251
|
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 block-level elements. |
|
12 |
// |
|
13 |
|
|
14 |
package blackfriday |
|
15 |
|
|
16 |
import ( |
|
17 |
"bytes" |
|
18 |
"strings" |
|
19 |
"unicode" |
|
20 |
) |
|
21 |
|
|
22 |
// Parse block-level data. |
|
23 |
// Note: this function and many that it calls assume that |
|
24 |
// the input buffer ends with a newline. |
|
25 |
func (p *parser) block(out *bytes.Buffer, data []byte) { |
|
26 |
if len(data) == 0 || data[len(data)-1] != '\n' { |
|
27 |
panic("block input is missing terminating newline") |
|
28 |
} |
|
29 |
|
|
30 |
// this is called recursively: enforce a maximum depth |
|
31 |
if p.nesting >= p.maxNesting { |
|
32 |
return |
|
33 |
} |
|
34 |
p.nesting++ |
|
35 |
|
|
36 |
// parse out one block-level construct at a time |
|
37 |
for len(data) > 0 { |
|
38 |
// prefixed header: |
|
39 |
// |
|
40 |
// # Header 1 |
|
41 |
// ## Header 2 |
|
42 |
// ... |
|
43 |
// ###### Header 6 |
|
44 |
if p.isPrefixHeader(data) { |
|
45 |
data = data[p.prefixHeader(out, data):] |
|
46 |
continue |
|
47 |
} |
|
48 |
|
|
49 |
// block of preformatted HTML: |
|
50 |
// |
|
51 |
// <div> |
|
52 |
// ... |
|
53 |
// </div> |
|
54 |
if data[0] == '<' { |
|
55 |
if i := p.html(out, data, true); i > 0 { |
|
56 |
data = data[i:] |
|
57 |
continue |
|
58 |
} |
|
59 |
} |
|
60 |
|
|
61 |
// title block |
|
62 |
// |
|
63 |
// % stuff |
|
64 |
// % more stuff |
|
65 |
// % even more stuff |
|
66 |
if p.flags&EXTENSION_TITLEBLOCK != 0 { |
|
67 |
if data[0] == '%' { |
|
68 |
if i := p.titleBlock(out, data, true); i > 0 { |
|
69 |
data = data[i:] |
|
70 |
continue |
|
71 |
} |
|
72 |
} |
|
73 |
} |
|
74 |
|
|
75 |
// blank lines. note: returns the # of bytes to skip |
|
76 |
if i := p.isEmpty(data); i > 0 { |
|
77 |
data = data[i:] |
|
78 |
continue |
|
79 |
} |
|
80 |
|
|
81 |
// indented code block: |
|
82 |
// |
|
83 |
// func max(a, b int) int { |
|
84 |
// if a > b { |
|
85 |
// return a |
|
86 |
// } |
|
87 |
// return b |
|
88 |
// } |
|
89 |
if p.codePrefix(data) > 0 { |
|
90 |
data = data[p.code(out, data):] |
|
91 |
continue |
|
92 |
} |
|
93 |
|
|
94 |
// fenced code block: |
|
95 |
// |
|
96 |
// ``` go info string here |
|
97 |
// func fact(n int) int { |
|
98 |
// if n <= 1 { |
|
99 |
// return n |
|
100 |
// } |
|
101 |
// return n * fact(n-1) |
|
102 |
// } |
|
103 |
// ``` |
|
104 |
if p.flags&EXTENSION_FENCED_CODE != 0 { |
|
105 |
if i := p.fencedCodeBlock(out, data, true); i > 0 { |
|
106 |
data = data[i:] |
|
107 |
continue |
|
108 |
} |
|
109 |
} |
|
110 |
|
|
111 |
// horizontal rule: |
|
112 |
// |
|
113 |
// ------ |
|
114 |
// or |
|
115 |
// ****** |
|
116 |
// or |
|
117 |
// ______ |
|
118 |
if p.isHRule(data) { |
|
119 |
p.r.HRule(out) |
|
120 |
var i int |
|
121 |
for i = 0; data[i] != '\n'; i++ { |
|
122 |
} |
|
123 |
data = data[i:] |
|
124 |
continue |
|
125 |
} |
|
126 |
|
|
127 |
// block quote: |
|
128 |
// |
|
129 |
// > A big quote I found somewhere |
|
130 |
// > on the web |
|
131 |
if p.quotePrefix(data) > 0 { |
|
132 |
data = data[p.quote(out, data):] |
|
133 |
continue |
|
134 |
} |
|
135 |
|
|
136 |
// table: |
|
137 |
// |
|
138 |
// Name | Age | Phone |
|
139 |
// ------|-----|--------- |
|
140 |
// Bob | 31 | 555-1234 |
|
141 |
// Alice | 27 | 555-4321 |
|
142 |
if p.flags&EXTENSION_TABLES != 0 { |
|
143 |
if i := p.table(out, data); i > 0 { |
|
144 |
data = data[i:] |
|
145 |
continue |
|
146 |
} |
|
147 |
} |
|
148 |
|
|
149 |
// an itemized/unordered list: |
|
150 |
// |
|
151 |
// * Item 1 |
|
152 |
// * Item 2 |
|
153 |
// |
|
154 |
// also works with + or - |
|
155 |
if p.uliPrefix(data) > 0 { |
|
156 |
data = data[p.list(out, data, 0):] |
|
157 |
continue |
|
158 |
} |
|
159 |
|
|
160 |
// a numbered/ordered list: |
|
161 |
// |
|
162 |
// 1. Item 1 |
|
163 |
// 2. Item 2 |
|
164 |
if p.oliPrefix(data) > 0 { |
|
165 |
data = data[p.list(out, data, LIST_TYPE_ORDERED):] |
|
166 |
continue |
|
167 |
} |
|
168 |
|
|
169 |
// definition lists: |
|
170 |
// |
|
171 |
// Term 1 |
|
172 |
// : Definition a |
|
173 |
// : Definition b |
|
174 |
// |
|
175 |
// Term 2 |
|
176 |
// : Definition c |
|
177 |
if p.flags&EXTENSION_DEFINITION_LISTS != 0 { |
|
178 |
if p.dliPrefix(data) > 0 { |
|
179 |
data = data[p.list(out, data, LIST_TYPE_DEFINITION):] |
|
180 |
continue |
|
181 |
} |
|
182 |
} |
|
183 |
|
|
184 |
// anything else must look like a normal paragraph |
|
185 |
// note: this finds underlined headers, too |
|
186 |
data = data[p.paragraph(out, data):] |
|
187 |
} |
|
188 |
|
|
189 |
p.nesting-- |
|
190 |
} |
|
191 |
|
|
192 |
func (p *parser) isPrefixHeader(data []byte) bool { |
|
193 |
if data[0] != '#' { |
|
194 |
return false |
|
195 |
} |
|
196 |
|
|
197 |
if p.flags&EXTENSION_SPACE_HEADERS != 0 { |
|
198 |
level := 0 |
|
199 |
for level < 6 && data[level] == '#' { |
|
200 |
level++ |
|
201 |
} |
|
202 |
if data[level] != ' ' { |
|
203 |
return false |
|
204 |
} |
|
205 |
} |
|
206 |
return true |
|
207 |
} |
|
208 |
|
|
209 |
func (p *parser) prefixHeader(out *bytes.Buffer, data []byte) int { |
|
210 |
level := 0 |
|
211 |
for level < 6 && data[level] == '#' { |
|
212 |
level++ |
|
213 |
} |
|
214 |
i := skipChar(data, level, ' ') |
|
215 |
end := skipUntilChar(data, i, '\n') |
|
216 |
skip := end |
|
217 |
id := "" |
|
218 |
if p.flags&EXTENSION_HEADER_IDS != 0 { |
|
219 |
j, k := 0, 0 |
|
220 |
