block.go (view raw)
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
14package blackfriday
15
16import (
17 "bytes"
18
19 "github.com/shurcooL/sanitized_anchor_name"
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.
25func (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
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.fencedCode(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 // anything else must look like a normal paragraph
170 // note: this finds underlined headers, too
171 data = data[p.paragraph(out, data):]
172 }
173
174 p.nesting--
175}
176
177func (p *parser) isPrefixHeader(data []byte) bool {
178 if data[0] != '#' {
179 return false
180 }
181
182 if p.flags&EXTENSION_SPACE_HEADERS != 0 {
183 level := 0
184 for level < 6 && data[level] == '#' {
185 level++
186 }
187 if data[level] != ' ' {
188 return false
189 }
190 }
191 return true
192}
193
194func (p *parser) prefixHeader(out *bytes.Buffer, data []byte) int {
195 level := 0
196 for level < 6 && data[level] == '#' {
197 level++
198 }
199 i, end := 0, 0
200 for i = level; data[i] == ' '; i++ {
201 }
202 for end = i; data[end] != '\n'; end++ {
203 }
204 skip := end
205 id := ""
206 if p.flags&EXTENSION_HEADER_IDS != 0 {
207 j, k := 0, 0
208 // find start/end of header id
209 for j = i; j < end-1 && (data[j] != '{' || data[j+1] != '#'); j++ {
210 }
211 for k = j + 1; k < end && data[k] != '}'; k++ {
212 }
213 // extract header id iff found
214 if j < end && k < end {
215 id = string(data[j+2 : k])
216 end = j
217 skip = k + 1
218 for end > 0 && data[end-1] == ' ' {
219 end--
220 }
221 }
222 }
223 for end > 0 && data[end-1] == '#' {
224 end--
225 }
226 for end > 0 && data[end-1] == ' ' {
227 end--
228 }
229 if end > i {
230 if id == "" && p.flags&EXTENSION_AUTO_HEADER_IDS != 0 {
231 id = sanitized_anchor_name.Create(string(data[i:end]))
232 }
233 work := func() bool {
234 p.inline(out, data[i:end])
235 return true
236 }
237 p.r.Header(out, work, level, id)
238 }
239 return skip
240}
241
242func (p *parser) isUnderlinedHeader(data []byte) int {
243 // test of level 1 header
244 if data[0] == '=' {
245 i := 1
246 for data[i] == '=' {
247 i++
248 }
249 for data[i] == ' ' {
250 i++
251 }
252 if data[i] == '\n' {
253 return 1
254 } else {
255 return 0
256 }
257 }
258
259 // test of level 2 header
260 if data[0] == '-' {
261 i := 1
262 for data[i] == '-' {
263 i++
264 }
265 for data[i] == ' ' {
266 i++
267 }
268 if data[i] == '\n' {
269 return 2
270 } else {
271 return 0
272 }
273 }
274
275 return 0
276}
277
278func (p *parser) titleBlock(out *bytes.Buffer, data []byte, doRender bool) int {
279 if data[0] != '%' {
280 return 0
281 }
282 splitData := bytes.Split(data, []byte("\n"))
283 var i int
284 for idx, b := range splitData {
285 if !bytes.HasPrefix(b, []byte("%")) {
286 i = idx // - 1
287 break
288 }
289 }
290
291 data = bytes.Join(splitData[0:i], []byte("\n"))
292 p.r.TitleBlock(out, data)
293
294 return len(data)
295}
296
297func (p *parser) html(out *bytes.Buffer, data []byte, doRender bool) int {
298 var i, j int
299
300 // identify the opening tag
301 if data[0] != '<' {
302 return 0
303 }
304 curtag, tagfound := p.htmlFindTag(data[1:])
305
306 // handle special cases
307 if !tagfound {
308 // check for an HTML comment
309 if size := p.htmlComment(out, data, doRender); size > 0 {
310 return size
311 }
312
313 // check for an <hr> tag
314 if size := p.htmlHr(out, data, doRender); size > 0 {
315 return size
316 }
317
318 // no special case recognized
