Merge pull request #629 from netlify/editor-fast-follow

editor fast follow

src/components/Widgets/Markdown/MarkdownControl/RawEditor/index.js

@@ -22,16 +22,14 @@ export default class RawEditor extends React.Component {
     this.setState({ editorState });
   }
 
-  onChange = debounce(this.props.onChange, 250);
-
   /**
    * When the document value changes, serialize from Slate's AST back to plain
    * text (which is Markdown) and pass that up as the new value.
    */
-  handleDocumentChange = (doc, editorState) => {
+  handleDocumentChange = debounce((doc, editorState) => {
     const value = Plain.serialize(editorState);
-    this.onChange(value);
-  };
+    this.props.onChange(value);
+  }, 150);
 
   /**
    * If a paste contains plain text, deserialize it to Slate's AST and insert

src/components/Widgets/Markdown/MarkdownControl/VisualEditor/__tests__/__snapshots__/parser.spec.js.snap

@@ -443,7 +443,7 @@ become anything else.",
         Object {
           "data": undefined,
           "kind": "text",
-          "text": " is supported. This *can be ",
+          "text": " is supported. This ",
         },
         Object {
           "data": undefined,
@@ -451,18 +451,37 @@ become anything else.",
           "ranges": Array [
             Object {
               "marks": Array [
+                Object {
+                  "type": "italic",
+                },
+              ],
+              "text": "can be ",
+            },
+            Object {
+              "marks": Array [
+                Object {
+                  "type": "italic",
+                },
                 Object {
                   "type": "bold",
                 },
               ],
               "text": "nested",
             },
+            Object {
+              "marks": Array [
+                Object {
+                  "type": "italic",
+                },
+              ],
+              "text": " like",
+            },
           ],
         },
         Object {
           "data": undefined,
           "kind": "text",
-          "text": " like* so.",
+          "text": " so.",
         },
       ],
       "type": "paragraph",
@@ -1288,12 +1307,18 @@ Object {
           "kind": "text",
           "text": "blue moon",
         },
+        Object {
+          "kind": "inline",
+          "nodes": Array [
             Object {
               "data": undefined,
               "kind": "text",
               "text": "
 ",
             },
+          ],
+          "type": "break",
+        },
         Object {
           "data": undefined,
           "kind": "text",

src/components/Widgets/Markdown/MarkdownControl/VisualEditor/components.js

@@ -1,4 +1,5 @@
 import React from 'react';
+import { List } from 'immutable';
 import cn from 'classnames';
 import styles from './index.css';
 
@@ -35,9 +36,28 @@ export const NODE_COMPONENTS = {
     <ol {...props.attributes} start={props.node.data.get('start') || 1}>{props.children}</ol>,
   'link': props => {
     const data = props.node.get('data');
+    const marks = data.get('marks');
     const url = data.get('url');
     const title = data.get('title');
-    return <a href={url} title={title} {...props.attributes}>{props.children}</a>;
+    const link = <a href={url} title={title} {...props.attributes}>{props.children}</a>;
+    const result = !marks ? link : marks.reduce((acc, mark) => {
+      const MarkComponent = MARK_COMPONENTS[mark.type];
+      return <MarkComponent>{acc}</MarkComponent>;
+    }, link);
+    return result;
+  },
+  'image': props => {
+    const data = props.node.get('data');
+    const marks = data.get('marks');
+    const url = data.get('url');
+    const title = data.get('title');
+    const alt = data.get('alt');
+    const image = <img src={url} title={title} alt={alt} {...props.attributes}/>;
+    const result = !marks ? image : marks.reduce((acc, mark) => {
+      const MarkComponent = MARK_COMPONENTS[mark.type];
+      return <MarkComponent>{acc}</MarkComponent>;
+    }, image);
+    return result;
   },
   'shortcode': props => {
     const { attributes, node, state: editorState } = props;

src/components/Widgets/Markdown/MarkdownControl/VisualEditor/index.js

@@ -44,14 +44,12 @@ export default class Editor extends Component {
     return state.transform().insertFragment(doc).apply();
   }
 
-  onChange = debounce(this.props.onChange, 250);
-
-  handleDocumentChange = (doc, editorState) => {
+  handleDocumentChange = debounce((doc, editorState) => {
     const raw = Raw.serialize(editorState, { terse: true });
     const plugins = this.state.shortcodePlugins;
     const markdown = slateToMarkdown(raw, plugins);
-    this.onChange(markdown);
-  };
+    this.props.onChange(markdown);
+  }, 150);
 
   hasMark = type => this.state.editorState.marks.some(mark => mark.type === type);
   hasBlock = type => this.state.editorState.blocks.some(node => node.type === type);

src/components/Widgets/Markdown/MarkdownControl/VisualEditor/plugins.js

@@ -1,3 +1,4 @@
+import { Text, Inline } from 'slate';
 import SlateSoftBreak from 'slate-soft-break';
 import EditList from 'slate-edit-list';
 import EditTable from 'slate-edit-table';
@@ -21,7 +22,13 @@ const SoftBreak = (options = {}) => ({
       return unwrapped.insertBlock(defaultBlock).apply();
     }
 
-    return state.transform().insertText('\n').apply();
+    const textNode = Text.createFromString('\n');
+    const breakNode = Inline.create({ type: 'break', nodes: [ textNode ] });
+    return state.transform()
+      .insertInline(breakNode)
+      .insertText('')
+      .collapseToStartOfNextText()
+      .apply();
   }
 });
 
@@ -32,7 +39,7 @@ const SoftBreakOpts = {
 
 export const SoftBreakConfigured = SoftBreak(SoftBreakOpts);
 
-export const ParagraphSoftBreakConfigured = SlateSoftBreak({ onlyIn: ['paragraph'], shift: true });
+export const ParagraphSoftBreakConfigured = SoftBreak({ onlyIn: ['paragraph'], shift: true });
 
 const BreakToDefaultBlock = ({ onlyIn = [], defaultBlock = 'paragraph' }) => ({
   onKeyDown(e, data, state) {

src/components/Widgets/Markdown/serializers/__tests__/remarkAllowHtmlEntities.spec.js

@@ -0,0 +1,24 @@
+import unified from 'unified';
+import markdownToRemark from 'remark-parse';
+import remarkAllowHtmlEntities from '../remarkAllowHtmlEntities';
+
+const process = markdown => {
+  const mdast = unified().use(markdownToRemark).use(remarkAllowHtmlEntities).parse(markdown);
+
+  /**
+   * The MDAST will look like:
+   *
+   * { type: 'root', children: [
+   *   { type: 'paragraph', children: [
+   *     // results here
+   *   ]}
+   * ]}
+   */
+  return mdast.children[0].children[0].value;
+};
+
+describe('remarkAllowHtmlEntities', () => {
+  it('should not decode HTML entities', () => {
+    expect(process('<div>')).toEqual('<div>');
+  });
+});

src/components/Widgets/Markdown/serializers/__tests__/remarkEscapeMarkdownEntities.spec.js

