gitoria

19.4 KB

1. // markdown.hl — copied from tickets.worldapi.org (markdown.hl, ticket tickets#6) for the repo homepage (gitoria#8), with tables, quotes and rules added. Originally: MARKDOWN IN TEXT FIELDS (mission 014; CONCEPT.md
2. // "Data formats": Markdown inside text fields). The summary, comment texts and state notes are
3. // Markdown; the ticket page renders them. Statics only, server realm.
4. //
5. // SAFE BY CONSTRUCTION: the text is parsed HERE into plain data — blocks of spans — and the
6. // View (components/markdown.hl) builds elements from it. Every character of the source ends
7. // up in a TEXT node (hl:web escapes it); there is no HTML string anywhere, so raw HTML in the
8. // source (`<script>`, `<img onerror>`) is shown as the text it is. A link's href is taken only
9. // if `safeHref` accepts it: http(s):, mailto:, a same-site path (`/…`, not `//…`) or `#…`;
10. // anything else (`javascript:`, `data:`, `vbscript:`, relative `foo`, quotes, spaces, control
11. // characters) stays plain text.
12. //
13. // WHAT IS UNDERSTOOD (the subset ticket #6 names — headings, lists, code, links, emphasis):
14. // blocks: `#`…`######` headings, `-`/`*`/`+` and `1.`/`1)` list items (one level; deeper
15. // indentation is flattened; a list may interrupt a paragraph — the import writes
16. // them that way), ``` / ~~~ fenced code, paragraphs (blank-line separated; their
17. // single line breaks are KEPT — the page showed them before, `white-space: pre-wrap`)
18. // inline: `code` (any backtick run), [text](url), <https://…>, bare http(s):// URLs,
19. // *em* _em_ **strong** __strong__ ***both***, backslash escapes of punctuation
20. // Also understood here (gitoria#8): `> block quotes` (inline text only), | tables |, horizontal rules.
21. // Not understood (shown as typed): images, nested lists, HTML.
22. //
23. // A block: { isP | isH1 | isH2 | isH3 | isUl | isOl | isCode | isQuote | isTable | isHr, head/rows (table: cells [{ spans }]), spans, items: [{ spans }],
24. // start (ol), code (text of a code block) }
25. // A span: { text, isText | isStrong | isEm | isStrongEm | isCode | isLink, href }
27. static NL = "
28. "
30. static isSpaceC = (c) => { return c == 32 || c == 9 || c == 10 || c == 13 }
31. static isAlnumC = (c) => { return (c >= 48 && c <= 57) || (c >= 65 && c <= 90) || (c >= 97 && c <= 122) || c > 127 }
32. static isPunctC = (c) => { return (c >= 33 && c <= 47) || (c >= 58 && c <= 64) || (c >= 91 && c <= 96) || (c >= 123 && c <= 126) }
34. // ---- links ------------------------------------------------------------------------------
35. // the href of a link, or null when it is not one we let through
36. static safeHref = (u) => {
37. if (u == null) { return null }
38. let t = ('' + u).trim()
39. if (t == '' || t.length > 2000) { return null }
40. let i = 0
41. while (i < t.length) {
42. let c = t.charCodeAt(i)
43. // no spaces / control characters, no quotes, no < > ` and no backslash
44. if (c <= 32 || c == 127 || c == 34 || c == 39 || c == 60 || c == 62 || c == 96 || c == 92) { return null }
45. i = i + 1
46. }
47. let low = t.toLowerCase()
48. if (low.startsWith('https://') && t.length > 8) { return t }
49. if (low.startsWith('http://') && t.length > 7) { return t }
50. if (low.startsWith('mailto:') && t.length > 7) { return t }
51. if (t.startsWith('/') && !t.startsWith('//')) { return t }