319 return 0
320 }
321
322 // look for an unindented matching closing tag
323 // followed by a blank line
324 found := false
325 /*
326 closetag := []byte("\n</" + curtag + ">")
327 j = len(curtag) + 1
328 for !found {
329 // scan for a closing tag at the beginning of a line
330 if skip := bytes.Index(data[j:], closetag); skip >= 0 {
331 j += skip + len(closetag)
332 } else {
333 break
334 }
335
336 // see if it is the only thing on the line
337 if skip := p.isEmpty(data[j:]); skip > 0 {
338 // see if it is followed by a blank line/eof
339 j += skip
340 if j >= len(data) {
341 found = true
342 i = j
343 } else {
344 if skip := p.isEmpty(data[j:]); skip > 0 {
345 j += skip
346 found = true
347 i = j
348 }
349 }
350 }
351 }
352 */
353
354 // if not found, try a second pass looking for indented match
355 // but not if tag is "ins" or "del" (following original Markdown.pl)
356 if !found && curtag != "ins" && curtag != "del" {
357 i = 1
358 for i < len(data) {
359 i++
360 for i < len(data) && !(data[i-1] == '<' && data[i] == '/') {
361 i++
362 }
363
364 if i+2+len(curtag) >= len(data) {
365 break
366 }
367
368 j = p.htmlFindEnd(curtag, data[i-1:])
369
370 if j > 0 {
371 i += j - 1
372 found = true
373 break
374 }
375 }
376 }
377
378 if !found {
379 return 0
380 }
381
382 // the end of the block has been found
383 if doRender {
384 // trim newlines
385 end := i
386 for end > 0 && data[end-1] == '\n' {
387 end--
388 }
389 p.r.BlockHtml(out, data[:end])
390 }
391
392 return i
393}
394
395// HTML comment, lax form
396func (p *parser) htmlComment(out *bytes.Buffer, data []byte, doRender bool) int {
397 if data[0] != '<' || data[1] != '!' || data[2] != '-' || data[3] != '-' {
398 return 0
399 }
400
401 i := 5
402
403 // scan for an end-of-comment marker, across lines if necessary
404 for i < len(data) && !(data[i-2] == '-' && data[i-1] == '-' && data[i] == '>') {
405 i++
406 }
407 i++
408
409 // no end-of-comment marker
410 if i >= len(data) {
411 return 0
412 }
413
414 // needs to end with a blank line
415 if j := p.isEmpty(data[i:]); j > 0 {
416 size := i + j
417 if doRender {
418 // trim trailing newlines
419 end := size
420 for end > 0 && data[end-1] == '\n' {
421 end--
422 }
423 p.r.BlockHtml(out, data[:end])
424 }
425 return size
426 }
427
428 return 0
429}
430
431// HR, which is the only self-closing block tag considered
432func (p *parser) htmlHr(out *bytes.Buffer, data []byte, doRender bool) int {
433 if data[0] != '<' || (data[1] != 'h' && data[1] != 'H') || (data[2] != 'r' && data[2] != 'R') {
434 return 0
435 }
436 if data[3] != ' ' && data[3] != '/' && data[3] != '>' {
437 // not an <hr> tag after all; at least not a valid one
438 return 0
439 }
440
441 i := 3
442 for data[i] != '>' && data[i] != '\n' {
443 i++
444 }
445
446 if data[i] == '>' {
447 i++
448 if j := p.isEmpty(data[i:]); j > 0 {
449 size := i + j
450 if doRender {
451 // trim newlines
452 end := size
453 for end > 0 && data[end-1] == '\n' {
454 end--
455 }
456 p.r.BlockHtml(out, data[:end])
457 }
458 return size
459 }
460 }
461
462 return 0
463}
464
465func (p *parser) htmlFindTag(data []byte) (string, bool) {
466 i := 0
467 for isalnum(data[i]) {
468 i++
469 }
470 key := string(data[:i])
471 if blockTags[key] {
472 return key, true
473 }
474 return "", false
475}
476
477func (p *parser) htmlFindEnd(tag string, data []byte) int {
478 // assume data[0] == '<' && data[1] == '/' already tested
479
480 // check if tag is a match
481 closetag := []byte("</" + tag + ">")
