|
1 // |
|
2 // Blackfriday Markdown Processor |
|
3 // Available at http://github.com/russross/blackfriday |
|
4 // |
|
5 // Copyright © 2011 Russ Ross <russ@russross.com>. |
|
6 // Distributed under the Simplified BSD License. |
|
7 // See README.md for details. |
|
8 // |
|
9 |
|
10 // |
|
11 // Functions to parse inline elements. |
|
12 // |
|
13 |
|
14 package blackfriday |
|
15 |
|
16 import ( |
|
17 "bytes" |
|
18 "regexp" |
|
19 "strconv" |
|
20 ) |
|
21 |
|
22 var ( |
|
23 urlRe = `((https?|ftp):\/\/|\/)[-A-Za-z0-9+&@#\/%?=~_|!:,.;\(\)]+` |
|
24 anchorRe = regexp.MustCompile(`^(<a\shref="` + urlRe + `"(\stitle="[^"<>]+")?\s?>` + urlRe + `<\/a>)`) |
|
25 ) |
|
26 |
|
27 // Functions to parse text within a block |
|
28 // Each function returns the number of chars taken care of |
|
29 // data is the complete block being rendered |
|
30 // offset is the number of valid chars before the current cursor |
|
31 |
|
32 func (p *parser) inline(out *bytes.Buffer, data []byte) { |
|
33 // this is called recursively: enforce a maximum depth |
|
34 if p.nesting >= p.maxNesting { |
|
35 return |
|
36 } |
|
37 p.nesting++ |
|
38 |
|
39 i, end := 0, 0 |
|
40 for i < len(data) { |
|
41 // copy inactive chars into the output |
|
42 for end < len(data) && p.inlineCallback[data[end]] == nil { |
|
43 end++ |
|
44 } |
|
45 |
|
46 p.r.NormalText(out, data[i:end]) |
|
47 |
|
48 if end >= len(data) { |
|
49 break |
|
50 } |
|
51 i = end |
|
52 |
|
53 // call the trigger |
|
54 handler := p.inlineCallback[data[end]] |
|
55 if consumed := handler(p, out, data, i); consumed == 0 { |
|
56 // no action from the callback; buffer the byte for later |
|
57 end = i + 1 |
|
58 } else { |
|
59 // skip past whatever the callback used |
|
60 i += consumed |
|
61 end = i |
|
62 } |
|
63 } |
|
64 |
|
65 p.nesting-- |
|
66 } |
|
67 |
|
68 // single and double emphasis parsing |
|
69 func emphasis(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
70 data = data[offset:] |
|
71 c := data[0] |
|
72 ret := 0 |
|
73 |
|
74 if len(data) > 2 && data[1] != c { |
|
75 // whitespace cannot follow an opening emphasis; |
|
76 // strikethrough only takes two characters '~~' |
|
77 if c == '~' || isspace(data[1]) { |
|
78 return 0 |
|
79 } |
|
80 if ret = helperEmphasis(p, out, data[1:], c); ret == 0 { |
|
81 return 0 |
|
82 } |
|
83 |
|
84 return ret + 1 |
|
85 } |
|
86 |
|
87 if len(data) > 3 && data[1] == c && data[2] != c { |
|
88 if isspace(data[2]) { |
|
89 return 0 |
|
90 } |
|
91 if ret = helperDoubleEmphasis(p, out, data[2:], c); ret == 0 { |
|
92 return 0 |
|
93 } |
|
94 |
|
95 return ret + 2 |
|
96 } |
|
97 |
|
98 if len(data) > 4 && data[1] == c && data[2] == c && data[3] != c { |
|
99 if c == '~' || isspace(data[3]) { |
|
100 return 0 |
|
101 } |
|
102 if ret = helperTripleEmphasis(p, out, data, 3, c); ret == 0 { |
|
103 return 0 |
|
104 } |
|
105 |
|
106 return ret + 3 |
|
107 } |
|
108 |
|
109 return 0 |
|
110 } |
|
111 |
|
112 func codeSpan(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
113 data = data[offset:] |
|
114 |
|
115 nb := 0 |
|
116 |
|
117 // count the number of backticks in the delimiter |
|
118 for nb < len(data) && data[nb] == '`' { |
|
119 nb++ |
|
120 } |
|
121 |
|
122 // find the next delimiter |
|
123 i, end := 0, 0 |
|
124 for end = nb; end < len(data) && i < nb; end++ { |
|
125 if data[end] == '`' { |
|
126 i++ |
|
127 } else { |
|
128 i = 0 |
|
129 } |
|
130 } |
|
131 |
|
132 // no matching delimiter? |
|
133 if i < nb && end >= len(data) { |
|
134 return 0 |
|
135 } |
|
136 |
|
137 // trim outside whitespace |
|
138 fBegin := nb |
|
139 for fBegin < end && data[fBegin] == ' ' { |
|
140 fBegin++ |
|
141 } |
|
142 |
|
143 fEnd := end - nb |
|
144 for fEnd > fBegin && data[fEnd-1] == ' ' { |
|
145 fEnd-- |
