeforeAttributeName(cp);
}
}
}
// Self-closing start tag state
//------------------------------------------------------------------
_stateSelfClosingStartTag(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
const token = this.currentToken;
token.selfClosing = true;
this.state = State.DATA;
this.emitCurrentTagToken();
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInTag);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.unexpectedSolidusInTag);
this.state = State.BEFORE_ATTRIBUTE_NAME;
this._stateBeforeAttributeName(cp);
}
}
}
// Bogus comment state
//------------------------------------------------------------------
_stateBogusComment(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentComment(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.data += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
default: {
token.data += String.fromCodePoint(cp);
}
}
}
// Markup declaration open state
//------------------------------------------------------------------
_stateMarkupDeclarationOpen(cp) {
if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.DASH_DASH, true)) {
this._createCommentToken(unicode_js_1.SEQUENCES.DASH_DASH.length + 1);
this.state = State.COMMENT_START;
}
else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.DOCTYPE, false)) {
// NOTE: Doctypes tokens are created without fixed offsets. We keep track of the moment a doctype *might* start here.
this.currentLocation = this.getCurrentLocation(unicode_js_1.SEQUENCES.DOCTYPE.length + 1);
this.state = State.DOCTYPE;
}
else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.CDATA_START, true)) {
if (this.inForeignNode) {
this.state = State.CDATA_SECTION;
}
else {
this._err(error_codes_js_1.ERR.cdataInHtmlContent);
this._createCommentToken(unicode_js_1.SEQUENCES.CDATA_START.length + 1);
this.currentToken.data = '[CDATA[';
this.state = State.BOGUS_COMMENT;
}
}
//NOTE: Sequence lookups can be abrupted by hibernation. In that case, lookup
//results are no longer valid and we will need to start over.
else if (!this._ensureHibernation()) {
this._err(error_codes_js_1.ERR.incorrectlyOpenedComment);
this._createCommentToken(2);
this.state = State.BOGUS_COMMENT;
this._stateBogusComment(cp);
}
}
// Comment start state
//------------------------------------------------------------------
_stateCommentStart(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.COMMENT_START_DASH;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptClosingOfEmptyComment);
this.state = State.DATA;
const token = this.currentToken;
this.emitCurrentComment(token);
break;
}
default: {
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// Comment start dash state
//------------------------------------------------------------------
_stateCommentStartDash(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.COMMENT_END;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptClosingOfEmptyComment);
this.state = State.DATA;
this.emitCurrentComment(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInComment);
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
default: {
token.data += '-';
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// Comment state
//------------------------------------------------------------------
_stateComment(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.COMMENT_END_DASH;
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
token.data += '<';
this.state = State.COMMENT_LESS_THAN_SIGN;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.data += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInComment);
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
default: {
token.data += String.fromCodePoint(cp);
}
}
}
// Comment less-than sign state
//------------------------------------------------------------------
_stateCommentLessThanSign(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
token.data += '!';
this.state = State.COMMENT_LESS_THAN_SIGN_BANG;
break;
}
case unicode_js_1.CODE_POINTS.LESS_THAN_SIGN: {
token.data += '<';
break;
}
default: {
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// Comment less-than sign bang state
//------------------------------------------------------------------
_stateCommentLessThanSignBang(cp) {
if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH;
}
else {
this.state = State.COMMENT;
this._stateComment(cp);
}
}
// Comment less-than sign bang dash state
//------------------------------------------------------------------
_stateCommentLessThanSignBangDash(cp) {
if (cp === unicode_js_1.CODE_POINTS.HYPHEN_MINUS) {
this.state = State.COMMENT_LESS_THAN_SIGN_BANG_DASH_DASH;
}
else {
this.state = State.COMMENT_END_DASH;
this._stateCommentEndDash(cp);
}
}
// Comment less-than sign bang dash dash state
//------------------------------------------------------------------
_stateCommentLessThanSignBangDashDash(cp) {
if (cp !== unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN && cp !== unicode_js_1.CODE_POINTS.EOF) {
this._err(error_codes_js_1.ERR.nestedComment);
}
this.state = State.COMMENT_END;
this._stateCommentEnd(cp);
}
// Comment end dash state
//------------------------------------------------------------------
_stateCommentEndDash(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
this.state = State.COMMENT_END;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInComment);
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
default: {
token.data += '-';
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// Comment end state
//------------------------------------------------------------------
_stateCommentEnd(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentComment(token);
break;
}
case unicode_js_1.CODE_POINTS.EXCLAMATION_MARK: {
this.state = State.COMMENT_END_BANG;
