all repos — grayfriday @ 8df342acd510645e8bbc608a209e788edd2667d6

blackfriday fork with a few changes

html.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//
 12// HTML rendering backend
 13//
 14//
 15
 16package blackfriday
 17
 18import (
 19	"bytes"
 20	"fmt"
 21	"regexp"
 22	"strconv"
 23	"strings"
 24)
 25
 26// Html renderer configuration options.
 27const (
 28	HTML_SKIP_HTML                = 1 << iota // skip preformatted HTML blocks
 29	HTML_SKIP_STYLE                           // skip embedded <style> elements
 30	HTML_SKIP_IMAGES                          // skip embedded images
 31	HTML_SKIP_LINKS                           // skip all links
 32	HTML_SANITIZE_OUTPUT                      // strip output of everything that's not known to be safe
 33	HTML_SAFELINK                             // only link to trusted protocols
 34	HTML_NOFOLLOW_LINKS                       // only link with rel="nofollow"
 35	HTML_TOC                                  // generate a table of contents
 36	HTML_OMIT_CONTENTS                        // skip the main contents (for a standalone table of contents)
 37	HTML_COMPLETE_PAGE                        // generate a complete HTML page
 38	HTML_GITHUB_BLOCKCODE                     // use github fenced code rendering rules
 39	HTML_USE_XHTML                            // generate XHTML output instead of HTML
 40	HTML_USE_SMARTYPANTS                      // enable smart punctuation substitutions
 41	HTML_SMARTYPANTS_FRACTIONS                // enable smart fractions (with HTML_USE_SMARTYPANTS)
 42	HTML_SMARTYPANTS_LATEX_DASHES             // enable LaTeX-style dashes (with HTML_USE_SMARTYPANTS)
 43)
 44
 45var (
 46	tags = []string{
 47		"b",
 48		"blockquote",
 49		"code",
 50		"del",
 51		"dd",
 52		"dl",
 53		"dt",
 54		"em",
 55		"h1",
 56		"h2",
 57		"h3",
 58		"h4",
 59		"h5",
 60		"h6",
 61		"i",
 62		"kbd",
 63		"li",
 64		"ol",
 65		"p",
 66		"pre",
 67		"s",
 68		"sup",
 69		"sub",
 70		"strong",
 71		"strike",
 72		"ul",
 73	}
 74	urlRe        = `((https?|ftp):\/\/|\/)[-A-Za-z0-9+&@#\/%?=~_|!:,.;\(\)]+`
 75	tagWhitelist = regexp.MustCompile(`^(<\/?(` + strings.Join(tags, "|") + `)>|<(br|hr)\s?\/?>)$`)
 76	anchorClean  = regexp.MustCompile(`^(<a\shref="` + urlRe + `"(\stitle="[^"<>]+")?\s?>|<\/a>)$`)
 77	imgClean     = regexp.MustCompile(`^(<img\ssrc="` + urlRe + `"(\swidth="\d{1,3}")?(\sheight="\d{1,3}")?(\salt="[^"<>]*")?(\stitle="[^"<>]*")?\s?\/?>)$`)
 78	// TODO: improve this regexp to catch all possible entities:
 79	htmlEntity = regexp.MustCompile(`&[a-z]{2,5};`)
 80)
 81
 82// Html is a type that implements the Renderer interface for HTML output.
 83//
 84// Do not create this directly, instead use the HtmlRenderer function.
 85type Html struct {
 86	flags    int    // HTML_* options
 87	closeTag string // how to end singleton tags: either " />\n" or ">\n"
 88	title    string // document title
 89	css      string // optional css file url (used with HTML_COMPLETE_PAGE)
 90
 91	// table of contents data
 92	tocMarker    int
 93	headerCount  int
 94	currentLevel int
 95	toc          *bytes.Buffer
 96
 97	smartypants *smartypantsRenderer
 98}
 99
100const (
101	xhtmlClose = " />\n"
102	htmlClose  = ">\n"
103)
104
105// HtmlRenderer creates and configures an Html object, which
106// satisfies the Renderer interface.
