@easyops-cn/docusaurus-search-local 0.23.2 → 0.24.1

package/CHANGELOG.md

@@ -2,6 +2,35 @@
 
 All notable changes to this project will be documented in this file. See [standard-version](https://github.com/conventional-changelog/standard-version) for commit guidelines.
 
+### [0.24.1](https://github.com/easyops-cn/docusaurus-search-local/compare/v0.24.0...v0.24.1) (2022-05-19)
+
+
+### Bug Fixes
+
+* fix indexing content between h1 and h2 ([f38b307](https://github.com/easyops-cn/docusaurus-search-local/commit/f38b307a669b582b8571fd2430b4454b49323c85))
+
+## [0.24.0](https://github.com/easyops-cn/docusaurus-search-local/compare/v0.23.3...v0.24.0) (2022-05-18)
+
+
+### Features
+
+* Add option to disable the default lunr stemmer ([ff8e9a2](https://github.com/easyops-cn/docusaurus-search-local/commit/ff8e9a2e3162dd847cbef11036772578c41a8828))
+
+
+### Bug Fixes
+
+* **deps:** update dependency fs-extra to v10 ([96c9220](https://github.com/easyops-cn/docusaurus-search-local/commit/96c92206bcff35b529400c19d5bcf66e1abd4d81))
+* Fix parsing of top level pages when using docs only mode ([ff8e9a2](https://github.com/easyops-cn/docusaurus-search-local/commit/ff8e9a2e3162dd847cbef11036772578c41a8828))
+* Fix passing of top sections for versions of docusaurus with breadcrumbs ([ff8e9a2](https://github.com/easyops-cn/docusaurus-search-local/commit/ff8e9a2e3162dd847cbef11036772578c41a8828))
+* Fix search term highlighting when reusing the same search string ([ff8e9a2](https://github.com/easyops-cn/docusaurus-search-local/commit/ff8e9a2e3162dd847cbef11036772578c41a8828))
+
+### [0.23.3](https://www.github.com/easyops-cn/docusaurus-search-local/compare/v0.23.2...v0.23.3) (2022-05-18)
+
+
+### Bug Fixes
+
+* Update getIndexHash.ts to support mdx and traverseAll ([#172](https://www.github.com/easyops-cn/docusaurus-search-local/issues/172)) ([717942e](https://www.github.com/easyops-cn/docusaurus-search-local/commit/717942e9caf36e37a0f119d9fdaa50e591339d9c))
+
 ### [0.23.2](https://www.github.com/easyops-cn/docusaurus-search-local/compare/v0.23.1...v0.23.2) (2022-04-25)
 
 

package/README.md

@@ -54,6 +54,7 @@ module.exports = {
         // ... Your options.
         // `hashed` is recommended as long-term-cache of index file is possible.
         hashed: true,
+        explicitSearchResultPath: true,
         // For Docs using Chinese, The `language` is recommended to set to:
         // ```
         // language: ["en", "zh"],
@@ -68,24 +69,25 @@ module.exports = {
 