// find start/end of header id |
|
221 |
for j = i; j < end-1 && (data[j] != '{' || data[j+1] != '#'); j++ { |
|
222 |
} |
|
223 |
for k = j + 1; k < end && data[k] != '}'; k++ { |
|
224 |
} |
|
225 |
// extract header id iff found |
|
226 |
if j < end && k < end { |
|
227 |
id = string(data[j+2 : k]) |
|
228 |
end = j |
|
229 |
skip = k + 1 |
|
230 |
for end > 0 && data[end-1] == ' ' { |
|
231 |
end-- |
|
232 |
} |
|
233 |
} |
|
234 |
} |
|
235 |
for end > 0 && data[end-1] == '#' { |
|
236 |
if isBackslashEscaped(data, end-1) { |
|
237 |
break |
|
238 |
} |
|
239 |
end-- |
|
240 |
} |
|
241 |
for end > 0 && data[end-1] == ' ' { |
|
242 |
end-- |
|
243 |
} |
|
244 |
if end > i { |
|
245 |
if id == "" && p.flags&EXTENSION_AUTO_HEADER_IDS != 0 { |
|
246 |
id = SanitizedAnchorName(string(data[i:end])) |
|
247 |
} |
|
248 |
work := func() bool { |
|
249 |
p.inline(out, data[i:end]) |
|
250 |
return true |
|
251 |
} |
|
252 |
p.r.Header(out, work, level, id) |
|
253 |
} |
|
254 |
return skip |
|
255 |
} |
|
256 |
|
|
257 |
func (p *parser) isUnderlinedHeader(data []byte) int { |
|
258 |
// test of level 1 header |
|
259 |
if data[0] == '=' { |
|
260 |
i := skipChar(data, 1, '=') |
|
261 |
i = skipChar(data, i, ' ') |
|
262 |
if data[i] == '\n' { |
|
263 |
return 1 |
|
264 |
} else { |
|
265 |
return 0 |
|
266 |
} |
|
267 |
} |
|
268 |
|
|
269 |
// test of level 2 header |
|
270 |
if data[0] == '-' { |
|
271 |
i := skipChar(data, 1, '-') |
|
272 |
i = skipChar(data, i, ' ') |
|
273 |
if data[i] == '\n' { |
|
274 |
return 2 |
|
275 |
} else { |
|
276 |
return 0 |
|
277 |
} |
|
278 |
} |
|
279 |
|
|
280 |
return 0 |
|
281 |
} |
|
282 |
|
|
283 |
func (p *parser) titleBlock(out *bytes.Buffer, data []byte, doRender bool) int { |
|
284 |
if data[0] != '%' { |
|
285 |
return 0 |
|
286 |
} |
|
287 |
splitData := bytes.Split(data, []byte("\n")) |
|
288 |
var i int |
|
289 |
for idx, b := range splitData { |
|
290 |
if !bytes.HasPrefix(b, []byte("%")) { |
|
291 |
i = idx // - 1 |
|
292 |
break |
|
293 |
} |
|
294 |
} |
|
295 |
|
|
296 |
data = bytes.Join(splitData[0:i], []byte("\n")) |
|
297 |
p.r.TitleBlock(out, data) |
|
298 |
|
|
299 |
return len(data) |
|
300 |
} |
|
301 |
|
|
302 |
func (p *parser) html(out *bytes.Buffer, data []byte, doRender bool) int { |
|
303 |
var i, j int |
|
304 |
|
|
305 |
// identify the opening tag |
|
306 |
if data[0] != '<' { |
|
307 |
return 0 |
|
308 |
} |
|
309 |
curtag, tagfound := p.htmlFindTag(data[1:]) |
|
310 |
|
|
311 |
// handle special cases |
|
312 |
if !tagfound { |
|
313 |
// check for an HTML comment |
|
314 |
if size := p.htmlComment(out, data, doRender); size > 0 { |
|
315 |
return size |
|
316 |
} |
|
317 |
|
|
318 |
// check for an <hr> tag |
|
319 |
if size := p.htmlHr(out, data, doRender); size > 0 { |
|
320 |
return size |
|
321 |
} |
|
322 |
|
|
323 |
// check for HTML CDATA |
|
324 |
if size := p.htmlCDATA(out, data, doRender); size > 0 { |
|
325 |
return size |
|
326 |
} |
|
327 |
|
|
328 |
// no special case recognized |
|
329 |
return 0 |
|
330 |
} |
|
331 |
|
|
332 |
// look for an unindented matching closing tag |
|
333 |
// followed by a blank line |
|
334 |
found := false |
|
335 |
/* |
|
336 |
closetag := []byte("\n</" + curtag + ">") |
|
337 |
j = len(curtag) + 1 |
|
338 |
for !found { |
|
339 |
// scan for a closing tag at the beginning of a line |
|
340 |
if skip := bytes.Index(data[j:], closetag); skip >= 0 { |
|
341 |
j += skip + len(closetag) |
|
342 |
} else { |
|
343 |
break |
|
344 |
} |
|
345 |
|
|
346 |
// see if it is the only thing on the line |
|
347 |
if skip := p.isEmpty(data[j:]); skip > 0 { |
|
348 |
// see if it is followed by a blank line/eof |
|
349 |
j += skip |
|
350 |
if j >= len(data) { |
|
351 |
found = true |
|
352 |
i = j |
|
353 |
} else { |
|
354 |
if skip := p.isEmpty(data[j:]); skip > 0 { |
|
355 |
j += skip |
|
356 |
found = true |
|
357 |
i = j |
|
358 |
} |
|
359 |
} |
|
360 |
} |
|
361 |
} |
|
362 |
*/ |
|
363 |
|
|
364 |
// if not found, try a second pass looking for indented match |
|
365 |
// but not if tag is "ins" or "del" (following original Markdown.pl) |
|
366 |
if !found && curtag != "ins" && curtag != "del" { |
|
367 |
i = 1 |
|
368 |
for i < len(data) { |
|
369 |
i++ |
|
370 |
for i < len(data) && !(data[i-1] == '<' && data[i] == '/') { |
|
371 |
i++ |
|
372 |
} |
|
373 |
|
|
374 |
if i+2+len(curtag) >= len(data) { |
|
375 |
break |
|
376 |
} |
|
377 |
|
|
378 |
j = p.htmlFindEnd(curtag, data[i-1:]) |
|
379 |
|
|
380 |
if j > 0 { |
|
381 |
i += j - 1 |
|
382 |
found = true |
|
383 |
break |
|
384 |
} |
|
385 |
} |
|
386 |
} |
|
387 |
|
|
388 |
if !found { |
|
389 |
return 0 |
|
390 |
} |
|
391 |
|
|
392 |
// the end of the block has been found |
|
393 |
if doRender { |
|
394 |
// trim newlines |
|
395 |
end := i |
|
396 |
for end > 0 && data[end-1] == '\n' { |
|
397 |
end-- |
|
398 |
} |
|
399 |
p.r.BlockHtml(out, data[:end]) |
|
400 |
} |
|
401 |
|
|
402 |
return i |
|
403 |
} |
|
404 |
|
|
405 |
func (p *parser) renderHTMLBlock(out *bytes.Buffer, data []byte, start int, doRender bool) int { |
|
406 |
// html block needs to end with a blank line |
|
407 |
if i := p.isEmpty(data[start:]); i > 0 { |
|
408 |
size := start + i |
|
409 |
if doRender { |
|
410 |
// trim trailing newlines |
|
411 |
end := size |
|
412 |
for end > 0 && data[end-1] == '\n' { |
|
413 |
end-- |
|
414 |
} |
|
415 |
p.r.BlockHtml(out, data[:end]) |