break;
}
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
token.data += '-';
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInComment);
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
default: {
token.data += '--';
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// Comment end bang state
//------------------------------------------------------------------
_stateCommentEndBang(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.HYPHEN_MINUS: {
token.data += '--!';
this.state = State.COMMENT_END_DASH;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.incorrectlyClosedComment);
this.state = State.DATA;
this.emitCurrentComment(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInComment);
this.emitCurrentComment(token);
this._emitEOFToken();
break;
}
default: {
token.data += '--!';
this.state = State.COMMENT;
this._stateComment(cp);
}
}
}
// DOCTYPE state
//------------------------------------------------------------------
_stateDoctype(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.BEFORE_DOCTYPE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.BEFORE_DOCTYPE_NAME;
this._stateBeforeDoctypeName(cp);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
this._createDoctypeToken(null);
const token = this.currentToken;
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingWhitespaceBeforeDoctypeName);
this.state = State.BEFORE_DOCTYPE_NAME;
this._stateBeforeDoctypeName(cp);
}
}
}
// Before DOCTYPE name state
//------------------------------------------------------------------
_stateBeforeDoctypeName(cp) {
if (isAsciiUpper(cp)) {
this._createDoctypeToken(String.fromCharCode(toAsciiLower(cp)));
this.state = State.DOCTYPE_NAME;
}
else
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
this._createDoctypeToken(unicode_js_1.REPLACEMENT_CHARACTER);
this.state = State.DOCTYPE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingDoctypeName);
this._createDoctypeToken(null);
const token = this.currentToken;
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
this._createDoctypeToken(null);
const token = this.currentToken;
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._createDoctypeToken(String.fromCodePoint(cp));
this.state = State.DOCTYPE_NAME;
}
}
}
// DOCTYPE name state
//------------------------------------------------------------------
_stateDoctypeName(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.AFTER_DOCTYPE_NAME;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.name += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
token.name += String.fromCodePoint(isAsciiUpper(cp) ? toAsciiLower(cp) : cp);
}
}
}
// After DOCTYPE name state
//------------------------------------------------------------------
_stateAfterDoctypeName(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.PUBLIC, false)) {
this.state = State.AFTER_DOCTYPE_PUBLIC_KEYWORD;
}
else if (this._consumeSequenceIfMatch(unicode_js_1.SEQUENCES.SYSTEM, false)) {
this.state = State.AFTER_DOCTYPE_SYSTEM_KEYWORD;
}
//NOTE: sequence lookup can be abrupted by hibernation. In that case lookup
//results are no longer valid and we will need to start over.
else if (!this._ensureHibernation()) {
this._err(error_codes_js_1.ERR.invalidCharacterSequenceAfterDoctypeName);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
}
// After DOCTYPE public keyword state
//------------------------------------------------------------------
_stateAfterDoctypePublicKeyword(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.BEFORE_DOCTYPE_PUBLIC_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypePublicKeyword);
token.publicId = '';
this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypePublicKeyword);
token.publicId = '';
this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingDoctypePublicIdentifier);
token.forceQuirks = true;
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypePublicIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// Before DOCTYPE public identifier state
//------------------------------------------------------------------
_stateBeforeDoctypePublicIdentifier(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
token.publicId = '';
this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
token.publicId = '';
this.state = State.DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingDoctypePublicIdentifier);
token.forceQuirks = true;
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypePublicIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// DOCTYPE public identifier (double-quoted) state
//------------------------------------------------------------------
_stateDoctypePublicIdentifierDoubleQuoted(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.publicId += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptDoctypePublicIdentifier);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
token.publicId += String.fromCodePoint(cp);
}
}
}
// DOCTYPE public identifier (single-quoted) state
//------------------------------------------------------------------
_stateDoctypePublicIdentifierSingleQuoted(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this.state = State.AFTER_DOCTYPE_PUBLIC_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.publicId += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptDoctypePublicIdentifier);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
token.publicId += String.fromCodePoint(cp);
}
}
}
// After DOCTYPE public identifier state
//------------------------------------------------------------------
_stateAfterDoctypePublicIdentifier(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this._err(error_codes_js_1.ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this._err(error_codes_js_1.ERR.missingWhitespaceBetweenDoctypePublicAndSystemIdentifiers);
