From 7b381da5bceee3c36be079e27a2bad7b3cc412c2 Mon Sep 17 00:00:00 2001
From: Titus Wormer <tituswormer@gmail.com>
Date: Sat, 30 Oct 2021 18:00:17 +0200
Subject: [PATCH] Fix to not escape some character in certain spans

Related to remarkjs/remark#885.
---
 lib/unsafe.js | 51 ++++++++++++++++++++++++++++++++++++++++-----------
 test/index.js | 45 +++++++++++++++++++++++++++++++++++++++------
 2 files changed, 79 insertions(+), 17 deletions(-)

diff --git a/lib/unsafe.js b/lib/unsafe.js
index 185f91a..4555d1a 100644
--- a/lib/unsafe.js
+++ b/lib/unsafe.js
@@ -2,6 +2,21 @@
  * @typedef {import('./types.js').Unsafe} Unsafe
  */
 
+/**
+ * List of constructs that occur in phrasing (paragraphs, headings), but cannot
+ * contain things like attention (emphasis, strong), images, or links.
+ * So they sort of cancel each other out.
+ * Note: could use a better name.
+ */
+const fullPhrasingSpans = [
+  'autolink',
+  'destinationLiteral',
+  'destinationRaw',
+  'reference',
+  'titleQuote',
+  'titleApostrophe'
+]
+
 /** @type {Array.<Unsafe>} */
 export const unsafe = [
   {character: '\t', after: '[\\r\\n]', inConstruct: 'phrasing'},
@@ -40,7 +55,12 @@ export const unsafe = [
   },
   // An exclamation mark can start an image, if it is followed by a link or
   // a link reference.
-  {character: '!', after: '\\[', inConstruct: 'phrasing'},
+  {
+    character: '!',
+    after: '\\[',
+    inConstruct: 'phrasing',
+    notInConstruct: fullPhrasingSpans
+  },
   // A quote can break out of a title.
   {character: '"', inConstruct: 'titleQuote'},
   // A number sign could start an ATX heading if it starts a line.
@@ -53,14 +73,20 @@ export const unsafe = [
   {character: "'", inConstruct: 'titleApostrophe'},
   // A left paren could break out of a destination raw.
   {character: '(', inConstruct: 'destinationRaw'},
-  {before: '\\]', character: '(', inConstruct: 'phrasing'},
+  // A left paren followed by `]` could make something into a link or image.
+  {
+    before: '\\]',
+    character: '(',
+    inConstruct: 'phrasing',
+    notInConstruct: fullPhrasingSpans
+  },
   // A right paren could start a list item or break out of a destination
   // raw.
   {atBreak: true, before: '\\d+', character: ')'},
   {character: ')', inConstruct: 'destinationRaw'},
   // An asterisk can start thematic breaks, list items, emphasis, strong.
   {atBreak: true, character: '*'},
-  {character: '*', inConstruct: 'phrasing'},
+  {character: '*', inConstruct: 'phrasing', notInConstruct: fullPhrasingSpans},
   // A plus sign could start a list item.
   {atBreak: true, character: '+'},
   // A dash can start thematic breaks, list items, and setext heading
@@ -75,7 +101,12 @@ export const unsafe = [
   // An autolink also starts with a letter.
   // Finally, it could break out of a destination literal.
   {atBreak: true, character: '<', after: '[!/?A-Za-z]'},
-  {character: '<', after: '[!/?A-Za-z]', inConstruct: 'phrasing'},
+  {
+    character: '<',
+    after: '[!/?A-Za-z]',
+    inConstruct: 'phrasing',
+    notInConstruct: fullPhrasingSpans
+  },
   {character: '<', inConstruct: 'destinationLiteral'},
   // An equals to can start setext heading underlines.
   {atBreak: true, character: '='},
@@ -86,7 +117,8 @@ export const unsafe = [
   // Question mark and at sign are not used in markdown for constructs.
   // A left bracket can start definitions, references, labels,
   {atBreak: true, character: '['},
-  {character: '[', inConstruct: ['phrasing', 'label', 'reference']},
+  {character: '[', inConstruct: 'phrasing', notInConstruct: fullPhrasingSpans},
+  {character: '[', inConstruct: ['label', 'reference']},
   // A backslash can start an escape (when followed by punctuation) or a
   // hard break (when followed by an eol).
   // Note: typical escapes are handled in `safe`!
@@ -96,18 +128,15 @@ export const unsafe = [
   // Caret is not used in markdown for constructs.
   // An underscore can start emphasis, strong, or a thematic break.
   {atBreak: true, character: '_'},
-  {character: '_', inConstruct: 'phrasing'},
+  {character: '_', inConstruct: 'phrasing', notInConstruct: fullPhrasingSpans},
   // A grave accent can start code (fenced or text), or it can break out of
   // a grave accent code fence.
   {atBreak: true, character: '`'},
   {
     character: '`',
-    inConstruct: [
-      'codeFencedLangGraveAccent',
-      'codeFencedMetaGraveAccent',
-      'phrasing'
-    ]
+    inConstruct: ['codeFencedLangGraveAccent', 'codeFencedMetaGraveAccent']
   },
+  {character: '`', inConstruct: 'phrasing', notInConstruct: fullPhrasingSpans},
   // Left brace, vertical bar, right brace are not used in markdown for
   // constructs.
   // A tilde can start code (fenced).
diff --git a/test/index.js b/test/index.js
index d40b904..322514d 100644
--- a/test/index.js
+++ b/test/index.js
@@ -790,6 +790,44 @@ test('code (flow)', (t) => {
     'should use an indent if the value is indented'
   )
 
+  t.equal(
+    to({type: 'link', url: 'a b![c](d*e_f[g_h`i', children: []}),
+    '[](<a b![c](d*e_f[g_h`i>)\n',
+    'should not escape unneeded characters in a `destinationLiteral`'
+  )
+
+  t.equal(
+    to({type: 'link', url: 'a![b](c*d_e[f_g`h<i</j', children: []}),
+    '[](a![b]\\(c*d_e[f_g`h<i</j)\n',
+    'should not escape unneeded characters in a `destinationRaw`'
+  )
+
+  t.equal(
+    to({
+      type: 'linkReference',
+      identifier: 'a![b](c*d_e[f_g`h<i</j',
+      referenceType: 'full',
+      children: []
+    }),
+    '[][a!\\[b\\](c*d_e\\[f_g`h<i</j]\n',
+    'should not escape unneeded characters in a `reference`'
+  )
+
+  t.equal(
+    to({type: 'link', url: '#', title: 'a![b](c*d_e[f_g`h<i</j', children: []}),
+    '[](# "a![b](c*d_e[f_g`h<i</j")\n',
+    'should not escape unneeded characters in a `title` (double quotes)'
+  )
+
+  t.equal(
+    to(
+      {type: 'link', url: '#', title: 'a![b](c*d_e[f_g`h<i</j', children: []},
+      {quote: "'"}
+    ),
+    "[](# 'a![b](c*d_e[f_g`h<i</j')\n",
+    'should not escape unneeded characters in a `title` (single quotes)'
+  )
+
   t.end()
 })
 
@@ -3195,12 +3233,7 @@ test('escape', (t) => {
                         children: [
                           {
                             type: 'paragraph',
-                            children: [
-                              {
-                                type: 'text',
-                                value: 'bar'
-                              }
-                            ]
+                            children: [{type: 'text', value: 'bar'}]
                           }
                         ]
                       }