chore(plugins) move useBR, tabReplace, and HTML merging into internal…

… plugins (#2873) - This begins the official deprecation of HTML merging and tabReplace. - `useBR` was previously deprecated.
highlightjs · Dec 15, 2020 · 03f2023 · 03f2023
1 parent 1846044
commit 03f2023
Show file tree

Hide file tree

Showing 8 changed files with 262 additions and 173 deletions.
diff --git a/.eslintrc.js b/.eslintrc.js
@@ -26,6 +26,7 @@ module.exports = {
     "init-declarations": ["error", "always"],
     "array-callback-return": "error",
     "block-scoped-var": "error",
+    "no-multiple-empty-lines": ["error", { max: 2 }],
     // we like our semi-colons
     semi: ["error", "always"],
     // our codebase doesn't do this at all, so disabled for now
@@ -57,7 +58,7 @@ module.exports = {
         // // languages are all over the map and we don't want to
         // // do a mass edit so turn off the most egregious rule violations
         // indent: "off",
-        "camelcase": 0,
+        camelcase: 0,
         "no-control-regex": 0,
         "no-useless-escape": 0,
         "comma-dangle": 1,

diff --git a/CHANGES.md b/CHANGES.md
@@ -29,6 +29,32 @@ Grammar improvements:
 - enh(dart) Fix empty doc-comment eating next line [Jan Pilzer][]
 - enh(asciidoc) Adds support for unconstrained bold syntax (#2869) [Guillaume Grossetie][]
 
+Deprecations (*these will be mentioned in every release*):
+
+- HTML "merging" is deprecated. (#2873) [Josh Goebel][]
+  - HTML inside `<pre>` blocks will no longer be magically merged back into the
+  highlighted code's HTML result - it will instead be silently removed.
+  - Consider [using a plugin][htmlPlugin] if you truly need this functionality
+  - Deprecated as of 10.5.0 - will be removed in v11.
+- `tabReplace` option deprecated. (#2873) [Josh Goebel][]
+  - **Consider:** Use the CSS `tab-size` property, or simply pre-process the
+    text yourself before rendering the initial HTML
+  - otherwise, [use a plugin][tabPlugin]
+  - Deprecated as of 10.5.0 - will be removed in v11.
+- `useBR` option deprecated. (#2559) [Josh Goebel][]
+  - **Recommended:** You really should just use the HTML `<pre>` tag
+  - or perhaps try CSS `white-space: pre;`
+  - otherwise, [use a plugin][brPlugin]
+  - Deprecated as of 10.3.0 - will be removed in v11.
+- `requireLanguage` API is deprecated, will be removed in v11.0.
+  - **Consider:** Use `getLanguage` (with custom error handling) or built-time dependencies.
+  - See [Library API](https://highlightjs.readthedocs.io/en/latest/api.html#requirelanguage-name) for more information.
+  - Deprecated as of 10.4.0 - will be removed in v11.
+
+[htmlPlugin]: https://github.com/highlightjs/highlight.js/issues/2889
+[tabPlugin]: https://github.com/highlightjs/highlight.js/issues/2874
+[brPlugin]: https://github.com/highlightjs/highlight.js/issues/2559
+
 [Martin Dørum]: https://github.com/mortie
 [Jan Pilzer]: https://github.com/Hirse
 [Oldes Huhuman]: https://github.com/Oldes
@@ -101,6 +127,7 @@ Parser:
 - enh(grammars) allow `classNameAliases` for more complex grammars [Josh Goebel][]
 - fix(vue): Language name now appears in CSS class (#2807) [Michael Rush][]
 - (chore) Clean up all regexs to be UTF-8 compliant/ready (#2759) [Josh Goebel][]
+- enh(grammars) allow `classNameAliases` for more complex grammars [Josh Goebel][]
 
 New Languages:
 

diff --git a/README.md b/README.md
@@ -87,14 +87,14 @@ Highlight.js supports over 180 different languages in the core library.  There a
 language plugins available for additional languages. You can find the full list of supported languages
 in [SUPPORTED_LANGUAGES.md][9].
 
-## Custom Initialization
+## Custom Scenarios
 
 When you need a bit more control over the initialization of
 highlight.js, you can use the [`highlightBlock`][3] and [`configure`][4]
-functions. This allows you to control *what* to highlight and *when*.
+functions. This allows you to better control *what* to highlight and *when*.
 
-Here’s an equivalent way to calling [`initHighlightingOnLoad`][1] using
-vanilla JS:
+Here’s the equivalent of calling [`initHighlightingOnLoad`][1] using
+only vanilla JS:
 
 ```js
 document.addEventListener('DOMContentLoaded', (event) => {
@@ -104,19 +104,45 @@ document.addEventListener('DOMContentLoaded', (event) => {
 });
 ```
 
-You can use any tags instead of `<pre><code>` to mark up your code. If
-you don't use a container that preserves line breaks you will need to
-configure highlight.js to use the `<br>` tag:
+Please refer to the documentation for [`configure`][4] options.
 