 ## Theme Options
 
-| Name                             | Type                                     | Default      | Description                                                                                                                                  |
-| -------------------------------- | ---------------------------------------- | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------- |
-| indexDocs                        | boolean                                  | `true`       | Whether to index docs.                                                                                                                       |
-| indexBlog                        | boolean                                  | `true`       | Whether to index blog.                                                                                                                       |
-| indexPages                       | boolean                                  | `false`      | Whether to index pages.                                                                                                                      |
-| docsRouteBasePath                | string \| string[]                       | `"/docs"`    | Base route path(s) of docs. Slash at beginning is not required.                                                                              |
-| blogRouteBasePath                | string \| string[]                       | `"/blog"`    | Base route path(s) of blog. Slash at beginning is not required.                                                                              |
-| language                         | string \| string[]                       | `"en"`       | All [lunr-languages](https://github.com/MihaiValentin/lunr-languages) supported languages, + `zh` 🔥.                                        |
-| hashed                           | boolean                                  | `false`      | Whether to add a hashed query when fetching index (based on the content hash of all indexed `*.md` in `docsDir` and `blogDir` if applicable) |
-| docsDir                          | string \| string[]                       | `"docs"`     | The dir(s) of docs to get the content hash, it's relative to the dir of your project.                                                        |
-| blogDir                          | string \| string[]                       | `"blog"`     | Just like the `docsDir` but applied to blog.                                                                                                 |
-| removeDefaultStopWordFilter      | boolean                                  | `false`      | Sometimes people (E.g., us) want to keep the English stop words as indexed, since they maybe are relevant in programming docs.               |
-| highlightSearchTermsOnTargetPage | boolean                                  | `false`      | Highlight search terms on target page.                                                                                                       |
-| searchResultLimits               | number                                   | `8`          | Limit the search results.                                                                                                                    |
-| searchResultContextMaxLength     | number                                   | `50`         | Set the max length of characters of each search result to show.                                                                              |
-| explicitSearchResultPath         | boolean                                  | false        | Whether an explicit path to a heading should be presented on a suggestion template.                                                          |
-| translations                     | TranslationMap                           | -            | Set translations of this theme, see [docs below](#translations).                                                                             |
-| ignoreFiles                      | string \| RegExp \| (string \| RegExp)[] | /**meta**\$/ | Set the match rules to ignore some files.                                                                                                    |
+| Name                             | Type                                     | Default      | Description                                                                                                                                                                                                                                                    |
+| -------------------------------- | ---------------------------------------- | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| indexDocs                        | boolean                                  | `true`       | Whether to index docs.                                                                                                                                                                                                                                         |
+| indexBlog                        | boolean                                  | `true`       | Whether to index blog.                                                                                                                                                                                                                                         |
+| indexPages                       | boolean                                  | `false`      | Whether to index pages.                                                                                                                                                                                                                                        |
+| docsRouteBasePath                | string \| string[]                       | `"/docs"`    | Base route path(s) of docs. Slash at beginning is not required. Note: for [docs-only mode](https://docusaurus.io/docs/docs-introduction#docs-only-mode), this needs to be the same as `routeBasePath` in your `@docusaurus/preset-classic` config e.g., `"/"`. |
+| blogRouteBasePath                | string \| string[]                       | `"/blog"`    | Base route path(s) of blog. Slash at beginning is not required.                                                                                                                                                                                                |
+| language                         | string \| string[]                       | `"en"`       | All [lunr-languages](https://github.com/MihaiValentin/lunr-languages) supported languages, + `zh` 🔥.                                                                                                                                                          |
+| hashed                           | boolean                                  | `false`      | Whether to add a hashed query when fetching index (based on the content hash of all indexed `*.md` in `docsDir` and `blogDir` if applicable)                                                                                                                   |
+| docsDir                          | string \| string[]                       | `"docs"`     | The dir(s) of docs to get the content hash, it's relative to the dir of your project.                                                                                                                                                                          |
+| blogDir                          | string \| string[]                       | `"blog"`     | Just like the `docsDir` but applied to blog.                                                                                                                                                                                                                   |
+| removeDefaultStopWordFilter      | boolean                                  | `false`      | Sometimes people (E.g., us) want to keep the English stop words as indexed, since they maybe are relevant in programming docs.                                                                                                                                 |
+| removeDefaultStemmer             | boolean                                  | `false`      | Enable this if you want to be able to search for any partial word at the cost of search performance.                                                                                                                                                           |
+| highlightSearchTermsOnTargetPage | boolean                                  | `false`      | Highlight search terms on target page.                                                                                                                                                                                                                         |
+| searchResultLimits               | number                                   | `8`          | Limit the search results.                                                                                                                                                                                                                                      |
+| searchResultContextMaxLength     | number                                   | `50`         | Set the max length of characters of each search result to show.                                                                                                                                                                                                |
+| explicitSearchResultPath         | boolean                                  | false        | Whether an explicit path to a heading should be presented on a suggestion template.                                                                                                                                                                            |
+| translations                     | TranslationMap                           | -            | Set translations of this theme, see [docs below](#translations).                                                                                                                                                                                               |
+| ignoreFiles                      | string \| RegExp \| (string \| RegExp)[] | /**meta**\$/ | Set the match rules to ignore some files.                                                                                                                                                                                                                      |
 