|
416 |
} |
|
417 |
return size |
|
418 |
} |
|
419 |
return 0 |
|
420 |
} |
|
421 |
|
|
422 |
// HTML comment, lax form |
|
423 |
func (p *parser) htmlComment(out *bytes.Buffer, data []byte, doRender bool) int { |
|
424 |
i := p.inlineHTMLComment(out, data) |
|
425 |
return p.renderHTMLBlock(out, data, i, doRender) |
|
426 |
} |
|
427 |
|
|
428 |
// HTML CDATA section |
|
429 |
func (p *parser) htmlCDATA(out *bytes.Buffer, data []byte, doRender bool) int { |
|
430 |
const cdataTag = "<![cdata[" |
|
431 |
const cdataTagLen = len(cdataTag) |
|
432 |
if len(data) < cdataTagLen+1 { |
|
433 |
return 0 |
|
434 |
} |
|
435 |
if !bytes.Equal(bytes.ToLower(data[:cdataTagLen]), []byte(cdataTag)) { |
|
436 |
return 0 |
|
437 |
} |
|
438 |
i := cdataTagLen |
|
439 |
// scan for an end-of-comment marker, across lines if necessary |
|
440 |
for i < len(data) && !(data[i-2] == ']' && data[i-1] == ']' && data[i] == '>') { |
|
441 |
i++ |
|
442 |
} |
|
443 |
i++ |
|
444 |
// no end-of-comment marker |
|
445 |
if i >= len(data) { |
|
446 |
return 0 |
|
447 |
} |
|
448 |
return p.renderHTMLBlock(out, data, i, doRender) |
|
449 |
} |
|
450 |
|
|
451 |
// HR, which is the only self-closing block tag considered |
|
452 |
func (p *parser) htmlHr(out *bytes.Buffer, data []byte, doRender bool) int { |
|
453 |
if data[0] != '<' || (data[1] != 'h' && data[1] != 'H') || (data[2] != 'r' && data[2] != 'R') { |
|
454 |
return 0 |
|
455 |
} |
|
456 |
if data[3] != ' ' && data[3] != '/' && data[3] != '>' { |
|
457 |
// not an <hr> tag after all; at least not a valid one |
|
458 |
return 0 |
|
459 |
} |
|
460 |
|
|
461 |
i := 3 |
|
462 |
for data[i] != '>' && data[i] != '\n' { |
|
463 |
i++ |
|
464 |
} |
|
465 |
|
|
466 |
if data[i] == '>' { |
|
467 |
return p.renderHTMLBlock(out, data, i+1, doRender) |
|
468 |
} |
|
469 |
|
|
470 |
return 0 |
|
471 |
} |
|
472 |
|
|
473 |
func (p *parser) htmlFindTag(data []byte) (string, bool) { |
|
474 |
i := 0 |
|
475 |
for isalnum(data[i]) { |
|
476 |
i++ |
|
477 |
} |
|
478 |
key := string(data[:i]) |
|
479 |
if _, ok := blockTags[key]; ok { |
|
480 |
return key, true |
|
481 |
} |
|
482 |
return "", false |
|
483 |
} |
|
484 |
|
|
485 |
func (p *parser) htmlFindEnd(tag string, data []byte) int { |
|
486 |
// assume data[0] == '<' && data[1] == '/' already tested |
|
487 |
|
|
488 |
// check if tag is a match |
|
489 |
closetag := []byte("</" + tag + ">") |
|
490 |
if !bytes.HasPrefix(data, closetag) { |
|
491 |
return 0 |
|
492 |
} |
|
493 |
i := len(closetag) |
|
494 |
|
|
495 |
// check that the rest of the line is blank |
|
496 |
skip := 0 |
|
497 |
if skip = p.isEmpty(data[i:]); skip == 0 { |
|
498 |
return 0 |
|
499 |
} |
|
500 |
i += skip |
|
501 |
skip = 0 |
|
502 |
|
|
503 |
if i >= len(data) { |
|
504 |
return i |
|
505 |
} |
|
506 |
|
|
507 |
if p.flags&EXTENSION_LAX_HTML_BLOCKS != 0 { |
|
508 |
return i |
|
509 |
} |
|
510 |
if skip = p.isEmpty(data[i:]); skip == 0 { |
|
511 |
// following line must be blank |
|
512 |
return 0 |
|
513 |
} |
|
514 |
|
|
515 |
return i + skip |
|
516 |
} |
|
517 |
|
|
518 |
func (*parser) isEmpty(data []byte) int { |
|
519 |
// it is okay to call isEmpty on an empty buffer |
|
520 |
if len(data) == 0 { |
|
521 |
return 0 |
|
522 |
} |
|
523 |
|
|
524 |
var i int |
|
525 |
for i = 0; i < len(data) && data[i] != '\n'; i++ { |
|
526 |
if data[i] != ' ' && data[i] != '\t' { |
|
527 |
return 0 |
|
528 |
} |
|
529 |
} |
|
530 |
return i + 1 |
|
531 |
} |
|
532 |
|
|
533 |
func (*parser) isHRule(data []byte) bool { |
|
534 |
i := 0 |
|
535 |
|
|
536 |
// skip up to three spaces |
|
537 |
for i < 3 && data[i] == ' ' { |
|
538 |
i++ |
|
539 |
} |
|
540 |
|
|
541 |
// look at the hrule char |
|
542 |
if data[i] != '*' && data[i] != '-' && data[i] != '_' { |
|
543 |
return false |
|
544 |
} |
|
545 |
c := data[i] |
|
546 |
|
|
547 |
// the whole line must be the char or whitespace |
|
548 |
n := 0 |
|
549 |
for data[i] != '\n' { |
|
550 |
switch { |
|
551 |
case data[i] == c: |
|
552 |
n++ |
|
553 |
case data[i] != ' ': |
|
554 |
return false |
|
555 |
} |
|
556 |
i++ |
|
557 |
} |
|
558 |
|
|
559 |
return n >= 3 |
|
560 |
} |
|
561 |
|
|
562 |
// isFenceLine checks if there's a fence line (e.g., ``` or ``` go) at the beginning of data, |
|
563 |
// and returns the end index if so, or 0 otherwise. It also returns the marker found. |
|
564 |
// If syntax is not nil, it gets set to the syntax specified in the fence line. |
|
565 |
// A final newline is mandatory to recognize the fence line, unless newlineOptional is true. |
|
566 |
func isFenceLine(data []byte, info *string, oldmarker string, newlineOptional bool) (end int, marker string) { |
|
567 |
i, size := 0, 0 |
|
568 |
|
|
569 |
// skip up to three spaces |
|
570 |
for i < len(data) && i < 3 && data[i] == ' ' { |
|
571 |
i++ |
|
572 |
} |
|
573 |
|
|
574 |
// check for the marker characters: ~ or ` |
|
575 |
if i >= len(data) { |
|
576 |
return 0, "" |
|
577 |
} |
|
578 |
if data[i] != '~' && data[i] != '`' { |
|
579 |
return 0, "" |
|
580 |
} |
|
581 |
|
|
582 |
c := data[i] |
|
583 |
|
|
584 |
// the whole line must be the same char or whitespace |
|
585 |
for i < len(data) && data[i] == c { |
|
586 |
size++ |
|
587 |
i++ |
|
588 |
} |
|
589 |
|
|
590 |
// the marker char must occur at least 3 times |
|
591 |
if size < 3 { |
|
592 |
return 0, "" |
|
593 |
} |
|
594 |
marker = string(data[i-size : i]) |
|
