@docusaurus/utils 2.0.0-beta.6f366f4b4 → 2.0.0-beta.7

package/src/index.ts

@@ -23,6 +23,11 @@ import resolvePathnameUnsafe from 'resolve-pathname';
 
 import {posixPath as posixPathImport} from './posixPath';
 import {simpleHash, docuHash} from './hashUtils';
+import {normalizeUrl} from './normalizeUrl';
+
+export * from './mdxUtils';
+export * from './normalizeUrl';
+export * from './tags';
 
 export const posixPath = posixPathImport;
 
@@ -190,80 +195,6 @@ export function getSubFolder(file: string, refDir: string): string | null {
   return match && match[1];
 }
 
-export function normalizeUrl(rawUrls: string[]): string {
-  const urls = [...rawUrls];
-  const resultArray = [];
-
-  let hasStartingSlash = false;
-  let hasEndingSlash = false;
-
-  // If the first part is a plain protocol, we combine it with the next part.
-  if (urls[0].match(/^[^/:]+:\/*$/) && urls.length > 1) {
-    const first = urls.shift();
-    urls[0] = first + urls[0];
-  }
-
-  // There must be two or three slashes in the file protocol,
-  // two slashes in anything else.
-  const replacement = urls[0].match(/^file:\/\/\//) ? '$1:///' : '$1://';
-  urls[0] = urls[0].replace(/^([^/:]+):\/*/, replacement);
-
-  // eslint-disable-next-line
-  for (let i = 0; i < urls.length; i++) {
-    let component = urls[i];
-
-    if (typeof component !== 'string') {
-      throw new TypeError(`Url must be a string. Received ${typeof component}`);
-    }
-
-    if (component === '') {
-      if (i === urls.length - 1 && hasEndingSlash) {
-        resultArray.push('/');
-      }
-      // eslint-disable-next-line
-      continue;
-    }
-
-    if (component !== '/') {
-      if (i > 0) {
-        // Removing the starting slashes for each component but the first.
-        component = component.replace(
-          /^[/]+/,
-          // Special case where the first element of rawUrls is empty ["", "/hello"] => /hello
-          component[0] === '/' && !hasStartingSlash ? '/' : '',
-        );
-      }
-
-      hasEndingSlash = component[component.length - 1] === '/';
-      // Removing the ending slashes for each component but the last.
-      // For the last component we will combine multiple slashes to a single one.
-      component = component.replace(/[/]+$/, i < urls.length - 1 ? '' : '/');
-    }
-
-    hasStartingSlash = true;
-    resultArray.push(component);
-  }
-
-  let str = resultArray.join('/');
-  // Each input component is now separated by a single slash
-  // except the possible first plain protocol part.
-
-  // Remove trailing slash before parameters or hash.
-  str = str.replace(/\/(\?|&|#[^!])/g, '$1');
-
-  // Replace ? in parameters with &.
-  const parts = str.split('?');
-  str = parts.shift() + (parts.length > 0 ? '?' : '') + parts.join('&');
-
-  // Dedupe forward slashes in the entire path, avoiding protocol slashes.
-  str = str.replace(/([^:]\/)\/+/g, '$1');
-
-  // Dedupe forward slashes at the beginning of the path.
-  str = str.replace(/^\/+/g, '/');
-
-  return str;
-}
-
 /**
  * Alias filepath relative to site directory, very useful so that we
  * don't expose user's site structure.
@@ -497,9 +428,7 @@ export function updateTranslationFileMessages(
 
 // Input: ## Some heading {#some-heading}
 // Output: {text: "## Some heading", id: "some-heading"}
-export function parseMarkdownHeadingId(
-  heading: string,
-): {
+export function parseMarkdownHeadingId(heading: string): {
   text: string;
   id?: string;
 } {

package/src/markdownLinks.ts

@@ -5,7 +5,7 @@
  * LICENSE file in the root directory of this source tree.
  */
 
-import {resolve} from 'url';
+import path from 'path';
 import {aliasedSitePath} from './index';
 
 export type ContentPaths = {
@@ -63,16 +63,27 @@ export function replaceMarkdownLinks<T extends ContentPaths>({
       // Replace it to correct html link.
       const mdLink = mdMatch[1];
 
