Improve handling of nested tags and blocks by MD033/no-inline-html (fixes #179).

helpers/helpers.js

@@ -115,6 +115,19 @@ module.exports.escapeForRegExp = function escapeForRegExp(str) {
   return str.replace(/[-/\\^$*+?.()|[\]{}]/g, "\\$&");
 };
 
+// Un-escapes Markdown content (simple algorithm; not a parser)
+const escapedMarkdownRe = /\\./g;
+module.exports.unescapeMarkdown =
+  function unescapeMarkdown(markdown, replacement) {
+    return markdown.replace(escapedMarkdownRe, (match) => {
+      const char = match[1];
+      if ("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~".includes(char)) {
+        return replacement || char;
+      }
+      return match;
+    });
+  };
+
 // Returns the indent for a token
 function indentFor(token) {
   const line = token.line.replace(/^[\s>]*(> |>)/, "");

lib/md033.js

@@ -2,10 +2,12 @@
 
 "use strict";
 
-const { addError, filterTokens, forEachInlineChild, newLineRe,
+const { addError, bareUrlRe, forEachLine, unescapeMarkdown } =
-  rangeFromRegExp } = require("../helpers");
+  require("../helpers");
+const { lineMetadata } = require("./cache");
 
-const htmlRe = /<[^>]*>/;
+const htmlElementRe = /<(\w+)(?:[^>]*)?>/g;
+const linkDestinationRe = /]\(\s*$/;
 
 module.exports = {
   "names": [ "MD033", "no-inline-html" ],
@@ -14,21 +16,21 @@ module.exports = {
   "function": function MD033(params, onError) {
     const allowedElements = (params.config.allowed_elements || [])
       .map((element) => element.toLowerCase());
-    function forToken(token) {
+    forEachLine(lineMetadata(), (line, lineIndex, inCode) => {
-      token.content.split(newLineRe)
+      let match = null;
-        .forEach((line, offset) => {
+      // eslint-disable-next-line no-unmodified-loop-condition
-          const allowed = (line.match(/<[^/\s>!]*/g) || [])
+      while (!inCode && (match = htmlElementRe.exec(line))) {
-            .filter((element) => element.length > 1)
+        const [ tag, element ] = match;
-            .map((element) => element.slice(1).toLowerCase())
+        if (!allowedElements.includes(element.toLowerCase()) &&
-            .filter((element) => !allowedElements.includes(element));
+          !tag.endsWith("\\>") && !bareUrlRe.test(tag)) {
-          if (allowed.length) {
+          const prefix = line.substring(0, match.index);
-            addError(onError, token.lineNumber + offset,
+          if (!linkDestinationRe.test(prefix) &&
-              "Element: " + allowed[0], null,
+            !unescapeMarkdown(prefix + "<", "_").endsWith("_")) {
-              rangeFromRegExp(token.line, htmlRe));
+            addError(onError, lineIndex + 1, "Element: " + element,
+              null, [ match.index + 1, tag.length ]);
           }
-        });
+        }
-    }
+      }
-    filterTokens(params, "html_block", forToken);
+    });
-    forEachInlineChild(params, "html_inline", forToken);
   }
 };

test/detailed-results-html-tags.json

@@ -0,0 +1,9 @@
+{
+  "default": true,
+  "MD033": {
+    "allowed_elements": [
+      "strong"
+    ]
+  },
+  "MD046": false
+}

test/detailed-results-html-tags.md

@@ -0,0 +1,81 @@
+# Detailed HTML Results
+
+Text
+
+<em>Block block</em>
+
+Text <em>inline inline</em> text
+
+Text
+
+<strong>Block block</strong>
+
+Text <strong>inline inline</strong> text
+
+Text
+
+<p>
+Block
+block <em>block</em> block
+block
+block <strong>block</strong> block
+block
+block <em>block</em> block <strong>block</strong> block
+block <strong>block</strong> block <em>block</em> block
+</p>
+
+Text
+
+<strong><em>Block</em> block</strong>
+
+Text <strong><em>inline</em> inline</strong> text
+
+Text
+
+<em><strong>Block</strong> block</em>
+
+Text <em><strong>inline</strong> inline</em> text
+
+Text
+
+Text <em>inline</em> text <strong>inline</strong> text <em>inline</em> text
+
+Text <strong>inline</strong> text <em>inline</em> text <strong>inline</strong> text
+
+Text
+
+\<not>Block block\</not>
+
+\\<problem>Block block\\</problem>
+
+<not\>Block block</not\>
+
+Text \<not>inline inline\</not> text
+
+Text \\<problem>inline inline\\</problem> text
+
+Text <not\>inline inline</not\> text
+
+Text
+
+> Text <em>inline inline</em> text
+> text <strong>inline inline</strong> text
+
+Text
+
+    Text <em>inline inline</em> text
+    text <strong>inline inline</strong> text
+
+Text
+
+```html
+Text <em>inline inline</em> text
+text <strong>inline inline</strong> text
+```
+
+Text
+
+Text <a href="#anchor">inline</a> text
+text <img src="src.png"/> text
+
+Text