|
146 } |
|
147 |
|
148 // render the code span |
|
149 if fBegin != fEnd { |
|
150 p.r.CodeSpan(out, data[fBegin:fEnd]) |
|
151 } |
|
152 |
|
153 return end |
|
154 |
|
155 } |
|
156 |
|
157 // newline preceded by two spaces becomes <br> |
|
158 // newline without two spaces works when EXTENSION_HARD_LINE_BREAK is enabled |
|
159 func lineBreak(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
160 // remove trailing spaces from out |
|
161 outBytes := out.Bytes() |
|
162 end := len(outBytes) |
|
163 eol := end |
|
164 for eol > 0 && outBytes[eol-1] == ' ' { |
|
165 eol-- |
|
166 } |
|
167 out.Truncate(eol) |
|
168 |
|
169 precededByTwoSpaces := offset >= 2 && data[offset-2] == ' ' && data[offset-1] == ' ' |
|
170 precededByBackslash := offset >= 1 && data[offset-1] == '\\' // see http://spec.commonmark.org/0.18/#example-527 |
|
171 precededByBackslash = precededByBackslash && p.flags&EXTENSION_BACKSLASH_LINE_BREAK != 0 |
|
172 |
|
173 if p.flags&EXTENSION_JOIN_LINES != 0 { |
|
174 return 1 |
|
175 } |
|
176 |
|
177 // should there be a hard line break here? |
|
178 if p.flags&EXTENSION_HARD_LINE_BREAK == 0 && !precededByTwoSpaces && !precededByBackslash { |
|
179 return 0 |
|
180 } |
|
181 |
|
182 if precededByBackslash && eol > 0 { |
|
183 out.Truncate(eol - 1) |
|
184 } |
|
185 p.r.LineBreak(out) |
|
186 return 1 |
|
187 } |
|
188 |
|
189 type linkType int |
|
190 |
|
191 const ( |
|
192 linkNormal linkType = iota |
|
193 linkImg |
|
194 linkDeferredFootnote |
|
195 linkInlineFootnote |
|
196 ) |
|
197 |
|
198 func isReferenceStyleLink(data []byte, pos int, t linkType) bool { |
|
199 if t == linkDeferredFootnote { |
|
200 return false |
|
201 } |
|
202 return pos < len(data)-1 && data[pos] == '[' && data[pos+1] != '^' |
|
203 } |
|
204 |
|
205 // '[': parse a link or an image or a footnote |
|
206 func link(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
207 // no links allowed inside regular links, footnote, and deferred footnotes |
|
208 if p.insideLink && (offset > 0 && data[offset-1] == '[' || len(data)-1 > offset && data[offset+1] == '^') { |
|
209 return 0 |
|
210 } |
|
211 |
|
212 var t linkType |
|
213 switch { |
|
214 // special case: ![^text] == deferred footnote (that follows something with |
|
215 // an exclamation point) |
|
216 case p.flags&EXTENSION_FOOTNOTES != 0 && len(data)-1 > offset && data[offset+1] == '^': |
|
217 t = linkDeferredFootnote |
|
218 // ![alt] == image |
|
219 case offset > 0 && data[offset-1] == '!': |
|
220 t = linkImg |
|
221 // ^[text] == inline footnote |
|
222 // [^refId] == deferred footnote |
|
223 case p.flags&EXTENSION_FOOTNOTES != 0: |
|
224 if offset > 0 && data[offset-1] == '^' { |
|
225 t = linkInlineFootnote |
|
226 } else if len(data)-1 > offset && data[offset+1] == '^' { |
|
227 t = linkDeferredFootnote |
|
228 } |
|
229 // [text] == regular link |
|
230 default: |
|
231 t = linkNormal |
|
232 } |
|
233 |
|
234 data = data[offset:] |
|
235 |
|
236 var ( |
|
237 i = 1 |
|
238 noteId int |
|
239 title, link, altContent []byte |
|
240 textHasNl = false |
|
241 ) |
|
242 |
|
243 if t == linkDeferredFootnote { |
|
244 i++ |
|
245 } |
|
246 |
|
247 brace := 0 |
|
248 |
|
249 // look for the matching closing bracket |
|
250 for level := 1; level > 0 && i < len(data); i++ { |
|
251 switch { |
|
252 case data[i] == '\n': |
|
253 textHasNl = true |
|
254 |
|
255 case data[i-1] == '\\': |
|
256 continue |
|
257 |
|
258 case data[i] == '[': |
|
259 level++ |
|
260 |
|
261 case data[i] == ']': |
|
262 level-- |
|
263 if level <= 0 { |
|
264 i-- // compensate for extra i++ in for loop |
|
265 } |
|
266 } |
|
267 } |
|
268 |
|
269 if i >= len(data) { |
|
270 return 0 |
|
271 } |
|
272 |
|
273 txtE := i |
|
274 i++ |
|
275 |
|
