Document.cpp 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644
  1. /*
  2. * Copyright (c) 2021-2022, Matthew Olsson <mattco@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <LibPDF/CommonNames.h>
  7. #include <LibPDF/Document.h>
  8. #include <LibPDF/Parser.h>
  9. #include <LibTextCodec/Decoder.h>
  10. namespace PDF {
  11. ByteString OutlineItem::to_byte_string(int indent) const
  12. {
  13. auto indent_str = ByteString::repeated(" "sv, indent + 1);
  14. StringBuilder child_builder;
  15. child_builder.append('[');
  16. for (auto& child : children)
  17. child_builder.appendff("{}\n", child->to_byte_string(indent + 1));
  18. child_builder.appendff("{}]", indent_str);
  19. StringBuilder builder;
  20. builder.append("OutlineItem {{\n"sv);
  21. builder.appendff("{}title={}\n", indent_str, title);
  22. builder.appendff("{}count={}\n", indent_str, count);
  23. builder.appendff("{}dest={}\n", indent_str, dest);
  24. builder.appendff("{}color={}\n", indent_str, color);
  25. builder.appendff("{}italic={}\n", indent_str, italic);
  26. builder.appendff("{}bold={}\n", indent_str, bold);
  27. builder.appendff("{}children={}\n", indent_str, child_builder.to_byte_string());
  28. builder.appendff("{}}}", ByteString::repeated(" "sv, indent));
  29. return builder.to_byte_string();
  30. }
  31. PDFErrorOr<Optional<ByteString>> InfoDict::title() const
  32. {
  33. return get_text(CommonNames::Title);
  34. }
  35. PDFErrorOr<Optional<ByteString>> InfoDict::author() const
  36. {
  37. return get_text(CommonNames::Author);
  38. }
  39. PDFErrorOr<Optional<ByteString>> InfoDict::subject() const
  40. {
  41. return get_text(CommonNames::Subject);
  42. }
  43. PDFErrorOr<Optional<ByteString>> InfoDict::keywords() const
  44. {
  45. return get_text(CommonNames::Keywords);
  46. }
  47. PDFErrorOr<Optional<ByteString>> InfoDict::creator() const
  48. {
  49. return get_text(CommonNames::Creator);
  50. }
  51. PDFErrorOr<Optional<ByteString>> InfoDict::producer() const
  52. {
  53. return get_text(CommonNames::Producer);
  54. }
  55. PDFErrorOr<Optional<ByteString>> InfoDict::creation_date() const
  56. {
  57. return get(CommonNames::CreationDate);
  58. }
  59. PDFErrorOr<Optional<ByteString>> InfoDict::modification_date() const
  60. {
  61. return get(CommonNames::ModDate);
  62. }
  63. PDFErrorOr<Optional<ByteString>> InfoDict::get_text(DeprecatedFlyString const& name) const
  64. {
  65. return TRY(get(name)).map(Document::text_string_to_utf8);
  66. }
  67. ByteString Document::text_string_to_utf8(ByteString const& text_string)
  68. {
  69. if (text_string.bytes().starts_with(Array<u8, 2> { 0xfe, 0xff })) {
  70. // The string is encoded in UTF16-BE
  71. return TextCodec::decoder_for("utf-16be"sv)->to_utf8(text_string).release_value_but_fixme_should_propagate_errors().to_byte_string();
  72. }
  73. if (text_string.bytes().starts_with(Array<u8, 3> { 239, 187, 191 })) {
  74. // The string is encoded in UTF-8.
  75. return text_string.substring(3);
  76. }
  77. return TextCodec::decoder_for("PDFDocEncoding"sv)->to_utf8(text_string).release_value_but_fixme_should_propagate_errors().to_byte_string();
  78. }
  79. PDFErrorOr<NonnullRefPtr<Document>> Document::create(ReadonlyBytes bytes)
  80. {
  81. auto parser = adopt_ref(*new DocumentParser({}, bytes));
  82. auto document = adopt_ref(*new Document(parser));
  83. document->m_version = TRY(parser->initialize());
  84. document->m_trailer = parser->trailer();
  85. document->m_catalog = TRY(parser->trailer()->get_dict(document, CommonNames::Root));
  86. if (document->m_trailer->contains(CommonNames::Encrypt)) {
  87. auto encryption_dict = TRY(document->m_trailer->get_dict(document, CommonNames::Encrypt));
  88. document->m_security_handler = TRY(SecurityHandler::create(document, encryption_dict));
  89. // Automatically attempt to unencrypt the document with the empty string. The
  90. // result is not important; it is the caller's responsibility to ensure the
  91. // document is unencrypted before calling initialize().