@@ -13,12 +13,36 @@ const process = text => {
 
 describe('remarkEscapeMarkdownEntities', () => {
   it('should escape common markdown entities', () => {
-    expect(process('*~`[_')).toEqual('\\*\\~\\`\\[\\_');
+    expect(process('*a*')).toEqual('\\*a\\*');
+    expect(process('**a**')).toEqual('\\*\\*a\\*\\*');
+    expect(process('***a***')).toEqual('\\*\\*\\*a\\*\\*\\*');
+    expect(process('_a_')).toEqual('\\_a\\_');
+    expect(process('__a__')).toEqual('\\_\\_a\\_\\_');
+    expect(process('~~a~~')).toEqual('\\~\\~a\\~\\~');
+    expect(process('[]')).toEqual('\\[]');
+    expect(process('[]()')).toEqual('\\[]()');
+    expect(process('[a](b)')).toEqual('\\[a](b)');
+    expect(process('[Test sentence.](https://www.example.com)'))
+      .toEqual('\\[Test sentence.](https://www.example.com)');
+    expect(process('![a](b)')).toEqual('!\\[a](b)');
+  });
+
+  it('should not escape inactive, single markdown entities', () => {
+    expect(process('a*b')).toEqual('a*b');
+    expect(process('_')).toEqual('_');
+    expect(process('~')).toEqual('~');
+    expect(process('[')).toEqual('[');
   });
 
   it('should escape leading markdown entities', () => {
     expect(process('#')).toEqual('\\#');
     expect(process('-')).toEqual('\\-');
+    expect(process('*')).toEqual('\\*');
+    expect(process('>')).toEqual('\\>');
+    expect(process('=')).toEqual('\\=');
+    expect(process('|')).toEqual('\\|');
+    expect(process('```')).toEqual('\\`\\``');
+    expect(process('    ')).toEqual('\\    ');
   });
 
   it('should escape leading markdown entities preceded by whitespace', () => {
@@ -30,4 +54,25 @@ describe('remarkEscapeMarkdownEntities', () => {
     expect(process('a# # b #')).toEqual('a# # b #');
     expect(process('a- - b -')).toEqual('a- - b -');
   });
+
+  it('should not escape html tags', () => {
+    expect(process('<a attr="**a**">')).toEqual('<a attr="**a**">');
+    expect(process('a b <c attr="**d**"> e')).toEqual('a b <c attr="**d**"> e');
+  });
+
+  it('should escape the contents of html blocks', () => {
+    expect(process('<div>*a*</div>')).toEqual('<div>\\*a\\*</div>');
+  });
+
+  it('should not escape the contents of preformatted html blocks', () => {
+    expect(process('<pre>*a*</pre>')).toEqual('<pre>*a*</pre>');
+    expect(process('<script>*a*</script>')).toEqual('<script>*a*</script>');
+    expect(process('<style>*a*</style>')).toEqual('<style>*a*</style>');
+    expect(process('<pre>\n*a*\n</pre>')).toEqual('<pre>\n*a*\n</pre>');
+    expect(process('a b <pre>*c*</pre> d e')).toEqual('a b <pre>*c*</pre> d e');
+  });
+
+  it('should not parse footnotes', () => {
+    expect(process('[^a]')).toEqual('\\[^a]');
+  });
 });

src/components/Widgets/Markdown/serializers/__tests__/remarkStripTrailingBreaks.spec.js

@@ -0,0 +1,24 @@
+import unified from 'unified';
+import u from 'unist-builder';
+import remarkStripTrailingBreaks from '../remarkStripTrailingBreaks';
+
+const process = children => {
+  const tree = u('root', children);
+  const strippedMdast = unified()
+    .use(remarkStripTrailingBreaks)
+    .runSync(tree);
+
+  return strippedMdast.children;
+};
+
+describe('remarkStripTrailingBreaks', () => {
+  it('should remove trailing breaks at the end of a block', () => {
+    expect(process([u('break')])).toEqual([]);
+    expect(process([u('break'), u('text', '\n  \n')])).toEqual([u('text', '\n  \n')]);
+    expect(process([u('text', 'a'), u('break')])).toEqual([u('text', 'a')]);
+  });
+
+  it('should not remove trailing breaks that are not at the end of a block', () => {
+    expect(process([u('break'), u('text', 'a')])).toEqual([u('break'), u('text', 'a')]);
+  });
+});

src/components/Widgets/Markdown/serializers/__tests__/slate.spec.js

@@ -0,0 +1,40 @@
+import { flow } from 'lodash';
+import { markdownToSlate, slateToMarkdown } from '../index';
+
+const process = flow([markdownToSlate, slateToMarkdown]);
+
+describe('slate', () => {
+  it('should distinguish between newlines and hard breaks', () => {
+    expect(process('a\n')).toEqual('a\n');
+  });
+
+  it('should not decode encoded html entities in inline code', () => {
+    expect(process('<code>&lt;div&gt;</code>')).toEqual('<code>&lt;div&gt;</code>\n');
+  });
+
+  it('should parse non-text children of mark nodes', () => {
+    expect(process('**a[b](c)d**')).toEqual('**a[b](c)d**\n');
+    expect(process('**[a](b)**')).toEqual('**[a](b)**\n');
+    expect(process('**![a](b)**')).toEqual('**![a](b)**\n');
+    expect(process('_`a`_')).toEqual('_`a`_\n');
+    expect(process('_`a`b_')).toEqual('_`a`b_\n');
+  });
+
+  it('should condense adjacent, identically styled text and inline nodes', () => {
+    expect(process('**a ~~b~~~~c~~**')).toEqual('**a ~~bc~~**\n');
+    expect(process('**a ~~b~~~~[c](d)~~**')).toEqual('**a ~~b[c](d)~~**\n');
+  });
+
+  it('should handle nested markdown entities', () => {
+    expect(process('**a**b**c**')).toEqual('**a**b**c**\n');
+    expect(process('**a _b_ c**')).toEqual('**a _b_ c**\n');
+  });
+
+  it('should parse inline images as images', () => {
+    expect(process('a ![b](c)')).toEqual('a ![b](c)\n');
+  });
+
+  it('should not escape markdown entities in html', () => {
+    expect(process('<span>*</span>')).toEqual('<span>*</span>\n');
+  });
+});

src/components/Widgets/Markdown/serializers/index.js

@@ -16,7 +16,9 @@ import remarkToSlate from './remarkSlate';
 import remarkSquashReferences from './remarkSquashReferences';
 import remarkImagesToText from './remarkImagesToText';
 import remarkShortcodes from './remarkShortcodes';
-import remarkEscapeMarkdownEntities from './remarkEscapeMarkdownEntities'
+import remarkEscapeMarkdownEntities from './remarkEscapeMarkdownEntities';
+import remarkStripTrailingBreaks from './remarkStripTrailingBreaks';
+import remarkAllowHtmlEntities from './remarkAllowHtmlEntities';
 import slateToRemark from './slateRemark';
 import registry from '../../../../lib/registry';
 