 ### Translations
 

package/dist/client/client/theme/SearchBar/SearchBar.jsx

@@ -141,7 +141,7 @@ export default function SearchBar({ handleSearchBarToggle, }) {
             mark.unmark();
             mark.mark(keywords);
         });
-    }, [location.search]);
+    }, [location.search, location.pathname]);
     const onInputFocus = useCallback(() => {
         focusAfterIndexLoaded.current = true;
         loadIndex();
@@ -177,8 +177,8 @@ export default function SearchBar({ handleSearchBarToggle, }) {
         };
     }, [isMac, onInputFocus]);
     return (<div className={clsx("navbar__search", styles.searchBarContainer, {
-        [styles.searchIndexLoading]: loading && inputChanged,
-    })}>
+            [styles.searchIndexLoading]: loading && inputChanged,
+        })}>
       <input placeholder={translations.search_placeholder} aria-label="Search" className="navbar__search-input" onMouseEnter={onInputMouseEnter} onFocus={onInputFocus} onBlur={onInputBlur} onChange={onInputChange} ref={searchBarRef}/>
       <LoadingRing className={styles.searchBarLoadingRing}/>
       <div className={styles.searchHintContainer}>

package/dist/client/client/theme/SearchBar/SuggestionTemplate.js

@@ -23,7 +23,7 @@ export function SuggestionTemplate({ document, type, page, metadata, tokens, isI
     const needsExplicitHitPath = !isInterOfTree && !isLastOfTree && explicitSearchResultPath;
     if (needsExplicitHitPath) {
         const pathItems = page
-            ? (page.b ?? []).concat(page.t).concat(document.s ?? [])
+            ? (page.b ?? []).concat(page.t).concat(!document.s || document.s === page.t ? [] : document.s)
             : document.b;
         wrapped.push(`<span class="${styles.hitPath}">${concatDocumentPath(pathItems ?? [])}</span>`);
     }

package/dist/client/client/theme/SearchPage/SearchPage.jsx

@@ -57,7 +57,10 @@ export default function SearchPage() {
     }, [baseUrl]);
     return (<Layout>
       <Head>
-        
+        {/*
+         We should not index search pages
+          See https://github.com/facebook/docusaurus/pull/3233
+        */}
         <meta property="robots" content="noindex, follow"/>
         <title>{pageTitle}</title>
       </Head>
@@ -72,12 +75,12 @@ export default function SearchPage() {
           </div>)}
 
         {searchResults &&
-        (searchResults.length > 0 ? (<p>
+            (searchResults.length > 0 ? (<p>
               {simpleTemplate(searchResults.length === 1
-            ? translations.count_documents_found
-            : translations.count_documents_found_plural, {
-            count: searchResults.length,
-        })}
+                    ? translations.count_documents_found
+                    : translations.count_documents_found_plural, {
+                    count: searchResults.length,
+                })}
             </p>) : process.env.NODE_ENV === "production" ? (<p>{translations.no_documents_were_found}</p>) : (<p>
               ⚠️ The search index is only available when you run docusaurus
               build!
@@ -85,7 +88,7 @@ export default function SearchPage() {
 
         <section>
           {searchResults &&
-        searchResults.map((item) => (<SearchResultItem key={item.document.i} searchResult={item}/>))}
+            searchResults.map((item) => (<SearchResultItem key={item.document.i} searchResult={item}/>))}
         </section>
       </div>
     </Layout>);
@@ -101,16 +104,16 @@ function SearchResultItem({ searchResult: { document, type, page, tokens, metada
     return (<article className={styles.searchResultItem}>
       <h2>
         <Link to={document.u + (document.h || "")} dangerouslySetInnerHTML={{
-        __html: isContent
-            ? highlight(articleTitle, tokens)
-            : highlightStemmed(articleTitle, getStemmedPositions(metadata, "t"), tokens, 100),
-    }}></Link>
+            __html: isContent
+                ? highlight(articleTitle, tokens)
+                : highlightStemmed(articleTitle, getStemmedPositions(metadata, "t"), tokens, 100),
+        }}></Link>
       </h2>
       {pathItems.length > 0 && (<p className={styles.searchResultItemPath}>
           {concatDocumentPath(pathItems)}
         </p>)}
       {isContent && (<p className={styles.searchResultItemSummary} dangerouslySetInnerHTML={{
-        __html: highlightStemmed(document.t, getStemmedPositions(metadata, "t"), tokens, 100),
-    }}/>)}
+                __html: highlightStemmed(document.t, getStemmedPositions(metadata, "t"), tokens, 100),
+            }}/>)}
     </article>);
 }

