feat: siteConfig.markdown.parseFrontMatter hook

packages/docusaurus-mdx-loader/src/loader.ts

@@ -8,7 +8,7 @@
 import fs from 'fs-extra';
 import logger from '@docusaurus/logger';
 import {
-  parseFrontMatter,
+  DEFAULT_PARSE_FRONT_MATTER,
   escapePath,
   getFileLoaderUtils,
   getWebpackLoaderCompilerName,
@@ -133,7 +133,7 @@ function extractContentTitleData(data: {
 
 export async function mdxLoader(
   this: LoaderContext<Options>,
-  fileString: string,
+  fileContent: string,
 ): Promise<void> {
   const compilerName = getWebpackLoaderCompilerName(this);
   const callback = this.async();
@@ -143,11 +143,15 @@ export async function mdxLoader(
 
   ensureMarkdownConfig(reqOptions);
 
-  const {frontMatter} = parseFrontMatter(fileString);
+  const {frontMatter} = await reqOptions.markdownConfig.parseFrontMatter({
+    filePath,
+    fileContent,
+    defaultParseFrontMatter: DEFAULT_PARSE_FRONT_MATTER,
+  });
   const mdxFrontMatter = validateMDXFrontMatter(frontMatter.mdx);
 
   const preprocessedContent = preprocessor({
-    fileContent: fileString,
+    fileContent,
     filePath,
     admonitions: reqOptions.admonitions,
     markdownConfig: reqOptions.markdownConfig,

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

@@ -8,6 +8,7 @@
 import {jest} from '@jest/globals';
 import path from 'path';
 import fs from 'fs-extra';
+import {DEFAULT_PARSE_FRONT_MATTER} from '@docusaurus/utils';
 import {DEFAULT_OPTIONS} from '../options';
 import {generateBlogPosts} from '../blogUtils';
 import {createBlogFeedFiles} from '../feed';
@@ -31,6 +32,8 @@ const DefaultI18N: I18n = {
   },
 };
 
+const markdown = {parseFrontMatter: DEFAULT_PARSE_FRONT_MATTER};
+
 function getBlogContentPaths(siteDir: string): BlogContentPaths {
   return {
     contentPath: path.resolve(siteDir, 'blog'),
@@ -72,6 +75,7 @@ describe.each(['atom', 'rss', 'json'])('%s', (feedType) => {
       baseUrl: '/',
       url: 'https://docusaurus.io',
       favicon: 'image/favicon.ico',
+      markdown,
     };
     const outDir = path.join(siteDir, 'build-snap');
 
@@ -110,6 +114,7 @@ describe.each(['atom', 'rss', 'json'])('%s', (feedType) => {
       baseUrl: '/myBaseUrl/',
       url: 'https://docusaurus.io',
       favicon: 'image/favicon.ico',
+      markdown,
     };
 
     // Build is quite difficult to mock, so we built the blog beforehand and
@@ -152,6 +157,7 @@ describe.each(['atom', 'rss', 'json'])('%s', (feedType) => {
       baseUrl: '/myBaseUrl/',
       url: 'https://docusaurus.io',
       favicon: 'image/favicon.ico',
+      markdown,
     };
 
     // Build is quite difficult to mock, so we built the blog beforehand and
@@ -204,6 +210,7 @@ describe.each(['atom', 'rss', 'json'])('%s', (feedType) => {
       baseUrl: '/myBaseUrl/',
       url: 'https://docusaurus.io',
       favicon: 'image/favicon.ico',
+      markdown,
     };
 
     // Build is quite difficult to mock, so we built the blog beforehand and

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

@@ -16,6 +16,7 @@ import type {
   LoadContext,
   I18n,
   Validate,
+  MarkdownConfig,
 } from '@docusaurus/types';
 import type {
   BlogPost,
@@ -24,6 +25,24 @@ import type {
   EditUrlFunction,
 } from '@docusaurus/plugin-content-blog';
 