595 |
|
|
596 |
// if this is the end marker, it must match the beginning marker |
|
597 |
if oldmarker != "" && marker != oldmarker { |
|
598 |
return 0, "" |
|
599 |
} |
|
600 |
|
|
601 |
// TODO(shurcooL): It's probably a good idea to simplify the 2 code paths here |
|
602 |
// into one, always get the info string, and discard it if the caller doesn't care. |
|
603 |
if info != nil { |
|
604 |
infoLength := 0 |
|
605 |
i = skipChar(data, i, ' ') |
|
606 |
|
|
607 |
if i >= len(data) { |
|
608 |
if newlineOptional && i == len(data) { |
|
609 |
return i, marker |
|
610 |
} |
|
611 |
return 0, "" |
|
612 |
} |
|
613 |
|
|
614 |
infoStart := i |
|
615 |
|
|
616 |
if data[i] == '{' { |
|
617 |
i++ |
|
618 |
infoStart++ |
|
619 |
|
|
620 |
for i < len(data) && data[i] != '}' && data[i] != '\n' { |
|
621 |
infoLength++ |
|
622 |
i++ |
|
623 |
} |
|
624 |
|
|
625 |
if i >= len(data) || data[i] != '}' { |
|
626 |
return 0, "" |
|
627 |
} |
|
628 |
|
|
629 |
// strip all whitespace at the beginning and the end |
|
630 |
// of the {} block |
|
631 |
for infoLength > 0 && isspace(data[infoStart]) { |
|
632 |
infoStart++ |
|
633 |
infoLength-- |
|
634 |
} |
|
635 |
|
|
636 |
for infoLength > 0 && isspace(data[infoStart+infoLength-1]) { |
|
637 |
infoLength-- |
|
638 |
} |
|
639 |
|
|
640 |
i++ |
|
641 |
} else { |
|
642 |
for i < len(data) && !isverticalspace(data[i]) { |
|
643 |
infoLength++ |
|
644 |
i++ |
|
645 |
} |
|
646 |
} |
|
647 |
|
|
648 |
*info = strings.TrimSpace(string(data[infoStart : infoStart+infoLength])) |
|
649 |
} |
|
650 |
|
|
651 |
i = skipChar(data, i, ' ') |
|
652 |
if i >= len(data) || data[i] != '\n' { |
|
653 |
if newlineOptional && i == len(data) { |
|
654 |
return i, marker |
|
655 |
} |
|
656 |
return 0, "" |
|
657 |
} |
|
658 |
|
|
659 |
return i + 1, marker // Take newline into account. |
|
660 |
} |
|
661 |
|
|
662 |
// fencedCodeBlock returns the end index if data contains a fenced code block at the beginning, |
|
663 |
// or 0 otherwise. It writes to out if doRender is true, otherwise it has no side effects. |
|
664 |
// If doRender is true, a final newline is mandatory to recognize the fenced code block. |
|
665 |
func (p *parser) fencedCodeBlock(out *bytes.Buffer, data []byte, doRender bool) int { |
|
666 |
var infoString string |
|
667 |
beg, marker := isFenceLine(data, &infoString, "", false) |
|
668 |
if beg == 0 || beg >= len(data) { |
|
669 |
return 0 |
|
670 |
} |
|
671 |
|
|
672 |
var work bytes.Buffer |
|
673 |
|
|
674 |
for { |
|
675 |
// safe to assume beg < len(data) |
|
676 |
|
|
677 |
// check for the end of the code block |
|
678 |
newlineOptional := !doRender |
|
679 |
fenceEnd, _ := isFenceLine(data[beg:], nil, marker, newlineOptional) |
|
680 |
if fenceEnd != 0 { |
|
681 |
beg += fenceEnd |
|
682 |
break |
|
683 |
} |
|
684 |
|
|
685 |
// copy the current line |
|
686 |
end := skipUntilChar(data, beg, '\n') + 1 |
|
687 |
|
|
688 |
// did we reach the end of the buffer without a closing marker? |
|
689 |
if end >= len(data) { |
|
690 |
return 0 |
|
691 |
} |
|
692 |
|
|
693 |
// verbatim copy to the working buffer |
|
694 |
if doRender { |
|
695 |
work.Write(data[beg:end]) |
|
696 |
} |
|
697 |
beg = end |
|
698 |
} |
|
699 |
|
|
700 |
if doRender { |
|
701 |
p.r.BlockCode(out, work.Bytes(), infoString) |
|
702 |
} |
|
703 |
|
|
704 |
return beg |
|
705 |
} |
|
706 |
|
|
707 |
func (p *parser) table(out *bytes.Buffer, data []byte) int { |
|
708 |
var header bytes.Buffer |
|
709 |
i, columns := p.tableHeader(&header, data) |
|
710 |
if i == 0 { |
|
711 |
return 0 |
|
712 |
} |
|
713 |
|
|
714 |
var body bytes.Buffer |
|
715 |
|
|
716 |
for i < len(data) { |
|
717 |
pipes, rowStart := 0, i |
|
718 |
for ; data[i] != '\n'; i++ { |
|
719 |
if data[i] == '|' { |
|
720 |
pipes++ |
|
721 |
} |
|
722 |
} |
|
723 |
|
|
724 |
if pipes == 0 { |
|
725 |
i = rowStart |
|
726 |
break |
|
727 |
} |
|
728 |
|
|
729 |
// include the newline in data sent to tableRow |
|
730 |
i++ |
|
731 |
p.tableRow(&body, data[rowStart:i], columns, false) |
|
732 |
} |
|
733 |
|
|
734 |
p.r.Table(out, header.Bytes(), body.Bytes(), columns) |
|
735 |
|
|
736 |
return i |
|
737 |
} |
|
738 |
|
|
739 |
// check if the specified position is preceded by an odd number of backslashes |
|
740 |
func isBackslashEscaped(data []byte, i int) bool { |
|
741 |
backslashes := 0 |
|
742 |
for i-backslashes-1 >= 0 && data[i-backslashes-1] == '\\' { |
|
743 |
backslashes++ |
|
744 |
} |
|
745 |
return backslashes&1 == 1 |
|
746 |
} |
|
747 |
|
|
748 |
func (p *parser) tableHeader(out *bytes.Buffer, data []byte) (size int, columns []int) { |
|
749 |
i := 0 |
|
750 |
colCount := 1 |
|
751 |
for i = 0; data[i] != '\n'; i++ { |
|
752 |
if data[i] == '|' && !isBackslashEscaped(data, i) { |
|
753 |
colCount++ |
|
754 |
} |
|
755 |
} |
|
756 |
|
|
757 |
// doesn't look like a table header |
|
758 |
if colCount == 1 { |
|
759 |
return |
|
760 |
} |
|
761 |
|
|
762 |
// include the newline in the data sent to tableRow |
|
763 |
header := data[:i+1] |
|
764 |
|
|
765 |
// column count ignores pipes at beginning or end of line |
|
766 |
if data[0] == '|' { |
|
767 |
colCount-- |
|
768 |
} |
|
769 |
if i > 2 && data[i-1] == '|' && !isBackslashEscaped(data, i-1) { |
|
770 |
colCount-- |
|
771 |
} |
|
772 |
|
|
773 |
columns = make([]int, colCount) |
|
774 |
|
|
775 |
// move on to the header underline |
|
776 |
i++ |
|
777 |