  92. document->m_security_handler->try_provide_user_password(""sv);
  93. }
  94. return document;
  95. }
  96. Document::Document(NonnullRefPtr<DocumentParser> const& parser)
  97. : m_parser(parser)
  98. {
  99. m_parser->set_document(this);
  100. }
  101. PDFErrorOr<void> Document::initialize()
  102. {
  103. if (m_security_handler)
  104. VERIFY(m_security_handler->has_user_password());
  105. TRY(build_page_tree());
  106. TRY(build_outline());
  107. return {};
  108. }
  109. PDFErrorOr<Value> Document::get_or_load_value(u32 index)
  110. {
  111. auto value = get_value(index);
  112. if (!value.has<Empty>()) // FIXME: Use Optional instead?
  113. return value;
  114. auto object = TRY(m_parser->parse_object_with_index(index));
  115. m_values.set(index, object);
  116. return object;
  117. }
  118. u32 Document::get_first_page_index() const
  119. {
  120. // FIXME: A PDF can have a different default first page, which
  121. // should be fetched and returned here
  122. return 0;
  123. }
  124. u32 Document::get_page_count() const
  125. {
  126. return m_page_object_indices.size();
  127. }
  128. static ErrorOr<void> collect_referenced_indices(Value const& value, Vector<int>& referenced_indices)
  129. {
  130. TRY(value.visit(
  131. [&](Empty const&) -> ErrorOr<void> { return {}; },
  132. [&](nullptr_t const&) -> ErrorOr<void> { return {}; },
  133. [&](bool const&) -> ErrorOr<void> { return {}; },
  134. [&](int const&) -> ErrorOr<void> { return {}; },
  135. [&](float const&) -> ErrorOr<void> { return {}; },
  136. [&](Reference const& ref) -> ErrorOr<void> {
  137. TRY(referenced_indices.try_append(ref.as_ref_index()));
  138. return {};
  139. },
  140. [&](NonnullRefPtr<Object> const& object) -> ErrorOr<void> {
  141. if (object->is<ArrayObject>()) {
  142. for (auto& element : object->cast<ArrayObject>()->elements())
  143. TRY(collect_referenced_indices(element, referenced_indices));
  144. } else if (object->is<DictObject>()) {
  145. for (auto& [key, value] : object->cast<DictObject>()->map()) {
  146. if (key != CommonNames::Parent)
  147. TRY(collect_referenced_indices(value, referenced_indices));
  148. }
  149. } else if (object->is<StreamObject>()) {
  150. for (auto& [key, value] : object->cast<StreamObject>()->dict()->map()) {
  151. if (key != CommonNames::Parent)
  152. TRY(collect_referenced_indices(value, referenced_indices));
  153. }
  154. }
  155. return {};
  156. }));
  157. return {};
  158. }
  159. static PDFErrorOr<void> dump_tree(Document& document, size_t index, HashTable<int>& seen)
  160. {
  161. if (seen.contains(index))
  162. return {};
  163. seen.set(index);
  164. auto const& value = TRY(document.get_or_load_value(index));
  165. outln("{} 0 obj", index);
  166. outln("{}", value.to_byte_string(0));
  167. outln("endobj");
  168. Vector<int> referenced_indices;
  169. TRY(collect_referenced_indices(value, referenced_indices));
  170. for (auto index : referenced_indices)
  171. TRY(dump_tree(document, index, seen));
  172. return {};
  173. }
  174. PDFErrorOr<void> Document::dump_page(u32 index)
  175. {
  176. VERIFY(index < m_page_object_indices.size());
  177. auto page_object_index = m_page_object_indices[index];