276 // skip any amount of whitespace or newline |
|
277 // (this is much more lax than original markdown syntax) |
|
278 for i < len(data) && isspace(data[i]) { |
|
279 i++ |
|
280 } |
|
281 |
|
282 switch { |
|
283 // inline style link |
|
284 case i < len(data) && data[i] == '(': |
|
285 // skip initial whitespace |
|
286 i++ |
|
287 |
|
288 for i < len(data) && isspace(data[i]) { |
|
289 i++ |
|
290 } |
|
291 |
|
292 linkB := i |
|
293 |
|
294 // look for link end: ' " ), check for new opening braces and take this |
|
295 // into account, this may lead for overshooting and probably will require |
|
296 // some fine-tuning. |
|
297 findlinkend: |
|
298 for i < len(data) { |
|
299 switch { |
|
300 case data[i] == '\\': |
|
301 i += 2 |
|
302 |
|
303 case data[i] == '(': |
|
304 brace++ |
|
305 i++ |
|
306 |
|
307 case data[i] == ')': |
|
308 if brace <= 0 { |
|
309 break findlinkend |
|
310 } |
|
311 brace-- |
|
312 i++ |
|
313 |
|
314 case data[i] == '\'' || data[i] == '"': |
|
315 break findlinkend |
|
316 |
|
317 default: |
|
318 i++ |
|
319 } |
|
320 } |
|
321 |
|
322 if i >= len(data) { |
|
323 return 0 |
|
324 } |
|
325 linkE := i |
|
326 |
|
327 // look for title end if present |
|
328 titleB, titleE := 0, 0 |
|
329 if data[i] == '\'' || data[i] == '"' { |
|
330 i++ |
|
331 titleB = i |
|
332 |
|
333 findtitleend: |
|
334 for i < len(data) { |
|
335 switch { |
|
336 case data[i] == '\\': |
|
337 i += 2 |
|
338 |
|
339 case data[i] == ')': |
|
340 break findtitleend |
|
341 |
|
342 default: |
|
343 i++ |
|
344 } |
|
345 } |
|
346 |
|
347 if i >= len(data) { |
|
348 return 0 |
|
349 } |
|
350 |
|
351 // skip whitespace after title |
|
352 titleE = i - 1 |
|
353 for titleE > titleB && isspace(data[titleE]) { |
|
354 titleE-- |
|
355 } |
|
356 |
|
357 // check for closing quote presence |
|
358 if data[titleE] != '\'' && data[titleE] != '"' { |
|
359 titleB, titleE = 0, 0 |
|
360 linkE = i |
|
361 } |
|
362 } |
|
363 |
|
364 // remove whitespace at the end of the link |
|
365 for linkE > linkB && isspace(data[linkE-1]) { |
|
366 linkE-- |
|
367 } |
|
368 |
|
369 // remove optional angle brackets around the link |
|
370 if data[linkB] == '<' { |
|
371 linkB++ |
|
372 } |
|
373 if data[linkE-1] == '>' { |
|
374 linkE-- |
|
375 } |
|
376 |
|
377 // build escaped link and title |
|
378 if linkE > linkB { |
|
379 link = data[linkB:linkE] |
|
380 } |
|
381 |
|
382 if titleE > titleB { |
|
383 title = data[titleB:titleE] |
|
384 } |
|
385 |
|
386 i++ |
|
387 |
|
388 // reference style link |
|
389 case isReferenceStyleLink(data, i, t): |
|
390 var id []byte |
|
391 altContentConsidered := false |
|
392 |
|
393 // look for the id |
|
394 i++ |
|
395 linkB := i |
|
396 for i < len(data) && data[i] != ']' { |
|
397 i++ |
|
398 } |
|
399 if i >= len(data) { |
|
400 return 0 |
|
401 } |
|
402 linkE := i |
|
403 |
|
404 // find the reference |
|
405 if linkB == linkE { |
|
406 if textHasNl { |
|
407 var b bytes.Buffer |
|
408 |
|
409 for j := 1; j < txtE; j++ { |
|
410 switch { |
|
411 case data[j] != '\n': |
|
412 b.WriteByte(data[j]) |
|
413 case data[j-1] != ' ': |
|
414 b.WriteByte(' ') |
|
415 } |
|
416 } |
|
417 |
|
418 id = b.Bytes() |
|
419 } else { |
|
420 id = data[1:txtE] |
|
421 altContentConsidered = true |
|
422 } |
|
423 } else { |
|
424 id = data[linkB:linkE] |
|
425 } |
|
426 |
|
427 // find the reference with matching id |
|
428 lr, ok := p.getRef(string(id)) |
|
429 if !ok { |
|
430 return 0 |
|
431 } |
|
432 |
|
433 // keep link and title from reference |
|
434 link = lr.link |
|
435 title = lr.title |
|
436 if altContentConsidered { |
|
437 altContent = lr.text |
|
438 } |
|
439 i++ |
|
440 |
|
441 // shortcut reference style link or reference or inline footnote |
|
442 default: |
|
443 var id []byte |
|
