markdown.go 9.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339
  1. // Copyright 2014 The Gogs Authors. All rights reserved.
  2. // Use of this source code is governed by a MIT-style
  3. // license that can be found in the LICENSE file.
  4. package base
  5. import (
  6. "bytes"
  7. "fmt"
  8. "io"
  9. "net/http"
  10. "path"
  11. "path/filepath"
  12. "regexp"
  13. "strings"
  14. "github.com/Unknwon/com"
  15. "github.com/russross/blackfriday"
  16. "golang.org/x/net/html"
  17. "github.com/gogits/gogs/modules/setting"
  18. )
  19. func isletter(c byte) bool {
  20. return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')
  21. }
  22. func isalnum(c byte) bool {
  23. return (c >= '0' && c <= '9') || isletter(c)
  24. }
  25. var validLinks = [][]byte{[]byte("http://"), []byte("https://"), []byte("ftp://"), []byte("mailto://")}
  26. func isLink(link []byte) bool {
  27. for _, prefix := range validLinks {
  28. if len(link) > len(prefix) && bytes.Equal(bytes.ToLower(link[:len(prefix)]), prefix) && isalnum(link[len(prefix)]) {
  29. return true
  30. }
  31. }
  32. return false
  33. }
  34. func IsMarkdownFile(name string) bool {
  35. name = strings.ToLower(name)
  36. switch filepath.Ext(name) {
  37. case ".md", ".markdown", ".mdown", ".mkd":
  38. return true
  39. }
  40. return false
  41. }
  42. func IsTextFile(data []byte) (string, bool) {
  43. contentType := http.DetectContentType(data)
  44. if strings.Index(contentType, "text/") != -1 {
  45. return contentType, true
  46. }
  47. return contentType, false
  48. }
  49. func IsImageFile(data []byte) (string, bool) {
  50. contentType := http.DetectContentType(data)
  51. if strings.Index(contentType, "image/") != -1 {
  52. return contentType, true
  53. }
  54. return contentType, false
  55. }
  56. // IsReadmeFile returns true if given file name suppose to be a README file.
  57. func IsReadmeFile(name string) bool {
  58. name = strings.ToLower(name)
  59. if len(name) < 6 {
  60. return false
  61. } else if len(name) == 6 {
  62. if name == "readme" {
  63. return true
  64. }
  65. return false
  66. }
  67. if name[:7] == "readme." {
  68. return true
  69. }
  70. return false
  71. }
  72. type CustomRender struct {
  73. blackfriday.Renderer
  74. urlPrefix string
  75. }
  76. func (options *CustomRender) Link(out *bytes.Buffer, link []byte, title []byte, content []byte) {
  77. if len(link) > 0 && !isLink(link) {
  78. if link[0] == '#' {
  79. // link = append([]byte(options.urlPrefix), link...)
  80. } else {
  81. link = []byte(path.Join(options.urlPrefix, string(link)))
  82. }
  83. }
  84. options.Renderer.Link(out, link, title, content)
  85. }
  86. var (
  87. svgSuffix = []byte(".svg")
  88. svgSuffixWithMark = []byte(".svg?")
  89. )
  90. func (options *CustomRender) Image(out *bytes.Buffer, link []byte, title []byte, alt []byte) {
  91. prefix := strings.Replace(options.urlPrefix, "/src/", "/raw/", 1)
  92. if len(link) > 0 {
  93. if isLink(link) {
  94. // External link with .svg suffix usually means CI status.
