parse: add support for emails without protocol

Closes GH-475.
Closes GH-479.

packages/remark-parse/lib/locate/email.js

@@ -0,0 +1,51 @@
+'use strict'
+
+var decimal = require('is-decimal')
+var alphabetical = require('is-alphabetical')
+
+var plusSign = 43 // '+'
+var dash = 45 // '-'
+var dot = 46 // '.'
+var underscore = 95 // '_'
+
+module.exports = locate
+
+// See: <https://github.github.com/gfm/#extended-email-autolink>
+function locate(value, fromIndex) {
+  var self = this
+  var at
+  var position
+
+  if (!this.options.gfm) {
+    return -1
+  }
+
+  at = value.indexOf('@', fromIndex)
+
+  if (at === -1) {
+    return -1
+  }
+
+  position = at
+
+  if (position === fromIndex || !isGfmAtext(value.charCodeAt(position - 1))) {
+    return locate.call(self, value, at + 1)
+  }
+
+  while (position > fromIndex && isGfmAtext(value.charCodeAt(position - 1))) {
+    position--
+  }
+
+  return position
+}
+
+function isGfmAtext(code) {
+  return (
+    decimal(code) ||
+    alphabetical(code) ||
+    code === plusSign ||
+    code === dash ||
+    code === dot ||
+    code === underscore
+  )
+}

packages/remark-parse/lib/parser.js

@@ -116,6 +116,7 @@ proto.inlineTokenizers = {
   escape: require('./tokenize/escape'),
   autoLink: require('./tokenize/auto-link'),
   url: require('./tokenize/url'),
+  email: require('./tokenize/email'),
   html: require('./tokenize/html-inline'),
   link: require('./tokenize/link'),
   reference: require('./tokenize/reference'),

packages/remark-parse/lib/tokenize/email.js

@@ -0,0 +1,114 @@
+'use strict'
+
+var decode = require('parse-entities')
+var decimal = require('is-decimal')
+var alphabetical = require('is-alphabetical')
+var locate = require('../locate/email')
+
+module.exports = email
+email.locator = locate
+email.notInLink = true
+
+var plusSign = 43 // '+'
+var dash = 45 // '-'
+var dot = 46 // '.'
+var atSign = 64 // '@'
+var underscore = 95 // '_'
+
+function email(eat, value, silent) {
+  var self = this
+  var gfm = self.options.gfm
+  var tokenizers = self.inlineTokenizers
+  var index = 0
+  var length = value.length
+  var firstDot = -1
+  var code
+  var content
+  var children
+  var exit
+
+  if (!gfm) {
+    return
+  }
+
+  code = value.charCodeAt(index)
+
+  while (
+    decimal(code) ||
+    alphabetical(code) ||
+    code === plusSign ||
+    code === dash ||
+    code === dot ||
+    code === underscore
+  ) {
+    code = value.charCodeAt(++index)
+  }
+
+  if (index === 0) {
+    return
+  }
+
+  if (code !== atSign) {
+    return
+  }
+
+  index++
+
+  while (index < length) {
+    code = value.charCodeAt(index)
+
+    if (
+      decimal(code) ||
+      alphabetical(code) ||
+      code === dash ||
+      code === dot ||
+      code === underscore
+    ) {
+      index++
+
+      if (firstDot === -1 && code === dot) {
+        firstDot = index
+      }
+
+      continue
+    }
+
+    break
+  }
+
+  if (
+    firstDot === -1 ||
+    firstDot === index ||
+    code === dash ||
+    code === underscore
+  ) {
+    return
+  }
+
+  if (code === dot) {
+    index--
+  }
+
+  content = value.slice(0, index)
+
+  /* istanbul ignore if - never used (yet) */
+  if (silent) {
+    return true
+  }
+
+  exit = self.enterLink()
+
+  // Temporarily remove all tokenizers except text in url.
+  self.inlineTokenizers = {text: tokenizers.text}
+  children = self.tokenizeInline(content, eat.now())
+  self.inlineTokenizers = tokenizers
+
+  exit()
+
+  return eat(content)({
+    type: 'link',
+    title: null,
+    url: 'mailto:' + decode(content, {nonTerminated: false}),
+    children: children
+  })
+}

packages/remark-parse/readme.md

@@ -279,6 +279,7 @@ Precedence of default inline methods is as follows:
 *   `escape`
 *   `autoLink`
 *   `url`
+*   `email`
 *   `html`
 *   `link`
 *   `reference`

test/fixtures/input/auto-link-url.text

@@ -5,3 +5,5 @@ Also, subdomain should be a part of the link (http://foo.example.com/(hello[worl
 So should this: mailto:foo@bar.com.
 
 And even with underscore http://domain.org/this_is_good.
+
+All links should work http://a.b, https://c.d, http://e.f, https://g.h.

test/fixtures/input/literal-email.text

@@ -0,0 +1,18 @@
+example1 (1234567@example.com)
+example1 (mailto:1234567@example.com)
+
+Lorem foo@bar.baz ipsum.
+
+alpha@bravo+charlie.delta isn’t valid, but echo+foxtrot@golf.hotel is.
+
+Valid: a.b-c_d@a.b
+
+Valid, but the dot is not part of the email: a.b-c_d@a.b.
+
+Not valid: a.b-c_d@a.b-
+
+Not valid: a.b-c_d@a.b_
+
+Not valid: alpha@bravo.
+
+<foo@example.com