444 |
|
445 // craft the id |
|
446 if textHasNl { |
|
447 var b bytes.Buffer |
|
448 |
|
449 for j := 1; j < txtE; j++ { |
|
450 switch { |
|
451 case data[j] != '\n': |
|
452 b.WriteByte(data[j]) |
|
453 case data[j-1] != ' ': |
|
454 b.WriteByte(' ') |
|
455 } |
|
456 } |
|
457 |
|
458 id = b.Bytes() |
|
459 } else { |
|
460 if t == linkDeferredFootnote { |
|
461 id = data[2:txtE] // get rid of the ^ |
|
462 } else { |
|
463 id = data[1:txtE] |
|
464 } |
|
465 } |
|
466 |
|
467 if t == linkInlineFootnote { |
|
468 // create a new reference |
|
469 noteId = len(p.notes) + 1 |
|
470 |
|
471 var fragment []byte |
|
472 if len(id) > 0 { |
|
473 if len(id) < 16 { |
|
474 fragment = make([]byte, len(id)) |
|
475 } else { |
|
476 fragment = make([]byte, 16) |
|
477 } |
|
478 copy(fragment, slugify(id)) |
|
479 } else { |
|
480 fragment = append([]byte("footnote-"), []byte(strconv.Itoa(noteId))...) |
|
481 } |
|
482 |
|
483 ref := &reference{ |
|
484 noteId: noteId, |
|
485 hasBlock: false, |
|
486 link: fragment, |
|
487 title: id, |
|
488 } |
|
489 |
|
490 p.notes = append(p.notes, ref) |
|
491 p.notesRecord[string(ref.link)] = struct{}{} |
|
492 |
|
493 link = ref.link |
|
494 title = ref.title |
|
495 } else { |
|
496 // find the reference with matching id |
|
497 lr, ok := p.getRef(string(id)) |
|
498 if !ok { |
|
499 return 0 |
|
500 } |
|
501 |
|
502 if t == linkDeferredFootnote && !p.isFootnote(lr) { |
|
503 lr.noteId = len(p.notes) + 1 |
|
504 p.notes = append(p.notes, lr) |
|
505 p.notesRecord[string(lr.link)] = struct{}{} |
|
506 } |
|
507 |
|
508 // keep link and title from reference |
|
509 link = lr.link |
|
510 // if inline footnote, title == footnote contents |
|
511 title = lr.title |
|
512 noteId = lr.noteId |
|
513 } |
|
514 |
|
515 // rewind the whitespace |
|
516 i = txtE + 1 |
|
517 } |
|
518 |
|
519 // build content: img alt is escaped, link content is parsed |
|
520 var content bytes.Buffer |
|
521 if txtE > 1 { |
|
522 if t == linkImg { |
|
523 content.Write(data[1:txtE]) |
|
524 } else { |
|
525 // links cannot contain other links, so turn off link parsing temporarily |
|
526 insideLink := p.insideLink |
|
527 p.insideLink = true |
|
528 p.inline(&content, data[1:txtE]) |
|
529 p.insideLink = insideLink |
|
530 } |
|
531 } |
|
532 |
|
533 var uLink []byte |
|
534 if t == linkNormal || t == linkImg { |
|
535 if len(link) > 0 { |
|
536 var uLinkBuf bytes.Buffer |
|
537 unescapeText(&uLinkBuf, link) |
|
538 uLink = uLinkBuf.Bytes() |
|
539 } |
|
540 |
|
541 // links need something to click on and somewhere to go |
|
542 if len(uLink) == 0 || (t == linkNormal && content.Len() == 0) { |
|
543 return 0 |
|
544 } |
|
545 } |
|
546 |
|
547 // call the relevant rendering function |
|
548 switch t { |
|
549 case linkNormal: |
|
550 if len(altContent) > 0 { |
|
551 p.r.Link(out, uLink, title, altContent) |
|
552 } else { |
|
553 p.r.Link(out, uLink, title, content.Bytes()) |
|
554 } |
|
555 |
|
556 case linkImg: |
|
557 outSize := out.Len() |
|
558 outBytes := out.Bytes() |
|
559 if outSize > 0 && outBytes[outSize-1] == '!' { |
|
560 out.Truncate(outSize - 1) |
|
561 } |
|
562 |
|
563 p.r.Image(out, uLink, title, content.Bytes()) |
|
564 |
|
565 case linkInlineFootnote: |
|
566 outSize := out.Len() |
|
567 outBytes := out.Bytes() |
|
568 if outSize > 0 && outBytes[outSize-1] == '^' { |
|
569 out.Truncate(outSize - 1) |
|
570 } |
|
571 |
|
572 p.r.FootnoteRef(out, link, noteId) |
|
573 |
|
574 case linkDeferredFootnote: |
|
575 p.r.FootnoteRef(out, link, noteId) |
|
576 |
|
577 default: |
|
578 return 0 |
|
579 } |
|
580 |
|
581 return i |
|
582 } |
|
583 |
|
584 func (p *parser) inlineHTMLComment(out *bytes.Buffer, data []byte) int { |
|
585 if len(data) < 5 { |
|
586 return 0 |
|
587 } |
|