if i >= len(data) { |
|
778 |
return |
|
779 |
} |
|
780 |
|
|
781 |
if data[i] == '|' && !isBackslashEscaped(data, i) { |
|
782 |
i++ |
|
783 |
} |
|
784 |
i = skipChar(data, i, ' ') |
|
785 |
|
|
786 |
// each column header is of form: / *:?-+:? *|/ with # dashes + # colons >= 3 |
|
787 |
// and trailing | optional on last column |
|
788 |
col := 0 |
|
789 |
for data[i] != '\n' { |
|
790 |
dashes := 0 |
|
791 |
|
|
792 |
if data[i] == ':' { |
|
793 |
i++ |
|
794 |
columns[col] |= TABLE_ALIGNMENT_LEFT |
|
795 |
dashes++ |
|
796 |
} |
|
797 |
for data[i] == '-' { |
|
798 |
i++ |
|
799 |
dashes++ |
|
800 |
} |
|
801 |
if data[i] == ':' { |
|
802 |
i++ |
|
803 |
columns[col] |= TABLE_ALIGNMENT_RIGHT |
|
804 |
dashes++ |
|
805 |
} |
|
806 |
for data[i] == ' ' { |
|
807 |
i++ |
|
808 |
} |
|
809 |
|
|
810 |
// end of column test is messy |
|
811 |
switch { |
|
812 |
case dashes < 3: |
|
813 |
// not a valid column |
|
814 |
return |
|
815 |
|
|
816 |
case data[i] == '|' && !isBackslashEscaped(data, i): |
|
817 |
// marker found, now skip past trailing whitespace |
|
818 |
col++ |
|
819 |
i++ |
|
820 |
for data[i] == ' ' { |
|
821 |
i++ |
|
822 |
} |
|
823 |
|
|
824 |
// trailing junk found after last column |
|
825 |
if col >= colCount && data[i] != '\n' { |
|
826 |
return |
|
827 |
} |
|
828 |
|
|
829 |
case (data[i] != '|' || isBackslashEscaped(data, i)) && col+1 < colCount: |
|
830 |
// something else found where marker was required |
|
831 |
return |
|
832 |
|
|
833 |
case data[i] == '\n': |
|
834 |
// marker is optional for the last column |
|
835 |
col++ |
|
836 |
|
|
837 |
default: |
|
838 |
// trailing junk found after last column |
|
839 |
return |
|
840 |
} |
|
841 |
} |
|
842 |
if col != colCount { |
|
843 |
return |
|
844 |
} |
|
845 |
|
|
846 |
p.tableRow(out, header, columns, true) |
|
847 |
size = i + 1 |
|
848 |
return |
|
849 |
} |
|
850 |
|
|
851 |
func (p *parser) tableRow(out *bytes.Buffer, data []byte, columns []int, header bool) { |
|
852 |
i, col := 0, 0 |
|
853 |
var rowWork bytes.Buffer |
|
854 |
|
|
855 |
if data[i] == '|' && !isBackslashEscaped(data, i) { |
|
856 |
i++ |
|
857 |
} |
|
858 |
|
|
859 |
for col = 0; col < len(columns) && i < len(data); col++ { |
|
860 |
for data[i] == ' ' { |
|
861 |
i++ |
|
862 |
} |
|
863 |
|
|
864 |
cellStart := i |
|
865 |
|
|
866 |
for (data[i] != '|' || isBackslashEscaped(data, i)) && data[i] != '\n' { |
|
867 |
i++ |
|
868 |
} |
|
869 |
|
|
870 |
cellEnd := i |
|
871 |
|
|
872 |
// skip the end-of-cell marker, possibly taking us past end of buffer |
|
873 |
i++ |
|
874 |
|
|
875 |
for cellEnd > cellStart && data[cellEnd-1] == ' ' { |
|
876 |
cellEnd-- |
|
877 |
} |
|
878 |
|
|
879 |
var cellWork bytes.Buffer |
|
880 |
p.inline(&cellWork, data[cellStart:cellEnd]) |
|
881 |
|
|
882 |
if header { |
|
883 |
p.r.TableHeaderCell(&rowWork, cellWork.Bytes(), columns[col]) |
|
884 |
} else { |
|
885 |
p.r.TableCell(&rowWork, cellWork.Bytes(), columns[col]) |
|
886 |
} |
|
887 |
} |
|
888 |
|
|
889 |
// pad it out with empty columns to get the right number |
|
890 |
for ; col < len(columns); col++ { |
|
891 |
if header { |
|
892 |
p.r.TableHeaderCell(&rowWork, nil, columns[col]) |
|
893 |
} else { |
|
894 |
p.r.TableCell(&rowWork, nil, columns[col]) |
|
895 |
} |
|
896 |
} |
|
897 |
|
|
898 |
// silently ignore rows with too many cells |
|
899 |
|
|
900 |
p.r.TableRow(out, rowWork.Bytes()) |
|
901 |
} |
|
902 |
|
|
903 |
// returns blockquote prefix length |
|
904 |
func (p *parser) quotePrefix(data []byte) int { |
|
905 |
i := 0 |
|
906 |
for i < 3 && data[i] == ' ' { |
|
907 |
i++ |
|
908 |
} |
|
909 |
if data[i] == '>' { |
|
910 |
if data[i+1] == ' ' { |
|
911 |
return i + 2 |
|
912 |
} |
|
913 |
return i + 1 |
|
914 |
} |
|
915 |
return 0 |
|
916 |
} |
|
917 |
|
|
918 |
// blockquote ends with at least one blank line |
|
919 |
// followed by something without a blockquote prefix |
|
920 |
func (p *parser) terminateBlockquote(data []byte, beg, end int) bool { |
|
921 |
if p.isEmpty(data[beg:]) <= 0 { |
|
922 |
return false |
|
923 |
} |
|
924 |
if end >= len(data) { |
|
925 |
return true |
|
926 |
} |
|
927 |
return p.quotePrefix(data[end:]) == 0 && p.isEmpty(data[end:]) == 0 |
|
928 |
} |
|
929 |
|
|
930 |
// parse a blockquote fragment |
|
931 |
func (p *parser) quote(out *bytes.Buffer, data []byte) int { |
|
932 |
var raw bytes.Buffer |
|
933 |
beg, end := 0, 0 |
|
934 |
for beg < len(data) { |
|
935 |
end = beg |
|
936 |
// Step over whole lines, collecting them. While doing that, check for |
|
937 |
// fenced code and if one's found, incorporate it altogether, |
|
938 |
// irregardless of any contents inside it |
|
939 |
for data[end] != '\n' { |
|
940 |
if p.flags&EXTENSION_FENCED_CODE != 0 { |
|
941 |
if i := p.fencedCodeBlock(out, data[end:], false); i > 0 { |
|
942 |
// -1 to compensate for the extra end++ after the loop: |
|
943 |
end += i - 1 |
|
944 |
break |
|
945 |
} |
|
946 |
} |
|
947 |
end++ |
|
948 |
} |
|
949 |
end++ |
|
950 |
|
|
951 |
if pre := p.quotePrefix(data[beg:]); pre > 0 { |
|
952 |
// skip the prefix |
|
953 |
beg += pre |
|
954 |
} else if p.terminateBlockquote(data, beg, end) { |
|
955 |
break |
|
956 |
} |
|
957 |
|
|
958 |
// this line is part of the blockquote |
|
959 |
raw.Write(data[beg:end]) |
|
960 |
beg = end |
|
961 |
} |
|
962 |
|
|
963 |
var cooked bytes.Buffer |
|
964 |
p.block(&cooked, raw.Bytes()) |
|
965 |
