TextNode.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470
  1. /*
  2. * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  3. * Copyright (c) 2022, Tobias Christiansen <tobyase@serenityos.org>
  4. *
  5. * SPDX-License-Identifier: BSD-2-Clause
  6. */
  7. #include <AK/CharacterTypes.h>
  8. #include <AK/StringBuilder.h>
  9. #include <LibUnicode/CharacterTypes.h>
  10. #include <LibWeb/DOM/Document.h>
  11. #include <LibWeb/Layout/BlockContainer.h>
  12. #include <LibWeb/Layout/InlineFormattingContext.h>
  13. #include <LibWeb/Layout/TextNode.h>
  14. #include <LibWeb/Painting/TextPaintable.h>
  15. namespace Web::Layout {
  16. TextNode::TextNode(DOM::Document& document, DOM::Text& text)
  17. : Node(document, &text)
  18. {
  19. }
  20. TextNode::~TextNode() = default;
  21. static bool is_all_whitespace(StringView string)
  22. {
  23. for (size_t i = 0; i < string.length(); ++i) {
  24. if (!is_ascii_space(string[i]))
  25. return false;
  26. }
  27. return true;
  28. }
  29. // https://w3c.github.io/mathml-core/#new-text-transform-values
  30. static DeprecatedString apply_math_auto_text_transform(DeprecatedString const& string)
  31. {
  32. // https://w3c.github.io/mathml-core/#italic-mappings
  33. auto map_code_point_to_italic = [](u32 code_point) -> u32 {
  34. switch (code_point) {
  35. case 0x0041:
  36. return 0x1D434;
  37. case 0x0042:
  38. return 0x1D435;
  39. case 0x0043:
  40. return 0x1D436;
  41. case 0x0044:
  42. return 0x1D437;
  43. case 0x0045:
  44. return 0x1D438;
  45. case 0x0046:
  46. return 0x1D439;
  47. case 0x0047:
  48. return 0x1D43A;
  49. case 0x0048:
  50. return 0x1D43B;
  51. case 0x0049:
  52. return 0x1D43C;
  53. case 0x004A:
  54. return 0x1D43D;
  55. case 0x004B:
  56. return 0x1D43E;
  57. case 0x004C:
  58. return 0x1D43F;
  59. case 0x004D:
  60. return 0x1D440;
  61. case 0x004E:
  62. return 0x1D441;
  63. case 0x004F:
  64. return 0x1D442;
  65. case 0x0050:
  66. return 0x1D443;
  67. case 0x0051:
  68. return 0x1D444;
  69. case 0x0052:
  70. return 0x1D445;
  71. case 0x0053:
  72. return 0x1D446;
  73. case 0x0054:
  74. return 0x1D447;
  75. case 0x0055:
  76. return 0x1D448;
  77. case 0x0056:
  78. return 0x1D449;
  79. case 0x0057:
  80. return 0x1D44A;
  81. case 0x0058:
  82. return 0x1D44B;
  83. case 0x0059:
  84. return 0x1D44C;
  85. case 0x005A:
  86. return 0x1D44D;
  87. case 0x0061:
  88. return 0x1D44E;
  89. case 0x0062:
  90. return 0x1D44F;
  91. case 0x0063:
  92. return 0x1D450;
  93. case 0x0064:
  94. return 0x1D451;
  95. case 0x0065:
  96. return 0x1D452;
  97. case 0x0066:
  98. return 0x1D453;
  99. case 0x0067:
  100. return 0x1D454;
  101. case 0x0068:
  102. return 0x0210E;
  103. case 0x0069:
  104. return 0x1D456;
  105. case 0x006A:
  106. return 0x1D457;
  107. case 0x006B:
  108. return 0x1D458;
  109. case 0x006C:
  110. return 0x1D459;
  111. case 0x006D:
  112. return 0x1D45A;
