Method parseSubsection

src/parser/xref-parser.ts:345–391

* Parse a single subsection of the xref table.

(entries: Map<number, XRefEntry>)

   * Parse a single subsection of the xref table.
   */
  private parseSubsection(entries: Map<number, XRefEntry>): void {
    // Read first object number and count
    const firstObjNum = this.readIntegerFromCurrent();

    if (firstObjNum === null) {
      throw new XRefParseError("Expected subsection start object number");
    }

    this.skipWhitespaceFromCurrent();

    const count = this.readIntegerFromCurrent();

    if (count === null) {
      throw new XRefParseError("Expected subsection entry count");
    }

    this.skipWhitespaceFromCurrent();

    // Read entries
    const parsedEntries: XRefEntry[] = [];

    for (let i = 0; i < count; i++) {
      parsedEntries.push(this.parseEntry());
    }

    // Detect off-by-one in subsection start: some malformed PDFs report
    // firstObjNum=1 when the entries actually start at object 0.
    // The free list head (generation 65535, type free) is always object 0,
    // so if we see it at position 1, correct it. (Same fix as pdf.js #3248/#7229)
    let correctedFirstObjNum = firstObjNum;

    if (
      firstObjNum === 1 &&
      parsedEntries.length > 0 &&
      parsedEntries[0].type === "free" &&
      parsedEntries[0].generation === 65535
    ) {
      correctedFirstObjNum = 0;
      console.warn(
        "XRef: corrected subsection start from 1 to 0 (free list head at wrong position)",
      );
    }

    for (let i = 0; i < parsedEntries.length; i++) {
      entries.set(correctedFirstObjNum + i, parsedEntries[i]);
    }
  }

  /**
   * Parse a single xref entry.