482 if !bytes.HasPrefix(data, closetag) {
483 return 0
484 }
485 i := len(closetag)
486
487 // check that the rest of the line is blank
488 skip := 0
489 if skip = p.isEmpty(data[i:]); skip == 0 {
490 return 0
491 }
492 i += skip
493 skip = 0
494
495 if i >= len(data) {
496 return i
497 }
498
499 if p.flags&EXTENSION_LAX_HTML_BLOCKS != 0 {
500 return i
501 }
502 if skip = p.isEmpty(data[i:]); skip == 0 {
503 // following line must be blank
504 return 0
505 }
506
507 return i + skip
508}
509
510func (p *parser) isEmpty(data []byte) int {
511 // it is okay to call isEmpty on an empty buffer
512 if len(data) == 0 {
513 return 0
514 }
515
516 var i int
517 for i = 0; i < len(data) && data[i] != '\n'; i++ {
518 if data[i] != ' ' && data[i] != '\t' {
519 return 0
520 }
521 }
522 return i + 1
523}
524
525func (p *parser) isHRule(data []byte) bool {
526 i := 0
527
528 // skip up to three spaces
529 for i < 3 && data[i] == ' ' {
530 i++
531 }
532
533 // look at the hrule char
534 if data[i] != '*' && data[i] != '-' && data[i] != '_' {
535 return false
536 }
537 c := data[i]
538
539 // the whole line must be the char or whitespace
540 n := 0
541 for data[i] != '\n' {
542 switch {
543 case data[i] == c:
544 n++
545 case data[i] != ' ':
546 return false
547 }
548 i++
549 }
550
551 return n >= 3
552}
553
554func (p *parser) isFencedCode(data []byte, syntax **string, oldmarker string) (skip int, marker string) {
555 i, size := 0, 0
556 skip = 0
557
558 // skip up to three spaces
559 for i < 3 && data[i] == ' ' {
560 i++
561 }
562
563 // check for the marker characters: ~ or `
564 if data[i] != '~' && data[i] != '`' {
565 return
566 }
567
568 c := data[i]
569
570 // the whole line must be the same char or whitespace
571 for data[i] == c {
572 size++
573 i++
574 }
575
576 // the marker char must occur at least 3 times
577 if size < 3 {
578 return
579 }
580 marker = string(data[i-size : i])
581
582 // if this is the end marker, it must match the beginning marker
583 if oldmarker != "" && marker != oldmarker {
584 return
585 }
586
587 if syntax != nil {
588 syn := 0
589
590 for data[i] == ' ' {
591 i++
592 }
593
594 syntaxStart := i
595
596 if data[i] == '{' {
597 i++
598 syntaxStart++
599
600 for data[i] != '}' && data[i] != '\n' {
601 syn++
602 i++
603 }
604
605 if data[i] != '}' {
606 return
607 }
608
609 // strip all whitespace at the beginning and the end
610 // of the {} block
611 for syn > 0 && isspace(data[syntaxStart]) {
612 syntaxStart++
613 syn--
614 }
615
616 for syn > 0 && isspace(data[syntaxStart+syn-1]) {
617 syn--
618 }
619
620 i++
621 } else {
622 for !isspace(data[i]) {
623 syn++
624 i++
625 }
626 }
627
628 language := string(data[syntaxStart : syntaxStart+syn])
629 *syntax = &language
630 }
631
632 for data[i] == ' ' {
633 i++
634 }
635 if data[i] != '\n' {
636 return
637 }
638
639 skip = i + 1
640 return
641}
642
643func (p *parser) fencedCode(out *bytes.Buffer, data []byte, doRender bool) int {
644 var lang *string
645 beg, marker := p.isFencedCode(data, &lang, "")
646 if beg == 0 || beg >= len(data) {
647 return 0
648 }
649
650 var work bytes.Buffer
651
652 for {
653 // safe to assume beg < len(data)
654
655 // check for the end of the code block
656 fenceEnd, _ := p.isFencedCode(data[beg:], nil, marker)
657 if fenceEnd != 0 {
658 beg += fenceEnd
659 break
660 }
661
662 // copy the current line
663 end := beg
664 for data[end] != '\n' {
665 end++
666 }
667 end++
668
669 // did we reach the end of the buffer without a closing marker?