p.r.BlockQuote(out, cooked.Bytes()) |
|
966 |
return end |
|
967 |
} |
|
968 |
|
|
969 |
// returns prefix length for block code |
|
970 |
func (p *parser) codePrefix(data []byte) int { |
|
971 |
if data[0] == ' ' && data[1] == ' ' && data[2] == ' ' && data[3] == ' ' { |
|
972 |
return 4 |
|
973 |
} |
|
974 |
return 0 |
|
975 |
} |
|
976 |
|
|
977 |
func (p *parser) code(out *bytes.Buffer, data []byte) int { |
|
978 |
var work bytes.Buffer |
|
979 |
|
|
980 |
i := 0 |
|
981 |
for i < len(data) { |
|
982 |
beg := i |
|
983 |
for data[i] != '\n' { |
|
984 |
i++ |
|
985 |
} |
|
986 |
i++ |
|
987 |
|
|
988 |
blankline := p.isEmpty(data[beg:i]) > 0 |
|
989 |
if pre := p.codePrefix(data[beg:i]); pre > 0 { |
|
990 |
beg += pre |
|
991 |
} else if !blankline { |
|
992 |
// non-empty, non-prefixed line breaks the pre |
|
993 |
i = beg |
|
994 |
break |
|
995 |
} |
|
996 |
|
|
997 |
// verbatim copy to the working buffeu |
|
998 |
if blankline { |
|
999 |
work.WriteByte('\n') |
|
1000 |
} else { |
|
1001 |
work.Write(data[beg:i]) |
|
1002 |
} |
|
1003 |
} |
|
1004 |
|
|
1005 |
// trim all the \n off the end of work |
|
1006 |
workbytes := work.Bytes() |
|
1007 |
eol := len(workbytes) |
|
1008 |
for eol > 0 && workbytes[eol-1] == '\n' { |
|
1009 |
eol-- |
|
1010 |
} |
|
1011 |
if eol != len(workbytes) { |
|
1012 |
work.Truncate(eol) |
|
1013 |
} |
|
1014 |
|
|
1015 |
work.WriteByte('\n') |
|
1016 |
|
|
1017 |
p.r.BlockCode(out, work.Bytes(), "") |
|
1018 |
|
|
1019 |
return i |
|
1020 |
} |
|
1021 |
|
|
1022 |
// returns unordered list item prefix |
|
1023 |
func (p *parser) uliPrefix(data []byte) int { |
|
1024 |
i := 0 |
|
1025 |
|
|
1026 |
// start with up to 3 spaces |
|
1027 |
for i < 3 && data[i] == ' ' { |
|
1028 |
i++ |
|
1029 |
} |
|
1030 |
|
|
1031 |
// need a *, +, or - followed by a space |
|
1032 |
if (data[i] != '*' && data[i] != '+' && data[i] != '-') || |
|
1033 |
data[i+1] != ' ' { |
|
1034 |
return 0 |
|
1035 |
} |
|
1036 |
return i + 2 |
|
1037 |
} |
|
1038 |
|
|
1039 |
// returns ordered list item prefix |
|
1040 |
func (p *parser) oliPrefix(data []byte) int { |
|
1041 |
i := 0 |
|
1042 |
|
|
1043 |
// start with up to 3 spaces |
|
1044 |
for i < 3 && data[i] == ' ' { |
|
1045 |
i++ |
|
1046 |
} |
|
1047 |
|
|
1048 |
// count the digits |
|
1049 |
start := i |
|
1050 |
for data[i] >= '0' && data[i] <= '9' { |
|
1051 |
i++ |
|
1052 |
} |
|
1053 |
|
|
1054 |
// we need >= 1 digits followed by a dot and a space |
|
1055 |
if start == i || data[i] != '.' || data[i+1] != ' ' { |
|
1056 |
return 0 |
|
1057 |
} |
|
1058 |
return i + 2 |
|
1059 |
} |
|
1060 |
|
|
1061 |
// returns definition list item prefix |
|
1062 |
func (p *parser) dliPrefix(data []byte) int { |
|
1063 |
i := 0 |
|
1064 |
|
|
1065 |
// need a : followed by a spaces |
|
1066 |
if data[i] != ':' || data[i+1] != ' ' { |
|
1067 |
return 0 |
|
1068 |
} |
|
1069 |
for data[i] == ' ' { |
|
1070 |
i++ |
|
1071 |
} |
|
1072 |
return i + 2 |
|
1073 |
} |
|
1074 |
|
|
1075 |
// parse ordered or unordered list block |
|
1076 |
func (p *parser) list(out *bytes.Buffer, data []byte, flags int) int { |
|
1077 |
i := 0 |
|
1078 |
flags |= LIST_ITEM_BEGINNING_OF_LIST |
|
1079 |
work := func() bool { |
|
1080 |
for i < len(data) { |
|
1081 |
skip := p.listItem(out, data[i:], &flags) |
|
1082 |
i += skip |
|
1083 |
|
|
1084 |
if skip == 0 || flags&LIST_ITEM_END_OF_LIST != 0 { |
|
1085 |
break |
|
1086 |
} |
|
1087 |
flags &= ^LIST_ITEM_BEGINNING_OF_LIST |
|
1088 |
} |
|
1089 |
return true |
|
1090 |
} |
|
1091 |
|
|
1092 |
p.r.List(out, work, flags) |
|
1093 |
return i |
|
1094 |
} |
|
1095 |
|
|
1096 |
// Parse a single list item. |
|
1097 |
// Assumes initial prefix is already removed if this is a sublist. |
|
1098 |
func (p *parser) listItem(out *bytes.Buffer, data []byte, flags *int) int { |
|
1099 |
// keep track of the indentation of the first line |
|
1100 |
itemIndent := 0 |
|
1101 |
for itemIndent < 3 && data[itemIndent] == ' ' { |
|
1102 |
itemIndent++ |
|
1103 |
} |
|
1104 |
|
|
1105 |
i := p.uliPrefix(data) |
|
1106 |
if i == 0 { |
|
1107 |
i = p.oliPrefix(data) |
|
1108 |
} |
|
1109 |
if i == 0 { |
|
1110 |
i = p.dliPrefix(data) |
|
1111 |
// reset definition term flag |
|
1112 |
if i > 0 { |
|
1113 |
*flags &= ^LIST_TYPE_TERM |
|
1114 |
} |
|
1115 |
} |
|
1116 |
if i == 0 { |
|
1117 |
// if in defnition list, set term flag and continue |
|
1118 |
if *flags&LIST_TYPE_DEFINITION != 0 { |
|
1119 |
*flags |= LIST_TYPE_TERM |
|
1120 |
} else { |
|
1121 |
return 0 |
|
1122 |
} |
|
1123 |
} |
|
1124 |
|
|
1125 |
// skip leading whitespace on first line |
|
1126 |
for data[i] == ' ' { |
|
1127 |
i++ |
|
1128 |
} |
|
1129 |
|
|
1130 |
// find the end of the line |
|
1131 |
line := i |
|
1132 |
for i > 0 && data[i-1] != '\n' { |
|
1133 |
i++ |
|
1134 |
} |
|
1135 |
|
|
1136 |
// get working buffer |
|
1137 |
var raw bytes.Buffer |
|
1138 |
|
|
1139 |
// put the first line into the working buffer |
|
1140 |
raw.Write(data[line:i]) |
|
1141 |
line = i |
|
1142 |
|
|
1143 |
// process the following lines |
|
1144 |
containsBlankLine := false |
|
1145 |
sublist := 0 |
|
1146 |
codeBlockMarker := "" |
|
1147 |
|
|
1148 |
gatherlines: |
|
1149 |
for line < len(data) { |
|
1150 |
i++ |
|
1151 |
|
|
1152 |
// find the end of this line |
|
1153 |
for data[i-1] != '\n' { |
|
1154 |
i++ |
|
1155 |
} |
|
1156 |
|
|
1157 |
// if it is an empty line, guess that it is part of this item |
