Fix to match GHs algorithm

Related to micromark/micromark-extension-gfm-autolink-literal#5.
Related to micromark/micromark-extension-gfm-autolink-literal#6.

.prettierignore

@@ -1,3 +1,4 @@
 coverage/
+*.html
 *.json
 *.md

.remarkignore

@@ -0,0 +1 @@
+test/

from-markdown.js

@@ -1,3 +1,9 @@
+var ccount = require('ccount')
+var findAndReplace = require('mdast-util-find-and-replace')
+var unicodePunctuation = require('micromark/dist/character/unicode-punctuation')
+var unicodeWhitespace = require('micromark/dist/character/unicode-whitespace')
+
+exports.transforms = [transformGfmAutolinkLiterals]
 exports.enter = {
   literalAutolink: enterLiteralAutolink,
   literalAutolinkEmail: enterLiteralAutolinkValue,
@@ -35,3 +41,117 @@ function exitLiteralAutolinkEmail(token) {
 function exitLiteralAutolink(token) {
   this.exit(token)
 }
+
+function transformGfmAutolinkLiterals(tree) {
+  findAndReplace(
+    tree,
+    [
+      [/(https?:\/\/|www(?=\.))([-.\w]+)([^ \t\r\n]*)/i, findUrl],
+      [/([-.\w+]+)@([-\w]+(?:\.[-\w]+)+)/, findEmail]
+    ],
+    {ignore: ['link', 'linkReference']}
+  )
+}
+
+function findUrl($0, protocol, domain, path, match) {
+  var prefix = ''
+  var parts
+  var result
+
+  // Not an expected previous character.
+  if (!previous(match)) {
+    return false
+  }
+
+  // Treat `www` as part of the domain.
+  if (/^w/i.test(protocol)) {
+    domain = protocol + domain
+    protocol = ''
+    prefix = 'http://'
+  }
+
+  if (!isCorrectDomain(domain)) {
+    return false
+  }
+
+  parts = splitUrl(domain + path)
+
+  if (!parts[0]) return false
+
+  result = {
+    type: 'link',
+    title: null,
+    url: prefix + protocol + parts[0],
+    children: [{type: 'text', value: protocol + parts[0]}]
+  }
+
+  if (parts[1]) {
+    result = [result, {type: 'text', value: parts[1]}]
+  }
+
+  return result
+}
+
+function findEmail($0, atext, label, match) {
+  // Not an expected previous character.
+  if (!previous(match, true) || /[_-]$/.test(label)) {
+    return false
+  }
+
+  return {
+    type: 'link',
+    title: null,
+    url: 'mailto:' + atext + '@' + label,
+    children: [{type: 'text', value: atext + '@' + label}]
+  }
+}
+
+function isCorrectDomain(domain) {
+  var parts = domain.split('.')
+
+  if (
+    parts.length < 2 ||
+    (parts[parts.length - 1] &&
+      (/_/.test(parts[parts.length - 1]) ||
+        !/[a-zA-Z\d]/.test(parts[parts.length - 1]))) ||
+    (parts[parts.length - 2] &&
+      (/_/.test(parts[parts.length - 2]) ||
+        !/[a-zA-Z\d]/.test(parts[parts.length - 2])))
+  ) {
+    return false
+  }
+
+  return true
+}
+
+function splitUrl(url) {
+  var trail = /[!"&'),.:;<>?\]}]+$/.exec(url)
+  var closingParenIndex
+  var openingParens
+  var closingParens
+
+  if (trail) {
+    url = url.slice(0, trail.index)
+    trail = trail[0]
+    closingParenIndex = trail.indexOf(')')
+    openingParens = ccount(url, '(')
+    closingParens = ccount(url, ')')
+
+    while (closingParenIndex !== -1 && openingParens > closingParens) {
+      url += trail.slice(0, closingParenIndex + 1)
+      trail = trail.slice(closingParenIndex + 1)
+      closingParenIndex = trail.indexOf(')')
+      closingParens++
+    }
+  }
+
+  return [url, trail]
+}
+
+function previous(match, email) {
+  var code = match.input.charCodeAt(match.index - 1)
+  return (
+    (code !== code || unicodeWhitespace(code) || unicodePunctuation(code)) &&
+    (!email || code !== 47)
+  )
+}

