FIX: don't create tags autolinks inside links (new engine)

app/assets/javascripts/pretty-text/engines/markdown-it/category-hashtag.js.es6

@@ -1,42 +1,97 @@
-import { inlineRegexRule } from 'pretty-text/engines/markdown-it/helpers';
+import { textReplace } from 'pretty-text/engines/markdown-it/helpers';
 
-function emitter(matches, state) {
-    const options = state.md.options.discourse;
-    const [hashtag, slug] = matches;
-    const categoryHashtagLookup = options.categoryHashtagLookup;
-    const result = categoryHashtagLookup && categoryHashtagLookup(slug);
+function addHashtag(buffer, matches, state) {
+  const options = state.md.options.discourse;
+  const [hashtag, slug] = matches;
+  const categoryHashtagLookup = options.categoryHashtagLookup;
+  const result = categoryHashtagLookup && categoryHashtagLookup(slug);
 
-    let token;
+  let token;
 
-    if (result) {
-      token = state.push('link_open', 'a', 1);
-      token.attrs = [['class', 'hashtag'], ['href', result[0]]];
-      token.block = false;
+  if (result) {
+    token = new state.Token('link_open', 'a', 1);
+    token.attrs = [['class', 'hashtag'], ['href', result[0]]];
+    token.block = false;
+    buffer.push(token);
 
-      token = state.push('text', '', 0);
-      token.content = '#';
+    token = new state.Token('text', '', 0);
+    token.content = '#';
+    buffer.push(token);
 
-      token = state.push('span_open', 'span', 1);
-      token.block = false;
+    token = new state.Token('span_open', 'span', 1);
+    token.block = false;
+    buffer.push(token);
 
-      token = state.push('text', '', 0);
-      token.content = result[1];
+    token = new state.Token('text', '', 0);
+    token.content = result[1];
+    buffer.push(token);
 
-      state.push('span_close', 'span', -1);
+    buffer.push(new state.Token('span_close', 'span', -1));
 
-      state.push('link_close', 'a', -1);
-    } else {
+    buffer.push(new state.Token('link_close', 'a', -1));
+  } else {
 
-      token = state.push('span_open', 'span', 1);
-      token.attrs = [['class', 'hashtag']];
+    token = new state.Token('span_open', 'span', 1);
+    token.attrs = [['class', 'hashtag']];
+    buffer.push(token);
 
-      token = state.push('text', '', 0);
-      token.content = hashtag;
+    token = new state.Token('text', '', 0);
+    token.content = hashtag;
+    buffer.push(token);
 
-      token = state.push('span_close', 'span', -1);
+    token = new state.Token('span_close', 'span', -1);
+    buffer.push(token);
+  }
+}
+
+const REGEX = /#([\w-:]{1,101})/gi;
+
+function allowedBoundary(content, index, utils) {
+  let code = content.charCodeAt(index);
+  return (utils.isWhiteSpace(code) || utils.isPunctChar(String.fromCharCode(code)));
+}
+
+function applyHashtag(content, state) {
+  let result = null,
+      match,
+      pos = 0;
+
+  while (match = REGEX.exec(content)) {
+    // check boundary
+    if (match.index > 0) {
+      if (!allowedBoundary(content, match.index-1, state.md.utils)) {
+        console.log("not allowed");
+        continue;
+      }
     }
 
-    return true;
+    // check forward boundary as well
+    if (match.index + match[0].length < content.length) {
+      if (!allowedBoundary(content, match.index + match[0].length, state.md.utils)) {
+        continue;
+      }
+    }
+
+    if (match.index > pos) {
+      result = result || [];
+      let token = new state.Token('text', '', 0);
+      token.content = content.slice(pos, match.index);
+      result.push(token);
+    }
+
+    result = result || [];
+    addHashtag(result, match, state);
+
+    pos = match.index + match[0].length;
+  }
+
+  if (result && pos < content.length) {
+    let token = new state.Token('text', '', 0);
+    token.content = content.slice(pos);
+    result.push(token);
+  }
+
+  return result;
 }
 
 export function setup(helper) {
@@ -45,14 +100,8 @@ export function setup(helper) {
 