  178. HashTable<int> seen;
  179. TRY(dump_tree(*this, page_object_index, seen));
  180. return {};
  181. }
  182. PDFErrorOr<Page> Document::get_page(u32 index)
  183. {
  184. VERIFY(index < m_page_object_indices.size());
  185. auto cached_page = m_pages.get(index);
  186. if (cached_page.has_value())
  187. return cached_page.value();
  188. auto page_object_index = m_page_object_indices[index];
  189. auto page_object = TRY(get_or_load_value(page_object_index));
  190. auto raw_page_object = TRY(resolve_to<DictObject>(page_object));
  191. RefPtr<DictObject> resources;
  192. auto maybe_resources_object = TRY(get_inheritable_object(CommonNames::Resources, raw_page_object));
  193. if (maybe_resources_object.has_value())
  194. resources = maybe_resources_object.value()->cast<DictObject>();
  195. else
  196. resources = make_object<DictObject>(HashMap<DeprecatedFlyString, Value> {});
  197. RefPtr<Object> contents;
  198. if (raw_page_object->contains(CommonNames::Contents))
  199. contents = TRY(raw_page_object->get_object(this, CommonNames::Contents));
  200. Rectangle media_box;
  201. auto maybe_media_box_object = TRY(get_inheritable_object(CommonNames::MediaBox, raw_page_object));
  202. if (maybe_media_box_object.has_value()) {
  203. auto media_box_array = maybe_media_box_object.value()->cast<ArrayObject>();
  204. media_box = Rectangle {
  205. media_box_array->at(0).to_float(),
  206. media_box_array->at(1).to_float(),
  207. media_box_array->at(2).to_float(),
  208. media_box_array->at(3).to_float(),
  209. };
  210. } else {
  211. // As most other libraries seem to do, we default to the standard
  212. // US letter size of 8.5" x 11" (612 x 792 Postscript units).
  213. media_box = Rectangle {
  214. 0, 0,
  215. 612, 792
  216. };
  217. }
  218. Rectangle crop_box;
  219. auto maybe_crop_box_object = TRY(get_inheritable_object(CommonNames::CropBox, raw_page_object));
  220. if (maybe_crop_box_object.has_value()) {
  221. auto crop_box_array = maybe_crop_box_object.value()->cast<ArrayObject>();
  222. crop_box = Rectangle {
  223. crop_box_array->at(0).to_float(),
  224. crop_box_array->at(1).to_float(),
  225. crop_box_array->at(2).to_float(),
  226. crop_box_array->at(3).to_float(),
  227. };
  228. } else {
  229. crop_box = media_box;
  230. }
  231. float user_unit = 1.0f;
  232. if (raw_page_object->contains(CommonNames::UserUnit))
  233. user_unit = raw_page_object->get_value(CommonNames::UserUnit).to_float();
  234. int rotate = 0;
  235. auto maybe_rotate = TRY(get_inheritable_value(CommonNames::Rotate, raw_page_object));
  236. if (maybe_rotate.has_value()) {
  237. rotate = TRY(resolve_to<int>(maybe_rotate.value()));
  238. VERIFY(rotate % 90 == 0);
  239. }
  240. Page page { resources.release_nonnull(), move(contents), media_box, crop_box, user_unit, rotate };
  241. m_pages.set(index, page);
  242. return page;
  243. }
  244. PDFErrorOr<Value> Document::resolve(Value const& value)
  245. {
  246. if (value.has<Reference>()) {
  247. // FIXME: Surely indirect PDF objects can't contain another indirect PDF object,
  248. // right? Unsure from the spec, but if they can, these return values would have
  249. // to be wrapped with another resolve() call.