52. if (t.startsWith('#')) { return t }
53. return null
54. }
56. static isUrlStart = (s, i) => {
57. let low = s.slice(i, i + 8).toLowerCase()
58. return low.startsWith('https://') || low.startsWith('http://')
59. }
61. // where a bare URL starting at i ends: up to whitespace or < > " `, minus trailing punctuation
62. // (and a closing parenthesis that has no opening one inside the URL)
63. static isUrlStop = (c) => { return isSpaceC(c) || c == 60 || c == 62 || c == 34 || c == 96 }
64. static urlEnd = (s, i) => {
65. let end = i
66. while (end < s.length && !isUrlStop(s.charCodeAt(end))) { end = end + 1 }
67. let trimming = true
68. while (trimming && end > i) {
69. let c = s.charCodeAt(end - 1)
70. if (c == 46 || c == 44 || c == 59 || c == 58 || c == 33 || c == 63 || c == 39 || c == 42 || c == 95) {
71. end = end - 1
72. } else if (c == 41) {
73. let opens = 0
74. let closes = 0
75. let k = i
76. while (k < end) {
77. if (s.charCodeAt(k) == 40) { opens = opens + 1 }
78. if (s.charCodeAt(k) == 41) { closes = closes + 1 }
79. k = k + 1
80. }
81. if (closes > opens) { end = end - 1 } else { trimming = false }
82. } else {
83. trimming = false
84. }
85. }
86. return end
87. }
89. // `[label](url)` at i → { label, href (null = not allowed), end } or null (not a link)
90. static linkAt = (s, i) => {
91. let j = i + 1
92. let found = -1
93. while (j < s.length && found < 0) {
94. let c = s.charCodeAt(j)
95. if (c == 91) { return null }
96. if (c == 93) { found = j }
97. j = j + 1
98. }
99. if (found < 0 || found == i + 1) { return null }
100. if (found + 1 >= s.length || s.charCodeAt(found + 1) != 40) { return null }
101. let k = s.indexOf(')', found + 2)
102. if (k < 0) { return null }
103. let target = s.slice(found + 2, k).trim()
104. // an optional title after a space is dropped: [x](url "title")
105. let sp = target.indexOf(' ')
106. if (sp > 0) { target = target.slice(0, sp) }
107. return { label = s.slice(i + 1, found) href = safeHref(target) end = k + 1 }
108. }
110. // ---- spans --------------------------------------------------------------------------------
111. static textSpan = (t, strong, em) => {
112. return { text = t isText = !strong && !em isStrong = strong && !em isEm = em && !strong isStrongEm = strong && em isCode = false isLink = false href = '' }
113. }
114. static codeSpan = (t) => {
115. return { text = t isText = false isStrong = false isEm = false isStrongEm = false isCode = true isLink = false href = '' }
116. }
117. static linkSpan = (t, h) => {
118. return { text = t isText = false isStrong = false isEm = false isStrongEm = false isCode = false isLink = true href = h }
119. }
121. // the index of a run of EXACTLY n characters c at or after `at`, or -1. `esc`: a character
122. // after a backslash does not count (emphasis; inside `code` a backslash is just a backslash)
123. static runAt = (s, at, c, n, esc) => {
124. let i = at
125. while (i < s.length) {
126. if (esc && s.charCodeAt(i) == 92) {
127. i = i + 2
128. } else if (s.charCodeAt(i) == c) {
129. let k = i
130. while (k < s.length && s.charCodeAt(k) == c) { k = k + 1 }
131. if (k - i == n) { return i }
132. i = k
133. } else {
134. i = i + 1
135. }
136. }
137. return -1
138. }
140. // is there a CLOSING delimiter run (exactly n of c, not after a space; `_` not before a
141. // letter/digit) after `at`?
