GitHub list parsing

lib/plugin/list.js

@@ -0,0 +1,265 @@
+var isUnorderedItem = function (line) {
+  var match = line.match(/^( *)[-|\*|\+] /)
+  return match
+}
+
+var isOrderedItem = function (line) {
+  var match = line.match(/^( *)([1-9][0-9]*)\. /)
+  return match
+}
+
+var prefixCodefence = function (line) {
+  if (line.length < 3) return 0
+
+    // This is how it looks in redcarpet and I didn't feel like making it better
+  var i = 0
+  if (line[0] === ' ') {
+    i++
+    if (line[1] === ' ') {
+      i++
+      if (line[2] === ' ') {
+        i++
+      }
+    }
+  }
+
+  if (i + 2 >= line.length || !(line[i] === '~' || line[i] === '`')) {
+    return 0
+  }
+
+    // Make sure fence extends to the end of the line
+  var fenceChar = line[i]
+  var n = 0
+  while (i < line.length && line[i] === fenceChar) {
+    n++
+    i++
+  }
+
+  if (n < 3) return 0
+
+  return i
+}
+
+var isCodefence = function (line) {
+  var i = prefixCodefence(line)
+  if (i === 0) return 0
+
+  while (i < line.length && line[i] === ' ') i++
+
+  var languageStart = i
+  var languageLength = 0
+  if (i < line.length && line[i] === '{') {
+    i++
+    languageStart++
+
+    while (i < line.length && line[i] !== '}' && line[i] !== '\n') {
+      i++
+      languageLength++
+    }
+
+    if (i === line.length || line[i] !== '}') return 0
+
+    while (languageLength > 0 && line[languageStart] === ' ') {
+      languageStart++
+      languageLength--
+    }
+
+    while (languageLength > 0 && line[languageStart + languageLength - 1] === ' ') {
+      languageLength++
+    }
+
+    i++
+  } else {
+    while (i < line.length && line[i] !== ' ') {
+      languageLength++
+      i++
+    }
+  }
+
+  while (i < line.length && line[i] !== '\n') {
+    if (line[i] !== ' ') return 0
+    i++
+  }
+
+  return i + 1
+}
+
+var hRule = function (line) {
+  var stripped = line.trim()
+  var c = stripped[0]
+  if (c !== '-' && c !== '*' && c !== '+') return false
+
+  var n = 0
+  for (var i = 0; i < stripped.length; i++) {
+    if (stripped[i] === c) n++
+    else if (stripped[i] !== ' ') return false
+  }
+
+  return n >= 3
+}
+
+module.exports = function (md, opts) {
+  var listItemParser = function (state, startLine, endLine, silent) {
+    var originalIndent = Math.min(state.sCount[state.line], 3)
+    var line = state.src.slice(state.bMarks[state.line], state.eMarks[state.line])
+    console.log(line)
+
+    var lineStart = isOrderedItem(line) || isUnorderedItem(line)
+
+    if (!lineStart) return false
+
+    var ordered = isOrderedItem(line)
+    var markChar
+    if (ordered) {
+      markChar = '.'
+    } else {
+      markChar = line.trim()[0]
+    }
+
+    state.bMarks[state.line] += lineStart[0].length
+
+    var unorderedPos
+    var orderedPos
+    var inEmpty = 0
+    var emptyInside = 0
+    var indent
+    var sublist
+    var inFence = false
+    var lineIndex = state.line + 1
+    var finished = false
+
+    while (lineIndex < endLine) {
+      line = state.src.slice(state.bMarks[lineIndex], state.eMarks[lineIndex])
+      console.log(line)
+
+      if (line.trim() === '') {
+        inEmpty = 1
+        lineIndex++
+        continue
+      }
+
+      indent = Math.min(4, state.sCount[lineIndex])
+
+      if (isCodefence(line) !== 0) inFence = !inFence
+
+      if (!inFence) {
+        unorderedPos = isUnorderedItem(line)
+        orderedPos = isOrderedItem(line)
+      }
+
+      if (inEmpty && ((ordered && unorderedPos) || (!ordered && orderedPos))) {
+        finished = true
+        break
+      }
+
+      // Check whether the line is a horizontal rule because the same characters are used for that
+      if ((unorderedPos && !hRule(line)) || orderedPos) {
+        if (inEmpty) emptyInside = 1
+        // Reached original level in the list. Break out so that the other elements get parsed in listParser
+        if (indent === originalIndent) break
+
+        if (!sublist) {
+          sublist = lineIndex
+        }
+      } else if (inEmpty && indent < 4 && line[0] !== '\t') {
+        finished = true
+        break