package.json

@@ -34,11 +34,17 @@
     "index.js",
     "to-markdown.js"
   ],
-  "dependencies": {},
+  "dependencies": {
+    "ccount": "^1.0.0",
+    "mdast-util-find-and-replace": "^1.1.0",
+    "micromark": "^2.11.3"
+  },
   "devDependencies": {
-    "mdast-util-from-markdown": "^0.8.0",
+    "hast-util-to-html": "^7.0.0",
+    "mdast-util-from-markdown": "^0.8.5",
+    "mdast-util-to-hast": "^10.0.0",
     "mdast-util-to-markdown": "^0.6.0",
-    "micromark-extension-gfm-autolink-literal": "^0.5.0",
+    "micromark-extension-gfm-autolink-literal": "^0.5.6",
     "nyc": "^15.0.0",
     "prettier": "^2.0.0",
     "remark-cli": "^9.0.0",
@@ -49,7 +55,7 @@
   "scripts": {
     "format": "remark . -qfo && prettier . -w --loglevel warn && xo --fix",
     "test-api": "node test",
-    "test-coverage": "nyc --reporter lcov tape test.js",
+    "test-coverage": "nyc --reporter lcov tape test/index.js",
     "test": "npm run format && npm run test-coverage"
   },
   "nyc": {
@@ -68,7 +74,13 @@
   },
   "xo": {
     "prettier": true,
-    "esnext": false
+    "esnext": false,
+    "rules": {
+      "max-params": "off",
+      "no-self-compare": "off",
+      "unicorn/prefer-includes": "off",
+      "unicorn/prefer-optional-catch-binding": "off"
+    }
   },
   "remarkConfig": {
     "plugins": [

readme.md

@@ -14,8 +14,8 @@ autolink literals in **[mdast][]**.
 When parsing (`from-markdown`), must be combined with
 [`micromark-extension-gfm-autolink-literal`][extension].
 
-You probably shouldn’t use this package directly, but instead use
-[`remark-gfm`][remark-gfm] with **[remark][]**.
+You might want to use this package through [`remark-gfm`][remark-gfm] with
+**[remark][]**.
 