107//
108// flags is a set of HTML_* options ORed together.
109// title is the title of the document, and css is a URL for the document's
110// stylesheet.
111// title and css are only used when HTML_COMPLETE_PAGE is selected.
112func HtmlRenderer(flags int, title string, css string) Renderer {
113	// configure the rendering engine
114	closeTag := htmlClose
115	if flags&HTML_USE_XHTML != 0 {
116		closeTag = xhtmlClose
117	}
118
119	return &Html{
120		flags:    flags,
121		closeTag: closeTag,
122		title:    title,
123		css:      css,
124
125		headerCount:  0,
126		currentLevel: 0,
127		toc:          new(bytes.Buffer),
128
129		smartypants: smartypants(flags),
130	}
131}
132
133// Using if statements is a bit faster than a switch statement. As the compiler
134// improves, this should be unnecessary this is only worthwhile because
135// attrEscape is the single largest CPU user in normal use.
136// Also tried using map, but that gave a ~3x slowdown.
137func escapeSingleChar(char byte) (string, bool) {
138	if char == '"' {
139		return "&quot;", true
140	}
141	if char == '&' {
142		return "&amp;", true
143	}
144	if char == '<' {
145		return "&lt;", true
146	}
147	if char == '>' {
148		return "&gt;", true
149	}
150	return "", false
151}
152
153func attrEscape(out *bytes.Buffer, src []byte) {
154	org := 0
155	for i, ch := range src {
156		if entity, ok := escapeSingleChar(ch); ok {
157			if i > org {
158				// copy all the normal characters since the last escape
159				out.Write(src[org:i])
160			}
161			org = i + 1
162			out.WriteString(entity)
163		}
164	}
165	if org < len(src) {
166		out.Write(src[org:])
167	}
168}
169
170func entityEscapeWithSkip(out *bytes.Buffer, src []byte, skipRanges [][]int) {
171	end := 0
172	for _, rang := range skipRanges {
173		attrEscape(out, src[end:rang[0]])
174		out.Write(src[rang[0]:rang[1]])
175		end = rang[1]
176	}
177	attrEscape(out, src[end:])
178}
179
180func (options *Html) GetFlags() int {
181	return options.flags
182}
183
184func (options *Html) Header(out *bytes.Buffer, text func() bool, level int, id string) {
185	marker := out.Len()
186	doubleSpace(out)
187
188	if id != "" {
189		out.WriteString(fmt.Sprintf("<h%d id=\"%s\">", level, id))
190	} else if options.flags&HTML_TOC != 0 {
191		// headerCount is incremented in htmlTocHeader
192		out.WriteString(fmt.Sprintf("<h%d id=\"toc_%d\">", level, options.headerCount))
193	} else {
194		out.WriteString(fmt.Sprintf("<h%d>", level))
195	}
196
197	tocMarker := out.Len()
198	if !text() {
199		out.Truncate(marker)
200		return
201	}
202
203	// are we building a table of contents?
204	if options.flags&HTML_TOC != 0 {
205		options.TocHeader(out.Bytes()[tocMarker:], level)
206	}
207
208	out.WriteString(fmt.Sprintf("</h%d>\n", level))
209}
210
211func (options *Html) BlockHtml(out *bytes.Buffer, text []byte) {
212	if options.flags&HTML_SKIP_HTML != 0 {
213		return
214	}
215
216	doubleSpace(out)
217	out.Write(text)
218	out.WriteByte('\n')
219}
220
221func (options *Html) HRule(out *bytes.Buffer) {
222	doubleSpace(out)
223	out.WriteString("<hr")
224	out.WriteString(options.closeTag)
225}
226
227func (options *Html) BlockCode(out *bytes.Buffer, text []byte, lang string) {
228	if options.flags&HTML_GITHUB_BLOCKCODE != 0 {
229		options.BlockCodeGithub(out, text, lang)
230	} else {
231		options.BlockCodeNormal(out, text, lang)
232	}
233}
234
235func (options *Html) BlockCodeNormal(out *bytes.Buffer, text []byte, lang string) {
236	doubleSpace(out)
237
238	// parse out the language names/classes
239	count := 0
240	for _, elt := range strings.Fields(lang) {
241		if elt[0] == '.' {
242			elt = elt[1:]
243		}
244		if len(elt) == 0 {
245			continue
246		}
247		if count == 0 {
248			out.WriteString("<pre><code class=\"")
249		} else {
250			out.WriteByte(' ')
251		}
252		attrEscape(out, []byte(elt))
253		count++
254	}
255
256	if count == 0 {
257		out.WriteString("<pre><code>")
258	} else {
259		out.WriteString("\">")
260	}
261
262	attrEscape(out, text)
263	out.WriteString("</code></pre>\n")
264}
265
266// GitHub style code block:
267//
268//              <pre lang="LANG"><code>
269//              ...
