paste-html.ts 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413
  1. import { EditorView } from '@codemirror/view'
  2. import { EditorSelection, Prec } from '@codemirror/state'
  3. export const pasteHtml = Prec.highest(
  4. EditorView.domEventHandlers({
  5. paste(event, view) {
  6. const { clipboardData } = event
  7. if (!clipboardData?.types.includes('text/html')) {
  8. return false
  9. }
  10. const html = clipboardData.getData('text/html').trim()
  11. if (html.length === 0) {
  12. return false
  13. }
  14. // convert the HTML to LaTeX
  15. try {
  16. const latex = htmlToLaTeX(html)
  17. view.dispatch(
  18. view.state.changeByRange(range => {
  19. return {
  20. range: EditorSelection.cursor(range.from + latex.length),
  21. changes: { from: range.from, to: range.to, insert: latex },
  22. }
  23. })
  24. )
  25. return true
  26. } catch {
  27. // fall back to the default paste handler
  28. return false
  29. }
  30. },
  31. })
  32. )
  33. const removeUnwantedElements = (
  34. documentElement: HTMLElement,
  35. selector: string
  36. ) => {
  37. for (const element of documentElement.querySelectorAll(selector)) {
  38. element.remove()
  39. }
  40. }
  41. const htmlToLaTeX = (html: string) => {
  42. const parser = new DOMParser()
  43. const { documentElement } = parser.parseFromString(html, 'text/html')
  44. // remove style elements
  45. removeUnwantedElements(documentElement, 'style')
  46. // protect special characters in non-LaTeX text nodes
  47. protectSpecialCharacters(documentElement)
  48. processMatchedElements(documentElement)
  49. const text = documentElement.textContent
  50. if (!text) {
  51. return ''
  52. }
  53. // normalise multiple newlines
  54. return text.replaceAll(/\n{2,}/g, '\n\n')
  55. }
  56. const protectSpecialCharacters = (documentElement: HTMLElement) => {
  57. for (const element of documentElement.childNodes) {
  58. const text = element.textContent
  59. if (text) {
  60. // if there are no code blocks, use backslash as an indicator of LaTeX code that shouldn't be protected
  61. if (
  62. element instanceof HTMLElement &&
  63. !element.querySelector('code') &&
  64. text.includes('\\')
  65. ) {
  66. continue
  67. }
  68. const walker = document.createTreeWalker(element, NodeFilter.SHOW_TEXT)
  69. for (let node = walker.nextNode(); node; node = walker.nextNode()) {
  70. const text = node.textContent
  71. if (text) {
  72. // leave text that's in a code block
  73. if (node.parentElement?.closest('code')) {
  74. continue
  75. }
  76. // replace non-backslash-prefixed characters
  77. node.textContent = text.replaceAll(
  78. /(^|[^\\])([#$%&~_^\\{}])/g,
  79. (_match, prefix: string, char: string) => `${prefix}\\${char}`
  80. )
  81. }
  82. }
  83. }
  84. }
  85. }
  86. const processMatchedElements = (documentElement: HTMLElement) => {
  87. for (const item of selectors) {
  88. for (const element of documentElement.querySelectorAll<any>(
  89. item.selector
  90. )) {
  91. if (!item.match || item.match(element)) {
  92. // start the markup
  93. if (item.start) {
  94. const start = document.createTextNode(item.start(element))
  95. if (item.inside) {
  96. element.prepend(start)
  97. } else {
  98. element.before(start)
  99. }
  100. }
  101. // end the markup
  102. if (item.end) {
  103. const end = document.createTextNode(item.end(element))
  104. if (item.inside) {
  105. element.append(end)
  106. } else {
  107. element.after(end)
  108. }
  109. }
  110. }
  111. }
  112. }
  113. }
  114. const matchingParents = (element: HTMLElement, selector: string) => {
  115. const matches = []
  116. for (
  117. let ancestor = element.parentElement?.closest(selector);
  118. ancestor;
  119. ancestor = ancestor.parentElement?.closest(selector)
  120. ) {
  121. matches.push(ancestor)
  122. }
  123. return matches
  124. }
  125. const tabular = (element: HTMLTableElement) => {
  126. const options = []
  127. // NOTE: only analysing cells in the first row
  128. const row = element.querySelector('tr')
  129. if (row) {
  130. // TODO: look for horizontal borders and insert \hline (or \toprule, \midrule, \bottomrule etc)?