-      const aliasedSource = (source: string) =>
-        aliasedSitePath(source, siteDir);
+      const sourcesToTry = [
+        path.resolve(path.dirname(filePath), decodeURIComponent(mdLink)),
+        `${contentPathLocalized}/${decodeURIComponent(mdLink)}`,
+        `${contentPath}/${decodeURIComponent(mdLink)}`,
+      ];
 
-      const permalink: string | undefined =
-        sourceToPermalink[aliasedSource(resolve(filePath, mdLink))] ||
-        sourceToPermalink[aliasedSource(`${contentPathLocalized}/${mdLink}`)] ||
-        sourceToPermalink[aliasedSource(`${contentPath}/${mdLink}`)];
+      const aliasedSourceMatch = sourcesToTry
+        .map((source) => aliasedSitePath(source, siteDir))
+        .find((source) => sourceToPermalink[source]);
+
+      const permalink: string | undefined = aliasedSourceMatch
+        ? sourceToPermalink[aliasedSourceMatch]
+        : undefined;
 
       if (permalink) {
-        modifiedLine = modifiedLine.replace(mdLink, permalink);
+        // MDX won't be happy if the permalink contains a space, we need to convert it to %20
+        const encodedPermalink = permalink
+          .split('/')
+          .map((part) => part.replace(/\s/g, '%20'))
+          .join('/');
+        modifiedLine = modifiedLine.replace(mdLink, encodedPermalink);
       } else {
         const brokenMarkdownLink: BrokenMarkdownLink<T> = {
           contentPaths,

package/src/markdownParser.ts

@@ -18,6 +18,7 @@ export function createExcerpt(fileString: string): string | undefined {
     // Remove Markdown alternate title
     .replace(/^[^\n]*\n[=]+/g, '')
     .split('\n');
+  let inCode = false;
 
   /* eslint-disable no-continue */
   // eslint-disable-next-line no-restricted-syntax
@@ -32,6 +33,14 @@ export function createExcerpt(fileString: string): string | undefined {
       continue;
     }
 
+    // Skip code block line.
+    if (fileLine.trim().startsWith('```')) {
+      inCode = !inCode;
+      continue;
+    } else if (inCode) {
+      continue;
+    }
+
     const cleanedLine = fileLine
       // Remove HTML tags.
       .replace(/<[^>]*>/g, '')
@@ -67,9 +76,7 @@ export function createExcerpt(fileString: string): string | undefined {
   return undefined;
 }
 
-export function parseFrontMatter(
-  markdownFileContent: string,
-): {
+export function parseFrontMatter(markdownFileContent: string): {
   frontMatter: Record<string, unknown>;
   content: string;
 } {
@@ -98,10 +105,11 @@ export function parseMarkdownContentTitle(
 
   const content = contentUntrimmed.trim();
 
-  const IMPORT_STATEMENT = /import\s+(([\w*{}\s\n,]+)from\s+)?["'\s]([@\w/_.-]+)["'\s];?|\n/
-    .source;
-  const REGULAR_TITLE = /(?<pattern>#\s*(?<title>[^#\n{]*)+[ \t]*(?<suffix>({#*[\w-]+})|#)?\n*?)/
-    .source;
+  const IMPORT_STATEMENT =
+    /import\s+(([\w*{}\s\n,]+)from\s+)?["'\s]([@\w/_.-]+)["'\s];?|\n/.source;
+  const REGULAR_TITLE =
+    /(?<pattern>#\s*(?<title>[^#\n{]*)+[ \t]*(?<suffix>({#*[\w-]+})|#)?\n*?)/
+      .source;
   const ALTERNATE_TITLE = /(?<pattern>\s*(?<title>[^\n]*)\s*\n[=]+)/.source;
 