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// Between DOCTYPE public and system identifiers state
//------------------------------------------------------------------
_stateBetweenDoctypePublicAndSystemIdentifiers(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// After DOCTYPE system keyword state
//------------------------------------------------------------------
_stateAfterDoctypeSystemKeyword(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
this.state = State.BEFORE_DOCTYPE_SYSTEM_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypeSystemKeyword);
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this._err(error_codes_js_1.ERR.missingWhitespaceAfterDoctypeSystemKeyword);
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// Before DOCTYPE system identifier state
//------------------------------------------------------------------
_stateBeforeDoctypeSystemIdentifier(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
token.systemId = '';
this.state = State.DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.missingDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.DATA;
this.emitCurrentDoctype(token);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.missingQuoteBeforeDoctypeSystemIdentifier);
token.forceQuirks = true;
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// DOCTYPE system identifier (double-quoted) state
//------------------------------------------------------------------
_stateDoctypeSystemIdentifierDoubleQuoted(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.QUOTATION_MARK: {
this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.systemId += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptDoctypeSystemIdentifier);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
token.systemId += String.fromCodePoint(cp);
}
}
}
// DOCTYPE system identifier (single-quoted) state
//------------------------------------------------------------------
_stateDoctypeSystemIdentifierSingleQuoted(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.APOSTROPHE: {
this.state = State.AFTER_DOCTYPE_SYSTEM_IDENTIFIER;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
token.systemId += unicode_js_1.REPLACEMENT_CHARACTER;
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this._err(error_codes_js_1.ERR.abruptDoctypeSystemIdentifier);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
token.systemId += String.fromCodePoint(cp);
}
}
}
// After DOCTYPE system identifier state
//------------------------------------------------------------------
_stateAfterDoctypeSystemIdentifier(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.SPACE:
case unicode_js_1.CODE_POINTS.LINE_FEED:
case unicode_js_1.CODE_POINTS.TABULATION:
case unicode_js_1.CODE_POINTS.FORM_FEED: {
// Ignore whitespace
break;
}
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInDoctype);
token.forceQuirks = true;
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default: {
this._err(error_codes_js_1.ERR.unexpectedCharacterAfterDoctypeSystemIdentifier);
this.state = State.BOGUS_DOCTYPE;
this._stateBogusDoctype(cp);
}
}
}
// Bogus DOCTYPE state
//------------------------------------------------------------------
_stateBogusDoctype(cp) {
const token = this.currentToken;
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.emitCurrentDoctype(token);
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.NULL: {
this._err(error_codes_js_1.ERR.unexpectedNullCharacter);
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this.emitCurrentDoctype(token);
this._emitEOFToken();
break;
}
default:
// Do nothing
}
}
// CDATA section state
//------------------------------------------------------------------
_stateCdataSection(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET: {
this.state = State.CDATA_SECTION_BRACKET;
break;
}
case unicode_js_1.CODE_POINTS.EOF: {
this._err(error_codes_js_1.ERR.eofInCdata);
this._emitEOFToken();
break;
}
default: {
this._emitCodePoint(cp);
}
}
}
// CDATA section bracket state
//------------------------------------------------------------------
_stateCdataSectionBracket(cp) {
if (cp === unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET) {
this.state = State.CDATA_SECTION_END;
}
else {
this._emitChars(']');
this.state = State.CDATA_SECTION;
this._stateCdataSection(cp);
}
}
// CDATA section end state
//------------------------------------------------------------------
_stateCdataSectionEnd(cp) {
switch (cp) {
case unicode_js_1.CODE_POINTS.GREATER_THAN_SIGN: {
this.state = State.DATA;
break;
}
case unicode_js_1.CODE_POINTS.RIGHT_SQUARE_BRACKET: {
this._emitChars(']');
break;
}
default: {
this._emitChars(']]');
this.state = State.CDATA_SECTION;
this._stateCdataSection(cp);
}
}
}
// Character reference state
//------------------------------------------------------------------
_stateCharacterReference() {
let length = this.entityDecoder.write(this.preprocessor.html, this.preprocessor.pos);
if (length < 0) {
if (this.preprocessor.lastChunkWritten) {
length = this.entityDecoder.end();
}
else {
// Wait for the rest of the entity.
this.active = false;
// Mark the entire buffer as read.
this.preprocessor.pos = this.preprocessor.html.length - 1;
this.consumedAfterSnapshot = 0;
this.preprocessor.endOfChunkHit = true;
return;
}
}
if (length === 0) {
// This was not a valid entity. Go back to the beginning, and
// figure out what to do.
this.preprocessor.pos = this.entityStartPos;
this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
this.state =
!this._isCharacterReferenceInAttribute() && isAsciiAlphaNumeric(this.preprocessor.peek(1))
? State.AMBIGUOUS_AMPERSAND
: this.returnState;
}
else {
// We successfully parsed an entity. Switch to the return state.