  131. const cells = [...row.childNodes].filter(
  132. element => element.nodeName === 'TD' || element.nodeName === 'TH'
  133. ) as Array<HTMLTableCellElement | HTMLTableHeaderCellElement>
  134. for (const cell of cells) {
  135. const { borderLeft, textAlign, borderRight } = cell.style
  136. if (borderLeft && borderLeft !== 'none') {
  137. // avoid duplicating when both left and right borders are defined
  138. if (options.length === 0 || options[options.length - 1] !== '|') {
  139. options.push('|')
  140. }
  141. }
  142. options.push(
  143. textAlign === 'left' ? 'l' : textAlign === 'right' ? 'r' : 'c'
  144. )
  145. if (borderRight && borderRight !== 'none') {
  146. options.push('|')
  147. }
  148. }
  149. }
  150. return options.join(' ')
  151. }
  152. const listDepth = (
  153. element: HTMLOListElement | HTMLUListElement | HTMLLIElement
  154. ): number => Math.max(0, matchingParents(element, 'ul,ol').length - 1)
  155. const listIndent = (
  156. element: HTMLOListElement | HTMLUListElement | HTMLLIElement
  157. ): string => '\t'.repeat(listDepth(element))
  158. type ElementSelector<T extends string, E extends HTMLElement = HTMLElement> = {
  159. selector: T
  160. match?: (element: E) => boolean
  161. start?: (element: E) => string
  162. end?: (element: E) => string
  163. inside?: boolean
  164. }
  165. const createSelector = <
  166. T extends string,
  167. E extends HTMLElement = T extends keyof HTMLElementTagNameMap
  168. ? HTMLElementTagNameMap[T]
  169. : HTMLElement
  170. >({
  171. selector,
  172. ...elementSelector
  173. }: ElementSelector<T, E>) => ({
  174. selector,
  175. ...elementSelector,
  176. })
  177. const headings = ['H1', 'H2', 'H3', 'H4', 'H5', 'H6']
  178. const isHeading = (element: HTMLElement | null) => {
  179. return element && headings.includes(element.nodeName)
  180. }
  181. const hasContent = (element: HTMLElement): boolean => {
  182. return Boolean(element.textContent && element.textContent.trim().length > 0)
  183. }
  184. export const selectors = [
  185. createSelector({
  186. selector: 'b',
  187. match: element =>
  188. element.style.fontWeight !== 'normal' &&
  189. !isHeading(element.parentElement) &&
  190. hasContent(element),
  191. start: () => '\\textbf{',
  192. end: () => '}',
  193. }),
  194. createSelector({
  195. selector: '*',
  196. match: element =>
  197. parseInt(element.style.fontWeight) > 400 && hasContent(element),
  198. start: () => '\\textbf{',
  199. end: () => '}',
  200. inside: true,
  201. }),
  202. createSelector({
  203. selector: 'i',
  204. match: element =>
  205. element.style.fontStyle !== 'normal' && hasContent(element),
  206. start: () => '\\textit{',
  207. end: () => '}',
  208. }),
  209. createSelector({
  210. selector: '*',
  211. match: element =>
  212. element.style.fontStyle === 'italic' && hasContent(element),
  213. start: () => '\\textit{',
  214. end: () => '}',
  215. }),
  216. createSelector({
  217. selector: 'sup',
  218. match: element => hasContent(element),
  219. start: () => '\\textsuperscript{',
  220. end: () => '}',
  221. }),
  222. createSelector({
  223. selector: 'span',
  224. match: element =>
  225. element.style.verticalAlign === 'super' && hasContent(element),
  226. start: () => '\\textsuperscript{',
  227. end: () => '}',
  228. }),
  229. createSelector({
  230. selector: 'sub',
  231. match: element => hasContent(element),
  232. start: () => '\\textsubscript{',
  233. end: () => '}',
  234. }),
  235. createSelector({
  236. selector: 'span',
  237. match: element =>
  238. element.style.verticalAlign === 'sub' && hasContent(element),
  239. start: () => '\\textsubscript{',
  240. end: () => '}',
  241. }),
  242. createSelector({
  243. selector: 'a',
  244. match: element => !!element.href && hasContent(element),
  245. start: (element: HTMLAnchorElement) => `\\href{${element.href}}{`,
  246. end: element => `}`,
  247. }),
  248. createSelector({
  249. selector: 'h1',
  250. match: element => !element.closest('table') && hasContent(element),
  251. start: () => `\n\n\\section{`,
  252. end: () => `}\n\n`,
  253. }),
  254. createSelector({
  255. selector: 'h2',
  256. match: element => !element.closest('table') && hasContent(element),
  257. start: () => `\n\n\\subsection{`,
  258. end: () => `}\n\n`,
  259. }),
  260. createSelector({
  261. selector: 'h3',
  262. match: element => !element.closest('table') && hasContent(element),
  263. start: () => `\n\n\\subsubsection{`,
  264. end: () => `}\n\n`,
  265. }),
  266. createSelector({
  267. selector: 'h4',
  268. match: element => !element.closest('table') && hasContent(element),
  269. start: () => `\n\n\\paragraph{`,
  270. end: () => `}\n\n`,
  271. }),
  272. createSelector({
  273. selector: 'h5',
  274. match: element => !element.closest('table') && hasContent(element),
  275. start: () => `\n\n\\subparagraph{`,
  276. end: () => `}\n\n`,
  277. }),
  278. // TODO: h6?