670 if end >= len(data) {
671 return 0
672 }
673
674 // verbatim copy to the working buffer
675 if doRender {
676 work.Write(data[beg:end])
677 }
678 beg = end
679 }
680
681 syntax := ""
682 if lang != nil {
683 syntax = *lang
684 }
685
686 if doRender {
687 p.r.BlockCode(out, work.Bytes(), syntax)
688 }
689
690 return beg
691}
692
693func (p *parser) table(out *bytes.Buffer, data []byte) int {
694 var header bytes.Buffer
695 i, columns := p.tableHeader(&header, data)
696 if i == 0 {
697 return 0
698 }
699
700 var body bytes.Buffer
701
702 for i < len(data) {
703 pipes, rowStart := 0, i
704 for ; data[i] != '\n'; i++ {
705 if data[i] == '|' {
706 pipes++
707 }
708 }
709
710 if pipes == 0 {
711 i = rowStart
712 break
713 }
714
715 // include the newline in data sent to tableRow
716 i++
717 p.tableRow(&body, data[rowStart:i], columns, false)
718 }
719
720 p.r.Table(out, header.Bytes(), body.Bytes(), columns)
721
722 return i
723}
724
725// check if the specified position is preceeded by an odd number of backslashes
726func isBackslashEscaped(data []byte, i int) bool {
727 backslashes := 0
728 for i-backslashes-1 >= 0 && data[i-backslashes-1] == '\\' {
729 backslashes++
730 }
731 return backslashes&1 == 1
732}
733
734func (p *parser) tableHeader(out *bytes.Buffer, data []byte) (size int, columns []int) {
735 i := 0
736 colCount := 1
737 for i = 0; data[i] != '\n'; i++ {
738 if data[i] == '|' && !isBackslashEscaped(data, i) {
739 colCount++
740 }
741 }
742
743 // doesn't look like a table header
744 if colCount == 1 {
745 return
746 }
747
748 // include the newline in the data sent to tableRow
749 header := data[:i+1]
750
751 // column count ignores pipes at beginning or end of line
752 if data[0] == '|' {
753 colCount--
754 }
755 if i > 2 && data[i-1] == '|' && !isBackslashEscaped(data, i-1) {
756 colCount--
757 }
758
759 columns = make([]int, colCount)
760
761 // move on to the header underline
762 i++
763 if i >= len(data) {
764 return
765 }
766
767 if data[i] == '|' && !isBackslashEscaped(data, i) {
768 i++
769 }
770 for data[i] == ' ' {
771 i++
772 }
773
774 // each column header is of form: / *:?-+:? *|/ with # dashes + # colons >= 3
775 // and trailing | optional on last column
776 col := 0
777 for data[i] != '\n' {
778 dashes := 0
779
780 if data[i] == ':' {
781 i++
782 columns[col] |= TABLE_ALIGNMENT_LEFT
783 dashes++
784 }
785 for data[i] == '-' {
786 i++
787 dashes++
788 }
789 if data[i] == ':' {
790 i++
791 columns[col] |= TABLE_ALIGNMENT_RIGHT
792 dashes++
793 }
794 for data[i] == ' ' {
795 i++
796 }
797
798 // end of column test is messy
799 switch {
800 case dashes < 3:
801 // not a valid column
802 return
803
804 case data[i] == '|' && !isBackslashEscaped(data, i):
