Method parseWithRecovery

src/parser/document-parser.ts:197–247

* Recovery parsing using brute-force when normal parsing fails.

()

   * Recovery parsing using brute-force when normal parsing fails.
   */
  private parseWithRecovery(): ParsedDocument {
    // Try to get version even if header is malformed
    let version = DEFAULT_VERSION;

    try {
      version = this.parseHeader();
    } catch {
      this.warnings.push("Could not parse header, using default version");
    }

    // Use brute-force parser to find objects
    const bruteForce = new BruteForceParser(this.scanner);

    const result = bruteForce.recover();

    if (result === null) {
      throw new UnrecoverableParseError("Could not recover PDF structure: no objects found");
    }

    this.warnings.push(...result.warnings);

    // Build xref from recovered entries
    const xref = new Map<number, XRefEntry>();

    for (const [key, recoveredEntry] of result.xref.entries()) {
      // Key is "objNum genNum"
      const [objNumStr, genNumStr] = key.split(" ");

      const objNum = parseInt(objNumStr, 10);
      const generation = parseInt(genNumStr, 10);

      if (recoveredEntry.type === "uncompressed") {
        xref.set(objNum, { type: "uncompressed", offset: recoveredEntry.offset, generation });
      } else {
        // Compressed entry from object stream
        xref.set(objNum, {
          type: "compressed",
          streamObjNum: recoveredEntry.streamObjNum,
          indexInStream: recoveredEntry.indexInStream,
        });
      }
    }

    // Build trailer from recovered root
    const trailer = new PdfDict([
      ["Root", result.trailer.Root],
      ["Size", new PdfNumber(result.trailer.Size)],
    ]);

    return this.buildDocument(version, xref, trailer, true);
  }

  /**
   * Parse PDF header and extract version.