package/dist/client/client/utils/__mocks__/proxiedGenerated.js

@@ -1,5 +1,6 @@
 export let language = ["en", "zh"];
 export let removeDefaultStopWordFilter = false;
+export let removeDefaultStemmer = false;
 export const indexHash = "abc";
 export const searchResultLimits = 8;
 export const searchResultContextMaxLength = 50;
@@ -20,3 +21,6 @@ export function __setLanguage(value) {
 export function __setRemoveDefaultStopWordFilter(value) {
     removeDefaultStopWordFilter = value;
 }
+export function __setRemoveDefaultStemmer(value) {
+    removeDefaultStemmer = value;
+}

package/dist/server/server/index.js

@@ -10,10 +10,10 @@ const postBuildFactory_1 = require("./utils/postBuildFactory");
 const generate_1 = require("./utils/generate");
 const PLUGIN_NAME = "@easyops-cn/docusaurus-search-local";
 function DocusaurusSearchLocalPlugin(context, options) {
-    const config = processPluginOptions_1.processPluginOptions(options, context.siteDir);
+    const config = (0, processPluginOptions_1.processPluginOptions)(options, context.siteDir);
     const dir = path_1.default.join(context.generatedFilesDir, PLUGIN_NAME, "default");
     fs_extra_1.default.ensureDirSync(dir);
-    generate_1.generate(config, dir);
+    (0, generate_1.generate)(config, dir);
     const themePath = path_1.default.resolve(__dirname, "../../client/client/theme");
     const pagePath = path_1.default.join(themePath, "SearchPage/index.js");
     return {
@@ -21,14 +21,14 @@ function DocusaurusSearchLocalPlugin(context, options) {
         getThemePath() {
             return themePath;
         },
-        postBuild: postBuildFactory_1.postBuildFactory(config),
+        postBuild: (0, postBuildFactory_1.postBuildFactory)(config),
         getPathsToWatch() {
             return [pagePath];
         },
         contentLoaded({ actions: { addRoute } }) {
             return tslib_1.__awaiter(this, void 0, void 0, function* () {
                 addRoute({
-                    path: utils_1.normalizeUrl([context.baseUrl, "search"]),
+                    path: (0, utils_1.normalizeUrl)([context.baseUrl, "search"]),
                     component: "@theme/SearchPage",
                     exact: true,
                 });

package/dist/server/server/utils/buildIndex.js

@@ -3,7 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.buildIndex = void 0;
 const tslib_1 = require("tslib");
 const lunr_1 = tslib_1.__importDefault(require("lunr"));
-function buildIndex(allDocuments, { language, removeDefaultStopWordFilter }) {
+function buildIndex(allDocuments, { language, removeDefaultStopWordFilter, removeDefaultStemmer, }) {
     if (language.length > 1 || language.some((item) => item !== "en")) {
         // eslint-disable-next-line @typescript-eslint/no-var-requires
         require("lunr-languages/lunr.stemmer.support")(lunr_1.default);
@@ -28,7 +28,7 @@ function buildIndex(allDocuments, { language, removeDefaultStopWordFilter }) {
     }
     return allDocuments.map((documents) => ({
         documents,
-        index: lunr_1.default(function () {
+        index: (0, lunr_1.default)(function () {
             if (language.length > 1) {
                 this.use(lunr_1.default.multiLanguage(...language));
             }
@@ -40,6 +40,9 @@ function buildIndex(allDocuments, { language, removeDefaultStopWordFilter }) {
                 // since they are almost all programming code.
                 this.pipeline.remove(lunr_1.default.stopWordFilter);
             }
+            if (removeDefaultStemmer) {
+                this.pipeline.remove(lunr_1.default.stemmer);
+            }
             // Override tokenizer when language `zh` is enabled.
             if (language.includes("zh")) {
                 this.tokenizer = lunr_1.default.zh.tokenizer;