805 // marker found, now skip past trailing whitespace
806 col++
807 i++
808 for data[i] == ' ' {
809 i++
810 }
811
812 // trailing junk found after last column
813 if col >= colCount && data[i] != '\n' {
814 return
815 }
816
817 case (data[i] != '|' || isBackslashEscaped(data, i)) && col+1 < colCount:
818 // something else found where marker was required
819 return
820
821 case data[i] == '\n':
822 // marker is optional for the last column
823 col++
824
825 default:
826 // trailing junk found after last column
827 return
828 }
829 }
830 if col != colCount {
831 return
832 }
833
834 p.tableRow(out, header, columns, true)
835 size = i + 1
836 return
837}
838
839func (p *parser) tableRow(out *bytes.Buffer, data []byte, columns []int, header bool) {
840 i, col := 0, 0
841 var rowWork bytes.Buffer
842
843 if data[i] == '|' && !isBackslashEscaped(data, i) {
844 i++
845 }
846
847 for col = 0; col < len(columns) && i < len(data); col++ {
848 for data[i] == ' ' {
849 i++
850 }
851
852 cellStart := i
853
854 for (data[i] != '|' || isBackslashEscaped(data, i)) && data[i] != '\n' {
855 i++
856 }
857
858 cellEnd := i
859
860 // skip the end-of-cell marker, possibly taking us past end of buffer
861 i++
862
863 for cellEnd > cellStart && data[cellEnd-1] == ' ' {
864 cellEnd--
865 }
866
867 var cellWork bytes.Buffer
868 p.inline(&cellWork, data[cellStart:cellEnd])
869
870 if header {
871 p.r.TableHeaderCell(&rowWork, cellWork.Bytes(), columns[col])
872 } else {
873 p.r.TableCell(&rowWork, cellWork.Bytes(), columns[col])
874 }
875 }
876
877 // pad it out with empty columns to get the right number
878 for ; col < len(columns); col++ {
879 if header {
880 p.r.TableHeaderCell(&rowWork, nil, columns[col])
881 } else {
882 p.r.TableCell(&rowWork, nil, columns[col])
883 }
884 }
885
886 // silently ignore rows with too many cells
887
888 p.r.TableRow(out, rowWork.Bytes())
889}
890
891// returns blockquote prefix length
892func (p *parser) quotePrefix(data []byte) int {
893 i := 0
894 for i < 3 && data[i] == ' ' {
895 i++
896 }
897 if data[i] == '>' {
898 if data[i+1] == ' ' {
899 return i + 2
900 }
901 return i + 1
902 }
903 return 0
904}
905
906// parse a blockquote fragment
907func (p *parser) quote(out *bytes.Buffer, data []byte) int {
908 var raw bytes.Buffer
909 beg, end := 0, 0
910 for beg < len(data) {
911 end = beg
912 for data[end] != '\n' {
913 end++
914 }
915 end++
916
917 if pre := p.quotePrefix(data[beg:]); pre > 0 {
918 // skip the prefix
919 beg += pre
920 } else if p.isEmpty(data[beg:]) > 0 &&
921 (end >= len(data) ||
922 (p.quotePrefix(data[end:]) == 0 && p.isEmpty(data[end:]) == 0)) {
923 // blockquote ends with at least one blank line
924 // followed by something without a blockquote prefix
925 break
926 }
927