+const markdown: MarkdownConfig = {
+  format: 'mdx',
+  mermaid: true,
+  mdx1Compat: {
+    comments: true,
+    headingIds: true,
+    admonitions: true,
+  },
+  parseFrontMatter: async (params) => {
+    // Reuse the default parser
+    const result = await params.defaultParseFrontMatter(params);
+    if (result.frontMatter.title === 'Complex Slug') {
+      result.frontMatter.custom_frontMatter = 'added by parseFrontMatter';
+    }
+    return result;
+  },
+};
+
 function findByTitle(
   blogPosts: BlogPost[],
   title: string,
@@ -81,6 +100,7 @@ const getPlugin = async (
     title: 'Hello',
     baseUrl: '/',
     url: 'https://docusaurus.io',
+    markdown,
   } as DocusaurusConfig;
   return pluginContentBlog(
     {
@@ -242,6 +262,7 @@ describe('blog plugin', () => {
         slug: '/hey/my super path/héllô',
         title: 'Complex Slug',
         tags: ['date', 'complex'],
+        custom_frontMatter: 'added by parseFrontMatter',
       },
       tags: [
         {

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

@@ -11,7 +11,7 @@ import _ from 'lodash';
 import logger from '@docusaurus/logger';
 import readingTime from 'reading-time';
 import {
-  parseMarkdownString,
+  parseMarkdownFile,
   normalizeUrl,
   aliasedSitePath,
   getEditUrl,
@@ -29,7 +29,7 @@ import {
 } from '@docusaurus/utils';
 import {validateBlogPostFrontMatter} from './frontMatter';
 import {type AuthorsMap, getAuthorsMap, getBlogPostAuthors} from './authors';
-import type {LoadContext} from '@docusaurus/types';
+import type {LoadContext, ParseFrontMatter} from '@docusaurus/types';
 import type {
   PluginOptions,
   ReadingTimeFunction,
@@ -180,18 +180,27 @@ function formatBlogPostDate(
   }
 }
 
-async function parseBlogPostMarkdownFile(blogSourceAbsolute: string) {
-  const markdownString = await fs.readFile(blogSourceAbsolute, 'utf-8');
+async function parseBlogPostMarkdownFile({
+  filePath,
+  parseFrontMatter,
+}: {
+  filePath: string;
+  parseFrontMatter: ParseFrontMatter;
+}) {
+  const fileContent = await fs.readFile(filePath, 'utf-8');
   try {
-    const result = parseMarkdownString(markdownString, {
+    const result = await parseMarkdownFile({
+      filePath,
+      fileContent,
+      parseFrontMatter,
       removeContentTitle: true,
     });
     return {
       ...result,
       frontMatter: validateBlogPostFrontMatter(result.frontMatter),
     };
   } catch (err) {
-    logger.error`Error while parsing blog post file path=${blogSourceAbsolute}.`;
+    logger.error`Error while parsing blog post file path=${filePath}.`;
     throw err;
   }
 }
@@ -207,7 +216,10 @@ async function processBlogSourceFile(
   authorsMap?: AuthorsMap,
 ): Promise<BlogPost | undefined> {
   const {
-    siteConfig: {baseUrl},
+    siteConfig: {
+      baseUrl,
+      markdown: {parseFrontMatter},
+    },
     siteDir,
     i18n,
   } = context;
@@ -228,7 +240,10 @@ async function processBlogSourceFile(
   const blogSourceAbsolute = path.join(blogDirPath, blogSourceRelative);
 
   const {frontMatter, content, contentTitle, excerpt} =
-    await parseBlogPostMarkdownFile(blogSourceAbsolute);
+    await parseBlogPostMarkdownFile({
+      filePath: blogSourceAbsolute,
+      parseFrontMatter,
+    });
 
   const aliasedSource = aliasedSitePath(blogSourceAbsolute, siteDir);
 

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

@@ -463,7 +463,7 @@ exports[`simple website content: data 1`] = `
   "frontMatter": {
     "title": "Custom Last Update",
     "last_update": {
-      "author": "Custom Author",
+      "author": "Custom Author (processed by parseFrontMatter)",
       "date": "1/1/2000"
     }
   }
@@ -686,7 +686,7 @@ exports[`simple website content: data 1`] = `
   "frontMatter": {
     "title": "Last Update Author Only",
     "last_update": {
-      "author": "Custom Author"
+      "author": "Custom Author (processed by parseFrontMatter)"
     }
   }
 }",