270//              </code></pre>
271//
272// Unlike other parsers, we store the language identifier in the <pre>,
273// and don't let the user generate custom classes.
274//
275// The language identifier in the <pre> block gets postprocessed and all
276// the code inside gets syntax highlighted with Pygments. This is much safer
277// than letting the user specify a CSS class for highlighting.
278//
279// Note that we only generate HTML for the first specifier.
280// E.g.
281//              ~~~~ {.python .numbered}        =>      <pre lang="python"><code>
282func (options *Html) BlockCodeGithub(out *bytes.Buffer, text []byte, lang string) {
283	doubleSpace(out)
284
285	// parse out the language name
286	count := 0
287	for _, elt := range strings.Fields(lang) {
288		if elt[0] == '.' {
289			elt = elt[1:]
290		}
291		if len(elt) == 0 {
292			continue
293		}
294		out.WriteString("<pre lang=\"")
295		attrEscape(out, []byte(elt))
296		out.WriteString("\"><code>")
297		count++
298		break
299	}
300
301	if count == 0 {
302		out.WriteString("<pre><code>")
303	}
304
305	attrEscape(out, text)
306	out.WriteString("</code></pre>\n")
307}
308
309func (options *Html) BlockQuote(out *bytes.Buffer, text []byte) {
310	doubleSpace(out)
311	out.WriteString("<blockquote>\n")
312	out.Write(text)
313	out.WriteString("</blockquote>\n")
314}
315
316func (options *Html) Table(out *bytes.Buffer, header []byte, body []byte, columnData []int) {
317	doubleSpace(out)
318	out.WriteString("<table>\n<thead>\n")
319	out.Write(header)
320	out.WriteString("</thead>\n\n<tbody>\n")
321	out.Write(body)
322	out.WriteString("</tbody>\n</table>\n")
323}
324
325func (options *Html) TableRow(out *bytes.Buffer, text []byte) {
326	doubleSpace(out)
327	out.WriteString("<tr>\n")
328	out.Write(text)
329	out.WriteString("\n</tr>\n")
330}
331
332func (options *Html) TableHeaderCell(out *bytes.Buffer, text []byte, align int) {
333	doubleSpace(out)
334	switch align {
335	case TABLE_ALIGNMENT_LEFT:
336		out.WriteString("<th align=\"left\">")
337	case TABLE_ALIGNMENT_RIGHT:
338		out.WriteString("<th align=\"right\">")
339	case TABLE_ALIGNMENT_CENTER:
340		out.WriteString("<th align=\"center\">")
341	default:
342		out.WriteString("<th>")
343	}
344
345	out.Write(text)
346	out.WriteString("</th>")
347}
348
349func (options *Html) TableCell(out *bytes.Buffer, text []byte, align int) {
350	doubleSpace(out)
351	switch align {
352	case TABLE_ALIGNMENT_LEFT:
353		out.WriteString("<td align=\"left\">")
354	case TABLE_ALIGNMENT_RIGHT:
355		out.WriteString("<td align=\"right\">")
356	case TABLE_ALIGNMENT_CENTER:
357		out.WriteString("<td align=\"center\">")
358	default:
359		out.WriteString("<td>")
360	}
361
362	out.Write(text)
363	out.WriteString("</td>")
364}
365
366func (options *Html) Footnotes(out *bytes.Buffer, text func() bool) {
367	out.WriteString("<div class=\"footnotes\">\n")
368	options.HRule(out)
369	options.List(out, text, LIST_TYPE_ORDERED)
370	out.WriteString("</div>\n")
371}
372
373func (options *Html) FootnoteItem(out *bytes.Buffer, name, text []byte, flags int) {
374	if flags&LIST_ITEM_CONTAINS_BLOCK != 0 || flags&LIST_ITEM_BEGINNING_OF_LIST != 0 {
375		doubleSpace(out)
376	}
377	out.WriteString(`<li id="fn:`)