this.state = this.returnState;
}
}
// Ambiguos ampersand state
//------------------------------------------------------------------
_stateAmbiguousAmpersand(cp) {
if (isAsciiAlphaNumeric(cp)) {
this._flushCodePointConsumedAsCharacterReference(cp);
}
else {
if (cp === unicode_js_1.CODE_POINTS.SEMICOLON) {
this._err(error_codes_js_1.ERR.unknownNamedCharacterReference);
}
this.state = this.returnState;
this._callState(cp);
}
}
}
exports.Tokenizer = Tokenizer;
+36
View File
@@ -0,0 +1,36 @@
import { ERR, type ParserError, type ParserErrorHandler } from '../common/error-codes.js';
export declare class Preprocessor {
private handler;
html: string;
pos: number;
private lastGapPos;
private gapStack;
private skipNextNewLine;
lastChunkWritten: boolean;
endOfChunkHit: boolean;
bufferWaterline: number;
private isEol;
private lineStartPos;
droppedBufferSize: number;
line: number;
constructor(handler: {
onParseError?: ParserErrorHandler | null;
});
/** The column on the current line. If we just saw a gap (eg. a surrogate pair), return the index before. */
get col(): number;
get offset(): number;
getError(code: ERR, cpOffset: number): ParserError;
private lastErrOffset;
private _err;
private _addGap;
private _processSurrogate;
willDropParsedChunk(): boolean;
dropParsedChunk(): void;
write(chunk: string, isLastChunk: boolean): void;
insertHtmlAtCurrentPos(chunk: string): void;
startsWith(pattern: string, caseSensitive: boolean): boolean;
peek(offset: number): number;
advance(): number;
private _checkForProblematicCharacters;
retreat(count: number): void;
}
+200
View File
@@ -0,0 +1,200 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.Preprocessor = void 0;
const unicode_js_1 = require("../common/unicode.js");
const error_codes_js_1 = require("../common/error-codes.js");
//Const
const DEFAULT_BUFFER_WATERLINE = 1 << 16;
//Preprocessor
//NOTE: HTML input preprocessing
//(see: http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#preprocessing-the-input-stream)
class Preprocessor {
constructor(handler) {
this.handler = handler;
this.html = '';
this.pos = -1;
// NOTE: Initial `lastGapPos` is -2, to ensure `col` on initialisation is 0
this.lastGapPos = -2;
this.gapStack = [];
this.skipNextNewLine = false;
this.lastChunkWritten = false;
this.endOfChunkHit = false;
this.bufferWaterline = DEFAULT_BUFFER_WATERLINE;
this.isEol = false;
this.lineStartPos = 0;
this.droppedBufferSize = 0;
this.line = 1;
//NOTE: avoid reporting errors twice on advance/retreat
this.lastErrOffset = -1;
}
/** The column on the current line. If we just saw a gap (eg. a surrogate pair), return the index before. */
get col() {
return this.pos - this.lineStartPos + Number(this.lastGapPos !== this.pos);
}
get offset() {
return this.droppedBufferSize + this.pos;
}
getError(code, cpOffset) {
const { line, col, offset } = this;
const startCol = col + cpOffset;
const startOffset = offset + cpOffset;
return {
code,
startLine: line,
endLine: line,
startCol,
endCol: startCol,
startOffset,
endOffset: startOffset,
};
}
_err(code) {
if (this.handler.onParseError && this.lastErrOffset !== this.offset) {
this.lastErrOffset = this.offset;
this.handler.onParseError(this.getError(code, 0));
}
}
_addGap() {
this.gapStack.push(this.lastGapPos);
this.lastGapPos = this.pos;
}
_processSurrogate(cp) {
//NOTE: try to peek a surrogate pair
if (this.pos !== this.html.length - 1) {
const nextCp = this.html.charCodeAt(this.pos + 1);
if ((0, unicode_js_1.isSurrogatePair)(nextCp)) {
//NOTE: we have a surrogate pair. Peek pair character and recalculate code point.
this.pos++;
//NOTE: add a gap that should be avoided during retreat
this._addGap();
return (0, unicode_js_1.getSurrogatePairCodePoint)(cp, nextCp);
}
}
//NOTE: we are at the end of a chunk, therefore we can't infer the surrogate pair yet.