588 if data[0] != '<' || data[1] != '!' || data[2] != '-' || data[3] != '-' { |
|
589 return 0 |
|
590 } |
|
591 i := 5 |
|
592 // scan for an end-of-comment marker, across lines if necessary |
|
593 for i < len(data) && !(data[i-2] == '-' && data[i-1] == '-' && data[i] == '>') { |
|
594 i++ |
|
595 } |
|
596 // no end-of-comment marker |
|
597 if i >= len(data) { |
|
598 return 0 |
|
599 } |
|
600 return i + 1 |
|
601 } |
|
602 |
|
603 // '<' when tags or autolinks are allowed |
|
604 func leftAngle(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
605 data = data[offset:] |
|
606 altype := LINK_TYPE_NOT_AUTOLINK |
|
607 end := tagLength(data, &altype) |
|
608 if size := p.inlineHTMLComment(out, data); size > 0 { |
|
609 end = size |
|
610 } |
|
611 if end > 2 { |
|
612 if altype != LINK_TYPE_NOT_AUTOLINK { |
|
613 var uLink bytes.Buffer |
|
614 unescapeText(&uLink, data[1:end+1-2]) |
|
615 if uLink.Len() > 0 { |
|
616 p.r.AutoLink(out, uLink.Bytes(), altype) |
|
617 } |
|
618 } else { |
|
619 p.r.RawHtmlTag(out, data[:end]) |
|
620 } |
|
621 } |
|
622 |
|
623 return end |
|
624 } |
|
625 |
|
626 // '\\' backslash escape |
|
627 var escapeChars = []byte("\\`*_{}[]()#+-.!:|&<>~") |
|
628 |
|
629 func escape(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
630 data = data[offset:] |
|
631 |
|
632 if len(data) > 1 { |
|
633 if bytes.IndexByte(escapeChars, data[1]) < 0 { |
|
634 return 0 |
|
635 } |
|
636 |
|
637 p.r.NormalText(out, data[1:2]) |
|
638 } |
|
639 |
|
640 return 2 |
|
641 } |
|
642 |
|
643 func unescapeText(ob *bytes.Buffer, src []byte) { |
|
644 i := 0 |
|
645 for i < len(src) { |
|
646 org := i |
|
647 for i < len(src) && src[i] != '\\' { |
|
648 i++ |
|
649 } |
|
650 |
|
651 if i > org { |
|
652 ob.Write(src[org:i]) |
|
653 } |
|
654 |
|
655 if i+1 >= len(src) { |
|
656 break |
|
657 } |
|
658 |
|
659 ob.WriteByte(src[i+1]) |
|
660 i += 2 |
|
661 } |
|
662 } |
|
663 |
|
664 // '&' escaped when it doesn't belong to an entity |
|
665 // valid entities are assumed to be anything matching &#?[A-Za-z0-9]+; |
|
666 func entity(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
667 data = data[offset:] |
|
668 |
|
669 end := 1 |
|
670 |
|
671 if end < len(data) && data[end] == '#' { |
|
672 end++ |
|
673 } |
|
674 |
|
675 for end < len(data) && isalnum(data[end]) { |
|
676 end++ |
|
677 } |
|
678 |
|
679 if end < len(data) && data[end] == ';' { |
|
680 end++ // real entity |
|
681 } else { |
|
682 return 0 // lone '&' |
|
683 } |
|
684 |
|
685 p.r.Entity(out, data[:end]) |
|
686 |
|
687 return end |
|
688 } |
|
689 |
|
690 func linkEndsWithEntity(data []byte, linkEnd int) bool { |
|
691 entityRanges := htmlEntity.FindAllIndex(data[:linkEnd], -1) |
|
692 return entityRanges != nil && entityRanges[len(entityRanges)-1][1] == linkEnd |
|
693 } |
|
694 |
|
695 func autoLink(p *parser, out *bytes.Buffer, data []byte, offset int) int { |
|
696 // quick check to rule out most false hits on ':' |
|
697 if p.insideLink || len(data) < offset+3 || data[offset+1] != '/' || data[offset+2] != '/' { |
|
698 return 0 |
|
699 } |
|
700 |
|
701 // Now a more expensive check to see if we're not inside an anchor element |
|
702 anchorStart := offset |
|
703 offsetFromAnchor := 0 |
|
704 for anchorStart > 0 && data[anchorStart] != '<' { |
|
705 anchorStart-- |
|
706 offsetFromAnchor++ |
|
707 } |
|
708 |
|
709 anchorStr := anchorRe.Find(data[anchorStart:]) |
|
710 if anchorStr != nil { |
|
711 out.Write(anchorStr[offsetFromAnchor:]) |
|
712 return len(anchorStr) - offsetFromAnchor |
|
713 } |
|
714 |
|
715 // scan backward for a word boundary |
|
716 rewind := 0 |
|
717 for offset-rewind > 0 && rewind <= 7 && isletter(data[offset-rewind-1]) { |
|
718 rewind++ |
|
719 } |
|