378	out.Write(slugify(name))
379	out.WriteString(`">`)
380	out.Write(text)
381	out.WriteString("</li>\n")
382}
383
384func (options *Html) List(out *bytes.Buffer, text func() bool, flags int) {
385	marker := out.Len()
386	doubleSpace(out)
387
388	if flags&LIST_TYPE_ORDERED != 0 {
389		out.WriteString("<ol>")
390	} else {
391		out.WriteString("<ul>")
392	}
393	if !text() {
394		out.Truncate(marker)
395		return
396	}
397	if flags&LIST_TYPE_ORDERED != 0 {
398		out.WriteString("</ol>\n")
399	} else {
400		out.WriteString("</ul>\n")
401	}
402}
403
404func (options *Html) ListItem(out *bytes.Buffer, text []byte, flags int) {
405	if flags&LIST_ITEM_CONTAINS_BLOCK != 0 || flags&LIST_ITEM_BEGINNING_OF_LIST != 0 {
406		doubleSpace(out)
407	}
408	out.WriteString("<li>")
409	out.Write(text)
410	out.WriteString("</li>\n")
411}
412
413func (options *Html) Paragraph(out *bytes.Buffer, text func() bool) {
414	marker := out.Len()
415	doubleSpace(out)
416
417	out.WriteString("<p>")
418	if !text() {
419		out.Truncate(marker)
420		return
421	}
422	out.WriteString("</p>\n")
423}
424
425func (options *Html) AutoLink(out *bytes.Buffer, link []byte, kind int) {
426	skipRanges := htmlEntity.FindAllIndex(link, -1)
427	if options.flags&HTML_SAFELINK != 0 && !isSafeLink(link) && kind != LINK_TYPE_EMAIL {
428		// mark it but don't link it if it is not a safe link: no smartypants
429		out.WriteString("<tt>")
430		entityEscapeWithSkip(out, link, skipRanges)
431		out.WriteString("</tt>")
432		return
433	}
434
435	out.WriteString("<a href=\"")
436	if kind == LINK_TYPE_EMAIL {
437		out.WriteString("mailto:")
438	}
439	entityEscapeWithSkip(out, link, skipRanges)
440	out.WriteString("\">")
441
442	// Pretty print: if we get an email address as
443	// an actual URI, e.g. `mailto:foo@bar.com`, we don't
444	// want to print the `mailto:` prefix
445	switch {
446	case bytes.HasPrefix(link, []byte("mailto://")):
447		attrEscape(out, link[len("mailto://"):])
448	case bytes.HasPrefix(link, []byte("mailto:")):
449		attrEscape(out, link[len("mailto:"):])
450	default:
451		entityEscapeWithSkip(out, link, skipRanges)
452	}
453
454	out.WriteString("</a>")
455}
456
457func (options *Html) CodeSpan(out *bytes.Buffer, text []byte) {
458	out.WriteString("<code>")
459	attrEscape(out, text)
460	out.WriteString("</code>")
461}
462
463func (options *Html) DoubleEmphasis(out *bytes.Buffer, text []byte) {
464	out.WriteString("<strong>")
465	out.Write(text)
466	out.WriteString("</strong>")
467}
468
469func (options *Html) Emphasis(out *bytes.Buffer, text []byte) {
470	if len(text) == 0 {
471		return
472	}
473	out.WriteString("<em>")
474	out.Write(text)
475	out.WriteString("</em>")
476}
477
478func (options *Html) Image(out *bytes.Buffer, link []byte, title []byte, alt []byte) {
479	if options.flags&HTML_SKIP_IMAGES != 0 {
480		return
481	}
482
483	out.WriteString("<img src=\"")
484	attrEscape(out, link)
485	out.WriteString("\" alt=\"")
486	if len(alt) > 0 {
487		attrEscape(out, alt)
488	}
489	if len(title) > 0 {
490		out.WriteString("\" title=\"")
491		attrEscape(out, title)
492	}
493
494	out.WriteByte('"')
495	out.WriteString(options.closeTag)
496	return
497}
498
499func (options *Html) LineBreak(out *bytes.Buffer) {