142. static hasCloser = (s, at, c, n) => {
143. let i = at
144. while (i < s.length) {
145. let r = runAt(s, i, c, n, true)
146. if (r < 0) { return false }
147. let prev = r > 0 ? s.charCodeAt(r - 1) : 32
148. let next = r + n < s.length ? s.charCodeAt(r + n) : 32
149. if (r > at && !isSpaceC(prev) && (c != 95 || !isAlnumC(next))) { return true }
150. i = r + n
151. }
152. return false
153. }
155. static inlineSpans = (src) => {
156. let out = []
157. if (src == null || src == '') { return out }
158. let s = '' + src
159. let n = s.length
160. let buf = ''
161. let strong = false
162. let em = false
163. let strongC = 0
164. let emC = 0
165. let i = 0
166. while (i < n) {
167. let c = s.charCodeAt(i)
168. if (c == 92 && i + 1 < n && isPunctC(s.charCodeAt(i + 1))) {
169. // a backslash escape: the next punctuation character as itself
170. buf = buf + s[i + 1]
171. i = i + 2
172. } else if (c == 96) {
173. // `code`: a run of backticks up to the same run
174. let k = i
175. while (k < n && s.charCodeAt(k) == 96) { k = k + 1 }
176. let close = runAt(s, k, 96, k - i, false)
177. if (close < 0) {
178. buf = buf + s.slice(i, k)
179. i = k
180. } else {
181. if (buf != '') { out.push(textSpan(buf, strong, em)) }
182. buf = ''
183. let code = s.slice(k, close).replaceAll(NL, ' ')
184. if (code.length > 2 && code.startsWith(' ') && code.endsWith(' ') && code.trim() != '') { code = code.slice(1, code.length - 1) }
185. out.push(codeSpan(code))
186. i = close + (k - i)
187. }
188. } else if (c == 91) {
189. // [label](url)
190. let l = linkAt(s, i)
191. if (l == null) {
192. buf = buf + '['
193. i = i + 1
194. } else if (l.href == null) {
195. buf = buf + s.slice(i, l.end)
196. i = l.end
197. } else {
198. if (buf != '') { out.push(textSpan(buf, strong, em)) }
199. buf = ''
200. out.push(linkSpan(l.label, l.href))
201. i = l.end
202. }
203. } else if (c == 60 && i + 1 < n && isUrlStart(s, i + 1)) {
204. // <https://…>
205. let e = s.indexOf('>', i + 1)
206. let url = e > 0 ? s.slice(i + 1, e) : ''
207. let h = safeHref(url)
208. if (e < 0 || h == null) {
209. buf = buf + '<'
210. i = i + 1
211. } else {
212. if (buf != '') { out.push(textSpan(buf, strong, em)) }
213. buf = ''
214. out.push(linkSpan(url, h))
215. i = e + 1
216. }
217. } else if ((c == 104 || c == 72) && (i == 0 || !isAlnumC(s.charCodeAt(i - 1))) && isUrlStart(s, i)) {
218. // a bare http(s):// URL
219. let e = urlEnd(s, i)
220. let url = s.slice(i, e)
221. let h = safeHref(url)
222. if (h == null) {
223. buf = buf + s[i]
224. i = i + 1
225. } else {
226. if (buf != '') { out.push(textSpan(buf, strong, em)) }
227. buf = ''
228. out.push(linkSpan(url, h))
229. i = e
230. }
231. } else if (c == 42 || c == 95) {
232. // * and _: emphasis (1), strong (2), both (3)
233. let k = i
234. while (k < n && s.charCodeAt(k) == c) { k = k + 1 }
235. let run = k - i
236. let prev = i > 0 ? s.charCodeAt(i - 1) : 32
237. let next = k < n ? s.charCodeAt(k) : 32
238. let wantStrong = run == 2 || run == 3
239. let wantEm = run == 1 || run == 3
240. let canOpen = !isSpaceC(next) && (c == 42 || !isAlnumC(prev))
241. let canClose = !isSpaceC(prev) && (c == 42 || !isAlnumC(next))
242. let closes = canClose && run <= 3 && (!wantStrong || (strong && strongC == c)) && (!wantEm || (em && emC == c))