  95. if bytes.HasSuffix(link, svgSuffix) || bytes.Contains(link, svgSuffixWithMark) {
  96. options.Renderer.Image(out, link, title, alt)
  97. return
  98. }
  99. } else {
  100. if link[0] != '/' {
  101. prefix += "/"
  102. }
  103. link = []byte(prefix + string(link))
  104. }
  105. }
  106. out.WriteString(`<a href="`)
  107. out.Write(link)
  108. out.WriteString(`">`)
  109. options.Renderer.Image(out, link, title, alt)
  110. out.WriteString("</a>")
  111. }
  112. var (
  113. MentionPattern = regexp.MustCompile(`(\s|^)@[0-9a-zA-Z_\.]+`)
  114. commitPattern = regexp.MustCompile(`(\s|^)https?.*commit/[0-9a-zA-Z]+(#+[0-9a-zA-Z-]*)?`)
  115. issueFullPattern = regexp.MustCompile(`(\s|^)https?.*issues/[0-9]+(#+[0-9a-zA-Z-]*)?`)
  116. issueIndexPattern = regexp.MustCompile(`( |^)#[0-9]+\b`)
  117. sha1CurrentPattern = regexp.MustCompile(`\b[0-9a-f]{40}\b`)
  118. )
  119. func cutoutVerbosePrefix(prefix string) string {
  120. count := 0
  121. for i := 0; i < len(prefix); i++ {
  122. if prefix[i] == '/' {
  123. count++
  124. }
  125. if count >= 3 {
  126. return prefix[:i]
  127. }
  128. }
  129. return prefix
  130. }
  131. func RenderIssueIndexPattern(rawBytes []byte, urlPrefix string, metas map[string]string) []byte {
  132. urlPrefix = cutoutVerbosePrefix(urlPrefix)
  133. ms := issueIndexPattern.FindAll(rawBytes, -1)
  134. for _, m := range ms {
  135. var space string
  136. m2 := m
  137. if m2[0] == ' ' {
  138. space = " "
  139. m2 = m2[1:]
  140. }
  141. if metas == nil {
  142. rawBytes = bytes.Replace(rawBytes, m, []byte(fmt.Sprintf(`%s<a href="%s/issues/%s">%s</a>`,
  143. space, urlPrefix, m2[1:], m2)), 1)
  144. } else {
  145. // Support for external issue tracker
  146. metas["index"] = string(m2[1:])
  147. rawBytes = bytes.Replace(rawBytes, m, []byte(fmt.Sprintf(`%s<a href="%s">%s</a>`,
  148. space, com.Expand(metas["format"], metas), m2)), 1)
  149. }
  150. }
  151. return rawBytes
  152. }
  153. func RenderSpecialLink(rawBytes []byte, urlPrefix string, metas map[string]string) []byte {
  154. ms := MentionPattern.FindAll(rawBytes, -1)
  155. for _, m := range ms {
  156. m = bytes.TrimSpace(m)
  157. rawBytes = bytes.Replace(rawBytes, m,
  158. []byte(fmt.Sprintf(`<a href="%s/%s">%s</a>`, setting.AppSubUrl, m[1:], m)), -1)
  159. }
  160. rawBytes = RenderIssueIndexPattern(rawBytes, urlPrefix, metas)
  161. rawBytes = RenderSha1CurrentPattern(rawBytes, urlPrefix)
  162. return rawBytes
  163. }
  164. func RenderSha1CurrentPattern(rawBytes []byte, urlPrefix string) []byte {
  165. ms := sha1CurrentPattern.FindAll(rawBytes, -1)
  166. for _, m := range ms {
  167. rawBytes = bytes.Replace(rawBytes, m, []byte(fmt.Sprintf(
  168. `<a href="%s/commit/%s"><code>%s</code></a>`, urlPrefix, m, ShortSha(string(m)))), -1)
  169. }
  170. return rawBytes
  171. }
  172. func RenderRawMarkdown(body []byte, urlPrefix string) []byte {
  173. htmlFlags := 0
  174. htmlFlags |= blackfriday.HTML_SKIP_STYLE
  175. htmlFlags |= blackfriday.HTML_OMIT_CONTENTS
  176. renderer := &CustomRender{
  177. Renderer: blackfriday.HtmlRenderer(htmlFlags, "", ""),
  178. urlPrefix: urlPrefix,
  179. }
  180. // set up the parser
  181. extensions := 0
  182. extensions |= blackfriday.EXTENSION_NO_INTRA_EMPHASIS
  183. extensions |= blackfriday.EXTENSION_TABLES
  184. extensions |= blackfriday.EXTENSION_FENCED_CODE
  185. extensions |= blackfriday.EXTENSION_AUTOLINK
  186. extensions |= blackfriday.EXTENSION_STRIKETHROUGH
  187. extensions |= blackfriday.EXTENSION_SPACE_HEADERS
  188. extensions |= blackfriday.EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK
  189. if setting.Markdown.EnableHardLineBreak {
  190. extensions |= blackfriday.EXTENSION_HARD_LINE_BREAK
  191. }
  192. body = blackfriday.Markdown(body, renderer, extensions)
  193. return body
  194. }
  195. var (
  196. leftAngleBracket = []byte("</")
  197. rightAngleBracket = []byte(">")
  198. )
  199. var noEndTags = []string{"img", "input", "br", "hr"}
  200. // PreProcessMarkdown renders full links of commits, issues and pulls to shorter version.
