1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-26 20:47:45 +00:00

LibPDF: Make PDF version accessible on Document

This commit is contained in:
Nico Weber 2023-07-11 12:37:13 -04:00 committed by Tim Flynn
parent 9174ffd7e6
commit ea89053c12
4 changed files with 26 additions and 15 deletions

View file

@ -79,8 +79,7 @@ PDFErrorOr<NonnullRefPtr<Document>> Document::create(ReadonlyBytes bytes)
auto parser = adopt_ref(*new DocumentParser({}, bytes)); auto parser = adopt_ref(*new DocumentParser({}, bytes));
auto document = adopt_ref(*new Document(parser)); auto document = adopt_ref(*new Document(parser));
TRY(parser->initialize()); document->m_version = TRY(parser->initialize());
document->m_trailer = parser->trailer(); document->m_trailer = parser->trailer();
document->m_catalog = TRY(parser->trailer()->get_dict(document, CommonNames::Root)); document->m_catalog = TRY(parser->trailer()->get_dict(document, CommonNames::Root));

View file

@ -122,6 +122,8 @@ public:
// need to handle the case where the user password is the empty string. // need to handle the case where the user password is the empty string.
PDFErrorOr<void> initialize(); PDFErrorOr<void> initialize();
Version version() const { return m_version; }
ALWAYS_INLINE RefPtr<SecurityHandler> const& security_handler() const { return m_security_handler; } ALWAYS_INLINE RefPtr<SecurityHandler> const& security_handler() const { return m_security_handler; }
ALWAYS_INLINE RefPtr<OutlineDict> const& outline() const { return m_outline; } ALWAYS_INLINE RefPtr<OutlineDict> const& outline() const { return m_outline; }
@ -188,6 +190,7 @@ private:
PDFErrorOr<NonnullRefPtr<Object>> find_in_key_value_array(NonnullRefPtr<ArrayObject> key_value_array, DeprecatedFlyString name); PDFErrorOr<NonnullRefPtr<Object>> find_in_key_value_array(NonnullRefPtr<ArrayObject> key_value_array, DeprecatedFlyString name);
NonnullRefPtr<DocumentParser> m_parser; NonnullRefPtr<DocumentParser> m_parser;
Version m_version;
RefPtr<DictObject> m_catalog; RefPtr<DictObject> m_catalog;
RefPtr<DictObject> m_trailer; RefPtr<DictObject> m_trailer;
Vector<u32> m_page_object_indices; Vector<u32> m_page_object_indices;

View file

@ -21,22 +21,25 @@ DocumentParser::DocumentParser(Document* document, ReadonlyBytes bytes)
{ {
} }
PDFErrorOr<void> DocumentParser::initialize() PDFErrorOr<Version> DocumentParser::initialize()
{ {
m_reader.set_reading_forwards(); m_reader.set_reading_forwards();
if (m_reader.remaining() == 0) if (m_reader.remaining() == 0)
return error("Empty PDF document"); return error("Empty PDF document");
auto maybe_error = parse_header(); auto maybe_version = parse_header();
if (maybe_error.is_error()) { if (maybe_version.is_error()) {
warnln("{}", maybe_error.error().message()); warnln("{}", maybe_version.error().message());
warnln("No valid PDF header detected, continuing anyway."); warnln("No valid PDF header detected, continuing anyway.");
maybe_version = Version { 1, 6 }; // ¯\_(ツ)_/¯
} }
auto const linearization_result = TRY(initialize_linearization_dict()); auto const linearization_result = TRY(initialize_linearization_dict());
if (linearization_result == LinearizationResult::NotLinearized) if (linearization_result == LinearizationResult::NotLinearized) {
return initialize_non_linearized_xref_table(); TRY(initialize_non_linearized_xref_table());
return maybe_version.value();
}
bool is_linearized = m_linearization_dictionary.has_value(); bool is_linearized = m_linearization_dictionary.has_value();
if (is_linearized) { if (is_linearized) {
@ -53,9 +56,11 @@ PDFErrorOr<void> DocumentParser::initialize()
} }
if (is_linearized) if (is_linearized)
return initialize_linearized_xref_table(); TRY(initialize_linearized_xref_table());
else
TRY(initialize_non_linearized_xref_table());
return initialize_non_linearized_xref_table(); return maybe_version.value();
} }
PDFErrorOr<Value> DocumentParser::parse_object_with_index(u32 index) PDFErrorOr<Value> DocumentParser::parse_object_with_index(u32 index)
@ -73,9 +78,8 @@ PDFErrorOr<Value> DocumentParser::parse_object_with_index(u32 index)
return indirect_value->value(); return indirect_value->value();
} }
PDFErrorOr<void> DocumentParser::parse_header() PDFErrorOr<Version> DocumentParser::parse_header()
{ {
// FIXME: Do something with the version?
m_reader.move_to(0); m_reader.move_to(0);
if (m_reader.remaining() < 8 || !m_reader.matches("%PDF-")) if (m_reader.remaining() < 8 || !m_reader.matches("%PDF-"))
return error("Not a PDF document"); return error("Not a PDF document");
@ -106,7 +110,7 @@ PDFErrorOr<void> DocumentParser::parse_header()
} }
} }
return {}; return Version { major_ver - '0', minor_ver - '0' };
} }
PDFErrorOr<DocumentParser::LinearizationResult> DocumentParser::initialize_linearization_dict() PDFErrorOr<DocumentParser::LinearizationResult> DocumentParser::initialize_linearization_dict()

View file

@ -10,6 +10,11 @@
namespace PDF { namespace PDF {
struct Version {
int major { 0 };
int minor { 0 };
};
class DocumentParser final : public RefCounted<DocumentParser> class DocumentParser final : public RefCounted<DocumentParser>
, public Parser { , public Parser {
public: public:
@ -23,7 +28,7 @@ public:
[[nodiscard]] ALWAYS_INLINE RefPtr<DictObject> const& trailer() const { return m_xref_table->trailer(); } [[nodiscard]] ALWAYS_INLINE RefPtr<DictObject> const& trailer() const { return m_xref_table->trailer(); }
// Parses the header and initializes the xref table and trailer // Parses the header and initializes the xref table and trailer
PDFErrorOr<void> initialize(); PDFErrorOr<Version> initialize();
bool can_resolve_references() { return m_xref_table; } bool can_resolve_references() { return m_xref_table; }
@ -77,7 +82,7 @@ private:
friend struct AK::Formatter<PageOffsetHintTable>; friend struct AK::Formatter<PageOffsetHintTable>;
friend struct AK::Formatter<PageOffsetHintTableEntry>; friend struct AK::Formatter<PageOffsetHintTableEntry>;
PDFErrorOr<void> parse_header(); PDFErrorOr<Version> parse_header();
PDFErrorOr<LinearizationResult> initialize_linearization_dict(); PDFErrorOr<LinearizationResult> initialize_linearization_dict();
PDFErrorOr<void> initialize_linearized_xref_table(); PDFErrorOr<void> initialize_linearized_xref_table();
PDFErrorOr<void> initialize_non_linearized_xref_table(); PDFErrorOr<void> initialize_non_linearized_xref_table();