-```js
-hljs.configure({useBR: true});
 
-document.querySelectorAll('div.code').forEach((block) => {
+### Using custom HTML elements for code blocks
+
+We strongly recommend `<pre><code>` wrapping for code blocks. It's quite
+semantic and "just works" out of the box with zero fiddling. It is possible to
+use other HTML elements (or combos), but you may need to pay special attention to
+preserving linebreaks.
+
+Let's say your markup for code blocks uses divs:
+
+```html
+<div class='code'>...</div>
+```
+
+To highlight such blocks manually:
+
+```js
+// first, find all the div.code blocks
+document.querySelectorAll('div.code').forEach(block => {
+  // then highlight each
   hljs.highlightBlock(block);
 });
 ```
 
-For other options refer to the documentation for [`configure`][4].
+Without using a tag that preserves linebreaks (like `pre`) you'll need some
+additional CSS to help preserve them.  You could also [pre and post-process line
+breaks with a plug-in][brPlugin], but *we recommend using CSS*.
+
+[brPlugin]: https://github.com/highlightjs/highlight.js/issues/2559
+
+To preserve linebreaks inside a `div` using CSS:
+
+```css
+div.code {
+  white-space: pre;
+}
+```
 
 
 ## Using with Vue.js

diff --git a/docs/plugin-api.rst b/docs/plugin-api.rst
@@ -101,8 +101,8 @@ Note: This callback does not fire from highlighting resulting from auto-language
 It returns nothing.
 
 
-after:highlightBlock({block, result})
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+after:highlightBlock({block, result, text})
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 This callback function is passed an object with two keys:
 
@@ -112,11 +112,14 @@ block
 result
   The result object returned by `highlight` or `highlightAuto`.
 
+text
+  The raw text that was to be highlighted.
+
 It returns nothing.
 
 
 before:highlightBlock({block, language})
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 This callback function is passed an object with two keys:
 

diff --git a/src/highlight.js b/src/highlight.js
@@ -13,11 +13,10 @@ import * as MODES from './lib/modes.js';
 import { compileLanguage } from './lib/mode_compiler.js';
 import * as packageJSON from '../package.json';
 import { BuildVuePlugin } from "./plugins/vue.js";
+import { mergeHTMLPlugin } from "./plugins/merge_html.js";
 
 const escape = utils.escapeHTML;
 const inherit = utils.inherit;
-
-const { nodeStream, mergeStreams } = utils;
 const NO_MATCH = Symbol("nomatch");
 
 /**
@@ -667,6 +666,32 @@ const HLJS = function(hljs) {
     return result.join(' ').trim();
   }
 
+  /** @type {HLJSPlugin} */
+  const brPlugin = {
+    "before:highlightBlock": ({ block }) => {
+      if (options.useBR) {
+        block.innerHTML = block.innerHTML.replace(/\n/g, '').replace(/<br[ /]*>/g, '\n');
+      }
+    },
+    "after:highlightBlock": ({ result }) => {
+      if (options.useBR) {
+        result.value = result.value.replace(/\n/g, "<br>");
+      }
+    }
+  };
+
+  const TAB_REPLACE_RE = /^(<[^>]+>|\t)+/gm;
+  /** @type {HLJSPlugin} */
+  const tabReplacePlugin = {
+    "after:highlightBlock": ({ result }) => {
+      if (options.tabReplace) {
+        result.value = result.value.replace(TAB_REPLACE_RE, (m) =>
+          m.replace(/\t/g, options.tabReplace)
+        );
+      }
+    }
+  };
+
   /**
    * Applies highlighting to a DOM node containing code. Accepts a DOM node and
    * two optional parameters for fixMarkup.
@@ -683,24 +708,11 @@ const HLJS = function(hljs) {
     fire("before:highlightBlock",
       { block: element, language: language });
 
-    if (options.useBR) {
-      node = document.createElement('div');
-      node.innerHTML = element.innerHTML.replace(/\n/g, '').replace(/<br[ /]*>/g, '\n');
-    } else {
-      node = element;
-    }
+    node = element;
     const text = node.textContent;
     const result = language ? highlight(language, text, true) : highlightAuto(text);
 
-    const originalStream = nodeStream(node);
-    if (originalStream.length) {
-      const resultNode = document.createElement('div');
-      resultNode.innerHTML = result.value;
-      result.value = mergeStreams(originalStream, nodeStream(resultNode), text);
-    }
-    result.value = fixMarkup(result.value);
-
-    fire("after:highlightBlock", { block: element, result: result });
+    fire("after:highlightBlock", { block: element, result, text });
 
     element.innerHTML = result.value;
     element.className = buildClassName(element.className, language, result.language);
