Update clearHtmlCommentText helper to match CommonMark (instead of HTML) specification (refs #361).

.eslintrc.json

@@ -36,6 +36,7 @@
     "id-length": "off",
     "indent": ["error", 2, { "SwitchCase": 1 }],
     "linebreak-style": "off",
+    "max-depth": "off",
     "max-lines": "off",
     "max-lines-per-function": "off",
     "max-params": ["error", 10],

demo/markdownlint-browser.js

@@ -115,31 +115,48 @@ module.exports.includesSorted = function includesSorted(array, element) {
 // Replaces the text of all properly-formatted HTML comments with whitespace
 // This preserves the line/column information for the rest of the document
 // Trailing whitespace is avoided with a '\' character in the last column
-// See https://www.w3.org/TR/html5/syntax.html#comments for details
+// https://spec.commonmark.org/0.29/#html-blocks
+// https://spec.commonmark.org/0.29/#html-comment
 var htmlCommentBegin = "<!--";
 var htmlCommentEnd = "-->";
 module.exports.clearHtmlCommentText = function clearHtmlCommentText(text) {
     var i = 0;
     while ((i = text.indexOf(htmlCommentBegin, i)) !== -1) {
-        var j = text.indexOf(htmlCommentEnd, i);
+        var j = text.indexOf(htmlCommentEnd, i + 2);
         if (j === -1) {
             // Un-terminated comments are treated as text
             break;
         }
-        var comment = text.slice(i + htmlCommentBegin.length, j);
+        // If the comment has content...
-        if ((comment.length > 0) &&
+        if (j > i + htmlCommentBegin.length) {
-            !comment.startsWith(">") &&
+            var k = i - 1;
-            !comment.startsWith("->") &&
+            while (text[k] === " ") {
-            !comment.endsWith("<!-") &&
+                k--;
-            !comment.includes("<!--") &&
+            }
-            // !comment.includes("-->") &&
+            // If comment is not within an indented code block...
-            !comment.includes("--!>") &&
+            if (k >= i - 4) {
-            (text.slice(i, j + htmlCommentEnd.length).search(inlineCommentRe) === -1)) {
+                var content = text.slice(i + htmlCommentBegin.length, j);
-            var blanks = comment
+                var isBlock = (k < 0) || (text[k] === "\n");
+                var isValid = isBlock ||
+                    (!content.startsWith(">") && !content.startsWith("->") &&
+                        !content.endsWith("-") && !content.includes("--"));
+                // If a valid block/inline comment...
+                if (isValid) {
+                    var inlineCommentIndex = text
+                        .slice(i, j + htmlCommentEnd.length)
+                        .search(inlineCommentRe);
+                    // If not a markdownlint inline directive...
+                    if (inlineCommentIndex === -1) {
+                        var blanks = content
                             .replace(/[^\r\n]/g, " ")
                             .replace(/ ([\r\n])/g, "\\$1");
-            text = text.slice(0, i + htmlCommentBegin.length) +
+                        text =
-                blanks + text.slice(j);
+                            text.slice(0, i + htmlCommentBegin.length) +
+                                blanks +
+                                text.slice(j);
+                    }
+                }
+            }
         }
         i = j + htmlCommentEnd.length;
     }

helpers/helpers.js

@@ -102,33 +102,48 @@ module.exports.includesSorted = function includesSorted(array, element) {
 // Replaces the text of all properly-formatted HTML comments with whitespace
 // This preserves the line/column information for the rest of the document
 // Trailing whitespace is avoided with a '\' character in the last column
-// See https://www.w3.org/TR/html5/syntax.html#comments for details
+// https://spec.commonmark.org/0.29/#html-blocks
+// https://spec.commonmark.org/0.29/#html-comment
 const htmlCommentBegin = "<!--";
 const htmlCommentEnd = "-->";
 module.exports.clearHtmlCommentText = function clearHtmlCommentText(text) {
   let i = 0;
   while ((i = text.indexOf(htmlCommentBegin, i)) !== -1) {
-    const j = text.indexOf(htmlCommentEnd, i);
+    const j = text.indexOf(htmlCommentEnd, i + 2);
     if (j === -1) {
       // Un-terminated comments are treated as text
       break;
     }
-    const comment = text.slice(i + htmlCommentBegin.length, j);
+    // If the comment has content...
-    if (
+    if (j > i + htmlCommentBegin.length) {
-      (comment.length > 0) &&
+      let k = i - 1;
-      !comment.startsWith(">") &&
+      while (text[k] === " ") {
-      !comment.startsWith("->") &&
+        k--;
-      !comment.endsWith("<!-") &&
+      }
-      !comment.includes("<!--") &&
+      // If comment is not within an indented code block...
-      // !comment.includes("-->") &&
+      if (k >= i - 4) {
-      !comment.includes("--!>") &&
+        const content = text.slice(i + htmlCommentBegin.length, j);
-      (text.slice(i, j + htmlCommentEnd.length).search(inlineCommentRe) === -1)
+        const isBlock = (k < 0) || (text[k] === "\n");
-    ) {
+        const isValid = isBlock ||
-      const blanks = comment
+          (!content.startsWith(">") && !content.startsWith("->") &&
+           !content.endsWith("-") && !content.includes("--"));
+        // If a valid block/inline comment...
+        if (isValid) {
+          const inlineCommentIndex = text
+            .slice(i, j + htmlCommentEnd.length)
+            .search(inlineCommentRe);
+          // If not a markdownlint inline directive...
+          if (inlineCommentIndex === -1) {
+            const blanks = content
               .replace(/[^\r\n]/g, " ")
               .replace(/ ([\r\n])/g, "\\$1");
-      text = text.slice(0, i + htmlCommentBegin.length) +
+            text =
-        blanks + text.slice(j);
+              text.slice(0, i + htmlCommentBegin.length) +
+              blanks +
+              text.slice(j);
+          }
+        }
+      }
     }
     i = j + htmlCommentEnd.length;
   }