+      } else if (inEmpty) {
+        emptyInside = 1
+        console.log('empty inside')
+      }
+
+      // Notice that this algorithm actually removes up to four spaces from the start of all lines it parses. This results in strange behavior such as decreases in indentation increasing the bullet level
+      state.bMarks[lineIndex] += indent
+      state.sCount[lineIndex] -= indent
+      lineIndex++
+    }
+
+    var token
+
+    // An internal empty line determines whether this list item is parsed as a block or inline. This is what causes most of the weird behavior
+    token = state.push('list_item_open', 'li', 1)
+    token.markup = markChar
+    token.map = [startLine, 0]
+    state.line = startLine
+    state.pos = state.bMarks[startLine]
+    var oldMax = state.lineMax
+    var oldPosMax = state.posMax
+    if (emptyInside || finished) {
+      if (sublist && sublist < endLine) {
+        state.lineMax = sublist
+        state.md.block.tokenize(state, startLine, sublist)
+        token = state.push('list_item_close', 'li', -1)
+        token.markup = markChar
+        state.line = sublist
+        state.lineMax = lineIndex
+        state.md.block.tokenize(state, sublist, lineIndex)
+      } else {
+        state.lineMax = lineIndex
+        state.md.block.tokenize(state, startLine, lineIndex)
+        token = state.push('list_item_close', 'li', -1)
+        token.markup = markChar
+      }
+    } else {
+      if (sublist && sublist < endLine) {
+        token = state.push('inline', '', 0)
+        token.content = state.src.slice(state.bMarks[startLine], state.bMarks[sublist]).trim()
+        token.map = [startLine, sublist]
+        token.children = []
+        token = state.push('list_item_close', 'li', -1)
+        token.markup = markChar
+        state.line = sublist
+        state.lineMax = lineIndex
+        state.md.block.tokenize(state, sublist, lineIndex)
+      } else {
+        token = state.push('inline', '', 0)
+        token.content = state.src.slice(state.bMarks[startLine], state.bMarks[lineIndex]).trim()
+        token.map = [startLine, lineIndex]
+        token.children = []
+        token = state.push('list_item_close', 'li', -1)
+        token.markup = markChar
+      }
+    }
+
+    state.lineMax = oldMax
+    state.posMax = oldPosMax
+    state.line = lineIndex
+
+    return !finished
+  }
+
+  var listParser = function (state, startLine, endLine, silent) {
+    var line = state.src.slice(state.bMarks[state.line], state.eMarks[state.line])
+    if (!isOrderedItem(line) && !isUnorderedItem(line)) return false
+
+    var token
+    var keepGoing = true
+
+    var ordered = isOrderedItem(line)
+
+    if (ordered) {
+      token = state.push('ordered_list_open', 'ol', 1)
+      token.markup = '.'
+    } else {
+      token = state.push('bullet_list_open', 'ul', 1)
+      token.markup = line.trim()[0]
+    }
+    token.map = [startLine, 0]
+
+    while (state.line < endLine && keepGoing) {
+      keepGoing = listItemParser(state, state.line, endLine, silent)
+    }
+
+    if (ordered) {
+      token = state.push('ordered_list_close', 'ol', -1)
+      token.markup = '.'
+    } else {
+      token = state.push('bullet_list_close', 'ul', -1)
+      token.markup = line.trim()[0]
+    }
+
+    return true
+  }
+
+  md.block.ruler.before('list', 'ghlist', listParser, ['paragraph', 'reference', 'blockquote'])
+}

lib/render.js

@@ -16,6 +16,7 @@ var youtube = require('./plugin/youtube')
 var cdnImages = require('./plugin/cdn')
 var badges = require('./plugin/badges')
 var packagize = require('./plugin/packagize')
+var ghlist = require('./plugin/list')
 var relaxedLinkRefs = require('./gfm/relaxed-link-reference')
 var githubHeadings = require('./gfm/indented-headings')
 var githubTables = require('./gfm/table')
@@ -81,6 +82,7 @@ render.getParser = function (options) {
     .use(youtube)
     .use(badges)
     .use(packagize, {package: options.package})
+    .use(ghlist)
 
   if (options.highlightSyntax) parser.use(codeWrap)
   if (options.serveImagesWithCDN) parser.use(cdnImages, {package: options.package})