|
1158 |
// and move on to the next line |
|
1159 |
if p.isEmpty(data[line:i]) > 0 { |
|
1160 |
containsBlankLine = true |
|
1161 |
raw.Write(data[line:i]) |
|
1162 |
line = i |
|
1163 |
continue |
|
1164 |
} |
|
1165 |
|
|
1166 |
// calculate the indentation |
|
1167 |
indent := 0 |
|
1168 |
for indent < 4 && line+indent < i && data[line+indent] == ' ' { |
|
1169 |
indent++ |
|
1170 |
} |
|
1171 |
|
|
1172 |
chunk := data[line+indent : i] |
|
1173 |
|
|
1174 |
if p.flags&EXTENSION_FENCED_CODE != 0 { |
|
1175 |
// determine if in or out of codeblock |
|
1176 |
// if in codeblock, ignore normal list processing |
|
1177 |
_, marker := isFenceLine(chunk, nil, codeBlockMarker, false) |
|
1178 |
if marker != "" { |
|
1179 |
if codeBlockMarker == "" { |
|
1180 |
// start of codeblock |
|
1181 |
codeBlockMarker = marker |
|
1182 |
} else { |
|
1183 |
// end of codeblock. |
|
1184 |
*flags |= LIST_ITEM_CONTAINS_BLOCK |
|
1185 |
codeBlockMarker = "" |
|
1186 |
} |
|
1187 |
} |
|
1188 |
// we are in a codeblock, write line, and continue |
|
1189 |
if codeBlockMarker != "" || marker != "" { |
|
1190 |
raw.Write(data[line+indent : i]) |
|
1191 |
line = i |
|
1192 |
continue gatherlines |
|
1193 |
} |
|
1194 |
} |
|
1195 |
|
|
1196 |
// evaluate how this line fits in |
|
1197 |
switch { |
|
1198 |
// is this a nested list item? |
|
1199 |
case (p.uliPrefix(chunk) > 0 && !p.isHRule(chunk)) || |
|
1200 |
p.oliPrefix(chunk) > 0 || |
|
1201 |
p.dliPrefix(chunk) > 0: |
|
1202 |
|
|
1203 |
if containsBlankLine { |
|
1204 |
// end the list if the type changed after a blank line |
|
1205 |
if indent <= itemIndent && |
|
1206 |
((*flags&LIST_TYPE_ORDERED != 0 && p.uliPrefix(chunk) > 0) || |
|
1207 |
(*flags&LIST_TYPE_ORDERED == 0 && p.oliPrefix(chunk) > 0)) { |
|
1208 |
|
|
1209 |
*flags |= LIST_ITEM_END_OF_LIST |
|
1210 |
break gatherlines |
|
1211 |
} |
|
1212 |
*flags |= LIST_ITEM_CONTAINS_BLOCK |
|
1213 |
} |
|
1214 |
|
|
1215 |
// to be a nested list, it must be indented more |
|
1216 |
// if not, it is the next item in the same list |
|
1217 |
if indent <= itemIndent { |
|
1218 |
break gatherlines |
|
1219 |
} |
|
1220 |
|
|
1221 |
// is this the first item in the nested list? |
|
1222 |
if sublist == 0 { |
|
1223 |
sublist = raw.Len() |
|
1224 |
} |
|
1225 |
|
|
1226 |
// is this a nested prefix header? |
|
1227 |
case p.isPrefixHeader(chunk): |
|
1228 |
// if the header is not indented, it is not nested in the list |
|
1229 |
// and thus ends the list |
|
1230 |
if containsBlankLine && indent < 4 { |
|
1231 |
*flags |= LIST_ITEM_END_OF_LIST |
|
1232 |
break gatherlines |
|
1233 |
} |
|
1234 |
*flags |= LIST_ITEM_CONTAINS_BLOCK |
|
1235 |
|
|
1236 |
// anything following an empty line is only part |
|
1237 |
// of this item if it is indented 4 spaces |
|
1238 |
// (regardless of the indentation of the beginning of the item) |
|
1239 |
case containsBlankLine && indent < 4: |
|
1240 |
if *flags&LIST_TYPE_DEFINITION != 0 && i < len(data)-1 { |
|
1241 |
// is the next item still a part of this list? |
|
1242 |
next := i |
|
1243 |
for data[next] != '\n' { |
|
1244 |
next++ |
|
1245 |
} |
|
1246 |
for next < len(data)-1 && data[next] == '\n' { |
|
1247 |
next++ |
|
1248 |
} |
|
1249 |
if i < len(data)-1 && data[i] != ':' && data[next] != ':' { |
|
1250 |
*flags |= LIST_ITEM_END_OF_LIST |
|
1251 |
} |
|
1252 |
} else { |
|
1253 |
*flags |= LIST_ITEM_END_OF_LIST |
|
1254 |
} |
|
1255 |
break gatherlines |
|
1256 |
|
|
1257 |
// a blank line means this should be parsed as a block |
|
1258 |
case containsBlankLine: |
|
1259 |
*flags |= LIST_ITEM_CONTAINS_BLOCK |
|
1260 |
} |
|
1261 |
|
|
1262 |
containsBlankLine = false |
|
1263 |
|
|
1264 |
// add the line into the working buffer without prefix |
|
1265 |
raw.Write(data[line+indent : i]) |
|
1266 |
|
|
1267 |
line = i |
|
1268 |
} |
|
1269 |
|
|
1270 |
// If reached end of data, the Renderer.ListItem call we're going to make below |
|
1271 |
// is definitely the last in the list. |
|
1272 |
if line >= len(data) { |
|
1273 |
*flags |= LIST_ITEM_END_OF_LIST |
|
1274 |
} |
|
1275 |
|
|
1276 |
rawBytes := raw.Bytes() |
|
1277 |
|
|
1278 |
// render the contents of the list item |
|
1279 |
var cooked bytes.Buffer |
|
1280 |
if *flags&LIST_ITEM_CONTAINS_BLOCK != 0 && *flags&LIST_TYPE_TERM == 0 { |
|
1281 |
// intermediate render of block item, except for definition term |
|
1282 |
if sublist > 0 { |
|
1283 |
p.block(&cooked, rawBytes[:sublist]) |
|
1284 |
p.block(&cooked, rawBytes[sublist:]) |
|
1285 |
} else { |
|
1286 |
p.block(&cooked, rawBytes) |
|
1287 |
} |
|
1288 |
} else { |
|
1289 |
// intermediate render of inline item |
|
1290 |
if sublist > 0 { |
|
1291 |
p.inline(&cooked, rawBytes[:sublist]) |
|
1292 |
p.block(&cooked, rawBytes[sublist:]) |
|
1293 |
} else { |
|
1294 |
p.inline(&cooked, rawBytes) |
|
1295 |
} |
|
1296 |
} |
|
1297 |
|
|
1298 |
// render the actual list item |
|
1299 |
cookedBytes := cooked.Bytes() |
|
1300 |
parsedEnd := len(cookedBytes) |
|
1301 |
|
|
1302 |
// strip trailing newlines |
|
1303 |
for parsedEnd > 0 && cookedBytes[parsedEnd-1] == '\n' { |
|
1304 |
parsedEnd-- |
|
1305 |
} |
|
1306 |
p.r.ListItem(out, cookedBytes[:parsedEnd], *flags) |
|
1307 |
|
|
1308 |
return line |
|
1309 |
} |
|
1310 |
|
|
1311 |
// render a single paragraph that has already been parsed out |
|
1312 |
func (p *parser) renderParagraph(out *bytes.Buffer, data []byte) { |
