LibPDF: Handle indirect reference resolving during parsing more robustly

If `Document::resolve()` was called during parsing, it'd change the
reader's current position, so the parsing code that called it would
then end up at an unexpected position in the file.

Parser.cpp already had special-case recovery when a stream's length
was stored in an indirect reference.

Commit ead02da98a ("/JBIG2Globals") in #23503 added another case
where we could resolve indirect reference during parsing, but wasn't
aware of having to save and restore the reader position for that.

Put the save/restore code in `DocumentParser::parse_object_with_index`
instead, right before the place that ultimately changes the reader's
position during `Document::resolve`. This fixes `/JBIG2Globals` and
lets us remove the special-case code for `/Length` handling.

Since this is kind of subtle, include a test.

Tests/LibPDF/CMakeLists.txt

@@ -12,6 +12,7 @@ set(TEST_FILES
     complex.pdf
     encoding.pdf
     encryption_nocopy.pdf
+    jbig2-globals.pdf
     linearized.pdf
     non-linearized.pdf
     oss-fuzz-testcase-62065.pdf

Tests/LibPDF/TestPDF.cpp

@@ -121,6 +121,18 @@ TEST_CASE(encrypted_object_stream)
     EXPECT_EQ(MUST(info_dict.creator()).value(), "Acrobat PDFMaker 9.1 voor Word");
 }
 
+TEST_CASE(resolve_indirect_reference_during_parsing)
+{
+    auto file = MUST(Core::MappedFile::map("jbig2-globals.pdf"sv));
+    auto document = MUST(PDF::Document::create(file->bytes()));
+    MUST(document->initialize());
+    EXPECT_EQ(document->get_page_count(), 1U);
+
+    auto jbig2_stream_value = MUST(document->get_or_load_value(5));
+    auto jbig2_stream = MUST(document->resolve_to<PDF::StreamObject>(jbig2_stream_value));
+    EXPECT_EQ(jbig2_stream->bytes().size(), 20'000U);
+}
+
 TEST_CASE(malformed_pdf_document)
 {
     Array test_inputs = {

Userland/Libraries/LibPDF/DocumentParser.cpp

@@ -70,11 +70,31 @@ PDFErrorOr<Value> DocumentParser::parse_object_with_index(u32 index)
     if (!m_xref_table->is_object_in_use(index))
         return nullptr;
 
+    // If this is called to resolve an indirect object reference while parsing another object,
+    // make sure to restore the current position after parsing the indirect object, so that the
+    // parser can keep parsing the original object stream afterwards.
+    // parse_compressed_object_with_index() also moves the reader's position, so this needs
+    // to be before the potential call to parse_compressed_object_with_index().
+    class SavePoint {
+    public:
+        SavePoint(Reader& reader)
+            : m_reader(reader)
+        {
+            m_reader.save();
+        }
+        ~SavePoint() { m_reader.load(); }
+
+    private:
+        Reader& m_reader;
+    };
+    SavePoint restore_current_position { m_reader };
+
     if (m_xref_table->is_object_compressed(index))
         // The object can be found in a object stream
         return parse_compressed_object_with_index(index);
 
     auto byte_offset = m_xref_table->byte_offset_for_object(index);
+
     m_reader.move_to(byte_offset);
     auto indirect_value = TRY(parse_indirect_value());
     VERIFY(indirect_value->index() == index);

Userland/Libraries/LibPDF/Parser.cpp

@@ -483,9 +483,7 @@ PDFErrorOr<NonnullRefPtr<StreamObject>> Parser::parse_stream(NonnullRefPtr<DictO
     auto maybe_length = dict->get(CommonNames::Length);
     if (maybe_length.has_value() && m_document->can_resolve_references()) {
         // The PDF writer has kindly provided us with the direct length of the stream
-        m_reader.save();
         auto length = TRY(m_document->resolve_to<int>(maybe_length.value()));
-        m_reader.load();
         bytes = m_reader.bytes().slice(m_reader.offset(), length);
         m_reader.move_by(length);
         m_reader.consume_whitespace();