Method parseHeader

src/parser/document-parser.ts:257–316

* Parse PDF header and extract version. * * Lenient handling (like pdf.js and PDFBox): * - Search first 1024 bytes for %PDF- * - Accept garbage before/after header * - Default to 1.4 if version unparseable

()

   * - Default to 1.4 if version unparseable
   */
  parseHeader(): string {
    const bytes = this.scanner.bytes;
    const searchLimit = Math.min(bytes.length, HEADER_SEARCH_LIMIT);

    // Search for %PDF- marker
    let headerPos = -1;
    for (let i = 0; i <= searchLimit - PDF_HEADER.length; i++) {
      if (this.matchesAt(i, PDF_HEADER)) {
        headerPos = i;
        break;
      }
    }

    if (headerPos === -1) {
      if (this.options.lenient) {
        this.warnings.push("PDF header not found, using default version");
        return DEFAULT_VERSION;
      }
      throw new StructureError("PDF header not found");
    }

    if (headerPos > 0) {
      this.warnings.push(`PDF header found at offset ${headerPos} (expected 0)`);
    }

    // Read version string after %PDF-
    const versionStart = headerPos + PDF_HEADER.length;
    let version = "";

    // Read characters until whitespace or non-version char (max 7 chars like pdf.js)
    for (let i = 0; i < 7 && versionStart + i < bytes.length; i++) {
      const byte = bytes[versionStart + i];

      // Stop at whitespace or control chars
      if (byte <= 0x20) {
        break;
      }

      version += String.fromCharCode(byte);
    }

    // Validate version format
    if (VERSION_PATTERN.test(version)) {
      return version;
    }

    // Try to extract just the version part (handle garbage after version)
    const match = version.match(/^(\d\.\d)/);
    if (match) {
      this.warnings.push(`Version string has garbage after it: ${version}`);
      return match[1];
    }

    if (this.options.lenient) {
      this.warnings.push(`Invalid PDF version: ${version}, using default`);
      return DEFAULT_VERSION;
    }