  113. case 0x006E:
  114. return 0x1D45B;
  115. case 0x006F:
  116. return 0x1D45C;
  117. case 0x0070:
  118. return 0x1D45D;
  119. case 0x0071:
  120. return 0x1D45E;
  121. case 0x0072:
  122. return 0x1D45F;
  123. case 0x0073:
  124. return 0x1D460;
  125. case 0x0074:
  126. return 0x1D461;
  127. case 0x0075:
  128. return 0x1D462;
  129. case 0x0076:
  130. return 0x1D463;
  131. case 0x0077:
  132. return 0x1D464;
  133. case 0x0078:
  134. return 0x1D465;
  135. case 0x0079:
  136. return 0x1D466;
  137. case 0x007A:
  138. return 0x1D467;
  139. case 0x0131:
  140. return 0x1D6A4;
  141. case 0x0237:
  142. return 0x1D6A5;
  143. case 0x0391:
  144. return 0x1D6E2;
  145. case 0x0392:
  146. return 0x1D6E3;
  147. case 0x0393:
  148. return 0x1D6E4;
  149. case 0x0394:
  150. return 0x1D6E5;
  151. case 0x0395:
  152. return 0x1D6E6;
  153. case 0x0396:
  154. return 0x1D6E7;
  155. case 0x0397:
  156. return 0x1D6E8;
  157. case 0x0398:
  158. return 0x1D6E9;
  159. case 0x0399:
  160. return 0x1D6EA;
  161. case 0x039A:
  162. return 0x1D6EB;
  163. case 0x039B:
  164. return 0x1D6EC;
  165. case 0x039C:
  166. return 0x1D6ED;
  167. case 0x039D:
  168. return 0x1D6EE;
  169. case 0x039E:
  170. return 0x1D6EF;
  171. case 0x039F:
  172. return 0x1D6F0;
  173. case 0x03A0:
  174. return 0x1D6F1;
  175. case 0x03A1:
  176. return 0x1D6F2;
  177. case 0x03F4:
  178. return 0x1D6F3;
  179. case 0x03A3:
  180. return 0x1D6F4;
  181. case 0x03A4:
  182. return 0x1D6F5;
  183. case 0x03A5:
  184. return 0x1D6F6;
  185. case 0x03A6:
  186. return 0x1D6F7;
  187. case 0x03A7:
  188. return 0x1D6F8;
  189. case 0x03A8:
  190. return 0x1D6F9;
  191. case 0x03A9:
  192. return 0x1D6FA;
  193. case 0x2207:
  194. return 0x1D6FB;
  195. case 0x03B1:
  196. return 0x1D6FC;
  197. case 0x03B2:
  198. return 0x1D6FD;
  199. case 0x03B3:
  200. return 0x1D6FE;
  201. case 0x03B4:
  202. return 0x1D6FF;
  203. case 0x03B5:
  204. return 0x1D700;
  205. case 0x03B6:
  206. return 0x1D701;
  207. case 0x03B7:
  208. return 0x1D702;
  209. case 0x03B8:
  210. return 0x1D703;
  211. case 0x03B9:
  212. return 0x1D704;
  213. case 0x03BA:
  214. return 0x1D705;
  215. case 0x03BB:
  216. return 0x1D706;
  217. case 0x03BC:
  218. return 0x1D707;
  219. case 0x03BD:
  220. return 0x1D708;
  221. case 0x03BE:
  222. return 0x1D709;
  223. case 0x03BF:
  224. return 0x1D70A;
  225. case 0x03C0:
  226. return 0x1D70B;
  227. case 0x03C1:
  228. return 0x1D70C;
  229. case 0x03C2:
  230. return 0x1D70D;
  231. case 0x03C3:
  232. return 0x1D70E;
  233. case 0x03C4:
  234. return 0x1D70F;
  235. case 0x03C5:
  236. return 0x1D710;
  237. case 0x03C6:
  238. return 0x1D711;
  239. case 0x03C7:
  240. return 0x1D712;
  241. case 0x03C8:
  242. return 0x1D713;
  243. case 0x03C9:
  244. return 0x1D714;
  245. case 0x2202:
  246. return 0x1D715;
  247. case 0x03F5:
  248. return 0x1D716;