243. let opens = !closes && canOpen && run <= 3 && (!wantStrong || !strong) && (!wantEm || !em) && hasCloser(s, k, c, run)
244. if (closes || opens) {
245. if (buf != '') { out.push(textSpan(buf, strong, em)) }
246. buf = ''
247. if (wantStrong) {
248. strong = opens
249. strongC = c
250. }
251. if (wantEm) {
252. em = opens
253. emC = c
254. }
255. } else {
256. buf = buf + s.slice(i, k)
257. }
258. i = k
259. } else {
260. buf = buf + s[i]
261. i = i + 1
262. }
263. }
264. if (buf != '') { out.push(textSpan(buf, strong, em)) }
265. return out
266. }
268. // ---- blocks -------------------------------------------------------------------------------
269. static noCr = (line) => { return line.length > 0 && line.charCodeAt(line.length - 1) == 13 ? line.slice(0, line.length - 1) : line }
270. static indentOf = (line) => {
271. let i = 0
272. while (i < line.length && (line.charCodeAt(i) == 32 || line.charCodeAt(i) == 9)) { i = i + 1 }
273. return i
274. }
276. // a heading line (≤ 3 spaces in, 1–6 '#', a space or the end) → { level, text } or null
277. static headingOf = (line) => {
278. if (indentOf(line) > 3) { return null }
279. let t = line.trim()
280. let k = 0
281. while (k < t.length && t.charCodeAt(k) == 35) { k = k + 1 }
282. if (k == 0 || k > 6) { return null }
283. if (k < t.length && t.charCodeAt(k) != 32 && t.charCodeAt(k) != 9) { return null }
284. let body = t.slice(k).trim()
285. // a closing run of '#' after a space goes
286. let e = body.length
287. while (e > 0 && body.charCodeAt(e - 1) == 35) { e = e - 1 }
288. if (e == 0) { body = '' } else if (e < body.length && (body.charCodeAt(e - 1) == 32 || body.charCodeAt(e - 1) == 9)) { body = body.slice(0, e).trim() }
289. return { level = k text = body }
290. }
292. // a list item line → { ordered, start, text } or null
293. static itemOf = (line) => {
294. let t = line.trim()
295. if (t.length >= 2 && (t.charCodeAt(0) == 45 || t.charCodeAt(0) == 42 || t.charCodeAt(0) == 43) && (t.charCodeAt(1) == 32 || t.charCodeAt(1) == 9)) {
296. return { ordered = false start = 1 text = t.slice(2).trim() }
297. }
298. if (t.length == 1 && (t == '-' || t == '*' || t == '+')) { return { ordered = false start = 1 text = '' } }
299. let k = 0
300. while (k < t.length && k < 9 && t.charCodeAt(k) >= 48 && t.charCodeAt(k) <= 57) { k = k + 1 }
301. if (k == 0 || k >= t.length) { return null }
302. let d = t.charCodeAt(k)
303. if (d != 46 && d != 41) { return null }
304. if (k + 1 < t.length && t.charCodeAt(k + 1) != 32 && t.charCodeAt(k + 1) != 9) { return null }
305. return { ordered = true start = toNumber(t.slice(0, k)) text = t.slice(k + 1).trim() }
306. }
308. // a code fence line (``` or ~~~, ≥ 3) → its run (e.g. '````') or null
309. static fenceOf = (line) => {
310. if (indentOf(line) > 3) { return null }
311. let t = line.trim()
312. if (t.length < 3) { return null }
313. let c = t.charCodeAt(0)
314. if (c != 96 && c != 126) { return null }
315. let k = 0
316. while (k < t.length && t.charCodeAt(k) == c) { k = k + 1 }
317. if (k < 3) { return null }
318. // an info string after ``` must not contain a backtick
319. if (c == 96 && t.slice(k).includes('`')) { return null }
320. return t.slice(0, k)
321. }
323. static paraBlock = (lines) => {
324. let spans = inlineSpans(lines.join(NL))