  250. return get_or_load_value(value.as_ref_index());
  251. }
  252. if (!value.has<NonnullRefPtr<Object>>())
  253. return value;
  254. auto& obj = value.get<NonnullRefPtr<Object>>();
  255. if (obj->is<IndirectValue>())
  256. return static_ptr_cast<IndirectValue>(obj)->value();
  257. return value;
  258. }
  259. PDFErrorOr<Optional<InfoDict>> Document::info_dict()
  260. {
  261. if (!trailer()->contains(CommonNames::Info))
  262. return OptionalNone {};
  263. return InfoDict(this, TRY(trailer()->get_dict(this, CommonNames::Info)));
  264. }
  265. PDFErrorOr<Vector<DeprecatedFlyString>> Document::read_filters(NonnullRefPtr<DictObject> dict)
  266. {
  267. Vector<DeprecatedFlyString> filters;
  268. // We may either get a single filter or an array of cascading filters
  269. auto filter_object = TRY(dict->get_object(this, CommonNames::Filter));
  270. if (filter_object->is<ArrayObject>()) {
  271. auto filter_array = filter_object->cast<ArrayObject>();
  272. for (size_t i = 0; i < filter_array->size(); ++i)
  273. filters.append(TRY(filter_array->get_name_at(this, i))->name());
  274. } else {
  275. filters.append(filter_object->cast<NameObject>()->name());
  276. }
  277. return filters;
  278. }
  279. PDFErrorOr<void> Document::build_page_tree()
  280. {
  281. auto page_tree = TRY(m_catalog->get_dict(this, CommonNames::Pages));
  282. return add_page_tree_node_to_page_tree(page_tree);
  283. }
  284. PDFErrorOr<void> Document::add_page_tree_node_to_page_tree(NonnullRefPtr<DictObject> const& page_tree)
  285. {
  286. auto kids_array = TRY(page_tree->get_array(this, CommonNames::Kids));
  287. auto page_count = page_tree->get(CommonNames::Count).value().get<int>();
  288. if (static_cast<size_t>(page_count) != kids_array->elements().size()) {
  289. // This page tree contains child page trees, so we recursively add
  290. // these pages to the overall page tree
  291. for (auto& value : *kids_array) {
  292. auto reference_index = value.as_ref_index();
  293. auto maybe_page_tree_node = TRY(m_parser->conditionally_parse_page_tree_node(reference_index));
  294. if (maybe_page_tree_node) {
  295. TRY(add_page_tree_node_to_page_tree(maybe_page_tree_node.release_nonnull()));
  296. } else {
  297. m_page_object_indices.append(reference_index);
  298. }
  299. }
  300. } else {
  301. // We know all of the kids are leaf nodes
  302. for (auto& value : *kids_array)
  303. m_page_object_indices.append(value.as_ref_index());
  304. }
  305. return {};
  306. }
  307. PDFErrorOr<NonnullRefPtr<Object>> Document::find_in_name_tree(NonnullRefPtr<DictObject> tree, DeprecatedFlyString name)
  308. {
  309. if (tree->contains(CommonNames::Kids)) {
  310. return find_in_name_tree_nodes(tree->get_array(CommonNames::Kids), name);
  311. }
  312. if (!tree->contains(CommonNames::Names))
  313. return Error { Error::Type::MalformedPDF, "name tree has neither Kids nor Names" };
  314. auto key_value_names_array = TRY(tree->get_array(this, CommonNames::Names));
  315. return find_in_key_value_array(key_value_names_array, name);
  316. }
  317. PDFErrorOr<NonnullRefPtr<Object>> Document::find_in_name_tree_nodes(NonnullRefPtr<ArrayObject> siblings, DeprecatedFlyString name)
  318. {
  319. for (size_t i = 0; i < siblings->size(); i++) {
  320. auto sibling = TRY(resolve_to<DictObject>(siblings->at(i)));
  321. auto limits = sibling->get_array(CommonNames::Limits);
  322. if (limits->size() != 2)
  323. return Error { Error::Type::MalformedPDF, "Expected 2-element Limits array" };
  324. auto start = limits->get_string_at(0);
  325. auto end = limits->get_string_at(1);
