boc/sie-import/parser.mjs

/**
 * SIE4 Parser
 * Parses SIE4 accounting file format into structured objects.
 */

/**
 * Tokenize a SIE line remainder (after the directive keyword).
 * Handles:
 *  - Quoted strings "like this"
 *  - Object lists {key val ...} → replaced with placeholder '{}'
 *  - Regular whitespace-separated tokens
 */
function tokenize(str) {
  const tokens = [];
  let i = 0;
  const s = str.trim();

  while (i < s.length) {
    // Skip whitespace
    if (s[i] === ' ' || s[i] === '\t') { i++; continue; }

    // Quoted string
    if (s[i] === '"') {
      let j = i + 1;
      while (j < s.length && s[j] !== '"') j++;
      tokens.push(s.slice(i + 1, j));
      i = j + 1;
      continue;
    }

    // Object list { ... } — treat whole block as one placeholder token
    if (s[i] === '{') {
      let depth = 0;
      let j = i;
      while (j < s.length) {
        if (s[j] === '{') depth++;
        else if (s[j] === '}') { depth--; if (depth === 0) break; }
        j++;
      }
      tokens.push('{}');
      i = j + 1;
      continue;
    }

    // Regular token
    let j = i;
    while (j < s.length && s[j] !== ' ' && s[j] !== '\t' && s[j] !== '"') j++;
    if (j > i) tokens.push(s.slice(i, j));
    i = j;
  }

  return tokens;
}

/**
 * Parse SIE4 file content (string) into a structured object.
 * @param {string} content  Raw SIE4 text (UTF-8 or already decoded)
 * @returns {object}
 */
export function parseSIE4(content) {
  const lines = content.replace(/\r\n/g, '\n').replace(/\r/g, '\n').split('\n');

  const result = {
    company: '',
    org_nr: '',
    fiscal_year: null,
    accounts: [],
    opening_balances: [],
    closing_balances: [],
    journal_entries: [],
  };

  let currentVer = null;
  let inVerBlock = false;

  for (const rawLine of lines) {
    const line = rawLine.trim();
    if (!line) continue;

    // Block delimiters
    if (line === '{') {
      inVerBlock = true;
      continue;
    }
    if (line === '}') {
      if (currentVer) {
        result.journal_entries.push(currentVer);
        currentVer = null;
      }
      inVerBlock = false;
      continue;
    }

    if (!line.startsWith('#')) continue;

    // Split directive from rest
    const spaceIdx = line.indexOf(' ');
    const directive = (spaceIdx >= 0 ? line.slice(0, spaceIdx) : line).toUpperCase();
    const rest = spaceIdx >= 0 ? line.slice(spaceIdx + 1) : '';
    const tokens = tokenize(rest);

    switch (directive) {
      case '#FNAMN':
        result.company = tokens[0] || '';
        break;

      case '#ORGNR':
        result.org_nr = tokens[0] || '';
        break;

      case '#RAR': {
        // #RAR year_offset from_date to_date
        const yearOffset = parseInt(tokens[0], 10);
        if (yearOffset === 0) {
          result.fiscal_year = {
            from: tokens[1] || '',
            to: tokens[2] || '',
          };
        }
        break;
      }

      case '#KONTO':
        // #KONTO account_nr account_name
        result.accounts.push({
          number: tokens[0] || '',
          name: tokens[1] || '',
        });
        break;

      case '#IB':
        // #IB year_offset account amount [quantity]
        result.opening_balances.push({
          year_offset: parseInt(tokens[0], 10) || 0,
          account: tokens[1] || '',
          amount: parseFloat(tokens[2]) || 0,
        });
        break;

      case '#UB':
        // #UB year_offset account amount [quantity]
        result.closing_balances.push({
          year_offset: parseInt(tokens[0], 10) || 0,
          account: tokens[1] || '',
          amount: parseFloat(tokens[2]) || 0,
        });
        break;

      case '#VER':
        // #VER series number date "description" [reg_date] [reg_by]
        currentVer = {
          series: tokens[0] || '',
          number: tokens[1] || '',
          date: tokens[2] || '',
          description: tokens[3] || '',
          transactions: [],
        };
        break;

      case '#TRANS':
        // #TRANS account {object_list} amount [date] ["description"] [quantity]
        // tokens: [account, {}, amount, date?, description?]
        if (currentVer) {
          currentVer.transactions.push({
            account: tokens[0] || '',
            amount: parseFloat(tokens[2]) || 0,
            date: tokens[3] || currentVer.date,
            description: tokens[4] || '',
          });
        }
        break;

      // Ignored directives (metadata only)
      case '#FLAGGA':
      case '#PROGRAM':
      case '#FORMAT':
      case '#GEN':
      case '#SIETYP':
      case '#VALUTA':
      case '#TAXAR':
      case '#OMFATTN':
      case '#KPTYP':
      case '#ENHET':
        break;

      default:
        break;
    }
  }

  // If file ended without closing brace for last VER (malformed), still capture it
  if (currentVer) {
    result.journal_entries.push(currentVer);
  }

  return result;
}