928 // this line is part of the blockquote
929 raw.Write(data[beg:end])
930 beg = end
931 }
932
933 var cooked bytes.Buffer
934 p.block(&cooked, raw.Bytes())
935 p.r.BlockQuote(out, cooked.Bytes())
936 return end
937}
938
939// returns prefix length for block code
940func (p *parser) codePrefix(data []byte) int {
941 if data[0] == ' ' && data[1] == ' ' && data[2] == ' ' && data[3] == ' ' {
942 return 4
943 }
944 return 0
945}
946
947func (p *parser) code(out *bytes.Buffer, data []byte) int {
948 var work bytes.Buffer
949
950 i := 0
951 for i < len(data) {
952 beg := i
953 for data[i] != '\n' {
954 i++
955 }
956 i++
957
958 blankline := p.isEmpty(data[beg:i]) > 0
959 if pre := p.codePrefix(data[beg:i]); pre > 0 {
960 beg += pre
961 } else if !blankline {
962 // non-empty, non-prefixed line breaks the pre
963 i = beg
964 break
965 }
966
967 // verbatim copy to the working buffeu
968 if blankline {
969 work.WriteByte('\n')
970 } else {
971 work.Write(data[beg:i])
972 }
973 }
974
975 // trim all the \n off the end of work
976 workbytes := work.Bytes()
977 eol := len(workbytes)
978 for eol > 0 && workbytes[eol-1] == '\n' {
979 eol--
980 }
981 if eol != len(workbytes) {
982 work.Truncate(eol)
983 }
984
985 work.WriteByte('\n')
986
987 p.r.BlockCode(out, work.Bytes(), "")
988
989 return i
990}
991
992// returns unordered list item prefix
993func (p *parser) uliPrefix(data []byte) int {
994 i := 0
995
996 // start with up to 3 spaces
997 for i < 3 && data[i] == ' ' {
998 i++
999 }
1000
1001 // need a *, +, or - followed by a space
1002 if (data[i] != '*' && data[i] != '+' && data[i] != '-') ||
1003 data[i+1] != ' ' {
1004 return 0
1005 }
1006 return i + 2
1007}
1008
1009// returns ordered list item prefix
1010func (p *parser) oliPrefix(data []byte) int {
1011 i := 0
1012
1013 // start with up to 3 spaces
1014 for i < 3 && data[i] == ' ' {
1015 i++
1016 }
1017
1018 // count the digits
1019 start := i
1020 for data[i] >= '0' && data[i] <= '9' {
1021 i++
1022 }
1023
1024 // we need >= 1 digits followed by a dot and a space
1025 if start == i || data[i] != '.' || data[i+1] != ' ' {
1026 return 0
1027 }
1028 return i + 2
1029}
1030
1031// parse ordered or unordered list block
1032func (p *parser) list(out *bytes.Buffer, data []byte, flags int) int {
1033 i := 0
1034 flags |= LIST_ITEM_BEGINNING_OF_LIST
1035 work := func() bool {
1036 for i < len(data) {
1037 skip := p.listItem(out, data[i:], &flags)
1038 i += skip
1039
1040 if skip == 0 || flags&LIST_ITEM_END_OF_LIST != 0 {
1041 break
1042 }
1043 flags &= ^LIST_ITEM_BEGINNING_OF_LIST
1044 }
1045 return true
1046 }
1047
1048 p.r.List(out, work, flags)
1049 return i
1050}
1051
1052// Parse a single list item.
1053// Assumes initial prefix is already removed if this is a sublist.