  326. if (start->string() <= name && end->string() >= name) {
  327. return find_in_name_tree(sibling, name);
  328. }
  329. }
  330. return Error { Error::Type::MalformedPDF, ByteString::formatted("Didn't find node in name tree containing name {}", name) };
  331. }
  332. PDFErrorOr<NonnullRefPtr<Object>> Document::find_in_key_value_array(NonnullRefPtr<ArrayObject> key_value_array, DeprecatedFlyString name)
  333. {
  334. if (key_value_array->size() % 2 == 1)
  335. return Error { Error::Type::MalformedPDF, "key/value array has dangling key" };
  336. for (size_t i = 0; i < key_value_array->size() / 2; i++) {
  337. auto key = key_value_array->get_string_at(2 * i);
  338. if (key->string() == name) {
  339. return key_value_array->get_object_at(this, 2 * i + 1);
  340. }
  341. }
  342. return Error { Error::Type::MalformedPDF, ByteString::formatted("Didn't find expected name {} in key/value array", name) };
  343. }
  344. PDFErrorOr<void> Document::build_outline()
  345. {
  346. if (!m_catalog->contains(CommonNames::Outlines))
  347. return {};
  348. auto outline_dict = TRY(m_catalog->get_dict(this, CommonNames::Outlines));
  349. if (!outline_dict->contains(CommonNames::First))
  350. return {};
  351. if (!outline_dict->contains(CommonNames::Last))
  352. return {};
  353. HashMap<u32, u32> page_number_by_index_ref;
  354. for (u32 page_number = 0; page_number < m_page_object_indices.size(); ++page_number) {
  355. page_number_by_index_ref.set(m_page_object_indices[page_number], page_number);
  356. }
  357. auto first_ref = outline_dict->get_value(CommonNames::First);
  358. auto children = TRY(build_outline_item_chain(first_ref, page_number_by_index_ref));
  359. m_outline = adopt_ref(*new OutlineDict());
  360. m_outline->children = move(children);
  361. if (outline_dict->contains(CommonNames::Count))
  362. m_outline->count = outline_dict->get_value(CommonNames::Count).get<int>();
  363. return {};
  364. }
  365. PDFErrorOr<Destination> Document::create_destination_from_parameters(NonnullRefPtr<ArrayObject> array, HashMap<u32, u32> const& page_number_by_index_ref)
  366. {
  367. auto page_ref = array->at(0);
  368. if (page_ref.has<nullptr_t>())
  369. return Destination { Destination::Type::XYZ, {}, {} };
  370. auto type_name = TRY(array->get_name_at(this, 1))->name();
  371. Vector<Optional<float>> parameters;
  372. TRY(parameters.try_ensure_capacity(array->size() - 2));
  373. for (size_t i = 2; i < array->size(); i++) {
  374. auto& param = array->at(i);
  375. if (param.has<nullptr_t>())
  376. parameters.unchecked_append({});
  377. else
  378. parameters.append(param.to_float());
  379. }
  380. Destination::Type type;
  381. if (type_name == CommonNames::XYZ) {
  382. type = Destination::Type::XYZ;
  383. } else if (type_name == CommonNames::Fit) {
  384. type = Destination::Type::Fit;
  385. } else if (type_name == CommonNames::FitH) {
  386. type = Destination::Type::FitH;
  387. } else if (type_name == CommonNames::FitV) {
  388. type = Destination::Type::FitV;
  389. } else if (type_name == CommonNames::FitR) {
  390. type = Destination::Type::FitR;
  391. } else if (type_name == CommonNames::FitB) {
  392. type = Destination::Type::FitB;
  393. } else if (type_name == CommonNames::FitBH) {
  394. type = Destination::Type::FitBH;
  395. } else if (type_name == CommonNames::FitBV) {
  396. type = Destination::Type::FitBV;
  397. } else {
  398. VERIFY_NOT_REACHED();
  399. }
  400. // The spec requires page_ref to be an indirect reference to a page object,
  401. // but in practice it's sometimes a page index.