@@ -63,7 +65,9 @@ export const markdownToRemark = markdown => {
    * Parse the Markdown string input to an MDAST.
    */
   const parsed = unified()
-    .use(markdownToRemarkPlugin, { fences: true, pedantic: true, commonmark: true })
+    .use(markdownToRemarkPlugin, { fences: true, commonmark: true })
+    .use(markdownToRemarkRemoveTokenizers, { inlineTokenizers: ['url'] })
+    .use(remarkAllowHtmlEntities)
     .parse(markdown);
 
   /**
@@ -79,6 +83,16 @@ export const markdownToRemark = markdown => {
 };
 
 
+/**
+ * Remove named tokenizers from the parser, effectively deactivating them.
+ */
+function markdownToRemarkRemoveTokenizers({ inlineTokenizers }) {
+  inlineTokenizers && inlineTokenizers.forEach(tokenizer => {
+    delete this.Parser.prototype.inlineTokenizers[tokenizer];
+  });
+}
+
+
 /**
  * Serialize an MDAST to a Markdown string.
  */
@@ -102,7 +116,6 @@ export const remarkToMarkdown = obj => {
   const remarkToMarkdownPluginOpts = {
     commonmark: true,
     fences: true,
-    pedantic: true,
     listItemIndent: '1',
 
     // Settings to emulate the defaults from the Prosemirror editor, not
@@ -117,6 +130,7 @@ export const remarkToMarkdown = obj => {
    */
   const escapedMdast = unified()
     .use(remarkEscapeMarkdownEntities)
+    .use(remarkStripTrailingBreaks)
     .runSync(mdast);
 
   const markdown = unified()

src/components/Widgets/Markdown/serializers/remarkAllowHtmlEntities.js

@@ -0,0 +1,59 @@
+export default function remarkAllowHtmlEntities() {
+  this.Parser.prototype.inlineTokenizers.text = text;
+
+  /**
+   * This is a port of the `remark-parse` text tokenizer, adapted to exclude
+   * HTML entity decoding.
+   */
+  function text(eat, value, silent) {
+    var self = this;
+    var methods;
+    var tokenizers;
+    var index;
+    var length;
+    var subvalue;
+    var position;
+    var tokenizer;
+    var name;
+    var min;
+    var now;
+
+    /* istanbul ignore if - never used (yet) */
+    if (silent) {
+      return true;
+    }
+
+    methods = self.inlineMethods;
+    length = methods.length;
+    tokenizers = self.inlineTokenizers;
+    index = -1;
+    min = value.length;
+
+    while (++index < length) {
+      name = methods[index];
+
+      if (name === 'text' || !tokenizers[name]) {
+        continue;
+      }
+
+      tokenizer = tokenizers[name].locator;
+
+      if (!tokenizer) {
+        eat.file.fail('Missing locator: `' + name + '`');
+      }
+
+      position = tokenizer.call(self, value, 1);
+
+      if (position !== -1 && position < min) {
+        min = position;
+      }
+    }
+
+    subvalue = value.slice(0, min);
+
+    eat(subvalue)({
+      type: 'text',
+      value: subvalue,
+    });
+  }
+};