  279. createSelector({
  280. selector: 'br',
  281. match: element => element.parentElement?.nodeName !== 'TD', // TODO: why?
  282. start: () => `\n\n`,
  283. }),
  284. createSelector({
  285. selector: 'code',
  286. match: element =>
  287. element.parentElement?.nodeName !== 'PRE' && hasContent(element),
  288. start: () => `\\verb|`,
  289. end: () => `|`,
  290. }),
  291. createSelector({
  292. selector: 'pre > code',
  293. match: element => hasContent(element),
  294. start: () => `\n\n\\begin{verbatim}\n`,
  295. end: () => `\n\\end{verbatim}\n\n`,
  296. }),
  297. createSelector({
  298. selector: 'table',
  299. start: element =>
  300. `\n\n\\begin{table}\n\\centering\n\\begin{tabular}{${tabular(
  301. element
  302. )}}\n`,
  303. end: () => `\n\\end{tabular}\n\\end{table}\n\n`,
  304. }),
  305. createSelector({
  306. selector: 'thead',
  307. start: () => `\n`,
  308. end: () => `\n`,
  309. }),
  310. createSelector({
  311. selector: 'tfoot',
  312. start: () => `\n`,
  313. end: () => `\n`,
  314. }),
  315. createSelector({
  316. selector: 'tbody',
  317. start: () => `\n`,
  318. end: () => `\n`,
  319. }),
  320. createSelector({
  321. selector: 'tr',
  322. match: element => element.nextElementSibling?.nodeName === 'TR',
  323. end: () => `\n`,
  324. }),
  325. createSelector({
  326. selector: 'tr > td:not(:last-child), tr > th:not(:last-child)',
  327. start: element => {
  328. const colspan = element.getAttribute('colspan')
  329. return colspan ? `\\multicolumn{${Number(colspan)}}{` : ''
  330. },
  331. end: element => {
  332. const colspan = element.getAttribute('colspan')
  333. return colspan ? `} & ` : ` & `
  334. },
  335. }),
  336. createSelector({
  337. selector: 'tr > td:last-child, tr > th:last-child',
  338. start: element => {
  339. const colspan = element.getAttribute('colspan')
  340. return colspan ? `\\multicolumn{${Number(colspan)}}{` : ''
  341. },
  342. end: element => {
  343. const colspan = element.getAttribute('colspan')
  344. return colspan ? `} \\\\` : ` \\\\`
  345. },
  346. }),
  347. createSelector({
  348. selector: 'table > caption',
  349. start: () => `\n\n\\caption{\\label{tab:example}`,
  350. end: () => `}\n\n`,
  351. }),
  352. createSelector({
  353. selector: 'ul',
  354. start: element => `\n\n${listIndent(element)}\\begin{itemize}`,
  355. end: element => `\n${listIndent(element)}\\end{itemize}\n`,
  356. }),
  357. createSelector({
  358. selector: 'ol',
  359. start: element => `\n\n${listIndent(element)}\\begin{enumerate}`,
  360. end: element => `\n${listIndent(element)}\\end{enumerate}\n`,
  361. }),
  362. createSelector({
  363. selector: 'li',
  364. start: element => `\n${listIndent(element)}\t\\item `,
  365. }),
  366. createSelector({
  367. selector: 'p',
  368. match: element =>
  369. element.nextElementSibling?.nodeName === 'P' && hasContent(element),
  370. end: () => '\n\n',
  371. }),
  372. ]