@@ -909,6 +921,10 @@ const HLJS = function(hljs) {
   // merge all the modes/regexs into our main object
   Object.assign(hljs, MODES);
 
+  // built-in plugins, likely to be moved out of core in the future
+  hljs.addPlugin(brPlugin); // slated to be removed in v11
+  hljs.addPlugin(mergeHTMLPlugin);
+  hljs.addPlugin(tabReplacePlugin);
   return hljs;
 };
 

diff --git a/src/lib/utils.js b/src/lib/utils.js
@@ -34,143 +34,3 @@ export function inherit(original, ...objects) {
   return /** @type {T} */ (result);
 }
 
-/* Stream merging */
-
-/**
- * @typedef Event
- * @property {'start'|'stop'} event
- * @property {number} offset
- * @property {Node} node
- */
-
-/**
- * @param {Node} node
- */
-function tag(node) {
-  return node.nodeName.toLowerCase();
-}
-
-/**
- * @param {Node} node
- */
-export function nodeStream(node) {
-  /** @type Event[] */
-  const result = [];
-  (function _nodeStream(node, offset) {
-    for (let child = node.firstChild; child; child = child.nextSibling) {
-      if (child.nodeType === 3) {
-        offset += child.nodeValue.length;
-      } else if (child.nodeType === 1) {
-        result.push({
-          event: 'start',
-          offset: offset,
-          node: child
-        });
-        offset = _nodeStream(child, offset);
-        // Prevent void elements from having an end tag that would actually
-        // double them in the output. There are more void elements in HTML
-        // but we list only those realistically expected in code display.
-        if (!tag(child).match(/br|hr|img|input/)) {
-          result.push({
-            event: 'stop',
-            offset: offset,
-            node: child
-          });
-        }
-      }
-    }
-    return offset;
-  })(node, 0);
-  return result;
-}
-
-/**
- * @param {any} original - the original stream
- * @param {any} highlighted - stream of the highlighted source
- * @param {string} value - the original source itself
- */
-export function mergeStreams(original, highlighted, value) {
-  let processed = 0;
-  let result = '';
-  const nodeStack = [];
-
-  function selectStream() {
-    if (!original.length || !highlighted.length) {
-      return original.length ? original : highlighted;
-    }
-    if (original[0].offset !== highlighted[0].offset) {
-      return (original[0].offset < highlighted[0].offset) ? original : highlighted;
-    }
-
-    /*
-    To avoid starting the stream just before it should stop the order is
-    ensured that original always starts first and closes last:
-
-    if (event1 == 'start' && event2 == 'start')
-      return original;
-    if (event1 == 'start' && event2 == 'stop')
-      return highlighted;
-    if (event1 == 'stop' && event2 == 'start')
-      return original;
-    if (event1 == 'stop' && event2 == 'stop')
-      return highlighted;
-
-    ... which is collapsed to:
-    */
-    return highlighted[0].event === 'start' ? original : highlighted;
-  }
-
-  /**
-   * @param {Node} node
-   */
-  function open(node) {
-    /** @param {Attr} attr */
-    function attributeString(attr) {
-      return ' ' + attr.nodeName + '="' + escapeHTML(attr.value) + '"';
-    }
-    // @ts-ignore
-    result += '<' + tag(node) + [].map.call(node.attributes, attributeString).join('') + '>';
-  }
-
-  /**
-   * @param {Node} node
-   */
-  function close(node) {
-    result += '</' + tag(node) + '>';
-  }
-
-  /**
-   * @param {Event} event
-   */
-  function render(event) {
-    (event.event === 'start' ? open : close)(event.node);
-  }
-
-  while (original.length || highlighted.length) {
-    let stream = selectStream();
-    result += escapeHTML(value.substring(processed, stream[0].offset));
-    processed = stream[0].offset;
-    if (stream === original) {
-      /*
-      On any opening or closing tag of the original markup we first close
-      the entire highlighted node stack, then render the original tag along
-      with all the following original tags at the same offset and then
-      reopen all the tags on the highlighted stack.
-      */
-      nodeStack.reverse().forEach(close);
-      do {
-        render(stream.splice(0, 1)[0]);
-        stream = selectStream();
-      } while (stream === original && stream.length && stream[0].offset === processed);
-      nodeStack.reverse().forEach(open);
-    } else {
-      if (stream[0].event === 'start') {
-        nodeStack.push(stream[0].node);
-      } else {
-        nodeStack.pop();
-      }
-      render(stream.splice(0, 1)[0]);
-    }
-  }
-  return result + escapeHTML(value.substr(processed));
-}