500	out.WriteString("<br")
501	out.WriteString(options.closeTag)
502}
503
504func (options *Html) Link(out *bytes.Buffer, link []byte, title []byte, content []byte) {
505	if options.flags&HTML_SKIP_LINKS != 0 {
506		// write the link text out but don't link it, just mark it with typewriter font
507		out.WriteString("<tt>")
508		attrEscape(out, content)
509		out.WriteString("</tt>")
510		return
511	}
512
513	if options.flags&HTML_SAFELINK != 0 && !isSafeLink(link) {
514		// write the link text out but don't link it, just mark it with typewriter font
515		out.WriteString("<tt>")
516		attrEscape(out, content)
517		out.WriteString("</tt>")
518		return
519	}
520
521	out.WriteString("<a href=\"")
522	attrEscape(out, link)
523	if len(title) > 0 {
524		out.WriteString("\" title=\"")
525		attrEscape(out, title)
526	}
527	if options.flags&HTML_NOFOLLOW_LINKS != 0 {
528		out.WriteString("\" rel=\"nofollow")
529	}
530	out.WriteString("\">")
531	out.Write(content)
532	out.WriteString("</a>")
533	return
534}
535
536func (options *Html) RawHtmlTag(out *bytes.Buffer, text []byte) {
537	if options.flags&HTML_SKIP_HTML != 0 {
538		return
539	}
540	if options.flags&HTML_SKIP_STYLE != 0 && isHtmlTag(text, "style") {
541		return
542	}
543	if options.flags&HTML_SKIP_LINKS != 0 && isHtmlTag(text, "a") {
544		return
545	}
546	if options.flags&HTML_SKIP_IMAGES != 0 && isHtmlTag(text, "img") {
547		return
548	}
549	out.Write(text)
550}
551
552func (options *Html) TripleEmphasis(out *bytes.Buffer, text []byte) {
553	out.WriteString("<strong><em>")
554	out.Write(text)
555	out.WriteString("</em></strong>")
556}
557
558func (options *Html) StrikeThrough(out *bytes.Buffer, text []byte) {
559	out.WriteString("<del>")
560	out.Write(text)
561	out.WriteString("</del>")
562}
563
564func (options *Html) FootnoteRef(out *bytes.Buffer, ref []byte, id int) {
565	slug := slugify(ref)
566	out.WriteString(`<sup class="footnote-ref" id="fnref:`)
567	out.Write(slug)
568	out.WriteString(`"><a rel="footnote" href="#fn:`)
569	out.Write(slug)
570	out.WriteString(`">`)
571	out.WriteString(strconv.Itoa(id))
572	out.WriteString(`</a></sup>`)
573}
574
575func (options *Html) Entity(out *bytes.Buffer, entity []byte) {
576	out.Write(entity)
577}
578
579func (options *Html) NormalText(out *bytes.Buffer, text []byte) {
580	if options.flags&HTML_USE_SMARTYPANTS != 0 {
581		options.Smartypants(out, text)
582	} else {
583		attrEscape(out, text)
584	}
585}
586
587func (options *Html) Smartypants(out *bytes.Buffer, text []byte) {
588	smrt := smartypantsData{false, false}
589
590	// first do normal entity escaping
591	var escaped bytes.Buffer
592	attrEscape(&escaped, text)
593	text = escaped.Bytes()
594
595	mark := 0
596	for i := 0; i < len(text); i++ {
597		if action := options.smartypants[text[i]]; action != nil {
598			if i > mark {
599				out.Write(text[mark:i])
600			}
601
602			previousChar := byte(0)
603			if i > 0 {
604				previousChar = text[i-1]
605			}
606			i += action(out, &smrt, previousChar, text[i:])
607			mark = i + 1
608		}
609	}
610
611	if mark < len(text) {
612		out.Write(text[mark:])
613	}
614}
615
616func (options *Html) DocumentHeader(out *bytes.Buffer) {