test/detailed-results-html-tags.results.json

@@ -0,0 +1,155 @@
+[
+    {
+        "lineNumber": 5,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 1, 4 ]
+    },
+    {
+        "lineNumber": 7,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 6, 4 ]
+    },
+    {
+        "lineNumber": 17,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: p",
+        "errorContext": null,
+        "errorRange": [ 1, 3 ]
+    },
+    {
+        "lineNumber": 19,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 7, 4 ]
+    },
+    {
+        "lineNumber": 23,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 7, 4 ]
+    },
+    {
+        "lineNumber": 24,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 36, 4 ]
+    },
+    {
+        "lineNumber": 29,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 9, 4 ]
+    },
+    {
+        "lineNumber": 31,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 14, 4 ]
+    },
+    {
+        "lineNumber": 35,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 1, 4 ]
+    },
+    {
+        "lineNumber": 37,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 6, 4 ]
+    },
+    {
+        "lineNumber": 41,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 6, 4 ]
+    },
+    {
+        "lineNumber": 43,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 35, 4 ]
+    },
+    {
+        "lineNumber": 49,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: problem",
+        "errorContext": null,
+        "errorRange": [ 3, 9 ]
+    },
+    {
+        "lineNumber": 55,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: problem",
+        "errorContext": null,
+        "errorRange": [ 8, 9 ]
+    },
+    {
+        "lineNumber": 61,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: em",
+        "errorContext": null,
+        "errorRange": [ 8, 4 ]
+    },
+    {
+        "lineNumber": 78,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: a",
+        "errorContext": null,
+        "errorRange": [ 6, 18 ]
+    },
+    {
+        "lineNumber": 79,
+        "ruleNames": [ "MD033", "no-inline-html" ],
+        "ruleDescription": "Inline HTML",
+        "ruleInformation": "https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/Rules.md#md033",
+        "errorDetail": "Element: img",
+        "errorContext": null,
+        "errorRange": [ 6, 20 ]
+    }
+]

test/empty-links.md

@@ -6,24 +6,56 @@
 
 [text](<>) {MD042}
 
+[text]( <> ) {MD042}
+
+[text](<> "title") {MD042}
+
+[text]( <> "title" ) {MD042}
+
 [text](#) {MD042}
 
+[text]( # ) {MD042}
+
+[text](# "title") {MD042}
+
+[text]( # "title" ) {MD042}
+
 [text][frag] {MD042}
 
+[text][ frag ] {MD042}
+
 [frag]: #
 
 ## Non-empty links
 
 [text](link)
 
+[text]( link )
+
 [text](link "title")
 
+[text]( link "title" )
+
 [text](<link>)
 
+[text]( <link> )
+
+[text](<link> "title")
+
+[text]( <link> "title" )
+
 [text](#frag)
 
+[text]( #frag )
+
+[text](#frag "title")
+
+[text]( #frag "title" )
+
 [text][ref]
 
+[text][ ref ]
+
 [ref]: link
 
 [text]

test/markdownlint-test.js

@@ -1406,6 +1406,68 @@ function clearHtmlCommentTextEmbedded(test) {
   test.done();
 };
 
+module.exports.unescapeMarkdown = function unescapeMarkdown(test) {
+  test.expect(7);
+  // Test cases from https://spec.commonmark.org/0.29/#backslash-escapes
+  const testCases = [
+    [
+      "\\!\\\"\\#\\$\\%\\&\\'\\(\\)\\*\\+\\,\\-\\.\\/\\:\\;" +
+        "\\<\\=\\>\\?\\@\\[\\\\\\]\\^\\_\\`\\{\\|\\}\\~",
+      "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~"
+    ],
+    [
+      "\\→\\A\\a\\ \\3\\φ\\«",
+      "\\→\\A\\a\\ \\3\\φ\\«"
+    ],
+    [
+      `\\*not emphasized*
+\\<br/> not a tag
+\\[not a link](/foo)
+\\\`not code\`
+1\\. not a list
+\\* not a list
+\\# not a heading
+\\[foo]: /url "not a reference"
+\\&ouml; not a character entity`,
+      `*not emphasized*
+<br/> not a tag
+[not a link](/foo)
+\`not code\`
+1. not a list
+* not a list
+# not a heading
+[foo]: /url "not a reference"
+&ouml; not a character entity`
+    ],
+    [
+      "\\\\*emphasis*",
+      "\\*emphasis*"
+    ],
+    [
+      `foo\\
+bar`,
+      `foo\\
+bar`
+    ],
+    [
+      "Text \\<",
+      "Text _",
+      "_"
+    ],
+    [
+      "Text \\\\<",
+      "Text _<",
+      "_"
+    ]
+  ];
+  testCases.forEach(function forTestCase(testCase) {
+    const [ markdown, expected, replacement ] = testCase;
+    const actual = helpers.unescapeMarkdown(markdown, replacement);
+    test.equal(actual, expected);
+  });
+  test.done();
+};
+
 module.exports.isBlankLine = function isBlankLine(test) {
   test.expect(25);
   const blankLines = [