src/components/Widgets/Markdown/serializers/remarkEscapeMarkdownEntities.js

@@ -1,3 +1,237 @@
+import { flow, partial, flatMap, flatten, map } from 'lodash';
+import { joinPatternSegments, combinePatterns, replaceWhen } from '../../../../lib/regexHelper';
+
+/**
+ * Reusable regular expressions segments.
+ */
+const patternSegments = {
+  /**
+   * Matches zero or more HTML attributes followed by the tag close bracket,
+   * which may be prepended by zero or more spaces.  The attributes can use
+   * single or double quotes and may be prepended by zero or more spaces.
+   */
+  htmlOpeningTagEnd: /(?: *\w+=(?:(?:"[^"]*")|(?:'[^']*')))* *>/,
+};
+
+
+/**
+ * Patterns matching substrings that should not be escaped. Array values must be
+ * joined before use.
+ */
+const nonEscapePatterns = {
+  /**
+   * HTML Tags
+   *
+   * Matches HTML opening tags and any attributes. Does not check for contents
+   * between tags or closing tags.
+   */
+  htmlTags: [
+    /**
+     * Matches the beginning of an HTML tag, excluding preformatted tag types.
+     */
+    /<(?!pre|style|script)[\w]+/,
+
+    /**
+     * Matches attributes.
+     */
+    patternSegments.htmlOpeningTagEnd,
+  ],
+
+
+  /**
+   * Preformatted HTML Blocks
+   *
+   * Matches HTML blocks with preformatted content. The content of these blocks,
+   * including the tags and attributes, should not be escaped at all.
+   */
+  preformattedHtmlBlocks: [
+    /**
+     * Matches the names of tags known to have preformatted content. The capture
+     * group is reused when matching the closing tag.
+     *
+     * NOTE: this pattern reuses a capture group, and could break if combined with
+     * other expressions using capture groups.
+     */
+    /<(pre|style|script)/,
+
+    /**
+     * Matches attributes.
+     */
+    patternSegments.htmlOpeningTagEnd,
+
+    /**
+     * Allow zero or more of any character (including line breaks) between the
+     * tags. Match lazily in case of subsequent blocks.
+     */
+    /(.|[\n\r])*?/,
+
+    /**
+     * Match closing tag via first capture group.
+     */
+    /<\/\1>/,
+  ],
+};
+
+
+/**
+ * Escape patterns
+ *
+ * Each escape pattern matches a markdown entity and captures up to two
+ * groups. These patterns must use one of the following formulas:
+ *
+ * - Single capture group followed by match content - /(...).../
+ *   The captured characters should be escaped and the remaining match should
+ *   remain unchanged.
+ *
+ * - Two capture groups surrounding matched content - /(...)...(...)/
+ *   The captured characters in both groups should be escaped and the matched
+ *   characters in between should remain unchanged.
+ */
+const escapePatterns = [
+  /**
+   * Emphasis/Bold - Asterisk
+   *
+   * Match strings surrounded by one or more asterisks on both sides.
+   */
+  /(\*+)[^\*]*(\1)/g,
+
+  /**
+   * Emphasis - Underscore
+   *
+   * Match strings surrounded by a single underscore on both sides followed by
+   * a word boundary. Remark disregards whether a word boundary exists at the
+   * beginning of an emphasis node.
+   */
+  /(_)[^_]+(_)\b/g,
+
+  /**
+   * Bold - Underscore
+   *
+   * Match strings surrounded by multiple underscores on both sides. Remark
+   * disregards the absence of word boundaries on either side of a bold node.
+   */
+  /(_{2,})[^_]*(\1)/g,
+
+  /**
+   * Strikethrough
+   *
+   * Match strings surrounded by multiple tildes on both sides.
+   */
+  /(~+)[^~]*(\1)/g,
+
+  /**
+   * Inline Code
+   *
+   * Match strings surrounded by backticks.
+   */
+  /(`+)[^`]*(\1)/g,
+
+  /**
+   * Links, Images, References, and Footnotes
+   *
+   * Match strings surrounded by brackets. This could be improved to
+   * specifically match only the exact syntax of each covered entity, but
+   * doing so through current approach would incur a considerable performance
+   * penalty.
+   */
+  /(\[)[^\]]*]/g,
+];
+
+
+/**
+ * Generate new non-escape expression. The non-escape expression matches
+ * substrings whose contents should not be processed for escaping.
+ */
+const joinedNonEscapePatterns = map(nonEscapePatterns, pattern => {
+  return new RegExp(joinPatternSegments(pattern));
+});
+const nonEscapePattern = combinePatterns(joinedNonEscapePatterns);
+
+
+/**
+ * Create chain of successive escape functions for various markdown entities.
+ */
+const escapeFunctions = escapePatterns.map(pattern => partial(escapeDelimiters, pattern));
+const escapeAll = flow(escapeFunctions);
+
+
+/**
+ * Executes both the `escapeCommonChars` and `escapeLeadingChars` functions.
+ */
+function escapeAllChars(text) {
+  const partiallyEscapedMarkdown = escapeCommonChars(text);
+  return escapeLeadingChars(partiallyEscapedMarkdown);
+}
+
+
+/**
+ * escapeLeadingChars
+ *
+ * Handles escaping for characters that must be positioned at the beginning of
+ * the string, such as headers and list items.
+ *
+ * Escapes '#', '*', '-', '>', '=', '|', and sequences of 3+ backticks or 4+
+ * spaces when found at the beginning of a string, preceded by zero or more
+ * whitespace characters.
+ */
+function escapeLeadingChars(text) {
+  return text.replace(/^\s*([-#*>=|]| {4,}|`{3,})/, '$`\\$1');
+}
+
+
+/**
+ * escapeCommonChars
+ *
+ * Escapes active markdown entities. See escape pattern groups for details on
+ * which entities are replaced.
+ */
+function escapeCommonChars(text) {
+  /**
+   * Generate new non-escape expression (must happen at execution time because
+   * we use `RegExp.exec`, which tracks it's own state internally).
+   */
+  const nonEscapeExpression = new RegExp(nonEscapePattern, 'gm');
+
+  /**
+   * Use `replaceWhen` to escape markdown entities only within substrings that
+   * are eligible for escaping.
+   */
+  return replaceWhen(nonEscapeExpression, escapeAll, text, true);
+}
+
+
+/**
+ * escapeDelimiters
+ *
+ * Executes `String.replace` for a given pattern, but only on the first two
+ * capture groups. Specifically intended for escaping opening (and optionally
+ * closing) markdown entities without escaping the content in between.
+ */
+function escapeDelimiters(pattern, text) {
+  return text.replace(pattern, (match, start, end) => {
+    const hasEnd = typeof end === 'string';
+    const matchSliceEnd = hasEnd ? match.length - end.length : match.length;
+    const content = match.slice(start.length, matchSliceEnd);
+    return `${escape(start)}${content}${hasEnd ? escape(end) : ''}`;
+  });
+}
+
+
+/**
+ * escape
+ *
+ * Simple replacement function for escaping markdown entities. Prepends every
+ * character in the received string with a backslash.
+ */
+function escape(delim) {
+  let result = '';
+  for (const char of delim) {
+    result += `\\${char}`;
+  }
+  return result;
+}
+
+
 /**
  * A Remark plugin for escaping markdown entities.
  *
@@ -13,22 +247,6 @@
  * stringification.
  */
 export default function remarkEscapeMarkdownEntities() {
-  /**
-   * Escape all occurrences of '[', '*', '_', '`', and '~'.
-   */
-  function escapeCommonChars(text) {
-    return text.replace(/[\[*_`~]/g, '\\$&');
-  }
-
-  /**
-   * Runs escapeCommonChars, and also escapes '#' and '-' when found at the
-   * beginning of any node's first child node.
-   */
-  function escapeAllChars(text) {
-    const partiallyEscapedMarkdown = escapeCommonChars(text);
-    return partiallyEscapedMarkdown.replace(/^\s*([#-])/, '$`\\$1');
-  }
-
   const transform = (node, index) => {
     const children = node.children && node.children.map(transform);
 

src/components/Widgets/Markdown/serializers/remarkImagesToText.js

@@ -1,18 +1,26 @@
 /**
  * Images must be parsed as shortcodes for asset proxying. This plugin converts
- * MDAST image nodes back to text to allow shortcode pattern matching.
+ * MDAST image nodes back to text to allow shortcode pattern matching. Note that
+ * this transformation only occurs for images that are the sole child of a top
+ * level paragraph - any other image is left alone and treated as an inline
+ * image.
  */
 export default function remarkImagesToText() {
   return transform;
 
   function transform(node) {
-    const children = node.children ? node.children.map(transform) : node.children;
-    if (node.type === 'image') {
-      const alt = node.alt || '';
-      const url = node.url || '';
-      const title = node.title ? ` "${node.title}"` : '';
-      return { type: 'text', value: `![${alt}](${url}${title})` };
+    const children = node.children.map(child => {
+      if (
+        child.type === 'paragraph'
+        && child.children.length === 1
+        && child.children[0].type === 'image'
+      ) {
+        const { alt = '', url = '', title = '' } = child.children[0];
+        const value = `![${alt}](${url}${title ? ' title' : ''})`;
+        child.children = [{ type: 'text', value }];
       }
+      return child;
+    });
     return { ...node, children };
   }
 }

src/components/Widgets/Markdown/serializers/remarkSlate.js

@@ -1,6 +1,33 @@
-import { get, isEmpty, isArray } from 'lodash';
+import { get, isEmpty, isArray, last, flatMap } from 'lodash';
 import u from 'unist-builder';
 