325. return { isP = true isH1 = false isH2 = false isH3 = false isUl = false isOl = false isCode = false isQuote = false isTable = false isHr = false spans = spans items = [] start = 1 code = '' head = [] rows = [] }
326. }
327. static headingBlock = (level, text) => {
328. let spans = inlineSpans(text)
329. return { isP = false isH1 = level == 1 isH2 = level == 2 isH3 = level >= 3 isUl = false isOl = false isCode = false isQuote = false isTable = false isHr = false spans = spans items = [] start = 1 code = '' head = [] rows = [] }
330. }
331. static codeBlock = (lines) => {
332. let text = lines.join(NL)
333. return { isP = false isH1 = false isH2 = false isH3 = false isUl = false isOl = false isCode = true isQuote = false isTable = false isHr = false spans = [] items = [] start = 1 code = text head = [] rows = [] }
334. }
335. static listBlock = (ordered, first, itemLines) => {
336. let items = []
337. for (lines of itemLines) {
338. let spans = inlineSpans(lines.join(NL))
339. items.push({ spans = spans })
340. }
341. return { isP = false isH1 = false isH2 = false isH3 = false isUl = !ordered isOl = ordered isCode = false isQuote = false isTable = false isHr = false spans = [] items = items start = first code = '' head = [] rows = [] }
342. }
344. static quoteBlock = (lines) => {
345. let spans = inlineSpans(lines.join(NL))
346. return { isP = false isH1 = false isH2 = false isH3 = false isUl = false isOl = false isCode = false isQuote = true isTable = false isHr = false spans = spans items = [] start = 1 code = '' head = [] rows = [] }
347. }
348. static hrBlock = () => {
349. return { isP = false isH1 = false isH2 = false isH3 = false isUl = false isOl = false isCode = false isQuote = false isTable = false isHr = true spans = [] items = [] start = 1 code = '' head = [] rows = [] }
350. }
351. static tableBlock = (head, rows) => {
352. return { isP = false isH1 = false isH2 = false isH3 = false isUl = false isOl = false isCode = false isQuote = false isTable = true isHr = false spans = [] items = [] start = 1 code = '' head = head rows = rows }
353. }
355. // a line of 3+ of one of - * _ (spaces between allowed) and nothing else
356. static isHrLine = (line) => {
357. if (indentOf(line) > 3) { return false }
358. let t = line.trim()
359. if (t.length < 3) { return false }
360. let c = t.charCodeAt(0)
361. if (c != 45 && c != 42 && c != 95) { return false }
362. let n = 0
363. let i = 0
364. while (i < t.length) {
365. let d = t.charCodeAt(i)
366. if (d == c) { n = n + 1 } else if (d != 32 && d != 9) { return false }
367. i = i + 1
368. }
369. return n >= 3
370. }
372. // `> text` → the text after the marker, or null
373. static quoteOf = (line) => {
374. if (indentOf(line) > 3) { return null }
375. let t = line.trim()
376. if (!t.startsWith('>')) { return null }
377. return t.slice(1).trim()
378. }
380. // the cells of a table row line: `| a | b |` → [a, b] as spans lists
381. static cellsOf = (line) => {
382. let t = line.trim()
383. if (t.startsWith('|')) { t = t.slice(1) }
384. if (t.endsWith('|')) { t = t.slice(0, t.length - 1) }
385. let out = []
386. for (c of t.split('|')) { out.push({ spans = inlineSpans(c.trim()) }) }
387. return out
388. }
390. // is this the `|---|:--:|` line under a table head?