else if (!this.lastChunkWritten) {
this.endOfChunkHit = true;
return unicode_js_1.CODE_POINTS.EOF;
}
//NOTE: isolated surrogate
this._err(error_codes_js_1.ERR.surrogateInInputStream);
return cp;
}
willDropParsedChunk() {
return this.pos > this.bufferWaterline;
}
dropParsedChunk() {
if (this.willDropParsedChunk()) {
this.html = this.html.substring(this.pos);
this.lineStartPos -= this.pos;
this.droppedBufferSize += this.pos;
this.pos = 0;
this.lastGapPos = -2;
this.gapStack.length = 0;
}
}
write(chunk, isLastChunk) {
if (this.html.length > 0) {
this.html += chunk;
}
else {
this.html = chunk;
}
this.endOfChunkHit = false;
this.lastChunkWritten = isLastChunk;
}
insertHtmlAtCurrentPos(chunk) {
this.html = this.html.substring(0, this.pos + 1) + chunk + this.html.substring(this.pos + 1);
this.endOfChunkHit = false;
}
startsWith(pattern, caseSensitive) {
// Check if our buffer has enough characters
if (this.pos + pattern.length > this.html.length) {
this.endOfChunkHit = !this.lastChunkWritten;
return false;
}
if (caseSensitive) {
return this.html.startsWith(pattern, this.pos);
}
for (let i = 0; i < pattern.length; i++) {
const cp = this.html.charCodeAt(this.pos + i) | 0x20;
if (cp !== pattern.charCodeAt(i)) {
return false;
}
}
return true;
}
peek(offset) {
const pos = this.pos + offset;
if (pos >= this.html.length) {
this.endOfChunkHit = !this.lastChunkWritten;
return unicode_js_1.CODE_POINTS.EOF;
}
const code = this.html.charCodeAt(pos);
return code === unicode_js_1.CODE_POINTS.CARRIAGE_RETURN ? unicode_js_1.CODE_POINTS.LINE_FEED : code;
}
advance() {
this.pos++;
//NOTE: LF should be in the last column of the line
if (this.isEol) {
this.isEol = false;
this.line++;
this.lineStartPos = this.pos;
}
if (this.pos >= this.html.length) {
this.endOfChunkHit = !this.lastChunkWritten;
return unicode_js_1.CODE_POINTS.EOF;
}
let cp = this.html.charCodeAt(this.pos);
//NOTE: all U+000D CARRIAGE RETURN (CR) characters must be converted to U+000A LINE FEED (LF) characters
if (cp === unicode_js_1.CODE_POINTS.CARRIAGE_RETURN) {
this.isEol = true;
this.skipNextNewLine = true;
return unicode_js_1.CODE_POINTS.LINE_FEED;
}
//NOTE: any U+000A LINE FEED (LF) characters that immediately follow a U+000D CARRIAGE RETURN (CR) character
//must be ignored.
if (cp === unicode_js_1.CODE_POINTS.LINE_FEED) {
this.isEol = true;
if (this.skipNextNewLine) {
// `line` will be bumped again in the recursive call.
this.line--;
this.skipNextNewLine = false;
this._addGap();
return this.advance();
}
}
this.skipNextNewLine = false;
if ((0, unicode_js_1.isSurrogate)(cp)) {
cp = this._processSurrogate(cp);
}
//OPTIMIZATION: first check if code point is in the common allowed
//range (ASCII alphanumeric, whitespaces, big chunk of BMP)
//before going into detailed performance cost validation.
const isCommonValidRange = this.handler.onParseError === null ||
(cp > 0x1f && cp < 0x7f) ||
cp === unicode_js_1.CODE_POINTS.LINE_FEED ||
cp === unicode_js_1.CODE_POINTS.CARRIAGE_RETURN ||
(cp > 0x9f && cp < 64976);
if (!isCommonValidRange) {
this._checkForProblematicCharacters(cp);
}
return cp;
}
_checkForProblematicCharacters(cp) {
if ((0, unicode_js_1.isControlCodePoint)(cp)) {
this._err(error_codes_js_1.ERR.controlCharacterInInputStream);
}
else if ((0, unicode_js_1.isUndefinedCodePoint)(cp)) {
this._err(error_codes_js_1.ERR.noncharacterInInputStream);
}
}
retreat(count) {
this.pos -= count;
while (this.pos < this.lastGapPos) {
this.lastGapPos = this.gapStack.pop();
this.pos--;
}
this.isEol = false;
}
}
exports.Preprocessor = Preprocessor;
+84
View File
@@ -0,0 +1,84 @@
import { DOCUMENT_MODE, type NS } from '../common/html.js';
import type { Attribute, Location, ElementLocation } from '../common/token.js';
import type { TreeAdapter, TreeAdapterTypeMap } from './interface.js';
export interface Document {
/** The name of the node. */
nodeName: '#document';
/**
* Document mode.