1054func (p *parser) listItem(out *bytes.Buffer, data []byte, flags *int) int {
1055 // keep track of the indentation of the first line
1056 itemIndent := 0
1057 for itemIndent < 3 && data[itemIndent] == ' ' {
1058 itemIndent++
1059 }
1060
1061 i := p.uliPrefix(data)
1062 if i == 0 {
1063 i = p.oliPrefix(data)
1064 }
1065 if i == 0 {
1066 return 0
1067 }
1068
1069 // skip leading whitespace on first line
1070 for data[i] == ' ' {
1071 i++
1072 }
1073
1074 // find the end of the line
1075 line := i
1076 for data[i-1] != '\n' {
1077 i++
1078 }
1079
1080 // get working buffer
1081 var raw bytes.Buffer
1082
1083 // put the first line into the working buffer
1084 raw.Write(data[line:i])
1085 line = i
1086
1087 // process the following lines
1088 containsBlankLine := false
1089 sublist := 0
1090
1091gatherlines:
1092 for line < len(data) {
1093 i++
1094
1095 // find the end of this line
1096 for data[i-1] != '\n' {
1097 i++
1098 }
1099
1100 // if it is an empty line, guess that it is part of this item
1101 // and move on to the next line
1102 if p.isEmpty(data[line:i]) > 0 {
1103 containsBlankLine = true
1104 line = i
1105 continue
1106 }
1107
1108 // calculate the indentation
1109 indent := 0
1110 for indent < 4 && line+indent < i && data[line+indent] == ' ' {
1111 indent++
1112 }
1113
1114 chunk := data[line+indent : i]
1115
1116 // evaluate how this line fits in
1117 switch {
1118 // is this a nested list item?
1119 case (p.uliPrefix(chunk) > 0 && !p.isHRule(chunk)) ||
1120 p.oliPrefix(chunk) > 0:
1121
1122 if containsBlankLine {
1123 *flags |= LIST_ITEM_CONTAINS_BLOCK
1124 }
1125
1126 // to be a nested list, it must be indented more
1127 // if not, it is the next item in the same list
1128 if indent <= itemIndent {
1129 break gatherlines
1130 }
1131
1132 // is this the first item in the the nested list?
1133 if sublist == 0 {
1134 sublist = raw.Len()
1135 }
1136
1137 // is this a nested prefix header?
1138 case p.isPrefixHeader(chunk):
1139 // if the header is not indented, it is not nested in the list
1140 // and thus ends the list
1141 if containsBlankLine && indent < 4 {
1142 *flags |= LIST_ITEM_END_OF_LIST
1143 break gatherlines
1144 }
1145 *flags |= LIST_ITEM_CONTAINS_BLOCK
1146
1147 // anything following an empty line is only part
1148 // of this item if it is indented 4 spaces
1149 // (regardless of the indentation of the beginning of the item)
1150 case containsBlankLine && indent < 4:
1151 *flags |= LIST_ITEM_END_OF_LIST
1152 break gatherlines
1153
1154 // a blank line means this should be parsed as a block
1155 case containsBlankLine:
1156 raw.WriteByte('\n')
1157 *flags |= LIST_ITEM_CONTAINS_BLOCK
1158 }
1159
1160 // if this line was preceeded by one or more blanks,
1161 // re-introduce the blank into the buffer
1162 if containsBlankLine {
1163 containsBlankLine = false
1164 raw.WriteByte('\n')
1165 }
1166
1167 // add the line into the working buffer without prefix
1168 raw.Write(data[line+indent : i])
1169
1170 line = i
1171 }
1172
1173 rawBytes := raw.Bytes()
1174
1175 // render the contents of the list item
1176 var cooked bytes.Buffer
1177 if *flags&LIST_ITEM_CONTAINS_BLOCK != 0 {
1178 // intermediate render of block li
1179 if sublist > 0 {
1180 p.block(&cooked, rawBytes[:sublist])
1181 p.block(&cooked, rawBytes[sublist:])
1182 } else {
1183 p.block(&cooked, rawBytes)
1184 }
1185 } else {
1186 // intermediate render of inline li
1187 if sublist > 0 {
1188 p.inline(&cooked, rawBytes[:sublist])
1189 p.block(&cooked, rawBytes[sublist:])
1190 } else {
1191 p.inline(&cooked, rawBytes)
