extractImages method

List<PdfImageInfo> extractImages({ int? fromPage, int? toPage, bool includeUnusedXObjects = false, })

Extracts images from the PDF with page range support.

If includeUnusedXObjects is true, every XObject of the page is considered, without checking whether it was used in the content stream.

Implementation

List<PdfImageInfo> extractImages({
  int? fromPage,
  int? toPage,
  bool includeUnusedXObjects = false,
}) {
  _ensureXrefParsed();

  final trailer = _trailerInfo ??
      PdfParserXref.readTrailerInfoFromReader(reader, xrefOffset);
  final rootObjId = trailer.rootObj;
  if (rootObjId == null) return const <PdfImageInfo>[];

  final rootObj = _getObjectNoStream(rootObjId) ?? _getObject(rootObjId);
  if (rootObj == null || rootObj.value is! PdfDictToken) {
    return const <PdfImageInfo>[];
  }

  final rootDict = rootObj.value as PdfDictToken;
  final pagesRef =
      PdfParserObjects.asRef(rootDict.values[PdfNameTokens.pages]);
  var pageRefs = pagesRef != null
      ? _collectPageRefs(pagesRef, maxPages: toPage)
      : <PdfRefToken>[];

  if ((_repairAttempted || pageRefs.isEmpty) && _allowRepair) {
    pageRefs = _collectPageRefsByScan(maxPages: toPage);
  }

  // Deduplicate by image object id across pages, matching mutool info behavior.
  final images = <PdfImageInfo>[];
  final seenImageRefs = <int>{};

  for (int i = 0; i < pageRefs.length; i++) {
    final pageIndex = i + 1;
    if (fromPage != null && pageIndex < fromPage) continue;
    if (toPage != null && pageIndex > toPage) continue;

    final pageRef = pageRefs[i];
    final pageObj =
        _getObjectNoStream(pageRef.obj) ?? _getObject(pageRef.obj);
    if (pageObj == null || pageObj.value is! PdfDictToken) continue;
    final pageDict = pageObj.value as PdfDictToken;

    final resDict = _resolvePageResources(pageDict);
    final usedXObjects = includeUnusedXObjects
        ? null
        : _extractXObjectNamesFromContent(pageDict).toSet();
    _collectImagesFromResources(
      resources: resDict,
      pageIndex: pageIndex,
      pageRef: PdfIndirectRef(pageRef.obj, pageRef.gen),
      out: images,
      seenImageRefs: seenImageRefs,
      allowedTopLevelXObjects: usedXObjects,
    );
  }

  // Last resort fallback for damaged structures where resources cannot be
  // resolved. Mapping by index is best-effort and less reliable than
  // resource traversal.
  if (images.isEmpty && _allowRepair && pageRefs.isNotEmpty) {
    final scanned = _collectImagesByScan();
    if (scanned.length == pageRefs.length) {
      for (var i = 0; i < scanned.length; i++) {
        final pageIndex = i + 1;
        if (fromPage != null && pageIndex < fromPage) continue;
        if (toPage != null && pageIndex > toPage) continue;
        final pageRef = pageRefs[i];
        final img = scanned[i];
        images.add(PdfImageInfo(
          pageIndex: pageIndex,
          pageRef: PdfIndirectRef(pageRef.obj, pageRef.gen),
          imageRef: img.imageRef,
          width: img.width,
          height: img.height,
          bitsPerComponent: img.bitsPerComponent,
          colorSpace: img.colorSpace,
          filter: img.filter,
        ));
      }
    } else {
      final firstPageRef = pageRefs.first;
      for (var i = 0; i < scanned.length; i++) {
        final pageIndex = i + 1;
        if (fromPage != null && pageIndex < fromPage) continue;
        if (toPage != null && pageIndex > toPage) continue;
        final img = scanned[i];
        images.add(PdfImageInfo(
          pageIndex: pageIndex,
          pageRef: PdfIndirectRef(firstPageRef.obj, firstPageRef.gen),
          imageRef: img.imageRef,
          width: img.width,
          height: img.height,
          bitsPerComponent: img.bitsPerComponent,
          colorSpace: img.colorSpace,
          filter: img.filter,
        ));
      }
    }
  }

  return images;
}