391. static isSepLine = (line) => {
392. let t = line.trim()
393. if (!t.includes('-') || !t.includes('|') && !t.includes(':')) { return false }
394. let i = 0
395. while (i < t.length) {
396. let d = t.charCodeAt(i)
397. if (d != 124 && d != 45 && d != 58 && d != 32 && d != 9) { return false }
398. i = i + 1
399. }
400. return true
401. }
403. // the Markdown text → blocks (see the header)
404. static markdownBlocks = (text) => {
405. let out = []
406. if (text == null || text == '') { return out }
407. let lines = ('' + text).split(NL)
408. let para = []
409. let inList = false
410. let listOrdered = false
411. let listStart = 1
412. let items = []
413. let blankBefore = false
414. let i = 0
415. while (i < lines.length) {
416. let line = noCr(lines[i])
417. let fence = fenceOf(line)
418. let hr = fence == null && isHrLine(line)
419. let q = fence == null && !hr ? quoteOf(line) : null
420. let isTab = fence == null && !hr && q == null && line.includes('|') && i + 1 < lines.length && isSepLine(noCr(lines[i + 1])) && noCr(lines[i + 1]).includes('|')
421. let h = fence == null && !hr && q == null && !isTab ? headingOf(line) : null
422. let it = fence == null && !hr && q == null && !isTab && h == null ? itemOf(line) : null
423. if (hr) {
424. if (para.length > 0) { out.push(paraBlock(para)) }
425. para = []
426. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
427. inList = false
428. items = []
429. out.push(hrBlock())
430. blankBefore = false
431. i = i + 1
432. } else if (q != null) {
433. if (para.length > 0) { out.push(paraBlock(para)) }
434. para = []
435. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
436. inList = false
437. items = []
438. let ql = []
439. let j = i
440. let more = true
441. while (j < lines.length && more) {
442. let x = quoteOf(noCr(lines[j]))
443. if (x == null) { more = false } else {
444. ql.push(x)
445. j = j + 1
446. }
447. }
448. out.push(quoteBlock(ql))
449. i = j
450. blankBefore = false
451. } else if (isTab) {
452. if (para.length > 0) { out.push(paraBlock(para)) }
453. para = []
454. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
455. inList = false
456. items = []
457. let head = cellsOf(line)
458. let rows = []
459. let j = i + 2
460. while (j < lines.length && noCr(lines[j]).trim() != '' && noCr(lines[j]).includes('|')) {
461. rows.push(cellsOf(noCr(lines[j])))
462. j = j + 1
463. }
464. out.push(tableBlock(head, rows))
465. i = j
466. blankBefore = false
467. } else if (fence != null) {
468. if (para.length > 0) { out.push(paraBlock(para)) }
469. para = []
470. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
471. inList = false
472. items = []
473. let code = []
474. let j = i + 1
475. let closed = false
476. let fc = fence.charCodeAt(0)
477. while (j < lines.length && !closed) {
478. let l = noCr(lines[j])
479. let f = fenceOf(l)
480. if (f != null && f.charCodeAt(0) == fc && f.length >= fence.length && l.trim() == f) {
481. closed = true
482. } else {
483. code.push(l)
484. }
485. j = j + 1
486. }
487. out.push(codeBlock(code))
488. i = j
489. blankBefore = false
490. } else if (line.trim() == '') {
491. if (para.length > 0) { out.push(paraBlock(para)) }
492. para = []
493. blankBefore = true
494. i = i + 1
495. } else if (h != null) {
496. if (para.length > 0) { out.push(paraBlock(para)) }
497. para = []
498. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
499. inList = false
500. items = []
501. out.push(headingBlock(h.level, h.text))
502. blankBefore = false
503. i = i + 1
504. } else if (it != null) {
505. if (para.length > 0) { out.push(paraBlock(para)) }
506. para = []
507. if (inList && listOrdered != it.ordered) {
508. out.push(listBlock(listOrdered, listStart, items))
509. inList = false
510. items = []
511. }
512. if (!inList) {
513. inList = true
514. listOrdered = it.ordered
515. listStart = it.start
516. items = []
517. }
518. items.push([it.text])
519. blankBefore = false
520. i = i + 1
521. } else if (inList && (!blankBefore || indentOf(line) >= 2)) {
522. // a wrapped item line (or an indented paragraph of the item): onto the last item
523. let last = items[items.length - 1]
524. last.push(line.trim())
525. items[items.length - 1] = last
526. blankBefore = false
527. i = i + 1
528. } else {
529. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
530. inList = false
531. items = []
532. para.push(line.trim())
533. blankBefore = false
534. i = i + 1
535. }
536. }
537. if (para.length > 0) { out.push(paraBlock(para)) }
538. if (inList) { out.push(listBlock(listOrdered, listStart, items)) }
539. return out
540. }