|
1313 |
if len(data) == 0 { |
|
1314 |
return |
|
1315 |
} |
|
1316 |
|
|
1317 |
// trim leading spaces |
|
1318 |
beg := 0 |
|
1319 |
for data[beg] == ' ' { |
|
1320 |
beg++ |
|
1321 |
} |
|
1322 |
|
|
1323 |
// trim trailing newline |
|
1324 |
end := len(data) - 1 |
|
1325 |
|
|
1326 |
// trim trailing spaces |
|
1327 |
for end > beg && data[end-1] == ' ' { |
|
1328 |
end-- |
|
1329 |
} |
|
1330 |
|
|
1331 |
work := func() bool { |
|
1332 |
p.inline(out, data[beg:end]) |
|
1333 |
return true |
|
1334 |
} |
|
1335 |
p.r.Paragraph(out, work) |
|
1336 |
} |
|
1337 |
|
|
1338 |
func (p *parser) paragraph(out *bytes.Buffer, data []byte) int { |
|
1339 |
// prev: index of 1st char of previous line |
|
1340 |
// line: index of 1st char of current line |
|
1341 |
// i: index of cursor/end of current line |
|
1342 |
var prev, line, i int |
|
1343 |
|
|
1344 |
// keep going until we find something to mark the end of the paragraph |
|
1345 |
for i < len(data) { |
|
1346 |
// mark the beginning of the current line |
|
1347 |
prev = line |
|
1348 |
current := data[i:] |
|
1349 |
line = i |
|
1350 |
|
|
1351 |
// did we find a blank line marking the end of the paragraph? |
|
1352 |
if n := p.isEmpty(current); n > 0 { |
|
1353 |
// did this blank line followed by a definition list item? |
|
1354 |
if p.flags&EXTENSION_DEFINITION_LISTS != 0 { |
|
1355 |
if i < len(data)-1 && data[i+1] == ':' { |
|
1356 |
return p.list(out, data[prev:], LIST_TYPE_DEFINITION) |
|
1357 |
} |
|
1358 |
} |
|
1359 |
|
|
1360 |
p.renderParagraph(out, data[:i]) |
|
1361 |
return i + n |
|
1362 |
} |
|
1363 |
|
|
1364 |
// an underline under some text marks a header, so our paragraph ended on prev line |
|
1365 |
if i > 0 { |
|
1366 |
if level := p.isUnderlinedHeader(current); level > 0 { |
|
1367 |
// render the paragraph |
|
1368 |
p.renderParagraph(out, data[:prev]) |
|
1369 |
|
|
1370 |
// ignore leading and trailing whitespace |
|
1371 |
eol := i - 1 |
|
1372 |
for prev < eol && data[prev] == ' ' { |
|
1373 |
prev++ |
|
1374 |
} |
|
1375 |
for eol > prev && data[eol-1] == ' ' { |
|
1376 |
eol-- |
|
1377 |
} |
|
1378 |
|
|
1379 |
// render the header |
|
1380 |
// this ugly double closure avoids forcing variables onto the heap |
|
1381 |
work := func(o *bytes.Buffer, pp *parser, d []byte) func() bool { |
|
1382 |
return func() bool { |
|
1383 |
pp.inline(o, d) |
|
1384 |
return true |
|
1385 |
} |
|
1386 |
}(out, p, data[prev:eol]) |
|
1387 |
|
|
1388 |
id := "" |
|
1389 |
if p.flags&EXTENSION_AUTO_HEADER_IDS != 0 { |
|
1390 |
id = SanitizedAnchorName(string(data[prev:eol])) |
|
1391 |
} |
|
1392 |
|
|
1393 |
p.r.Header(out, work, level, id) |
|
1394 |
|
|
1395 |
// find the end of the underline |
|
1396 |
for data[i] != '\n' { |
|
1397 |
i++ |
|
1398 |
} |
|
1399 |
return i |
|
1400 |
} |
|
1401 |
} |
|
1402 |
|
|
1403 |
// if the next line starts a block of HTML, then the paragraph ends here |
|
1404 |
if p.flags&EXTENSION_LAX_HTML_BLOCKS != 0 { |
|
1405 |
if data[i] == '<' && p.html(out, current, false) > 0 { |
|
1406 |
// rewind to before the HTML block |
|
1407 |
p.renderParagraph(out, data[:i]) |
|
1408 |
return i |
|
1409 |
} |
|
1410 |
} |
|
1411 |
|
|
1412 |
// if there's a prefixed header or a horizontal rule after this, paragraph is over |
|
1413 |
if p.isPrefixHeader(current) || p.isHRule(current) { |
|
1414 |
p.renderParagraph(out, data[:i]) |
|
1415 |
return i |
|
1416 |
} |
|
1417 |
|
|
1418 |
// if there's a fenced code block, paragraph is over |
|
1419 |
if p.flags&EXTENSION_FENCED_CODE != 0 { |
|
1420 |
if p.fencedCodeBlock(out, current, false) > 0 { |
|
1421 |
p.renderParagraph(out, data[:i]) |
|
1422 |
return i |
|
1423 |
} |
|
1424 |
} |
|
1425 |
|
|
1426 |
// if there's a definition list item, prev line is a definition term |
|
1427 |
if p.flags&EXTENSION_DEFINITION_LISTS != 0 { |
|
1428 |
if p.dliPrefix(current) != 0 { |
|
1429 |
return p.list(out, data[prev:], LIST_TYPE_DEFINITION) |
|
1430 |
} |
|
1431 |
} |
|
1432 |
|
|
1433 |
// if there's a list after this, paragraph is over |
|
1434 |
if p.flags&EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK != 0 { |
|
1435 |
if p.uliPrefix(current) != 0 || |
|
1436 |
p.oliPrefix(current) != 0 || |
|
1437 |
p.quotePrefix(current) != 0 || |
|
1438 |
p.codePrefix(current) != 0 { |
|
1439 |
p.renderParagraph(out, data[:i]) |
|
1440 |
return i |
|
1441 |
} |
|
1442 |
} |
|
1443 |
|
|
1444 |
// otherwise, scan to the beginning of the next line |
|
1445 |
for data[i] != '\n' { |
|
1446 |
i++ |
|
1447 |
} |
|
1448 |
i++ |
|
1449 |
} |
|
1450 |
|
|
1451 |
p.renderParagraph(out, data[:i]) |
|
1452 |
return i |
|
1453 |
} |
|
1454 |
|
|
1455 |
// SanitizedAnchorName returns a sanitized anchor name for the given text. |
|
1456 |
// |
|
1457 |
// It implements the algorithm specified in the package comment. |
|
1458 |
func SanitizedAnchorName(text string) string { |
|
1459 |
var anchorName []rune |
|
1460 |
futureDash := false |
|
1461 |
for _, r := range text { |
|
1462 |
switch { |
|
1463 |
case unicode.IsLetter(r) || unicode.IsNumber(r): |
|
1464 |
if futureDash && len(anchorName) > 0 { |
|
1465 |
anchorName = append(anchorName, '-') |
|
1466 |
} |
|
1467 |
futureDash = false |
|
1468 |
anchorName = append(anchorName, unicode.ToLower(r)) |
|
1469 |
default: |
|
1470 |
futureDash = true |
|
1471 |
} |
|
1472 |
} |
|
1473 |
return string(anchorName) |
|
1474 |
} |