  402. Optional<u32> page_number;
  403. if (page_ref.has<int>())
  404. page_number = page_ref.get<int>();
  405. else
  406. page_number = page_number_by_index_ref.get(page_ref.as_ref_index());
  407. return Destination { type, page_number, parameters };
  408. }
  409. PDFErrorOr<Optional<NonnullRefPtr<Object>>> Document::get_inheritable_object(DeprecatedFlyString const& name, NonnullRefPtr<DictObject> object)
  410. {
  411. if (!object->contains(name)) {
  412. if (!object->contains(CommonNames::Parent))
  413. return { OptionalNone() };
  414. auto parent = TRY(object->get_dict(this, CommonNames::Parent));
  415. return get_inheritable_object(name, parent);
  416. }
  417. return TRY(object->get_object(this, name));
  418. }
  419. PDFErrorOr<Optional<Value>> Document::get_inheritable_value(DeprecatedFlyString const& name, NonnullRefPtr<DictObject> object)
  420. {
  421. if (!object->contains(name)) {
  422. if (!object->contains(CommonNames::Parent))
  423. return { OptionalNone() };
  424. auto parent = TRY(object->get_dict(this, CommonNames::Parent));
  425. return get_inheritable_value(name, parent);
  426. }
  427. return object->get(name);
  428. }
  429. PDFErrorOr<Destination> Document::create_destination_from_dictionary_entry(NonnullRefPtr<Object> const& entry, HashMap<u32, u32> const& page_number_by_index_ref)
  430. {
  431. if (entry->is<ArrayObject>()) {
  432. auto entry_array = entry->cast<ArrayObject>();
  433. return create_destination_from_parameters(entry_array, page_number_by_index_ref);
  434. }
  435. auto entry_dictionary = entry->cast<DictObject>();
  436. auto d_array = MUST(entry_dictionary->get_array(this, CommonNames::D));
  437. return create_destination_from_parameters(d_array, page_number_by_index_ref);
  438. }
  439. PDFErrorOr<Destination> Document::create_destination_from_object(NonnullRefPtr<Object> const& dest_obj, HashMap<u32, u32> const& page_number_by_index_ref)
  440. {
  441. // PDF 1.7 spec, "8.2.1 Destinations"
  442. if (dest_obj->is<ArrayObject>()) {
  443. auto dest_arr = dest_obj->cast<ArrayObject>();
  444. return TRY(create_destination_from_parameters(dest_arr, page_number_by_index_ref));
  445. }
  446. if (dest_obj->is<NameObject>() || dest_obj->is<StringObject>()) {
  447. DeprecatedFlyString dest_name;
  448. if (dest_obj->is<NameObject>())
  449. dest_name = dest_obj->cast<NameObject>()->name();
  450. else
  451. dest_name = dest_obj->cast<StringObject>()->string();
  452. if (auto dests_value = m_catalog->get(CommonNames::Dests); dests_value.has_value()) {
  453. auto dests = TRY(resolve_to<DictObject>(dests_value.value()));
  454. auto entry = MUST(dests->get_object(this, dest_name));
  455. return TRY(create_destination_from_dictionary_entry(entry, page_number_by_index_ref));
  456. }
  457. if (auto names_value = m_catalog->get(CommonNames::Names); names_value.has_value()) {
  458. auto names = TRY(resolve_to<DictObject>(names_value.release_value()));
  459. if (!names->contains(CommonNames::Dests))
  460. return Error { Error::Type::MalformedPDF, "Missing Dests key in document catalogue's Names dictionary" };
  461. auto dest_obj = TRY(find_in_name_tree(TRY(names->get_dict(this, CommonNames::Dests)), dest_name));
  462. return TRY(create_destination_from_dictionary_entry(dest_obj, page_number_by_index_ref));
  463. }
  464. }
  465. return Error { Error::Type::MalformedPDF, "Malformed outline destination" };
  466. }
  467. PDFErrorOr<NonnullRefPtr<OutlineItem>> Document::build_outline_item(NonnullRefPtr<DictObject> const& outline_item_dict, HashMap<u32, u32> const& page_number_by_index_ref)
  468. {
  469. auto outline_item = adopt_ref(*new OutlineItem {});
  470. if (outline_item_dict->contains(CommonNames::First)) {
  471. VERIFY(outline_item_dict->contains(CommonNames::Last));
  472. auto first_ref = outline_item_dict->get_value(CommonNames::First);