package/dist/server/server/utils/debug.js

@@ -3,7 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.debugWarn = exports.debugInfo = exports.debugVerbose = void 0;
 const tslib_1 = require("tslib");
 const debug_1 = tslib_1.__importDefault(require("debug"));
-const debug = debug_1.default("search-local");
+const debug = (0, debug_1.default)("search-local");
 exports.debugVerbose = debug.extend("verbose");
 exports.debugInfo = debug.extend("info");
 exports.debugWarn = debug.extend("warn");

package/dist/server/server/utils/generate.js

@@ -6,8 +6,8 @@ const fs_1 = tslib_1.__importDefault(require("fs"));
 const path_1 = tslib_1.__importDefault(require("path"));
 const getIndexHash_1 = require("./getIndexHash");
 function generate(config, dir) {
-    const { language, removeDefaultStopWordFilter, highlightSearchTermsOnTargetPage, searchResultLimits, searchResultContextMaxLength, explicitSearchResultPath, translations, } = config;
-    const indexHash = getIndexHash_1.getIndexHash(config);
+    const { language, removeDefaultStopWordFilter, removeDefaultStemmer, highlightSearchTermsOnTargetPage, searchResultLimits, searchResultContextMaxLength, explicitSearchResultPath, translations, } = config;
+    const indexHash = (0, getIndexHash_1.getIndexHash)(config);
     const contents = [
         `import lunr from ${JSON.stringify(require.resolve("lunr"))};`,
     ];
@@ -28,6 +28,7 @@ function generate(config, dir) {
     }
     contents.push(`export const language = ${JSON.stringify(language)};`);
     contents.push(`export const removeDefaultStopWordFilter = ${JSON.stringify(removeDefaultStopWordFilter)};`);
+    contents.push(`export const removeDefaultStemmer = ${JSON.stringify(removeDefaultStemmer)};`);
     if (highlightSearchTermsOnTargetPage) {
         contents.push(`export { default as Mark } from ${JSON.stringify(require.resolve("mark.js"))}`);
     }

package/dist/server/server/utils/getIndexHash.js

@@ -22,7 +22,7 @@ function getIndexHash(config) {
                     console.warn(`Warn: \`${dirField}\` is not a directory: "${dir}".`);
                 }
                 else {
-                    files.push(...klaw_sync_1.default(dir, { nodir: true, filter: markdownFilter }));
+                    files.push(...(0, klaw_sync_1.default)(dir, { nodir: true, filter: markdownFilter, traverseAll: true }));
                 }
             }
         }
@@ -35,18 +35,18 @@ function getIndexHash(config) {
         // since the index maybe changed between versions.
         // eslint-disable-next-line @typescript-eslint/no-var-requires
         const pluginVersion = require(path_1.default.resolve(__dirname, "../../../../package.json")).version;
-        debug_1.debugInfo("using @easyops-cn/docusaurus-search-local v%s", pluginVersion);
+        (0, debug_1.debugInfo)("using @easyops-cn/docusaurus-search-local v%s", pluginVersion);
         md5sum.update(pluginVersion, "utf8");
         for (const item of files) {
             md5sum.update(fs_1.default.readFileSync(item.path));
         }
         const indexHash = md5sum.digest("hex").substring(0, 8);
-        debug_1.debugInfo("the index hash is %j", indexHash);
+        (0, debug_1.debugInfo)("the index hash is %j", indexHash);
         return indexHash;
     }
     return null;
 }
 exports.getIndexHash = getIndexHash;
 function markdownFilter(item) {
-    return item.path.endsWith(".md");
+    return item.path.endsWith(".md") || item.path.endsWith(".mdx");
 }

package/dist/server/server/utils/parse.js

@@ -16,8 +16,8 @@ function parse(html, type, url) {
             .remove();
     }
     if (type === "page") {
-        return parsePage_1.parsePage($, url);
+        return (0, parsePage_1.parsePage)($, url);
     }
-    return parseDocument_1.parseDocument($);
+    return (0, parseDocument_1.parseDocument)($);
 }
 exports.parse = parse;