packages/docusaurus-plugin-content-docs/src/__tests__/docs.test.ts

@@ -567,14 +567,14 @@ describe('simple site', () => {
       description: 'Custom last update',
       frontMatter: {
         last_update: {
-          author: 'Custom Author',
+          author: 'Custom Author (processed by parseFrontMatter)',
           date: '1/1/2000',
         },
         title: 'Custom Last Update',
       },
       lastUpdatedAt: new Date('1/1/2000').getTime() / 1000,
       formattedLastUpdatedAt: 'Jan 1, 2000',
-      lastUpdatedBy: 'Custom Author',
+      lastUpdatedBy: 'Custom Author (processed by parseFrontMatter)',
       sidebarPosition: undefined,
       tags: [],
       unlisted: false,
@@ -607,13 +607,13 @@ describe('simple site', () => {
       description: 'Only custom author, so it will still use the date from Git',
       frontMatter: {
         last_update: {
-          author: 'Custom Author',
+          author: 'Custom Author (processed by parseFrontMatter)',
         },
         title: 'Last Update Author Only',
       },
       lastUpdatedAt: 1539502055,
       formattedLastUpdatedAt: 'Oct 14, 2018',
-      lastUpdatedBy: 'Custom Author',
+      lastUpdatedBy: 'Custom Author (processed by parseFrontMatter)',
       sidebarPosition: undefined,
       tags: [],
       unlisted: false,
@@ -685,7 +685,7 @@ describe('simple site', () => {
       description: 'Custom last update',
       frontMatter: {
         last_update: {
-          author: 'Custom Author',
+          author: 'Custom Author (processed by parseFrontMatter)',
           date: '1/1/2000',
         },
         title: 'Custom Last Update',

packages/docusaurus-plugin-content-docs/src/docs.ts

@@ -15,7 +15,7 @@ import {
   getFolderContainingFile,
   getContentPathList,
   normalizeUrl,
-  parseMarkdownString,
+  parseMarkdownFile,
   posixPath,
   Globby,
   normalizeFrontMatterTags,
@@ -140,13 +140,23 @@ async function doProcessDocMetadata({
   env: DocEnv;
 }): Promise<DocMetadataBase> {
   const {source, content, contentPath, filePath} = docFile;
-  const {siteDir, i18n} = context;
+  const {
+    siteDir,
+    i18n,
+    siteConfig: {
+      markdown: {parseFrontMatter},
+    },
+  } = context;
 
   const {
     frontMatter: unsafeFrontMatter,
     contentTitle,
     excerpt,
-  } = parseMarkdownString(content);
+  } = await parseMarkdownFile({
+    filePath,
+    fileContent: content,
+    parseFrontMatter,
+  });
   const frontMatter = validateDocFrontMatter(unsafeFrontMatter);
 
   const {

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

@@ -14,7 +14,9 @@ exports[`docusaurus-plugin-content-pages loads simple pages 1`] = `
   },
   {
     "description": "Markdown index page",
-    "frontMatter": {},
+    "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
+    },
     "permalink": "/hello/",
     "source": "@site/src/pages/hello/index.md",
     "title": "Index",
@@ -24,6 +26,7 @@ exports[`docusaurus-plugin-content-pages loads simple pages 1`] = `
   {
     "description": "my MDX page",
     "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
       "description": "my MDX page",
       "title": "MDX page",
     },
@@ -40,7 +43,9 @@ exports[`docusaurus-plugin-content-pages loads simple pages 1`] = `
   },
   {
     "description": "translated Markdown page",
-    "frontMatter": {},
+    "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
+    },
     "permalink": "/hello/translatedMd",
     "source": "@site/src/pages/hello/translatedMd.md",
     "title": undefined,
@@ -69,7 +74,9 @@ exports[`docusaurus-plugin-content-pages loads simple pages with french translat
   },
   {
     "description": "Markdown index page",
-    "frontMatter": {},
+    "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
+    },
     "permalink": "/fr/hello/",
     "source": "@site/src/pages/hello/index.md",
     "title": "Index",
@@ -79,6 +86,7 @@ exports[`docusaurus-plugin-content-pages loads simple pages with french translat
   {
     "description": "my MDX page",
     "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
       "description": "my MDX page",
       "title": "MDX page",
     },
@@ -95,7 +103,9 @@ exports[`docusaurus-plugin-content-pages loads simple pages with french translat
   },
   {
     "description": "translated Markdown page (fr)",
-    "frontMatter": {},
+    "frontMatter": {
+      "custom_frontMatter": "added by parseFrontMatter",
+    },
     "permalink": "/fr/hello/translatedMd",
     "source": "@site/i18n/fr/docusaurus-plugin-content-pages/hello/translatedMd.md",
     "title": undefined,