  473. auto children = TRY(build_outline_item_chain(first_ref, page_number_by_index_ref));
  474. for (auto& child : children) {
  475. child->parent = outline_item;
  476. }
  477. outline_item->children = move(children);
  478. }
  479. outline_item->title = text_string_to_utf8(TRY(outline_item_dict->get_string(this, CommonNames::Title))->string());
  480. if (outline_item_dict->contains(CommonNames::Count))
  481. outline_item->count = outline_item_dict->get_value(CommonNames::Count).get<int>();
  482. if (outline_item_dict->contains(CommonNames::Dest)) {
  483. auto dest_obj = TRY(outline_item_dict->get_object(this, CommonNames::Dest));
  484. outline_item->dest = TRY(create_destination_from_object(dest_obj, page_number_by_index_ref));
  485. } else if (outline_item_dict->contains(CommonNames::A)) {
  486. // PDF 1.7 spec, "8.5 Actions"
  487. auto action_dict = TRY(outline_item_dict->get_dict(this, CommonNames::A));
  488. if (action_dict->contains(CommonNames::S)) {
  489. // PDF 1.7 spec, "TABLE 8.48 Action types"
  490. auto action_type = TRY(action_dict->get_name(this, CommonNames::S))->name();
  491. if (action_type == "GoTo") {
  492. // PDF 1.7 spec, "Go-To Actions"
  493. if (action_dict->contains(CommonNames::D)) {
  494. auto dest_obj = TRY(action_dict->get_object(this, CommonNames::D));
  495. outline_item->dest = TRY(create_destination_from_object(dest_obj, page_number_by_index_ref));
  496. }
  497. } else {
  498. dbgln("Unhandled action type {}", action_type);
  499. }
  500. }
  501. }
  502. if (outline_item_dict->contains(CommonNames::C)) {
  503. auto color_array = TRY(outline_item_dict->get_array(this, CommonNames::C));
  504. auto r = static_cast<int>(255.0f * color_array->at(0).to_float());
  505. auto g = static_cast<int>(255.0f * color_array->at(1).to_float());
  506. auto b = static_cast<int>(255.0f * color_array->at(2).to_float());
  507. outline_item->color = Color(r, g, b);
  508. }
  509. if (outline_item_dict->contains(CommonNames::F)) {
  510. auto bitfield = outline_item_dict->get_value(CommonNames::F).get<int>();
  511. outline_item->italic = bitfield & 0x1;
  512. outline_item->bold = bitfield & 0x2;
  513. }
  514. return outline_item;
  515. }
  516. PDFErrorOr<Vector<NonnullRefPtr<OutlineItem>>> Document::build_outline_item_chain(Value const& first_ref, HashMap<u32, u32> const& page_number_by_index_ref)
  517. {
  518. // We used to receive a last_ref parameter, which was what the parent of this chain
  519. // thought was this chain's last child. There are documents out there in the wild
  520. // where this cross-references don't match though, and it seems like simply following
  521. // the /First and /Next links is the way to go to construct the whole Outline
  522. // (we already ignore the /Parent attribute too, which can also be out of sync).
  523. VERIFY(first_ref.has<Reference>());
  524. Vector<NonnullRefPtr<OutlineItem>> children;
  525. auto first_value = TRY(get_or_load_value(first_ref.as_ref_index())).get<NonnullRefPtr<Object>>();
  526. auto first_dict = first_value->cast<DictObject>();
  527. auto first = TRY(build_outline_item(first_dict, page_number_by_index_ref));
  528. children.append(first);
  529. auto current_child_dict = first_dict;
  530. u32 current_child_index = first_ref.as_ref_index();
  531. while (current_child_dict->contains(CommonNames::Next)) {
  532. auto next_child_dict_ref = current_child_dict->get_value(CommonNames::Next);
  533. current_child_index = next_child_dict_ref.as_ref_index();
  534. auto next_child_value = TRY(get_or_load_value(current_child_index)).get<NonnullRefPtr<Object>>();
  535. auto next_child_dict = next_child_value->cast<DictObject>();
  536. auto next_child = TRY(build_outline_item(next_child_dict, page_number_by_index_ref));
  537. children.append(next_child);
  538. current_child_dict = move(next_child_dict);
  539. }
  540. return children;
  541. }
  542. }