617	if options.flags&HTML_COMPLETE_PAGE == 0 {
618		return
619	}
620
621	ending := ""
622	if options.flags&HTML_USE_XHTML != 0 {
623		out.WriteString("<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.0 Transitional//EN\" ")
624		out.WriteString("\"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd\">\n")
625		out.WriteString("<html xmlns=\"http://www.w3.org/1999/xhtml\">\n")
626		ending = " /"
627	} else {
628		out.WriteString("<!DOCTYPE html>\n")
629		out.WriteString("<html>\n")
630	}
631	out.WriteString("<head>\n")
632	out.WriteString("  <title>")
633	options.NormalText(out, []byte(options.title))
634	out.WriteString("</title>\n")
635	out.WriteString("  <meta name=\"GENERATOR\" content=\"Blackfriday Markdown Processor v")
636	out.WriteString(VERSION)
637	out.WriteString("\"")
638	out.WriteString(ending)
639	out.WriteString(">\n")
640	out.WriteString("  <meta charset=\"utf-8\"")
641	out.WriteString(ending)
642	out.WriteString(">\n")
643	if options.css != "" {
644		out.WriteString("  <link rel=\"stylesheet\" type=\"text/css\" href=\"")
645		attrEscape(out, []byte(options.css))
646		out.WriteString("\"")
647		out.WriteString(ending)
648		out.WriteString(">\n")
649	}
650	out.WriteString("</head>\n")
651	out.WriteString("<body>\n")
652
653	options.tocMarker = out.Len()
654}
655
656func (options *Html) DocumentFooter(out *bytes.Buffer) {
657	// finalize and insert the table of contents
658	if options.flags&HTML_TOC != 0 {
659		options.TocFinalize()
660
661		// now we have to insert the table of contents into the document
662		var temp bytes.Buffer
663
664		// start by making a copy of everything after the document header
665		temp.Write(out.Bytes()[options.tocMarker:])
666
667		// now clear the copied material from the main output buffer
668		out.Truncate(options.tocMarker)
669
670		// corner case spacing issue
671		if options.flags&HTML_COMPLETE_PAGE != 0 {
672			out.WriteByte('\n')
673		}
674
675		// insert the table of contents
676		out.WriteString("<nav>\n")
677		out.Write(options.toc.Bytes())
678		out.WriteString("</nav>\n")
679
680		// corner case spacing issue
681		if options.flags&HTML_COMPLETE_PAGE == 0 && options.flags&HTML_OMIT_CONTENTS == 0 {
682			out.WriteByte('\n')
683		}
684
685		// write out everything that came after it
686		if options.flags&HTML_OMIT_CONTENTS == 0 {
687			out.Write(temp.Bytes())
688		}
689	}
690
691	if options.flags&HTML_COMPLETE_PAGE != 0 {
692		out.WriteString("\n</body>\n")
693		out.WriteString("</html>\n")
694	}
695
696}
697
698func (options *Html) TocHeader(text []byte, level int) {
699	for level > options.currentLevel {
700		switch {
701		case bytes.HasSuffix(options.toc.Bytes(), []byte("</li>\n")):
702			// this sublist can nest underneath a header
703			size := options.toc.Len()
704			options.toc.Truncate(size - len("</li>\n"))
705
706		case options.currentLevel > 0:
707			options.toc.WriteString("<li>")
708		}
709		if options.toc.Len() > 0 {
710			options.toc.WriteByte('\n')
711		}
712		options.toc.WriteString("<ul>\n")
713		options.currentLevel++
714	}
715
716	for level < options.currentLevel {
717		options.toc.WriteString("</ul>")
718		if options.currentLevel > 1 {
719			options.toc.WriteString("</li>\n")
720		}