*
* @see {@link DOCUMENT_MODE} */
mode: DOCUMENT_MODE;
/** The node's children. */
childNodes: ChildNode[];
/** Comment source code location info. Available if location info is enabled. */
sourceCodeLocation?: Location | null;
}
export interface DocumentFragment {
/** The name of the node. */
nodeName: '#document-fragment';
/** The node's children. */
childNodes: ChildNode[];
/** Comment source code location info. Available if location info is enabled. */
sourceCodeLocation?: Location | null;
}
export interface Element {
/** Element tag name. Same as {@link tagName}. */
nodeName: string;
/** Element tag name. Same as {@link nodeName}. */
tagName: string;
/** List of element attributes. */
attrs: Attribute[];
/** Element namespace. */
namespaceURI: NS;
/** Element source code location info, with attributes. Available if location info is enabled. */
sourceCodeLocation?: ElementLocation | null;
/** Parent node. */
parentNode: ParentNode | null;
/** The node's children. */
childNodes: ChildNode[];
}
export interface CommentNode {
/** The name of the node. */
nodeName: '#comment';
/** Parent node. */
parentNode: ParentNode | null;
/** Comment text. */
data: string;
/** Comment source code location info. Available if location info is enabled. */
sourceCodeLocation?: Location | null;
}
export interface TextNode {
nodeName: '#text';
/** Parent node. */
parentNode: ParentNode | null;
/** Text content. */
value: string;
/** Comment source code location info. Available if location info is enabled. */
sourceCodeLocation?: Location | null;
}
export interface Template extends Element {
nodeName: 'template';
tagName: 'template';
/** The content of a `template` tag. */
content: DocumentFragment;
}
export interface DocumentType {
/** The name of the node. */
nodeName: '#documentType';
/** Parent node. */
parentNode: ParentNode | null;
/** Document type name. */
name: string;
/** Document type public identifier. */
publicId: string;
/** Document type system identifier. */
systemId: string;
/** Comment source code location info. Available if location info is enabled. */
sourceCodeLocation?: Location | null;
}
export type ParentNode = Document | DocumentFragment | Element | Template;
export type ChildNode = Element | Template | CommentNode | TextNode | DocumentType;
export type Node = ParentNode | ChildNode;
export type DefaultTreeAdapterMap = TreeAdapterTypeMap<Node, ParentNode, ChildNode, Document, DocumentFragment, Element, CommentNode, TextNode, Template, DocumentType>;
export declare const defaultTreeAdapter: TreeAdapter<DefaultTreeAdapterMap>;
+176
View File
@@ -0,0 +1,176 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.defaultTreeAdapter = void 0;
const html_js_1 = require("../common/html.js");
exports.defaultTreeAdapter = {
//Node construction
createDocument() {
return {
nodeName: '#document',
mode: html_js_1.DOCUMENT_MODE.NO_QUIRKS,
childNodes: [],
};
},
createDocumentFragment() {
return {
nodeName: '#document-fragment',
childNodes: [],
};
},
createElement(tagName, namespaceURI, attrs) {
return {
nodeName: tagName,
tagName,
attrs,
namespaceURI,
childNodes: [],
parentNode: null,
};
},
createCommentNode(data) {
return {
nodeName: '#comment',
data,
parentNode: null,
};
},
createTextNode(value) {
return {
nodeName: '#text',
value,
parentNode: null,
};
},
//Tree mutation
appendChild(parentNode, newNode) {
parentNode.childNodes.push(newNode);
newNode.parentNode = parentNode;
},
insertBefore(parentNode, newNode, referenceNode) {
const insertionIdx = parentNode.childNodes.indexOf(referenceNode);
parentNode.childNodes.splice(insertionIdx, 0, newNode);
newNode.parentNode = parentNode;
},
setTemplateContent(templateElement, contentElement) {
templateElement.content = contentElement;
},
getTemplateContent(templateElement) {
return templateElement.content;
},
setDocumentType(document, name, publicId, systemId) {
const doctypeNode = document.childNodes.find((node) => node.nodeName === '#documentType');
if (doctypeNode) {
doctypeNode.name = name;
doctypeNode.publicId = publicId;
doctypeNode.systemId = systemId;
}
else {
const node = {
nodeName: '#documentType',
name,
publicId,
systemId,
parentNode: null,
};
exports.defaultTreeAdapter.appendChild(document, node);
}
},
setDocumentMode(document, mode) {
document.mode = mode;
},
getDocumentMode(document) {
return document.mode;
},
detachNode(node) {
if (node.parentNode) {
const idx = node.parentNode.childNodes.indexOf(node);
node.parentNode.childNodes.splice(idx, 1);
node.parentNode = null;
}
},
insertText(parentNode, text) {
if (parentNode.childNodes.length > 0) {
const prevNode = parentNode.childNodes[parentNode.childNodes.length - 1];
if (exports.defaultTreeAdapter.isTextNode(prevNode)) {
prevNode.value += text;
return;
}
}
exports.defaultTreeAdapter.appendChild(parentNode, exports.defaultTreeAdapter.createTextNode(text));
},
insertTextBefore(parentNode, text, referenceNode) {
const prevNode = parentNode.childNodes[parentNode.childNodes.indexOf(referenceNode) - 1];