 ## Install
 

test/algorithm-2.html

@@ -0,0 +1,86 @@
+<p>[https://</p>
+<p>[https://a</p>
+<p>[https://.</p>
+<p>[<a href="https://a">https://a</a>.</p>
+<p>[<a href="https://a">https://a</a>..</p>
+<p>[<a href="https://a..b">https://a..b</a></p>
+<p>[<a href="https://a.b">https://a.b</a></p>
+<p>[<a href="https://a.b">https://a.b</a>.</p>
+<p>[<a href="https://a.b">https://a.b</a>..</p>
+<p>[<a href="https://a.b.c">https://a.b.c</a></p>
+<p>[<a href="https://a.b..c">https://a.b..c</a></p>
+<p>[https://a.b_.c</p>
+<p>[<a href="https://a_.b.c">https://a_.b.c</a></p>
+<p>[https://a_.b_.c</p>
+<p>[<a href="https://a.b%C2%A9">https://a.b©</a></p>
+<p>[http://點看.com</p>
+<hr>
+<p>[<a href="http://a.b/c">http://a.b/c</a> (space)</p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>!</p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>"</p>
+<p>[<a href="http://a.b/c#">http://a.b/c#</a></p>
+<p>[<a href="http://a.b/c$">http://a.b/c$</a></p>
+<p>[<a href="http://a.b/c%25">http://a.b/c%</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>&amp;</p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>'</p>
+<p>[<a href="http://a.b/c(">http://a.b/c(</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>)</p>
+<p>[<a href="http://a.b/c*">http://a.b/c*</a></p>
+<p>[<a href="http://a.b/c+">http://a.b/c+</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>,</p>
+<p>[<a href="http://a.b/c-">http://a.b/c-</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>.</p>
+<p>[<a href="http://a.b/c/">http://a.b/c/</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>:</p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>;</p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>&lt;</p>
+<p>[<a href="http://a.b/c=">http://a.b/c=</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>?</p>
+<p>[<a href="http://a.b/c@">http://a.b/c@</a></p>
+<p>[<a href="http://a.b/c%5B">http://a.b/c[</a></p>
+<p>[<a href="http://a.b/c%5C">http://a.b/c\</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>]</p>
+<p>[<a href="http://a.b/c%5E">http://a.b/c^</a></p>
+<p>[<a href="http://a.b/c_">http://a.b/c_</a></p>
+<p>[<a href="http://a.b/c%60">http://a.b/c`</a></p>
+<p>[<a href="http://a.b/c%7B">http://a.b/c{</a></p>
+<p>[<a href="http://a.b/c%7C">http://a.b/c|</a></p>
+<p>[<a href="http://a.b/c">http://a.b/c</a>}</p>
+<p>[<a href="http://a.b/c~">http://a.b/c~</a></p>
+<hr>
+<p>[<a href="http://www">www</a>.</p>
+<p>[<a href="http://www.a">www.a</a></p>
+<p>[<a href="http://www">www</a>..</p>
+<p>[<a href="http://www.a">www.a</a>.</p>
+<p>[<a href="http://www.a">www.a</a>..</p>
+<p>[<a href="http://www.a..b">www.a..b</a></p>
+<p>[<a href="http://www.a.b">www.a.b</a></p>
+<p>[<a href="http://www.a.b">www.a.b</a>.</p>
+<p>[<a href="http://www.a.b">www.a.b</a>..</p>
+<p>[<a href="http://www.a.b.c">www.a.b.c</a></p>
+<p>[<a href="http://www.a.b..c">www.a.b..c</a></p>
+<p>[www.a.b_.c</p>
+<p>[<a href="http://www.a_.b.c">www.a_.b.c</a></p>
+<p>[www.a_.b_.c</p>
+<p>[<a href="http://www.a.b%C2%A9">www.a.b©</a></p>
+<p>[<a href="http://www.%E9%BB%9E%E7%9C%8B.com">www.點看.com</a></p>
+<hr>
+<p>[<a href="mailto:a@b.c">a@b.c</a></p>
+<p>[<a href="mailto:a@b.ca">a@b.ca</a></p>
+<p>[<a href="mailto:a@b.c">a@b.c</a>.</p>
+<p>[<a href="mailto:a@b.ca">a@b.ca</a>.</p>
+<p>[<a href="mailto:a@b.ca">a@b.ca</a>..</p>
+<p>[<a href="mailto:a@b.ca">a@b.ca</a>..b</p>
+<p>[<a href="mailto:a@b.ca.b">a@b.ca.b</a></p>
+<p>[<a href="mailto:a@b.ca.b">a@b.ca.b</a>.</p>
+<p>[<a href="mailto:a@b.ca.b">a@b.ca.b</a>..</p>
+<p>[<a href="mailto:a@b.ca.b.c">a@b.ca.b.c</a></p>
+<p>[<a href="mailto:a@b.ca.b">a@b.ca.b</a>..c</p>
+<p>[<a href="mailto:a@b.ca.b_.c">a@b.ca.b_.c</a></p>
+<p>[<a href="mailto:a@b.ca_.b.c">a@b.ca_.b.c</a></p>
+<p>[<a href="mailto:a@b.ca_.b_.c">a@b.ca_.b_.c</a></p>
+<p>[<a href="mailto:a@b.ca.b">a@b.ca.b</a>©</p>
+<p>[a@b點看.com</p>
+<p>[點看@b.com</p>