  249. case 0x03D1:
  250. return 0x1D717;
  251. case 0x03F0:
  252. return 0x1D718;
  253. case 0x03D5:
  254. return 0x1D719;
  255. case 0x03F1:
  256. return 0x1D71A;
  257. case 0x03D6:
  258. return 0x1D71B;
  259. default:
  260. return code_point;
  261. }
  262. };
  263. StringBuilder builder(string.length());
  264. for (auto code_point : Utf8View(string))
  265. builder.append_code_point(map_code_point_to_italic(code_point));
  266. return builder.to_deprecated_string();
  267. }
  268. static ErrorOr<DeprecatedString> apply_text_transform(DeprecatedString const& string, CSS::TextTransform text_transform)
  269. {
  270. switch (text_transform) {
  271. case CSS::TextTransform::Uppercase:
  272. return Unicode::to_unicode_uppercase_full(string);
  273. case CSS::TextTransform::Lowercase:
  274. return Unicode::to_unicode_lowercase_full(string);
  275. case CSS::TextTransform::None:
  276. return string;
  277. case CSS::TextTransform::MathAuto:
  278. return apply_math_auto_text_transform(string);
  279. case CSS::TextTransform::Capitalize: {
  280. return TRY(Unicode::to_unicode_titlecase_full(string, {}, Unicode::TrailingCodePointTransformation::PreserveExisting)).to_deprecated_string();
  281. }
  282. case CSS::TextTransform::FullSizeKana:
  283. case CSS::TextTransform::FullWidth:
  284. // FIXME: Implement these!
  285. return string;
  286. }
  287. VERIFY_NOT_REACHED();
  288. }
  289. void TextNode::invalidate_text_for_rendering()
  290. {
  291. m_text_for_rendering = {};
  292. }
  293. DeprecatedString const& TextNode::text_for_rendering() const
  294. {
  295. if (m_text_for_rendering.is_null())
  296. const_cast<TextNode*>(this)->compute_text_for_rendering();
  297. return m_text_for_rendering;
  298. }
  299. // NOTE: This collapses whitespace into a single ASCII space if the CSS white-space property tells us to.
  300. void TextNode::compute_text_for_rendering()
  301. {
  302. bool collapse = [](CSS::WhiteSpace white_space) {
  303. switch (white_space) {
  304. case CSS::WhiteSpace::Normal:
  305. case CSS::WhiteSpace::Nowrap:
  306. case CSS::WhiteSpace::PreLine:
  307. return true;
  308. case CSS::WhiteSpace::Pre:
  309. case CSS::WhiteSpace::PreWrap:
  310. return false;
  311. }
  312. VERIFY_NOT_REACHED();
  313. }(computed_values().white_space());
  314. if (dom_node().is_editable() && !dom_node().is_uninteresting_whitespace_node())
  315. collapse = false;
  316. auto data = apply_text_transform(dom_node().data().to_deprecated_string(), computed_values().text_transform()).release_value_but_fixme_should_propagate_errors();
  317. if (dom_node().is_password_input()) {
  318. m_text_for_rendering = DeprecatedString::repeated('*', data.length());
  319. return;
  320. }
  321. if (!collapse || data.is_empty()) {
  322. m_text_for_rendering = data;
  323. return;
  324. }
  325. // NOTE: A couple fast returns to avoid unnecessarily allocating a StringBuilder.