+/**
+ * A Remark plugin for converting an MDAST to Slate Raw AST. Remark plugins
+ * return a `transform` function that receives the MDAST as it's first argument.
+ */
+export default function remarkToSlate() {
+  return transform;
+}
+
+function transform(node) {
+
+  /**
+   * Call `transform` recursively on child nodes.
+   *
+   * If a node returns a falsey value, filter it out. Some nodes do not
+   * translate from MDAST to Slate, such as definitions for link/image
+   * references or footnotes.
+   */
+  const children = !['strong', 'emphasis', 'delete'].includes(node.type)
+    && !isEmpty(node.children)
+    && flatMap(node.children, transform).filter(val => val);
+
+  /**
+   * Run individual nodes through the conversion factory.
+   */
+  return convertNode(node, children);
+}
+
 /**
  * Map of MDAST node types to Slate node types.
  */
@@ -47,7 +74,7 @@ function createBlock(type, nodes, props = {}) {
 /**
  * Create a Slate Block node.
  */
-function createInline(type, nodes, props = {}) {
+function createInline(type, props = {}, nodes) {
   return { kind: 'inline', type, nodes, ...props };
 }
 
@@ -63,8 +90,7 @@ function createText(value, data) {
   return {...node, text: value };
 }
 
-function convertMarkNode(node, parentMarks = []) {
-
+function processMarkNode(node, parentMarks = []) {
   /**
    * Add the current node's mark type to the marks collected from parent
    * mark nodes, if any.
@@ -72,34 +98,69 @@ function convertMarkNode(node, parentMarks = []) {
   const markType = markMap[node.type];
   const marks = markType ? [...parentMarks, { type: markMap[node.type] }] : parentMarks;
 
-  /**
-   * Set an array to collect sections of text.
-   */
-  const ranges = [];
-
-  node.children && node.children.forEach(childNode => {
-
+  const children = flatMap(node.children, childNode => {
+    switch (childNode.type) {
       /**
        * If a text node is a direct child of the current node, it should be
        * set aside as a range, and all marks that have been collected in the
        * `marks` array should apply to that specific range.
        */
-    if (['html', 'text'].includes(childNode.type)) {
-      ranges.push({ text: childNode.value, marks });
-      return;
+      case 'html':
+      case 'text':
+        return { text: childNode.value, marks };
+
+      /**
+       * MDAST inline code nodes don't have children, just a text value, similar
+       * to a text node, so it receives the same treatment as a text node, but we
+       * first add the inline code mark to the marks array.
+       */
+      case 'inlineCode': {
+        const childMarks = [ ...marks, { type: markMap['inlineCode'] } ];
+        return { text: childNode.value, marks: childMarks };
       }
 
       /**
-     * Any non-text child node should be processed as a parent node. The
-     * recursive results should be pushed into the ranges array. This way,
-     * every MDAST nested text structure becomes a flat array of ranges
-     * that can serve as the value of a single Slate Raw text node.
+       * Process nested style nodes. The recursive results should be pushed into
+       * the ranges array. This way, every MDAST nested text structure becomes a
+       * flat array of ranges that can serve as the value of a single Slate Raw
+       * text node.
        */
-    const nestedRanges = convertMarkNode(childNode, marks);
-    ranges.push(...nestedRanges);
+      case 'strong':
+      case 'emphasis':
+      case 'delete':
+        return processMarkNode(childNode, marks);
+
+      /**
+       * Remaining nodes simply need mark data added to them, and to then be
+       * added into the cumulative children array.
+       */
+      default:
+        return { ...childNode, data: { marks } };
+    }
   });
 
-  return ranges;
+  return children;
+}
+
+function convertMarkNode(node) {
+  const slateNodes = processMarkNode(node);
+
+  const convertedSlateNodes = slateNodes.reduce((acc, node, idx, nodes) => {
+    const lastConvertedNode = last(acc);
+    if (node.text && lastConvertedNode && lastConvertedNode.ranges) {
+      lastConvertedNode.ranges.push(node);
+    }
+    else if (node.text) {
+      acc.push(createText([node]));
+    }
+    else {
+      acc.push(transform(node));
+    }
+
+    return acc;
+  }, []);
+
+  return convertedSlateNodes;
 }
 
 /**
@@ -186,7 +247,7 @@ function convertNode(node, nodes) {
     case 'strong':
     case 'emphasis':
     case 'delete': {
-      return createText(convertMarkNode(node));
+      return convertMarkNode(node);
     }
 
     /**
@@ -239,7 +300,8 @@ function convertNode(node, nodes) {
      * line breaks within a text node.
      */
     case 'break': {
-      return createText('\n');
+      const textNode = createText('\n');
+      return createInline('break', {}, [ textNode ]);
     }
 
     /**
@@ -258,11 +320,25 @@ function convertNode(node, nodes) {
      * schema references them in the data object.
      */
     case 'link': {
-      const { title, url } = node;
-      const data = { title, url };
-      return createInline(typeMap[type], nodes, { data });
+      const { title, url, data } = node;
+      const newData = { ...data, title, url };
+      return createInline(typeMap[type], { data: newData }, nodes);
     }
 
+    /**
+     * Images
+     *
+     * Identical to link nodes except for the lack of child nodes and addition
+     * of alt attribute data MDAST stores the link attributes directly on the
+     * node, while our Slate schema references them in the data object.
+     */
+    case 'image': {
+      const { title, url, alt, data } = node;
+      const newData = { ...data, title, alt, url };
+      return createInline(typeMap[type], { isVoid: true, data: newData });
+    }
+
+
     /**
      * Tables
      *
@@ -275,29 +351,3 @@ function convertNode(node, nodes) {
     }
   }
 }
-
-
-/**
- * A Remark plugin for converting an MDAST to Slate Raw AST. Remark plugins
- * return a `transform` function that receives the MDAST as it's first argument.
- */
-export default function remarkToSlate() {
-  function transform(node) {
-
-    /**
-     * Call `transform` recursively on child nodes.
-     *
-     * If a node returns a falsey value, filter it out. Some nodes do not
-     * translate from MDAST to Slate, such as definitions for link/image
-     * references or footnotes.
-     */
-    const children = !isEmpty(node.children) && node.children.map(transform).filter(val => val);
-
-    /**
-     * Run individual nodes through the conversion factory.
-     */
-    return convertNode(node, children);
-  }
-
-  return transform;
-}