if (prevNode && exports.defaultTreeAdapter.isTextNode(prevNode)) {
prevNode.value += text;
}
else {
exports.defaultTreeAdapter.insertBefore(parentNode, exports.defaultTreeAdapter.createTextNode(text), referenceNode);
}
},
adoptAttributes(recipient, attrs) {
const recipientAttrsMap = new Set(recipient.attrs.map((attr) => attr.name));
for (let j = 0; j < attrs.length; j++) {
if (!recipientAttrsMap.has(attrs[j].name)) {
recipient.attrs.push(attrs[j]);
}
}
},
//Tree traversing
getFirstChild(node) {
return node.childNodes[0];
},
getChildNodes(node) {
return node.childNodes;
},
getParentNode(node) {
return node.parentNode;
},
getAttrList(element) {
return element.attrs;
},
//Node data
getTagName(element) {
return element.tagName;
},
getNamespaceURI(element) {
return element.namespaceURI;
},
getTextNodeContent(textNode) {
return textNode.value;
},
getCommentNodeContent(commentNode) {
return commentNode.data;
},
getDocumentTypeNodeName(doctypeNode) {
return doctypeNode.name;
},
getDocumentTypeNodePublicId(doctypeNode) {
return doctypeNode.publicId;
},
getDocumentTypeNodeSystemId(doctypeNode) {
return doctypeNode.systemId;
},
//Node types
isTextNode(node) {
return node.nodeName === '#text';
},
isCommentNode(node) {
return node.nodeName === '#comment';
},
isDocumentTypeNode(node) {
return node.nodeName === '#documentType';
},
isElementNode(node) {
return Object.prototype.hasOwnProperty.call(node, 'tagName');
},
// Source code location
setNodeSourceCodeLocation(node, location) {
node.sourceCodeLocation = location;
},
getNodeSourceCodeLocation(node) {
return node.sourceCodeLocation;
},
updateNodeSourceCodeLocation(node, endLocation) {
node.sourceCodeLocation = Object.assign(Object.assign({}, node.sourceCodeLocation), endLocation);
},
};
+255
View File
@@ -0,0 +1,255 @@
import type { DOCUMENT_MODE, NS } from '../common/html.js';
import type { Attribute, ElementLocation } from '../common/token.js';
export interface TreeAdapterTypeMap<Node = unknown, ParentNode = unknown, ChildNode = unknown, Document = unknown, DocumentFragment = unknown, Element = unknown, CommentNode = unknown, TextNode = unknown, Template = unknown, DocumentType = unknown> {
node: Node;
parentNode: ParentNode;
childNode: ChildNode;
document: Document;
documentFragment: DocumentFragment;
element: Element;
commentNode: CommentNode;
textNode: TextNode;
template: Template;
documentType: DocumentType;
}
/**
* Tree adapter is a set of utility functions that provides minimal required abstraction layer beetween parser and a specific AST format.
* Note that `TreeAdapter` is not designed to be a general purpose AST manipulation library. You can build such library
* on top of existing `TreeAdapter` or use one of the existing libraries from npm.
*
* @see Have a look at the default tree adapter for reference.
*/
export interface TreeAdapter<T extends TreeAdapterTypeMap = TreeAdapterTypeMap> {
/**
* Copies attributes to the given element. Only attributes that are not yet present in the element are copied.
*
* @param recipient - Element to copy attributes into.
* @param attrs - Attributes to copy.
*/
adoptAttributes(recipient: T['element'], attrs: Attribute[]): void;
/**
* Appends a child node to the given parent node.
*
* @param parentNode - Parent node.
* @param newNode - Child node.
*/
appendChild(parentNode: T['parentNode'], newNode: T['childNode']): void;
/**
* Creates a comment node.
*
* @param data - Comment text.
*/
createCommentNode(data: string): T['commentNode'];
/**
* Creates a text node.
*
* @param value - Text.
*/
createTextNode(value: string): T['textNode'];
/**
* Creates a document node.
*/
createDocument(): T['document'];
/**
* Creates a document fragment node.
*/
createDocumentFragment(): T['documentFragment'];
/**
* Creates an element node.
*
* @param tagName - Tag name of the element.
* @param namespaceURI - Namespace of the element.
* @param attrs - Attribute name-value pair array. Foreign attributes may contain `namespace` and `prefix` fields as well.
*/
createElement(tagName: string, namespaceURI: NS, attrs: Attribute[]): T['element'];
/**
* Removes a node from its parent.
*
* @param node - Node to remove.
*/
detachNode(node: T['childNode']): void;
/**
* Returns the given element's attributes in an array, in the form of name-value pairs.
* Foreign attributes may contain `namespace` and `prefix` fields as well.
*
* @param element - Element.
*/
getAttrList(element: T['element']): Attribute[];
/**
* Returns the given node's children in an array.
*
* @param node - Node.
*/
getChildNodes(node: T['parentNode']): T['childNode'][];
/**
* Returns the given comment node's content.
*
* @param commentNode - Comment node.
*/
getCommentNodeContent(commentNode: T['commentNode']): string;
/**
* Returns [document mode](https://dom.spec.whatwg.org/#concept-document-limited-quirks).
*
* @param document - Document node.
*/
getDocumentMode(document: T['document']): DOCUMENT_MODE;
/**
* Returns the given document type node's name.