720 if rewind > 6 { // longest supported protocol is "mailto" which has 6 letters |
|
721 return 0 |
|
722 } |
|
723 |
|
724 origData := data |
|
725 data = data[offset-rewind:] |
|
726 |
|
727 if !isSafeLink(data) { |
|
728 return 0 |
|
729 } |
|
730 |
|
731 linkEnd := 0 |
|
732 for linkEnd < len(data) && !isEndOfLink(data[linkEnd]) { |
|
733 linkEnd++ |
|
734 } |
|
735 |
|
736 // Skip punctuation at the end of the link |
|
737 if (data[linkEnd-1] == '.' || data[linkEnd-1] == ',') && data[linkEnd-2] != '\\' { |
|
738 linkEnd-- |
|
739 } |
|
740 |
|
741 // But don't skip semicolon if it's a part of escaped entity: |
|
742 if data[linkEnd-1] == ';' && data[linkEnd-2] != '\\' && !linkEndsWithEntity(data, linkEnd) { |
|
743 linkEnd-- |
|
744 } |
|
745 |
|
746 // See if the link finishes with a punctuation sign that can be closed. |
|
747 var copen byte |
|
748 switch data[linkEnd-1] { |
|
749 case '"': |
|
750 copen = '"' |
|
751 case '\'': |
|
752 copen = '\'' |
|
753 case ')': |
|
754 copen = '(' |
|
755 case ']': |
|
756 copen = '[' |
|
757 case '}': |
|
758 copen = '{' |
|
759 default: |
|
760 copen = 0 |
|
761 } |
|
762 |
|
763 if copen != 0 { |
|
764 bufEnd := offset - rewind + linkEnd - 2 |
|
765 |
|
766 openDelim := 1 |
|
767 |
|
768 /* Try to close the final punctuation sign in this same line; |
|
769 * if we managed to close it outside of the URL, that means that it's |
|
770 * not part of the URL. If it closes inside the URL, that means it |
|
771 * is part of the URL. |
|
772 * |
|
773 * Examples: |
|
774 * |
|
775 * foo http://www.pokemon.com/Pikachu_(Electric) bar |
|
776 * => http://www.pokemon.com/Pikachu_(Electric) |
|
777 * |
|
778 * foo (http://www.pokemon.com/Pikachu_(Electric)) bar |
|
779 * => http://www.pokemon.com/Pikachu_(Electric) |
|
780 * |
|
781 * foo http://www.pokemon.com/Pikachu_(Electric)) bar |
|
782 * => http://www.pokemon.com/Pikachu_(Electric)) |
|
783 * |
|
784 * (foo http://www.pokemon.com/Pikachu_(Electric)) bar |
|
785 * => foo http://www.pokemon.com/Pikachu_(Electric) |
|
786 */ |
|
787 |
|
788 for bufEnd >= 0 && origData[bufEnd] != '\n' && openDelim != 0 { |
|
789 if origData[bufEnd] == data[linkEnd-1] { |
|
790 openDelim++ |
|
791 } |
|
792 |
|
793 if origData[bufEnd] == copen { |
|
794 openDelim-- |
|
795 } |
|
796 |
|
797 bufEnd-- |
|
798 } |
|
799 |
|
800 if openDelim == 0 { |
|
801 linkEnd-- |
|
802 } |
|
803 } |
|
804 |
|
805 // we were triggered on the ':', so we need to rewind the output a bit |
|
806 if out.Len() >= rewind { |
|
807 out.Truncate(len(out.Bytes()) - rewind) |
|
808 } |
|
809 |
|
810 var uLink bytes.Buffer |
|
811 unescapeText(&uLink, data[:linkEnd]) |
|
812 |
|
813 if uLink.Len() > 0 { |
|
814 p.r.AutoLink(out, uLink.Bytes(), LINK_TYPE_NORMAL) |
|
815 } |
|
816 |
|
817 return linkEnd - rewind |
|
818 } |
|
819 |
|
820 func isEndOfLink(char byte) bool { |
|
821 return isspace(char) || char == '<' |
|
822 } |
|
823 |
|
824 var validUris = [][]byte{[]byte("http://"), []byte("https://"), []byte("ftp://"), []byte("mailto://")} |
|
825 var validPaths = [][]byte{[]byte("/"), []byte("./"), []byte("../")} |
|
826 |
|
827 func isSafeLink(link []byte) bool { |
|
828 for _, path := range validPaths { |
|
829 if len(link) >= len(path) && bytes.Equal(link[:len(path)], path) { |
|
830 if len(link) == len(path) { |
|
831 return true |
|
832 } else if isalnum(link[len(path)]) { |
|
833 return true |
|
834 } |
|
835 } |
|
836 } |
|
837 |
|
838 for _, prefix := range validUris { |
|
839 // TODO: handle unicode here |
|
840 // case-insensitive prefix test |
|
841 if len(link) > len(prefix) && bytes.Equal(bytes.ToLower(link[:len(prefix)]), prefix) && isalnum(link[len(prefix)]) { |
|
842 return true |
|
843 } |
|
844 } |
|
845 |
|
846 return false |
|
847 } |
|
848 |
|