  326. if (data.length() == 1) {
  327. if (is_ascii_space(data[0])) {
  328. static DeprecatedString s_single_space_string = " ";
  329. m_text_for_rendering = s_single_space_string;
  330. } else {
  331. m_text_for_rendering = data;
  332. }
  333. return;
  334. }
  335. bool contains_space = false;
  336. for (auto& c : data) {
  337. if (is_ascii_space(c)) {
  338. contains_space = true;
  339. break;
  340. }
  341. }
  342. if (!contains_space) {
  343. m_text_for_rendering = data;
  344. return;
  345. }
  346. StringBuilder builder(data.length());
  347. size_t index = 0;
  348. auto skip_over_whitespace = [&index, &data] {
  349. while (index < data.length() && is_ascii_space(data[index]))
  350. ++index;
  351. };
  352. while (index < data.length()) {
  353. if (is_ascii_space(data[index])) {
  354. builder.append(' ');
  355. ++index;
  356. skip_over_whitespace();
  357. } else {
  358. builder.append(data[index]);
  359. ++index;
  360. }
  361. }
  362. m_text_for_rendering = builder.to_deprecated_string();
  363. }
  364. TextNode::ChunkIterator::ChunkIterator(StringView text, bool wrap_lines, bool respect_linebreaks)
  365. : m_wrap_lines(wrap_lines)
  366. , m_respect_linebreaks(respect_linebreaks)
  367. , m_utf8_view(text)
  368. , m_iterator(m_utf8_view.begin())
  369. {
  370. }
  371. Optional<TextNode::Chunk> TextNode::ChunkIterator::next()
  372. {
  373. if (m_iterator == m_utf8_view.end())
  374. return {};
  375. auto start_of_chunk = m_iterator;
  376. while (m_iterator != m_utf8_view.end()) {
  377. if (m_respect_linebreaks && *m_iterator == '\n') {
  378. // Newline encountered, and we're supposed to preserve them.
  379. // If we have accumulated some code points in the current chunk, commit them now and continue with the newline next time.
  380. if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false); result.has_value())
  381. return result.release_value();
  382. // Otherwise, commit the newline!
  383. ++m_iterator;
  384. auto result = try_commit_chunk(start_of_chunk, m_iterator, true);
  385. VERIFY(result.has_value());
  386. return result.release_value();
  387. }
  388. if (m_wrap_lines) {
  389. if (is_ascii_space(*m_iterator)) {
  390. // Whitespace encountered, and we're allowed to break on whitespace.
  391. // If we have accumulated some code points in the current chunk, commit them now and continue with the whitespace next time.
  392. if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false); result.has_value())
  393. return result.release_value();
  394. // Otherwise, commit the whitespace!
  395. ++m_iterator;
  396. if (auto result = try_commit_chunk(start_of_chunk, m_iterator, false); result.has_value())
  397. return result.release_value();
  398. continue;
  399. }
  400. }
  401. ++m_iterator;
  402. }
  403. if (start_of_chunk != m_utf8_view.end()) {
  404. // Try to output whatever's left at the end of the text node.
  405. if (auto result = try_commit_chunk(start_of_chunk, m_utf8_view.end(), false); result.has_value())
  406. return result.release_value();
  407. }
  408. return {};
  409. }
  410. Optional<TextNode::Chunk> TextNode::ChunkIterator::try_commit_chunk(Utf8View::Iterator const& start, Utf8View::Iterator const& end, bool has_breaking_newline) const
  411. {
  412. auto byte_offset = m_utf8_view.byte_offset_of(start);
  413. auto byte_length = m_utf8_view.byte_offset_of(end) - byte_offset;
  414. if (byte_length > 0) {
  415. auto chunk_view = m_utf8_view.substring_view(byte_offset, byte_length);
  416. return Chunk {
  417. .view = chunk_view,
  418. .start = byte_offset,
  419. .length = byte_length,
  420. .has_breaking_newline = has_breaking_newline,
  421. .is_all_whitespace = is_all_whitespace(chunk_view.as_string()),
  422. };
  423. }
  424. return {};
  425. }
  426. JS::GCPtr<Painting::Paintable> TextNode::create_paintable() const
  427. {
  428. return Painting::TextPaintable::create(*this);
  429. }
  430. }