test/html-comments.md

@@ -0,0 +1,65 @@
+# HTML Comments
+
+## Block Comments
+
+<https://spec.commonmark.org/0.29/#html-blocks>
+
+<!-- *comment * -->
+
+<!-- *comment * -->text
+
+   <!-- *comment * -->
+
+    <!-- *code * -->
+
+<!-- *comment *
+*comment * -->
+
+<!-- *comment *
+*comment *
+*comment * -->
+
+<!--> *{MD037} * -->
+
+<!---> *{MD037} * -->
+
+<!-- *comment * --->
+
+<!-- -- *comment * -->
+
+<!-- *comment * -- -->
+
+## Inline Comments
+
+<https://spec.commonmark.org/0.29/#html-comment>
+
+t<!-- *comment * -->
+
+t<!-- *comment * -->text
+
+   t<!-- *comment * -->
+
+    t<!-- *code * -->
+
+t<!-- *comment *
+*comment * -->
+
+t<!-- *comment *
+*comment *
+*comment * -->
+
+t<!--> *{MD037} * -->
+
+t<!---> *{MD037} * -->
+
+t<!-- *{MD037} * --->
+
+t<!-- -- *{MD037} * -->
+
+t<!-- *{MD037} * -- -->
+
+## Notes
+
+It's important that the rule used above is one that calls
+`helpers.forEachLine` so `markdown-it` doesn't ignore any
+incorrectly-remaining comment blocks.

test/ignore-comments.md

@@ -18,11 +18,11 @@ Hard	tab
 Hard	tab
 -->
 
-<!--
+Text <!--
 Hard	tab {MD010}
 Invalid--!>comment
 Hard	tab {MD010}
--->
+--> text
 
 Te<!-- Hard	tab -->xt
 

test/markdownlint-test-helpers.js

@@ -9,6 +9,12 @@ const helpers = require("../helpers");
 test("clearHtmlCommentTextValid", (t) => {
   t.plan(1);
   const validComments = [
+    "<!-->",
+    "<!--->",
+    "<!---->",
+    "<!-- comment -->",
+    " <!-- comment -->",
+    "  <!-- comment -->",
     "<!-- text -->",
     "<!--text-->",
     "<!-- -->",
@@ -47,6 +53,12 @@ test("clearHtmlCommentTextValid", (t) => {
     "text"
   ];
   const validResult = [
+    "<!-->",
+    "<!--->",
+    "<!---->",
+    "<!--         -->",
+    " <!--         -->",
+    "  <!--         -->",
     "<!--      -->",
     "<!--    -->",
     "<!-- -->",
@@ -100,22 +112,12 @@ test("clearHtmlCommentTextInvalid", (t) => {
     "<!-->text-->",
     "<!--->text-->",
     "<!---->",
-    // Restrictions from specification
     "<!-->-->",
     "<!-->t-->",
     "<!--->-->",
     "<!--->t-->",
-    "<!--<!--t-->",
+    "<!---->t-->",
-    "<!--t<!---->",
+    "    <!-- indented code block -->"
-    "<!--t<!--t-->",
-    // "<!---->t-->",
-    // "<!--t-->-->",
-    // "<!--t-->t-->",
-    "<!----!>t-->",
-    "<!--t--!>-->",
-    "<!--t--!>t-->",
-    "<!--<!--->",
-    "<!--t<!--->"
   ];
   const actual = helpers.clearHtmlCommentText(invalidComments.join("\n"));
   const expected = invalidComments.join("\n");