849 // return the length of the given tag, or 0 is it's not valid |
|
850 func tagLength(data []byte, autolink *int) int { |
|
851 var i, j int |
|
852 |
|
853 // a valid tag can't be shorter than 3 chars |
|
854 if len(data) < 3 { |
|
855 return 0 |
|
856 } |
|
857 |
|
858 // begins with a '<' optionally followed by '/', followed by letter or number |
|
859 if data[0] != '<' { |
|
860 return 0 |
|
861 } |
|
862 if data[1] == '/' { |
|
863 i = 2 |
|
864 } else { |
|
865 i = 1 |
|
866 } |
|
867 |
|
868 if !isalnum(data[i]) { |
|
869 return 0 |
|
870 } |
|
871 |
|
872 // scheme test |
|
873 *autolink = LINK_TYPE_NOT_AUTOLINK |
|
874 |
|
875 // try to find the beginning of an URI |
|
876 for i < len(data) && (isalnum(data[i]) || data[i] == '.' || data[i] == '+' || data[i] == '-') { |
|
877 i++ |
|
878 } |
|
879 |
|
880 if i > 1 && i < len(data) && data[i] == '@' { |
|
881 if j = isMailtoAutoLink(data[i:]); j != 0 { |
|
882 *autolink = LINK_TYPE_EMAIL |
|
883 return i + j |
|
884 } |
|
885 } |
|
886 |
|
887 if i > 2 && i < len(data) && data[i] == ':' { |
|
888 *autolink = LINK_TYPE_NORMAL |
|
889 i++ |
|
890 } |
|
891 |
|
892 // complete autolink test: no whitespace or ' or " |
|
893 switch { |
|
894 case i >= len(data): |
|
895 *autolink = LINK_TYPE_NOT_AUTOLINK |
|
896 case *autolink != 0: |
|
897 j = i |
|
898 |
|
899 for i < len(data) { |
|
900 if data[i] == '\\' { |
|
901 i += 2 |
|
902 } else if data[i] == '>' || data[i] == '\'' || data[i] == '"' || isspace(data[i]) { |
|
903 break |
|
904 } else { |
|
905 i++ |
|
906 } |
|
907 |
|
908 } |
|
909 |
|
910 if i >= len(data) { |
|
911 return 0 |
|
912 } |
|
913 if i > j && data[i] == '>' { |
|
914 return i + 1 |
|
915 } |
|
916 |
|
917 // one of the forbidden chars has been found |
|
918 *autolink = LINK_TYPE_NOT_AUTOLINK |
|
919 } |
|
920 |
|
921 // look for something looking like a tag end |
|
922 for i < len(data) && data[i] != '>' { |
|
923 i++ |
|
924 } |
|
925 if i >= len(data) { |
|
926 return 0 |
|
927 } |
|
928 return i + 1 |
|
929 } |
|
930 |
|
931 // look for the address part of a mail autolink and '>' |
|
932 // this is less strict than the original markdown e-mail address matching |
|
933 func isMailtoAutoLink(data []byte) int { |
|
934 nb := 0 |
|
935 |
|
936 // address is assumed to be: [-@._a-zA-Z0-9]+ with exactly one '@' |
|
937 for i := 0; i < len(data); i++ { |
|
938 if isalnum(data[i]) { |
|
939 continue |
|
940 } |
|
941 |
|
942 switch data[i] { |
|
943 case '@': |
|
944 nb++ |
|
945 |
|
946 case '-', '.', '_': |
|
947 // Do nothing. |
|
948 |
|
949 case '>': |
|
950 if nb == 1 { |
|
951 return i + 1 |
|
952 } else { |
|
953 return 0 |
|
954 } |
|
955 default: |
|
956 return 0 |
|
957 } |
|
958 } |
|
959 |
|
960 return 0 |
|
961 } |
|
962 |
|
963 // look for the next emph char, skipping other constructs |
|
964 func helperFindEmphChar(data []byte, c byte) int { |
|
965 i := 0 |
|
966 |
|
967 for i < len(data) { |
|
968 for i < len(data) && data[i] != c && data[i] != '`' && data[i] != '[' { |
|
969 i++ |
|
970 } |
|
971 if i >= len(data) { |
|
972 return 0 |
|
973 } |
|
974 // do not count escaped chars |
|
975 if i != 0 && data[i-1] == '\\' { |
|
976 i++ |
|
977 continue |
|
978 } |
|
979 if data[i] == c { |
|
980 return i |
|
981 } |
|
982 |
|
983 if data[i] == '`' { |
|
984 // skip a code span |
|
985 tmpI := 0 |
|
986 i++ |
|
987 for i < len(data) && data[i] != '`' { |
|
988 if tmpI == 0 && data[i] == c { |
|
989 tmpI = i |
|
990 } |
|
991 i++ |
|
992 } |
|
993 if i >= len(data) { |
|
994 return tmpI |
|
995 } |
|
996 i++ |
|
997 } else if data[i] == '[' { |
|
998 // skip a link |
|
999 tmpI := 0 |
|
1000 i++ |
|
1001 for i < len(data) && data[i] != ']' { |
|
1002 if tmpI == 0 && data[i] == c { |
|
1003 tmpI = i |
|
1004 } |
|
1005 i++ |
|
1006 } |