src/components/Widgets/Markdown/serializers/remarkStripTrailingBreaks.js

@@ -0,0 +1,56 @@
+import mdastToString from 'mdast-util-to-string';
+
+/**
+ * Removes break nodes that are at the end of a block.
+ *
+ * When a trailing double space or backslash is encountered at the end of a
+ * markdown block, Remark will interpret the character(s) literally, as only
+ * break entities followed by text qualify as breaks. A manually created MDAST,
+ * however, may have such entities, and users of visual editors shouldn't see
+ * these artifacts in resulting markdown.
+ */
+export default function remarkStripTrailingBreaks() {
+  const transform = node => {
+    if (node.children) {
+      node.children = node.children
+        .map((child, idx, children) => {
+
+          /**
+           * Only touch break nodes. Convert all subsequent nodes to their text
+           * value and exclude the break node if no non-whitespace characters
+           * are found.
+           */
+          if (child.type === 'break') {
+            const subsequentNodes = children.slice(idx + 1);
+
+            /**
+             * Create a small MDAST so that mdastToString can process all
+             * siblings as children of one node rather than making multiple
+             * calls.
+             */
+            const fragment = { type: 'root', children: subsequentNodes };
+            const subsequentText = mdastToString(fragment);
+            return subsequentText.trim() ? child : null;
+          }
+
+          /**
+           * Always return the child if not a break.
+           */
+          return child;
+        })
+
+        /**
+         * Because some break nodes may be excluded, we filter out the resulting
+         * null values.
+         */
+        .filter(child => child)
+
+        /**
+         * Recurse through the MDAST by transforming each individual child node.
+         */
+        .map(transform);
+    }
+    return node;
+  };
+  return transform;
+};

src/components/Widgets/Markdown/serializers/slateRemark.js

@@ -1,4 +1,4 @@
-import { get, isEmpty, concat, without, flatten, flatMap, initial } from 'lodash';
+import { get, isEmpty, concat, without, flatten, flatMap, initial, last, difference, reverse, sortBy } from 'lodash';
 import u from 'unist-builder';
 
 /**
@@ -21,6 +21,7 @@ const typeMap = {
   'table': 'table',
   'table-row': 'tableRow',
   'table-cell': 'tableCell',
+  'break': 'break',
   'thematic-break': 'thematicBreak',
   'link': 'link',
   'image': 'image',
@@ -37,6 +38,108 @@ const markMap = {
   code: 'inlineCode',
 };
 
+let shortcodePlugins;
+
+export default function slateToRemark(raw, opts) {
+  /**
+   * Set shortcode plugins in outer scope.
+   */
+  ({ shortcodePlugins } = opts);
+
+  /**
+   * The Slate Raw AST generally won't have a top level type, so we set it to
+   * "root" for clarity.
+   */
+  raw.type = 'root';
+
+  return transform(raw);
+}
+
+
+/**
+ * The transform function mimics the approach of a Remark plugin for
+ * conformity with the other serialization functions. This function converts
+ * Slate nodes to MDAST nodes, and recursively calls itself to process child
+ * nodes to arbitrary depth.
+ */
+function transform(node) {
+  /**
+   * Combine adjacent text and inline nodes before processing so they can
+   * share marks.
+   */
+  const combinedChildren = node.nodes && combineTextAndInline(node.nodes);
+
+  /**
+   * Call `transform` recursively on child nodes, and flatten the resulting
+   * array.
+   */
+  const children = !isEmpty(combinedChildren) && flatMap(combinedChildren, transform);
+
+  /**
+   * Run individual nodes through conversion factories.
+   */
+  return ['text'].includes(node.kind)
+    ? convertTextNode(node)
+    : convertNode(node, children, shortcodePlugins);
+}
+
+
+/**
+ * Includes inline nodes as ranges in adjacent text nodes where appropriate, so
+ * that mark node combining logic can apply to both text and inline nodes. This
+ * is necessary because Slate doesn't allow inline nodes to have marks while
+ * inline nodes in MDAST may be nested within mark nodes. Treating them as if
+ * they were text is a bit of a necessary hack.
+ */
+function combineTextAndInline(nodes) {
+  return nodes.reduce((acc, node, idx, nodes) => {
+    const prevNode = last(acc);
+    const prevNodeRanges = get(prevNode, 'ranges');
+    const data = node.data || {};
+
+    /**
+     * If the previous node has ranges and the current node has marks in data
+     * (only happens when we place them on inline nodes here in the parser), or
+     * the current node also has ranges (because the previous node was
+     * originally an inline node that we've already squashed into a range)
+     * combine the current node into the previous.
+     */
+    if (!isEmpty(prevNodeRanges) && !isEmpty(data.marks)) {
+      prevNodeRanges.push({ node, marks: data.marks });
+      return acc;
+    }
+
+    if (!isEmpty(prevNodeRanges) && !isEmpty(node.ranges)) {
+      prevNode.ranges = prevNodeRanges.concat(node.ranges);
+      return acc;
+    }
+
+    /**
+     * Break nodes contain a single child text node with a newline character
+     * for visual purposes in the editor, but Remark break nodes have no
+     * children, so we remove the child node here.
+     */
+    if (node.type === 'break') {
+      acc.push({ kind: 'inline', type: 'break' });
+      return acc;
+    }
+
+    /**
+     * Convert remaining inline nodes to standalone text nodes with ranges.
+     */
+    if (node.kind === 'inline') {
+      acc.push({ kind: 'text', ranges: [{ node, marks: data.marks }] });
+      return acc;
+    }
+
+    /**
+     * Only remaining case is an actual text node, can be pushed as is.
+     */
+    acc.push(node);
+    return acc;
+  }, []);
+}
+
 
 /**
  * Slate treats inline code decoration as a standard mark, but MDAST does
@@ -56,24 +159,6 @@ function processCodeMark(markTypes) {
 }
 
 
-/**
- * Returns an array of one or more MDAST text nodes of the given type, derived
- * from the text received. Certain transformations, such as line breaks, cause
- * multiple nodes to be returned.
- */
-function createTextNodes(text, type = 'html') {
-  /**
-   * Split the text string at line breaks, then map each substring to an array
-   * pair consisting of an MDAST text node followed by a break node. This will
-   * result in nested arrays, so we use `flatMap` to produce a flattened array,
-   * and `initial` to leave off the superfluous trailing break.
-   */
-  const brokenText = text.split('\n');
-  const toPair = str => [u(type, str), u('break')];
-  return initial(flatMap(brokenText, toPair));
-}
-
-
 /**
  * Wraps a text node in one or more mark nodes by placing the text node in an
  * array and using that as the `children` value of a mark node. The resulting
@@ -142,20 +227,29 @@ function wrapTextWithMarks(textNode, markTypes) {
  * replaced with multiple MDAST nodes, so the resulting array must be flattened.
  */
 function convertTextNode(node) {
-
   /**
-   * If the Slate text node has no "ranges" property, just return an equivalent
-   * MDAST node.
+   * If the Slate text node has a "ranges" property, translate the Slate AST to
+   * a nested MDAST structure. Otherwise, just return an equivalent MDAST text
+   * node.
    */
-  if (!node.ranges) {
-    return createTextNodes(node.text);
+  if (node.ranges) {
+    const processedRanges = node.ranges.map(processRanges);
+    const condensedNodes = processedRanges.reduce(condenseNodesReducer, { nodes: [] });
+    return condensedNodes.nodes;
   }
 