   helper.registerPlugin(md=>{
 
-    const rule = inlineRegexRule(md, {
-      start: '#',
-      matcher: /^#([\w-:]{1,101})/i,
-      skipInLink: true,
-      maxLength: 102,
-      emitter: emitter
-    });
-
-    md.inline.ruler.push('category-hashtag', rule);
+    md.core.ruler.push('category-hashtag', state => textReplace(
+      state, applyHashtag, true /* skip all links */
+    ));
   });
 }

app/assets/javascripts/pretty-text/engines/markdown-it/helpers.js.es6

@@ -12,6 +12,7 @@ export default null;
 //   matcher: /^#([\w-:]{1,101})/i,
 //   emitter: emitter
 // });
+
 export function inlineRegexRule(md, options) {
 
   const start = options.start.charCodeAt(0);
@@ -27,7 +28,7 @@ export function inlineRegexRule(md, options) {
     // test prev
     if (pos > 0) {
       let prev = state.src.charCodeAt(pos-1);
-      if (!md.utils.isSpace(prev) && !md.utils.isPunctChar(String.fromCharCode(prev))) {
+      if (!md.utils.isWhiteSpace(prev) && !md.utils.isPunctChar(String.fromCharCode(prev))) {
         return false;
       }
     }
@@ -38,10 +39,10 @@ export function inlineRegexRule(md, options) {
       for(i=state.tokens.length-1;i>=0;i--) {
         let token = state.tokens[i];
         let type = token.type;
-        if (type === 'link_open' || (type === 'html_inline' && token.content.substr(0,2) === "<a")) {
+        if (type === 'link_open' || (type === 'html_inline' && token.content.substr(0,2).toLowerCase() === "<a")) {
           return false;
         }
-        if (type.block || type === 'link_close' || (type === 'html_inline' && token.content.substr(0,3).toLowerCase() === "</a>")) {
+        if (type.block || type === 'link_close' || (type === 'html_inline' && token.content.substr(0,4).toLowerCase() === "</a>")) {
           break;
         }
       }
@@ -75,10 +76,10 @@ export function inlineRegexRule(md, options) {
 
 // based off https://github.com/markdown-it/markdown-it-emoji/blob/master/dist/markdown-it-emoji.js
 //
-export function textReplace(state, callback) {
+export function textReplace(state, callback, skipAllLinks) {
   var i, j, l, tokens, token,
         blockTokens = state.tokens,
-        autolinkLevel = 0;
+        linkLevel = 0;
 
   for (j = 0, l = blockTokens.length; j < l; j++) {
     if (blockTokens[j].type !== 'inline') { continue; }
@@ -89,11 +90,23 @@ export function textReplace(state, callback) {
     for (i = tokens.length - 1; i >= 0; i--) {
       token = tokens[i];
 
-      if (token.type === 'link_open' || token.type === 'link_close') {
-        if (token.info === 'auto') { autolinkLevel -= token.nesting; }
+      if (skipAllLinks) {
+        if (token.type === 'link_open' || token.type === 'link_close') {
+          linkLevel -= token.nesting;
+        } else if (token.type === 'html_inline') {
+          if (token.content.substr(0,2).toLowerCase() === "<a") {
+            linkLevel++;
+          } else if (token.content.substr(0,4).toLowerCase() === "</a>") {
+            linkLevel--;
+          }
+        }
+      } else {
+        if (token.type === 'link_open' || token.type === 'link_close') {
+          if (token.info === 'auto') { linkLevel -= token.nesting; }
+        }
       }
 
-      if (token.type === 'text' && autolinkLevel === 0) {
+      if (token.type === 'text' && linkLevel === 0) {
         let split;
         if(split = callback(token.content, state)) {
           // replace current node

spec/components/pretty_text_spec.rb

@@ -561,6 +561,12 @@ HTML
       HTML
 
       expect(cooked).to eq(html.strip)
+
+      # ensure it does not fight with the autolinker
+      expect(PrettyText.cook(' http://somewhere.com/#known')).not_to include('hashtag')
+      expect(PrettyText.cook(' http://somewhere.com/?#known')).not_to include('hashtag')
+      expect(PrettyText.cook(' http://somewhere.com/?abc#known')).not_to include('hashtag')
+
     end
 
     it "can handle mixed lists" do