  201. func PreProcessMarkdown(rawHTML []byte, urlPrefix string) []byte {
  202. ms := commitPattern.FindAll(rawHTML, -1)
  203. for _, m := range ms {
  204. m = bytes.TrimSpace(m)
  205. i := strings.Index(string(m), "commit/")
  206. j := strings.Index(string(m), "#")
  207. if j == -1 {
  208. j = len(m)
  209. }
  210. rawHTML = bytes.Replace(rawHTML, m, []byte(fmt.Sprintf(
  211. ` <code><a href="%s">%s</a></code>`, m, ShortSha(string(m[i+7:j])))), -1)
  212. }
  213. ms = issueFullPattern.FindAll(rawHTML, -1)
  214. for _, m := range ms {
  215. m = bytes.TrimSpace(m)
  216. i := strings.Index(string(m), "issues/")
  217. j := strings.Index(string(m), "#")
  218. if j == -1 {
  219. j = len(m)
  220. }
  221. rawHTML = bytes.Replace(rawHTML, m, []byte(fmt.Sprintf(
  222. ` <a href="%s">#%s</a>`, m, ShortSha(string(m[i+7:j])))), -1)
  223. }
  224. return rawHTML
  225. }
  226. // PostProcessMarkdown treats different types of HTML differently,
  227. // and only renders special links for plain text blocks.
  228. func PostProcessMarkdown(rawHtml []byte, urlPrefix string, metas map[string]string) []byte {
  229. startTags := make([]string, 0, 5)
  230. var buf bytes.Buffer
  231. tokenizer := html.NewTokenizer(bytes.NewReader(rawHtml))
  232. OUTER_LOOP:
  233. for html.ErrorToken != tokenizer.Next() {
  234. token := tokenizer.Token()
  235. switch token.Type {
  236. case html.TextToken:
  237. buf.Write(RenderSpecialLink([]byte(token.String()), urlPrefix, metas))
  238. case html.StartTagToken:
  239. buf.WriteString(token.String())
  240. tagName := token.Data
  241. // If this is an excluded tag, we skip processing all output until a close tag is encountered.
  242. if strings.EqualFold("a", tagName) || strings.EqualFold("code", tagName) || strings.EqualFold("pre", tagName) {
  243. stackNum := 1
  244. for html.ErrorToken != tokenizer.Next() {
  245. token = tokenizer.Token()
  246. // Copy the token to the output verbatim
  247. buf.WriteString(token.String())
  248. if token.Type == html.StartTagToken {
  249. stackNum++
  250. }
  251. // If this is the close tag to the outer-most, we are done
  252. if token.Type == html.EndTagToken && strings.EqualFold(tagName, token.Data) {
  253. stackNum--
  254. if stackNum == 0 {
  255. break
  256. }
  257. }
  258. }
  259. continue OUTER_LOOP
  260. }
  261. if !com.IsSliceContainsStr(noEndTags, token.Data) {
  262. startTags = append(startTags, token.Data)
  263. }
  264. case html.EndTagToken:
  265. if len(startTags) == 0 {
  266. buf.WriteString(token.String())
  267. break
  268. }
  269. buf.Write(leftAngleBracket)
  270. buf.WriteString(startTags[len(startTags)-1])
  271. buf.Write(rightAngleBracket)
  272. startTags = startTags[:len(startTags)-1]
  273. default:
  274. buf.WriteString(token.String())
  275. }
  276. }
  277. if io.EOF == tokenizer.Err() {
  278. return buf.Bytes()
  279. }
  280. // If we are not at the end of the input, then some other parsing error has occurred,
  281. // so return the input verbatim.
  282. return rawHtml
  283. }
  284. func RenderMarkdown(rawBytes []byte, urlPrefix string, metas map[string]string) []byte {
  285. result := PreProcessMarkdown(rawBytes, urlPrefix)
  286. result = RenderRawMarkdown(result, urlPrefix)
  287. result = PostProcessMarkdown(result, urlPrefix, metas)
  288. result = Sanitizer.SanitizeBytes(result)
  289. return result
  290. }
  291. func RenderMarkdownString(raw, urlPrefix string, metas map[string]string) string {
  292. return string(RenderMarkdown([]byte(raw), urlPrefix, metas))
  293. }