+  if (node.kind === 'inline') {
+    return transform(node);
+  }
+
+  return u('html', node.text);
+}
+
+
 /**
-   * If there is no "text" property, convert the text range(s) to an array of
-   * one or more nested MDAST nodes.
+ * Process Slate node ranges in preparation for MDAST transformation.
  */
-  const textNodes = node.ranges.map(range => {
+function processRanges(range) {
   /**
    * Get an array of the mark types, converted to their MDAST equivalent
    * types.
@@ -163,29 +257,104 @@ function convertTextNode(node) {
   const { marks = [], text } = range;
   const markTypes = marks.map(mark => markMap[mark.type]);
 
+  if (typeof range.text === 'string') {
     /**
      * Code marks must be removed from the marks array, and the presence of a
      * code mark changes the text node type that should be used.
      */
     const { filteredMarkTypes, textNodeType } = processCodeMark(markTypes);
+    return { text, marks: filteredMarkTypes, textNodeType };
+  }
+
+  return { node: range.node, marks: markTypes };
+}
+
 
 /**
-     * Create the base text node.
+ * Slate's AST doesn't group adjacent text nodes with the same marks - a
+ * change in marks from letter to letter, even if some are in common, results
+ * in a separate range. For example, given "**a_b_**", transformation to and
+ * from Slate's AST will result in "**a****_b_**".
+ *
+ * MDAST treats styling entities as distinct nodes that contain children, so a
+ * "strong" node can contain a plain text node with a sibling "emphasis" node,
+ * which contains more text. This reducer serves to create an optimized nested
+ * MDAST without the typical redundancies that Slate's AST would produce if
+ * transformed as-is. The reducer can be called recursively to produce nested
+ * structures.
  */
-    const textNodes = createTextNodes(text, textNodeType);
+function condenseNodesReducer(acc, node, idx, nodes) {
+  /**
+   * Skip any nodes that are being processed as children of an MDAST node
+   * through recursive calls.
+   */
+  if (typeof acc.nextIndex === 'number' && acc.nextIndex > idx) {
+    return acc;
+  }
+
+  /**
+   * Processing for nodes with marks.
+   */
+  if (node.marks && node.marks.length > 0) {
+    /**
+     * For each mark on the current node, get the number of consecutive nodes
+     * (starting with this one) that have the mark. Whichever mark covers the
+     * most nodes is used as the parent node, and the nodes with that mark are
+     * processed as children. If the greatest number of consecutive nodes is
+     * tied between multiple marks, there is no priority as to which goes
+     * first.
+     */
+    const markLengths = node.marks.map(mark => getMarkLength(mark, nodes.slice(idx)));
+    const parentMarkLength = last(sortBy(markLengths, 'length'));
+    const { markType: parentType, length: parentLength } = parentMarkLength;
+
+    /**
+     * Since this and any consecutive nodes with the parent mark are going to
+     * be processed as children of the parent mark, this reducer should simply
+     * return the accumulator until after the last node to be covered by the
+     * new parent node. Here we set the next index that should be processed,
+     * if any.
+     */
+    const newNextIndex = idx + parentLength;
+
+    /**
+     * Get the set of nodes that should be processed as children of the new
+     * parent mark node, run each through the reducer as children of the
+     * parent node, and create the parent MDAST node with the resulting
+     * children.
+     */
+    const children = nodes.slice(idx, newNextIndex);
+    const denestedChildren = children.map(child => ({ ...child, marks: without(child.marks, parentType) }));
+    const mdastChildren = denestedChildren.reduce(condenseNodesReducer, { nodes: [], parentType }).nodes;
+    const mdastNode = u(parentType, mdastChildren);
+
+    return { ...acc, nodes: [ ...acc.nodes, mdastNode ], nextIndex: newNextIndex };
+  }
+
+  /**
+   * Create the base text node, and pass in the array of mark types as data
+   * (helpful when optimizing/condensing the final structure).
+   */
+  const baseNode = typeof node.text === 'string'
+    ? u(node.textNodeType, { marks: node.marks }, node.text)
+    : transform(node.node);
 
   /**
    * Recursively wrap the base text node in the individual mark nodes, if
    * any exist.
    */
-    return textNodes.map(textNode => wrapTextWithMarks(textNode, filteredMarkTypes));
-  });
+  return { ...acc, nodes: [ ...acc.nodes, baseNode ] };
+}
+
 
 /**
-   * Since each range will be mapped into an array, we flatten the result to
-   * return a single array of all nodes.
+ * Get the number of consecutive Slate nodes containing a given mark beginning
+ * from the first received node.
  */
-  return flatten(textNodes);
+function getMarkLength(markType, nodes) {
+  let length = 0;
+  while(nodes[length] && nodes[length].marks.includes(markType)) { ++length; }
+  return { markType, length };
 }
 
 
@@ -267,8 +436,8 @@ function convertNode(node, children, shortcodePlugins) {
      */
     case 'code': {
       const value = get(node.nodes, [0, 'text']);
-      const lang = get(node.data, 'lang');
-      return u(typeMap[node.type], { lang }, value);
+      const { lang, ...data } = get(node, 'data', {});
+      return u(typeMap[node.type], { lang, data }, value);
     }
 
     /**
@@ -288,11 +457,11 @@ function convertNode(node, children, shortcodePlugins) {
     }
 
     /**
-     * Thematic Breaks
+     * Breaks
      *
-     * Thematic breaks don't have children. We parse them separately for
-     * clarity.
+     * Breaks don't have children. We parse them separately for clarity.
      */
+    case 'break':
     case 'thematic-break': {
       return u(typeMap[node.type]);
     }
@@ -304,45 +473,28 @@ function convertNode(node, children, shortcodePlugins) {
      * the node for both Slate and Remark schemas.
      */
     case 'link': {
-      const { url, title } = get(node, 'data', {});
-      return u(typeMap[node.type], { url, title }, children);
+      const { url, title, ...data } = get(node, 'data', {});
+      return u(typeMap[node.type], { url, title, data }, children);
     }
 
+    /**
+     * Images
+     *
+     * This transformation is almost identical to that of links, except for the
+     * lack of child nodes and addition of `alt` attribute data. Currently the
+     * CMS handles block images by shortcode, so this case will only apply to
+     * inline images, which currently can only occur through raw markdown
+     * insertion.
+     */
+    case 'image': {
+      const { url, title, alt, ...data } = get(node, 'data', {});
+      return u(typeMap[node.type], { url, title, alt, data });
+    }
+
+
     /**
      * No default case is supplied because an unhandled case should never
      * occur. In the event that it does, let the error throw (for now).
      */
   }
 }
-
-
-export default function slateToRemark(raw, { shortcodePlugins }) {
-  /**
-   * The transform function mimics the approach of a Remark plugin for
-   * conformity with the other serialization functions. This function converts
-   * Slate nodes to MDAST nodes, and recursively calls itself to process child
-   * nodes to arbitrary depth.
-   */
-  function transform(node) {
-
-    /**
-     * Call `transform` recursively on child nodes, and flatten the resulting
-     * array.
-     */
-    const children = !isEmpty(node.nodes) && flatten(node.nodes.map(transform));
-
-    /**
-     * Run individual nodes through conversion factories.
-     */
-    return node.kind === 'text' ? convertTextNode(node) : convertNode(node, children, shortcodePlugins);
-  }
-
-  /**
-   * The Slate Raw AST generally won't have a top level type, so we set it to
-   * "root" for clarity.
-   */
-  raw.type = 'root';
-
-  const mdast = transform(raw);
-  return mdast;
-}

src/components/Widgets/ObjectControl.js

@@ -21,6 +21,13 @@ export default class ObjectControl extends Component {
     className: PropTypes.string,
   };
 
+  /**
+   * In case the `onChange` function is frozen by a child widget implementation,
+   * e.g. when debounced, always get the latest object value instead of usin
+   * `this.props.value` directly.
+   */
+  getObjectValue = () => this.props.value;
+
   controlFor(field) {
     const { onAddAsset, onRemoveAsset, getAsset, value, onChange } = this.props;
     if (field.get('widget') === 'hidden') {
@@ -38,7 +45,7 @@ export default class ObjectControl extends Component {
             field,
             value: fieldValue,
             onChange: (val, metadata) => {
-              onChange((value || Map()).set(field.get('name'), val), metadata);
+              onChange((this.getObjectValue() || Map()).set(field.get('name'), val), metadata);
             },
             onAddAsset,
             onRemoveAsset,

src/lib/regexHelper.js

@@ -0,0 +1,145 @@
+import { last } from 'lodash';
+
+/**
+ * Joins an array of regular expressions into a single expression, without
+ * altering the received expressions.
+ */
+export function joinPatternSegments(patterns) {
+  return patterns.map(p => p.source).join('');
+}
+
+
+/**
+ * Combines an array of regular expressions into a single expression, wrapping
+ * each in a non-capturing group and interposing alternation characters (|) so
+ * that each expression is executed separately.
+ */
+export function combinePatterns(patterns, flags = '') {
+  return patterns.map(p => `(?:${p.source})`).join('|');
+}
+
+
+/**
+ * Modify substrings within a string if they match a (global) pattern. Can be
+ * inverted to only modify non-matches.
+ *
+ * params:
+ * matchPattern - regexp - a regular expression to check for matches
+ * replaceFn - function - a replacement function that receives a matched
+ *   substring and returns a replacement substring
+ * text - string - the string to process
+ * invertMatchPattern - boolean - if true, non-matching substrings are modified
+ *   instead of matching substrings
+ */
+export function replaceWhen(matchPattern, replaceFn, text, invertMatchPattern) {
+  /**
+   * Splits the string into an array of objects with the following shape:
+   *
+   * {
+   *   index: number - the index of the substring within the string
+   *   text: string - the substring
+   *   match: boolean - true if the substring matched `matchPattern`
+   * }
+   *
+   * Loops through matches via recursion (`RegExp.exec` tracks the loop
+   * internally).
+   */
+  function split(exp, text, acc) {
+    /**
+     * Get the next match starting from the end of the last match or start of
+     * string.
+     */
+    const match = exp.exec(text);
+    const lastEntry = last(acc);
+
+    /**
+     * `match` will be null if there are no matches.
+     */
+    if (!match) return acc;
+
+    /**
+     * If the match is at the beginning of the input string, normalize to a data
+     * object with the `match` flag set to `true`, and add to the accumulator.
+     */
+    if (match.index === 0) {
+      addSubstring(acc, 0, match[0], true);
+    }
+
+    /**
+     * If there are no entries in the accumulator, convert the substring before
+     * the match to a data object (without the `match` flag set to true) and
+     * push to the accumulator, followed by a data object for the matching
+     * substring.
+     */
+    else if (!lastEntry) {
+      addSubstring(acc, 0, match.input.slice(0, match.index));
+      addSubstring(acc, match.index, match[0], true);
+    }
+
+    /**
+     * If the last entry in the accumulator immediately preceded the current
+     * matched substring in the original string, just add the data object for
+     * the matching substring to the accumulator.
+     */
+    else if (match.index === lastEntry.index + lastEntry.text.length) {
+      addSubstring(acc, match.index, match[0], true);
+    }
+
+    /**
+     * Convert the substring before the match to a data object (without the
+     * `match` flag set to true), followed by a data object for the matching
+     * substring.
+     */
+    else {
+      const nextIndex = lastEntry.index + lastEntry.text.length;
+      const nextText = match.input.slice(nextIndex, match.index);
+      addSubstring(acc, nextIndex, nextText);
+      addSubstring(acc, match.index, match[0], true);
+    }
+
+    /**
+     * Continue executing the expression.
+     */
+    return split(exp, text, acc);
+  }
+
+  /**
+   * Factory for converting substrings to data objects and adding to an output
+   * array.
+   */
+  function addSubstring(arr, index, text, match = false) {
+    arr.push({ index, text, match });
+  }
+
+  /**
+   * Split the input string to an array of data objects, each representing a
+   * matching or non-matching string.
+   */
+  const acc = split(matchPattern, text, []);
+
+  /**
+   * Process the trailing substring after the final match, if one exists.
+   */
+  const lastEntry = last(acc);
+  if (!lastEntry) return replaceFn(text);
+
+  const nextIndex = lastEntry.index + lastEntry.text.length;
+  if (text.length > nextIndex) {
+    acc.push({ index: nextIndex, text: text.slice(nextIndex) });
+  }
+
+  /**
+   * Map the data objects in the accumulator to their string values, modifying
+   * matched strings with the replacement function. Modifies non-matches if
+   * `invertMatchPattern` is truthy.
+   */
+  const replacedText = acc.map(entry => {
+    const isMatch = invertMatchPattern ? !entry.match : entry.match;
+    return isMatch ? replaceFn(entry.text) : entry.text;
+  });
+
+  /**
+   * Return the joined string.
+   */
+  return replacedText.join('');
+}