   const regularTitleMatch = new RegExp(
@@ -141,9 +149,8 @@ export function parseMarkdownString(
   options?: {removeContentTitle?: boolean},
 ): ParsedMarkdown {
   try {
-    const {frontMatter, content: contentWithoutFrontMatter} = parseFrontMatter(
-      markdownFileContent,
-    );
+    const {frontMatter, content: contentWithoutFrontMatter} =
+      parseFrontMatter(markdownFileContent);
 
     const {content, contentTitle} = parseMarkdownContentTitle(
       contentWithoutFrontMatter,
@@ -174,7 +181,7 @@ export async function parseMarkdownFile(
   const markdownString = await fs.readFile(source, 'utf-8');
   try {
     return parseMarkdownString(markdownString, options);
-  } catch (e) {
+  } catch (e: any) {
     throw new Error(
       `Error while parsing Markdown file ${source}: "${e.message}".`,
     );

package/src/mdxUtils.ts

@@ -0,0 +1,32 @@
+/**
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+import React from 'react';
+import ReactDOMServer from 'react-dom/server';
+import MDX from '@mdx-js/runtime';
+import removeImports from 'remark-mdx-remove-imports';
+import removeExports from 'remark-mdx-remove-exports';
+
+/**
+ * Transform mdx text to plain html text
+ * Initially created to convert MDX blog posts to HTML for the RSS feed
+ * without import/export nodes
+ *
+ * TODO not ideal implementation, won't work well with MDX elements!
+ * TODO theme+global site config should be able to declare MDX comps in scope for rendering the RSS feeds
+ * see also https://github.com/facebook/docusaurus/issues/4625
+ */
+export function mdxToHtml(
+  mdxStr: string,
+  // TODO allow providing components/scope here, see https://github.com/mdx-js/mdx/tree/v1.6.13/packages/runtime
+): string {
+  return ReactDOMServer.renderToString(
+    React.createElement(MDX, {remarkPlugins: [removeImports, removeExports]}, [
+      mdxStr,
+    ]),
+  );
+}

package/src/normalizeUrl.ts

@@ -0,0 +1,80 @@
+/**
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+export function normalizeUrl(rawUrls: string[]): string {
+  const urls = [...rawUrls];
+  const resultArray = [];
+
+  let hasStartingSlash = false;
+  let hasEndingSlash = false;
+
+  // If the first part is a plain protocol, we combine it with the next part.
+  if (urls[0].match(/^[^/:]+:\/*$/) && urls.length > 1) {
+    const first = urls.shift();
+    urls[0] = first + urls[0];
+  }
+
+  // There must be two or three slashes in the file protocol,
+  // two slashes in anything else.
+  const replacement = urls[0].match(/^file:\/\/\//) ? '$1:///' : '$1://';
+  urls[0] = urls[0].replace(/^([^/:]+):\/*/, replacement);
+
+  // eslint-disable-next-line
+  for (let i = 0; i < urls.length; i++) {
+    let component = urls[i];
+
+    if (typeof component !== 'string') {
+      throw new TypeError(`Url must be a string. Received ${typeof component}`);
+    }
+
+    if (component === '') {
+      if (i === urls.length - 1 && hasEndingSlash) {
+        resultArray.push('/');
+      }
+      // eslint-disable-next-line
+      continue;
+    }
+
+    if (component !== '/') {
+      if (i > 0) {
+        // Removing the starting slashes for each component but the first.
+        component = component.replace(
+          /^[/]+/,
+          // Special case where the first element of rawUrls is empty ["", "/hello"] => /hello
+          component[0] === '/' && !hasStartingSlash ? '/' : '',
+        );
+      }
+
+      hasEndingSlash = component[component.length - 1] === '/';
+      // Removing the ending slashes for each component but the last.
+      // For the last component we will combine multiple slashes to a single one.
+      component = component.replace(/[/]+$/, i < urls.length - 1 ? '' : '/');
+    }
+
+    hasStartingSlash = true;
+    resultArray.push(component);
+  }
+
+  let str = resultArray.join('/');
+  // Each input component is now separated by a single slash
+  // except the possible first plain protocol part.
+
+  // Remove trailing slash before parameters or hash.
+  str = str.replace(/\/(\?|&|#[^!])/g, '$1');
+
+  // Replace ? in parameters with &.
+  const parts = str.split('?');
+  str = parts.shift() + (parts.length > 0 ? '?' : '') + parts.join('&');
+
+  // Dedupe forward slashes in the entire path, avoiding protocol slashes.
+  str = str.replace(/([^:]\/)\/+/g, '$1');
+
+  // Dedupe forward slashes at the beginning of the path.
+  str = str.replace(/^\/+/g, '/');
+
+  return str;
+}