|
1007 i++ |
|
1008 for i < len(data) && (data[i] == ' ' || data[i] == '\n') { |
|
1009 i++ |
|
1010 } |
|
1011 if i >= len(data) { |
|
1012 return tmpI |
|
1013 } |
|
1014 if data[i] != '[' && data[i] != '(' { // not a link |
|
1015 if tmpI > 0 { |
|
1016 return tmpI |
|
1017 } else { |
|
1018 continue |
|
1019 } |
|
1020 } |
|
1021 cc := data[i] |
|
1022 i++ |
|
1023 for i < len(data) && data[i] != cc { |
|
1024 if tmpI == 0 && data[i] == c { |
|
1025 return i |
|
1026 } |
|
1027 i++ |
|
1028 } |
|
1029 if i >= len(data) { |
|
1030 return tmpI |
|
1031 } |
|
1032 i++ |
|
1033 } |
|
1034 } |
|
1035 return 0 |
|
1036 } |
|
1037 |
|
1038 func helperEmphasis(p *parser, out *bytes.Buffer, data []byte, c byte) int { |
|
1039 i := 0 |
|
1040 |
|
1041 // skip one symbol if coming from emph3 |
|
1042 if len(data) > 1 && data[0] == c && data[1] == c { |
|
1043 i = 1 |
|
1044 } |
|
1045 |
|
1046 for i < len(data) { |
|
1047 length := helperFindEmphChar(data[i:], c) |
|
1048 if length == 0 { |
|
1049 return 0 |
|
1050 } |
|
1051 i += length |
|
1052 if i >= len(data) { |
|
1053 return 0 |
|
1054 } |
|
1055 |
|
1056 if i+1 < len(data) && data[i+1] == c { |
|
1057 i++ |
|
1058 continue |
|
1059 } |
|
1060 |
|
1061 if data[i] == c && !isspace(data[i-1]) { |
|
1062 |
|
1063 if p.flags&EXTENSION_NO_INTRA_EMPHASIS != 0 { |
|
1064 if !(i+1 == len(data) || isspace(data[i+1]) || ispunct(data[i+1])) { |
|
1065 continue |
|
1066 } |
|
1067 } |
|
1068 |
|
1069 var work bytes.Buffer |
|
1070 p.inline(&work, data[:i]) |
|
1071 p.r.Emphasis(out, work.Bytes()) |
|
1072 return i + 1 |
|
1073 } |
|
1074 } |
|
1075 |
|
1076 return 0 |
|
1077 } |
|
1078 |
|
1079 func helperDoubleEmphasis(p *parser, out *bytes.Buffer, data []byte, c byte) int { |
|
1080 i := 0 |
|
1081 |
|
1082 for i < len(data) { |
|
1083 length := helperFindEmphChar(data[i:], c) |
|
1084 if length == 0 { |
|
1085 return 0 |
|
1086 } |
|
1087 i += length |
|
1088 |
|
1089 if i+1 < len(data) && data[i] == c && data[i+1] == c && i > 0 && !isspace(data[i-1]) { |
|
1090 var work bytes.Buffer |
|
1091 p.inline(&work, data[:i]) |
|
1092 |
|
1093 if work.Len() > 0 { |
|
1094 // pick the right renderer |
|
1095 if c == '~' { |
|
1096 p.r.StrikeThrough(out, work.Bytes()) |
|
1097 } else { |
|
1098 p.r.DoubleEmphasis(out, work.Bytes()) |
|
1099 } |
|
1100 } |
|
1101 return i + 2 |
|
1102 } |
|
1103 i++ |
|
1104 } |
|
1105 return 0 |
|
1106 } |
|
1107 |
|
1108 func helperTripleEmphasis(p *parser, out *bytes.Buffer, data []byte, offset int, c byte) int { |
|
1109 i := 0 |
|
1110 origData := data |
|
1111 data = data[offset:] |
|
1112 |
|
1113 for i < len(data) { |
|
1114 length := helperFindEmphChar(data[i:], c) |
|
1115 if length == 0 { |
|
1116 return 0 |
|
1117 } |
|
1118 i += length |
|
1119 |
|
1120 // skip whitespace preceded symbols |
|
1121 if data[i] != c || isspace(data[i-1]) { |
|
1122 continue |
|
1123 } |
|
1124 |
|
1125 switch { |
|
1126 case i+2 < len(data) && data[i+1] == c && data[i+2] == c: |
|
1127 // triple symbol found |
|
1128 var work bytes.Buffer |
|
1129 |
|
1130 p.inline(&work, data[:i]) |
|
1131 if work.Len() > 0 { |
|
1132 p.r.TripleEmphasis(out, work.Bytes()) |
|
1133 } |
|
1134 return i + 3 |
|
1135 case (i+1 < len(data) && data[i+1] == c): |
|
1136 // double symbol found, hand over to emph1 |
|
1137 length = helperEmphasis(p, out, origData[offset-2:], c) |
|
1138 if length == 0 { |
|
1139 return 0 |
|
1140 } else { |
|
1141 return length - 2 |
|
1142 } |
|
1143 default: |
|
1144 // single symbol found, hand over to emph2 |
|
1145 length = helperDoubleEmphasis(p, out, origData[offset-1:], c) |
|
1146 if length == 0 { |
|
1147 return 0 |
|
1148 } else { |
|
1149 return length - 1 |
|
1150 } |
|
1151 } |
|
1152 } |
|
1153 return 0 |
|
1154 } |