721		options.currentLevel--
722	}
723
724	options.toc.WriteString("<li><a href=\"#toc_")
725	options.toc.WriteString(strconv.Itoa(options.headerCount))
726	options.toc.WriteString("\">")
727	options.headerCount++
728
729	options.toc.Write(text)
730
731	options.toc.WriteString("</a></li>\n")
732}
733
734func (options *Html) TocFinalize() {
735	for options.currentLevel > 1 {
736		options.toc.WriteString("</ul></li>\n")
737		options.currentLevel--
738	}
739
740	if options.currentLevel > 0 {
741		options.toc.WriteString("</ul>\n")
742	}
743}
744
745func isHtmlTag(tag []byte, tagname string) bool {
746	found, _ := findHtmlTagPos(tag, tagname)
747	return found
748}
749
750// Look for a character, but ignore it when it's in any kind of quotes, it
751// might be JavaScript
752func skipUntilCharIgnoreQuotes(html []byte, start int, char byte) int {
753	inSingleQuote := false
754	inDoubleQuote := false
755	inGraveQuote := false
756	i := start
757	for i < len(html) {
758		switch {
759		case html[i] == char && !inSingleQuote && !inDoubleQuote && !inGraveQuote:
760			return i
761		case html[i] == '\'':
762			inSingleQuote = !inSingleQuote
763		case html[i] == '"':
764			inDoubleQuote = !inDoubleQuote
765		case html[i] == '`':
766			inGraveQuote = !inGraveQuote
767		}
768		i++
769	}
770	return start
771}
772
773func findHtmlTagPos(tag []byte, tagname string) (bool, int) {
774	i := 0
775	if i < len(tag) && tag[0] != '<' {
776		return false, -1
777	}
778	i++
779	i = skipSpace(tag, i)
780
781	if i < len(tag) && tag[i] == '/' {
782		i++
783	}
784
785	i = skipSpace(tag, i)
786	j := 0
787	for ; i < len(tag); i, j = i+1, j+1 {
788		if j >= len(tagname) {
789			break
790		}
791
792		if strings.ToLower(string(tag[i]))[0] != tagname[j] {
793			return false, -1
794		}
795	}
796
797	if i == len(tag) {
798		return false, -1
799	}
800
801	rightAngle := skipUntilCharIgnoreQuotes(tag, i, '>')
802	if rightAngle > i {
803		return true, rightAngle
804	}
805
806	return false, -1
807}
808
809func sanitizeHtml(html []byte) []byte {
810	var result []byte
811	for string(html) != "" {
812		skip, tag, rest := findHtmlTag(html)
813		html = rest
814		result = append(result, skip...)
815		result = append(result, sanitizeTag(tag)...)
816	}
817	return append(result, []byte("\n")...)
818}
819
820func sanitizeTag(tag []byte) []byte {
821	if tagWhitelist.Match(tag) || anchorClean.Match(tag) || imgClean.Match(tag) {
822		return tag
823	}
824	return []byte("")
825}
826
827func skipUntilChar(text []byte, start int, char byte) int {
828	i := start
829	for i < len(text) && text[i] != char {
830		i++
831	}
832	return i
833}
834
835func findHtmlTag(html []byte) (skip, tag, rest []byte) {
836	start := skipUntilChar(html, 0, '<')
837	rightAngle := skipUntilCharIgnoreQuotes(html, start, '>')
838	if rightAngle > start {
839		skip = html[0:start]
840		tag = html[start : rightAngle+1]
841		rest = html[rightAngle+1:]
842		return
843	}
844
845	return []byte(""), []byte(""), []byte("")
846}
847
848func skipSpace(tag []byte, i int) int {
849	for i < len(tag) && isspace(tag[i]) {
850		i++
851	}
852	return i
853}
854
855func doubleSpace(out *bytes.Buffer) {
856	if out.Len() > 0 {
857		out.WriteByte('\n')
858	}
859}