*
* @param doctypeNode - Document type node.
*/
getDocumentTypeNodeName(doctypeNode: T['documentType']): string;
/**
* Returns the given document type node's public identifier.
*
* @param doctypeNode - Document type node.
*/
getDocumentTypeNodePublicId(doctypeNode: T['documentType']): string;
/**
* Returns the given document type node's system identifier.
*
* @param doctypeNode - Document type node.
*/
getDocumentTypeNodeSystemId(doctypeNode: T['documentType']): string;
/**
* Returns the first child of the given node.
*
* @param node - Node.
*/
getFirstChild(node: T['parentNode']): T['childNode'] | null;
/**
* Returns the given element's namespace.
*
* @param element - Element.
*/
getNamespaceURI(element: T['element']): NS;
/**
* Returns the given node's source code location information.
*
* @param node - Node.
*/
getNodeSourceCodeLocation(node: T['node']): ElementLocation | undefined | null;
/**
* Returns the given node's parent.
*
* @param node - Node.
*/
getParentNode(node: T['node']): T['parentNode'] | null;
/**
* Returns the given element's tag name.
*
* @param element - Element.
*/
getTagName(element: T['element']): string;
/**
* Returns the given text node's content.
*
* @param textNode - Text node.
*/
getTextNodeContent(textNode: T['textNode']): string;
/**
* Returns the `<template>` element content element.
*
* @param templateElement - `<template>` element.
*/
getTemplateContent(templateElement: T['template']): T['documentFragment'];
/**
* Inserts a child node to the given parent node before the given reference node.
*
* @param parentNode - Parent node.
* @param newNode - Child node.
* @param referenceNode - Reference node.
*/
insertBefore(parentNode: T['parentNode'], newNode: T['childNode'], referenceNode: T['childNode']): void;
/**
* Inserts text into a node. If the last child of the node is a text node, the provided text will be appended to the
* text node content. Otherwise, inserts a new text node with the given text.
*
* @param parentNode - Node to insert text into.
* @param text - Text to insert.
*/
insertText(parentNode: T['parentNode'], text: string): void;
/**
* Inserts text into a sibling node that goes before the reference node. If this sibling node is the text node,
* the provided text will be appended to the text node content. Otherwise, inserts a new sibling text node with
* the given text before the reference node.
*
* @param parentNode - Node to insert text into.
* @param text - Text to insert.
* @param referenceNode - Node to insert text before.
*/
insertTextBefore(parentNode: T['parentNode'], text: string, referenceNode: T['childNode']): void;
/**
* Determines if the given node is a comment node.
*
* @param node - Node.
*/
isCommentNode(node: T['node']): node is T['commentNode'];
/**
* Determines if the given node is a document type node.
*
* @param node - Node.
*/
isDocumentTypeNode(node: T['node']): node is T['documentType'];
/**
* Determines if the given node is an element.
*
* @param node - Node.
*/
isElementNode(node: T['node']): node is T['element'];
/**
* Determines if the given node is a text node.
*
* @param node - Node.
*/
isTextNode(node: T['node']): node is T['textNode'];
/**
* Sets the [document mode](https://dom.spec.whatwg.org/#concept-document-limited-quirks).
*
* @param document - Document node.
* @param mode - Document mode.
*/
setDocumentMode(document: T['document'], mode: DOCUMENT_MODE): void;
/**
* Sets the document type. If the `document` already contains a document type node, the `name`, `publicId` and `systemId`
* properties of this node will be updated with the provided values. Otherwise, creates a new document type node
* with the given properties and inserts it into the `document`.
*
* @param document - Document node.
* @param name - Document type name.
* @param publicId - Document type public identifier.
* @param systemId - Document type system identifier.
*/
setDocumentType(document: T['document'], name: string, publicId: string, systemId: string): void;
/**
* Attaches source code location information to the node.
*
* @param node - Node.
*/
setNodeSourceCodeLocation(node: T['node'], location: ElementLocation | null): void;
/**
* Updates the source code location information of the node.
*
* @param node - Node.
*/
updateNodeSourceCodeLocation(node: T['node'], location: Partial<ElementLocation>): void;
/**
* Sets the `<template>` element content element.
*
* @param templateElement - `<template>` element.
* @param contentElement - Content element.
*/
setTemplateContent(templateElement: T['template'], contentElement: T['documentFragment']): void;
/**
* Optional callback for elements being pushed to the stack of open elements.
*
* @param element The element being pushed to the stack of open elements.
*/
onItemPush?: (item: T['element']) => void;
/**
* Optional callback for elements being popped from the stack of open elements.
*
* @param item The element being popped.
*/
onItemPop?: (item: T['element'], newTop: T['parentNode']) => void;
}
+2
View File
@@ -0,0 +1,2 @@
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });