refactor(content-blog): replace reading-time with Intl.Segmenter API (#11138)

Co-authored-by: sebastien <lorber.sebastien@gmail.com>

Author: shreedharbhat98

packages/docusaurus-plugin-content-blog/package.json

@@ -43,7 +43,6 @@
     "feed": "^4.2.2",
     "fs-extra": "^11.1.1",
     "lodash": "^4.17.21",
-    "reading-time": "^1.5.0",
     "schema-dts": "^1.1.2",
     "srcset": "^4.0.0",
     "tslib": "^2.6.0",

packages/docusaurus-plugin-content-blog/src/__tests__/__snapshots__/index.test.ts.snap

@@ -150,7 +150,7 @@ exports[`blog plugin process blog posts load content 2`] = `
         "title": "Another With Tag",
       },
       "permalink": "/blog/simple/slug/another",
-      "readingTime": 0.015,
+      "readingTime": 0.02,
       "source": "@site/blog/another-simple-slug-with-tags.md",
       "tags": [
         {

packages/docusaurus-plugin-content-blog/src/__tests__/feed.test.ts

@@ -120,7 +120,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: {atom: null, rss: null},
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -164,7 +164,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: {atom: null, rss: null},
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -220,7 +220,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: {atom: null, rss: null},
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -267,7 +267,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: {atom: null, rss: null},
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -314,7 +314,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: {atom: null, rss: null},
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -360,7 +360,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           xslt: true,
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',
@@ -409,7 +409,7 @@ describe.each(['atom', 'rss', 'json'] as const)('%s', (feedType) => {
           },
         },
         readingTime: ({content, defaultReadingTime}) =>
-          defaultReadingTime({content}),
+          defaultReadingTime({content, locale: 'en'}),
         truncateMarker: /<!--\s*truncate\s*-->/,
         onInlineTags: 'ignore',
         onInlineAuthors: 'ignore',

packages/docusaurus-plugin-content-blog/src/__tests__/index.test.ts

@@ -211,7 +211,7 @@ describe('blog plugin', () => {
     ).toEqual({
       editUrl: `${BaseEditUrl}/blog/2018-12-14-Happy-First-Birthday-Slash.md`,
       permalink: '/blog/2018/12/14/Happy-First-Birthday-Slash',
-      readingTime: 0.015,
+      readingTime: 0.02,
       source: path.posix.join(
         '@site',
         path.posix.join('i18n', 'en', 'docusaurus-plugin-content-blog'),
@@ -276,7 +276,7 @@ describe('blog plugin', () => {
     }).toEqual({
       editUrl: `${BaseEditUrl}/blog/complex-slug.md`,
       permalink: '/blog/hey/my super path/héllô',
-      readingTime: 0.015,
+      readingTime: 0.02,
       source: path.posix.join('@site', PluginPath, 'complex-slug.md'),
       title: 'Complex Slug',
       description: `complex url slug`,
@@ -318,7 +318,7 @@ describe('blog plugin', () => {
     }).toEqual({
       editUrl: `${BaseEditUrl}/blog/simple-slug.md`,
       permalink: '/blog/simple/slug',
-      readingTime: 0.015,
+      readingTime: 0.02,
       source: path.posix.join('@site', PluginPath, 'simple-slug.md'),
       title: 'Simple Slug',
       description: `simple url slug`,

packages/docusaurus-plugin-content-blog/src/__tests__/readingTime.test.ts

@@ -9,48 +9,46 @@ import {calculateReadingTime} from '../readingTime';
 
 describe('calculateReadingTime', () => {
   it('calculates reading time for empty content', () => {
-    expect(calculateReadingTime('')).toBe(0);
+    expect(calculateReadingTime('', 'en')).toBe(0);
   });
 
   it('calculates reading time for short content', () => {
     const content = 'This is a short test content.';
-    expect(calculateReadingTime(content)).toBe(0.03);
+    expect(calculateReadingTime(content, 'en')).toBe(0.03);
   });
 
   it('calculates reading time for long content', () => {
     const content = 'This is a test content. '.repeat(100);
-    expect(calculateReadingTime(content)).toBe(2.5);
+    expect(calculateReadingTime(content, 'en')).toBe(2.5);
   });
 
   it('respects custom words per minute', () => {
     const content = 'This is a test content. '.repeat(100);
-    expect(calculateReadingTime(content, {wordsPerMinute: 100})).toBe(5);
+    expect(calculateReadingTime(content, 'en', {wordsPerMinute: 100})).toBe(5);
   });
 
   it('handles content with special characters', () => {
     const content = 'Hello! How are you? This is a test...';
-    expect(calculateReadingTime(content)).toBe(0.04);
+    expect(calculateReadingTime(content, 'en')).toBe(0.04);
   });
 
   it('handles content with multiple lines', () => {
-    const content = `This is line 1.
-    This is line 2.
-    This is line 3.`;
-    expect(calculateReadingTime(content)).toBe(0.06);
+    const content = `This is line 1.\n    This is line 2.\n    This is line 3.`;
+    expect(calculateReadingTime(content, 'en')).toBe(0.06);
   });
 
   it('handles content with HTML tags', () => {
     const content = '<p>This is a <strong>test</strong> content.</p>';
-    expect(calculateReadingTime(content)).toBe(0.025);
+    expect(calculateReadingTime(content, 'en')).toBe(0.05);
   });
 
   it('handles content with markdown', () => {
     const content = '# Title\n\nThis is **bold** and *italic* text.';
-    expect(calculateReadingTime(content)).toBe(0.04);
+    expect(calculateReadingTime(content, 'en')).toBe(0.04);
   });
 
   it('handles CJK content', () => {
     const content = '你好，世界！这是一段测试内容。';
-    expect(calculateReadingTime(content)).toBe(0.06);
+    expect(calculateReadingTime(content, 'zh')).toBe(0.04);
   });
 });

packages/docusaurus-plugin-content-blog/src/blogUtils.ts

@@ -210,8 +210,8 @@ async function parseBlogPostMarkdownFile({
   }
 }
 
-const defaultReadingTime: ReadingTimeFunction = ({content, options}) =>
-  calculateReadingTime(content, options);
+const defaultReadingTime: ReadingTimeFunction = ({content, locale, options}) =>
+  calculateReadingTime(content, locale, options);
 
 async function processBlogSourceFile(
   blogSourceRelative: string,
@@ -373,6 +373,7 @@ async function processBlogSourceFile(
             content,
             frontMatter,
             defaultReadingTime,
+            locale: i18n.currentLocale,
           })
         : undefined,
       hasTruncateMarker: truncateMarker.test(content),

packages/docusaurus-plugin-content-blog/src/options.ts

@@ -63,7 +63,8 @@ export const DEFAULT_OPTIONS: PluginOptions = {
   path: 'blog',
   editLocalizedFiles: false,
   authorsMapPath: 'authors.yml',
-  readingTime: ({content, defaultReadingTime}) => defaultReadingTime({content}),
+  readingTime: ({content, defaultReadingTime, locale}) =>
+    defaultReadingTime({content, locale}),
   sortPosts: 'descending',
   showLastUpdateTime: false,
   showLastUpdateAuthor: false,

packages/docusaurus-plugin-content-blog/src/plugin-content-blog.d.ts

@@ -387,15 +387,10 @@ declare module '@docusaurus/plugin-content-blog' {
   };
 
   /**
-   * Duplicate from ngryman/reading-time to keep stability of API.
+   * Options for reading time calculation using Intl.Segmenter.
    */
   type ReadingTimeOptions = {
     wordsPerMinute?: number;
-    /**
-     * @param char The character to be matched.
-     * @returns `true` if this character is a word bound.
-     */
-    wordBound?: (char: string) => boolean;
   };
 
   /**
@@ -405,24 +400,22 @@ declare module '@docusaurus/plugin-content-blog' {
   export type ReadingTimeFunction = (params: {
     /** Markdown content. */
     content: string;
+    /** Locale for word segmentation. */
+    locale: string;
     /** Front matter. */
     frontMatter?: BlogPostFrontMatter & {[key: string]: unknown};
-    /** Options accepted by ngryman/reading-time. */
+    /** Options for reading time calculation. */
     options?: ReadingTimeOptions;
   }) => number;
 
   /**
-   * @returns The reading time directly plugged into metadata. `undefined` to
-   * hide reading time for a specific post.
+   * @returns The reading time directly plugged into metadata.
+   * `undefined` to hide reading time for a specific post.
    */
   export type ReadingTimeFunctionOption = (
-    /**
-     * The `options` is not provided by the caller; the user can inject their
-     * own option values into `defaultReadingTime`
-     */
     params: Required<Omit<Parameters<ReadingTimeFunction>[0], 'options'>> & {
       /**
-       * The default reading time implementation from ngryman/reading-time.
+       * The default reading time implementation.
        */
       defaultReadingTime: ReadingTimeFunction;
     },

packages/docusaurus-plugin-content-blog/src/readingTime.ts

@@ -5,25 +5,45 @@
  * LICENSE file in the root directory of this source tree.
  */
 
-import readingTime from 'reading-time';
-
 const DEFAULT_WORDS_PER_MINUTE = 200;
 
-interface ReadingTimeOptions {
-  wordsPerMinute?: number;
-  wordBound?: (char: string) => boolean;
+/**
+ * Counts the number of words in a string using Intl.Segmenter.
+ * @param content The text content to count words in.
+ * @param locale The locale to use for segmentation.
+ */
+function countWords(content: string, locale: string): number {
+  if (!content) {
+    return 0;
+  }
+  const segmenter = new Intl.Segmenter(locale, {granularity: 'word'});
+  let wordCount = 0;
+  for (const {isWordLike} of segmenter.segment(content)) {
+    if (isWordLike) {
+      wordCount += 1;
+    }
+  }
+  return wordCount;
 }
 
 /**
- * Calculates the reading time for a given content string.
- * Uses the reading-time package under the hood.
+ * Calculates the reading time for a given content string using Intl.Segmenter.
+ * @param content The text content to calculate reading time for.
+ * @param locale Required locale string for Intl.Segmenter
+ * @param options Options for reading time calculation.
+ *   - wordsPerMinute: number of words per minute (default 200)
+ * @returns Estimated reading time in minutes (float, rounded to 2 decimals)
  */
 export function calculateReadingTime(
   content: string,
-  options: ReadingTimeOptions = {},
+  locale: string,
+  options?: {wordsPerMinute?: number},
 ): number {
-  const wordsPerMinute = options.wordsPerMinute ?? DEFAULT_WORDS_PER_MINUTE;
-  const {wordBound} = options;
-  return readingTime(content, {wordsPerMinute, ...(wordBound && {wordBound})})
-    .minutes;
+  const wordsPerMinute = options?.wordsPerMinute ?? DEFAULT_WORDS_PER_MINUTE;
+  const words = countWords(content, locale);
+  if (words === 0) {
+    return 0;
+  }
+  // Calculate reading time in minutes and round to 2 decimal places
+  return Math.round((words / wordsPerMinute) * 100) / 100;
 }

website/_dogfooding/dogfooding.config.ts

@@ -104,7 +104,11 @@ export const dogfoodingPluginInstances: PluginConfig[] = [
       readingTime: ({content, frontMatter, defaultReadingTime}) =>
         frontMatter.hide_reading_time
           ? undefined
-          : defaultReadingTime({content, options: {wordsPerMinute: 5}}),
+          : defaultReadingTime({
+              content,
+              locale: 'en',
+              options: {wordsPerMinute: 5},
+            }),
       onInlineTags: 'warn',
       onInlineAuthors: 'ignore',
       onUntruncatedBlogPosts: 'ignore',

website/docs/api/plugins/plugin-content-blog.mdx

@@ -109,17 +109,18 @@ type EditUrlFunction = (params: {
 ```ts
 type ReadingTimeOptions = {
   wordsPerMinute: number;
-  wordBound: (char: string) => boolean;
 };
 
 type ReadingTimeCalculator = (params: {
   content: string;
+  locale: string;
   frontMatter?: BlogPostFrontMatter & Record<string, unknown>;
   options?: ReadingTimeOptions;
 }) => number;
 
 type ReadingTimeFn = (params: {
   content: string;
+  locale: string;
   frontMatter: BlogPostFrontMatter & Record<string, unknown>;
   defaultReadingTime: ReadingTimeCalculator;
 }) => number | undefined;