1192 }
1193 }
1194
1195 // render the actual list item
1196 cookedBytes := cooked.Bytes()
1197 parsedEnd := len(cookedBytes)
1198
1199 // strip trailing newlines
1200 for parsedEnd > 0 && cookedBytes[parsedEnd-1] == '\n' {
1201 parsedEnd--
1202 }
1203 p.r.ListItem(out, cookedBytes[:parsedEnd], *flags)
1204
1205 return line
1206}
1207
1208// render a single paragraph that has already been parsed out
1209func (p *parser) renderParagraph(out *bytes.Buffer, data []byte) {
1210 if len(data) == 0 {
1211 return
1212 }
1213
1214 // trim leading spaces
1215 beg := 0
1216 for data[beg] == ' ' {
1217 beg++
1218 }
1219
1220 // trim trailing newline
1221 end := len(data) - 1
1222
1223 // trim trailing spaces
1224 for end > beg && data[end-1] == ' ' {
1225 end--
1226 }
1227
1228 work := func() bool {
1229 p.inline(out, data[beg:end])
1230 return true
1231 }
1232 p.r.Paragraph(out, work)
1233}
1234
1235func (p *parser) paragraph(out *bytes.Buffer, data []byte) int {
1236 // prev: index of 1st char of previous line
1237 // line: index of 1st char of current line
1238 // i: index of cursor/end of current line
1239 var prev, line, i int
1240
1241 // keep going until we find something to mark the end of the paragraph
1242 for i < len(data) {
1243 // mark the beginning of the current line
1244 prev = line
1245 current := data[i:]
1246 line = i
1247
1248 // did we find a blank line marking the end of the paragraph?
1249 if n := p.isEmpty(current); n > 0 {
1250 p.renderParagraph(out, data[:i])
1251 return i + n
1252 }
1253
1254 // an underline under some text marks a header, so our paragraph ended on prev line
1255 if i > 0 {
1256 if level := p.isUnderlinedHeader(current); level > 0 {
1257 // render the paragraph
1258 p.renderParagraph(out, data[:prev])
1259
1260 // ignore leading and trailing whitespace
1261 eol := i - 1
1262 for prev < eol && data[prev] == ' ' {
1263 prev++
1264 }
1265 for eol > prev && data[eol-1] == ' ' {
1266 eol--
1267 }
1268
1269 // render the header
1270 // this ugly double closure avoids forcing variables onto the heap
1271 work := func(o *bytes.Buffer, pp *parser, d []byte) func() bool {
1272 return func() bool {
1273 pp.inline(o, d)
1274 return true
1275 }
1276 }(out, p, data[prev:eol])
1277
1278 id := ""
1279 if p.flags&EXTENSION_AUTO_HEADER_IDS != 0 {
1280 id = sanitized_anchor_name.Create(string(data[prev:eol]))
1281 }
1282
1283 p.r.Header(out, work, level, id)
1284
1285 // find the end of the underline
1286 for data[i] != '\n' {
1287 i++
1288 }
1289 return i
1290 }
1291 }
1292
1293 // if the next line starts a block of HTML, then the paragraph ends here
1294 if p.flags&EXTENSION_LAX_HTML_BLOCKS != 0 {
1295 if data[i] == '<' && p.html(out, current, false) > 0 {
1296 // rewind to before the HTML block
1297 p.renderParagraph(out, data[:i])
1298 return i
1299 }
1300 }
1301
1302 // if there's a prefixed header or a horizontal rule after this, paragraph is over
1303 if p.isPrefixHeader(current) || p.isHRule(current) {
1304 p.renderParagraph(out, data[:i])
1305 return i
1306 }
1307
1308 // if there's a list after this, paragraph is over
1309 if p.flags&EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK != 0 {
1310 if p.uliPrefix(current) != 0 ||
1311 p.oliPrefix(current) != 0 ||
1312 p.quotePrefix(current) != 0 ||
1313 p.codePrefix(current) != 0 {
1314 p.renderParagraph(out, data[:i])
1315 return i
1316 }
1317 }
1318
1319 // otherwise, scan to the beginning of the next line
1320 for data[i] != '\n' {
1321 i++
1322 }
1323 i++
1324 }
1325
1326 p.renderParagraph(out, data[:i])
1327 return i
1328}