package/src/tags.ts

@@ -0,0 +1,100 @@
+/**
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+import {kebabCase, uniq, uniqBy} from 'lodash';
+import {normalizeUrl} from './normalizeUrl';
+
+export type Tag = {
+  label: string;
+  permalink: string;
+};
+
+export type FrontMatterTag = string | Tag;
+
+export function normalizeFrontMatterTag(
+  tagsPath: string,
+  frontMatterTag: FrontMatterTag,
+): Tag {
+  function toTagObject(tagString: string): Tag {
+    return {
+      label: tagString,
+      permalink: kebabCase(tagString),
+    };
+  }
+
+  // TODO maybe make ensure the permalink is valid url path?
+  function normalizeTagPermalink(permalink: string): string {
+    // note: we always apply tagsPath on purpose
+    // for versioned docs, v1/doc.md and v2/doc.md tags with custom permalinks don't lead to the same created page
+    // tagsPath is different for each doc version
+    return normalizeUrl([tagsPath, permalink]);
+  }
+
+  const tag: Tag =
+    typeof frontMatterTag === 'string'
+      ? toTagObject(frontMatterTag)
+      : frontMatterTag;
+
+  return {
+    label: tag.label,
+    permalink: normalizeTagPermalink(tag.permalink),
+  };
+}
+
+export function normalizeFrontMatterTags(
+  tagsPath: string,
+  frontMatterTags: FrontMatterTag[] | undefined,
+): Tag[] {
+  const tags =
+    frontMatterTags?.map((tag) => normalizeFrontMatterTag(tagsPath, tag)) ?? [];
+
+  return uniqBy(tags, (tag) => tag.permalink);
+}
+
+export type TaggedItemGroup<Item> = {
+  tag: Tag;
+  items: Item[];
+};
+
+// Permits to group docs/blogPosts by tag (provided by FrontMatter)
+// Note: groups are indexed by permalink, because routes must be unique in the end
+// Labels may vary on 2 md files but they are normalized.
+// Docs with label='some label' and label='some-label' should end-up in the same group/page in the end
+// We can't create 2 routes /some-label because one would override the other
+export function groupTaggedItems<Item>(
+  items: Item[],
+  getItemTags: (item: Item) => Tag[],
+): Record<string, TaggedItemGroup<Item>> {
+  const result: Record<string, TaggedItemGroup<Item>> = {};
+
+  function handleItemTag(item: Item, tag: Tag) {
+    // Init missing tag groups
+    // TODO: it's not really clear what should be the behavior if 2 items have the same tag but the permalink is different for each
+    // For now, the first tag found wins
+    result[tag.permalink] = result[tag.permalink] ?? {
+      tag,
+      items: [],
+    };
+
+    // Add item to group
+    result[tag.permalink].items.push(item);
+  }
+
+  items.forEach((item) => {
+    getItemTags(item).forEach((tag) => {
+      handleItemTag(item, tag);
+    });
+  });
+
+  // If user add twice the same tag to a md doc (weird but possible),
+  // we don't want the item to appear twice in the list...
+  Object.values(result).forEach((group) => {
+    group.items = uniq(group.items);
+  });
+
+  return result;
+}