package/dist/server/server/utils/parseDocument.js

@@ -1,8 +1,10 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.parseDocument = void 0;
+const utils_common_1 = require("@docusaurus/utils-common");
 const getCondensedText_1 = require("./getCondensedText");
 const HEADINGS = "h1, h2, h3";
+// const SUB_HEADINGS = "h2, h3";
 function parseDocument($) {
     const $pageTitle = $("article h1").first();
     const pageTitle = $pageTitle.text();
@@ -32,22 +34,48 @@ function parseDocument($) {
         // <a aria-hidden="true" tabindex="-1" class="hash-link" href="#first-subheader" title="Direct link to heading">#</a>
         const title = $h.contents().not("a.hash-link").text().trim();
         const hash = $h.find("a.hash-link").attr("href") || "";
-        let $sectionElements;
+        // Find all content between h1 and h2/h3,
+        // which is considered as the content section of page title.
+        let $sectionElements = $([]);
         if ($h.is($pageTitle)) {
-            const $firstElement = $("article")
-                .children() // div.markdown, header
-                .not("header") // div.markdown
-                .children() // h1, p, p, h2, ...
-                .first(); // h1 || p
-            if ($firstElement.filter(HEADINGS).length > 0) {
-                return;
+            const $header = $h.parent();
+            let $firstElement;
+            if ($header.is("header")) {
+                $firstElement = $header;
+            }
+            else {
+                $firstElement = $h;
+            }
+            const blogPost = $(`#${utils_common_1.blogPostContainerID}`);
+            if (blogPost.length) {
+                // Simplify blog post.
+                $firstElement = blogPost.children().first();
+                $sectionElements = $firstElement.nextUntil(HEADINGS).addBack();
+            }
+            else {
+                const $nextElements = $firstElement.nextAll();
+                const $headings = $nextElements.filter(HEADINGS);
+                if ($headings.length) {
+                    $sectionElements = $firstElement.nextUntil(HEADINGS);
+                }
+                else {
+                    for (const next of $nextElements.get()) {
+                        const $heading = $(next).find(HEADINGS);
+                        if ($heading.length) {
+                            $sectionElements = $sectionElements.add($heading.first().prevAll());
+                            break;
+                        }
+                        else {
+                            $sectionElements = $sectionElements.add(next);
+                        }
+                    }
+                }
             }
-            $sectionElements = $firstElement.nextUntil(HEADINGS).addBack();
         }
         else {
             $sectionElements = $h.nextUntil(HEADINGS);
         }
-        const content = getCondensedText_1.getCondensedText($sectionElements.get(), $);
+        const content = (0, getCondensedText_1.getCondensedText)($sectionElements.get(), $);
         sections.push({
             title,
             hash,

package/dist/server/server/utils/parsePage.js

@@ -12,7 +12,7 @@ function parsePage($, url) {
     const pageTitle = $pageTitle.text();
     const $main = $("main");
     if ($main.length === 0) {
-        debug_1.debugWarn("page has no <main>, therefore no content was indexed for this page %o", url);
+        (0, debug_1.debugWarn)("page has no <main>, therefore no content was indexed for this page %o", url);
     }
     return {
         pageTitle,
@@ -20,7 +20,7 @@ function parsePage($, url) {
             {
                 title: pageTitle,
                 hash: "",
-                content: $main.length > 0 ? getCondensedText_1.getCondensedText($main.get(0), $).trim() : "",
+                content: $main.length > 0 ? (0, getCondensedText_1.getCondensedText)($main.get(0), $).trim() : "",
             },
         ],
         breadcrumb: [],

package/dist/server/server/utils/postBuildFactory.js

@@ -13,16 +13,16 @@ const writeFileAsync = util_1.default.promisify(fs_1.default.writeFile);
 function postBuildFactory(config) {
     return function postBuild(buildData) {
         return tslib_1.__awaiter(this, void 0, void 0, function* () {
-            debug_1.debugInfo("gathering documents");
-            const data = processDocInfos_1.processDocInfos(buildData, config);
-            debug_1.debugInfo("parsing documents");
+            (0, debug_1.debugInfo)("gathering documents");
+            const data = (0, processDocInfos_1.processDocInfos)(buildData, config);
+            (0, debug_1.debugInfo)("parsing documents");
             // Give every index entry a unique id so that the index does not need to store long URLs.
-            const allDocuments = yield scanDocuments_1.scanDocuments(data);
-            debug_1.debugInfo("building index");
-            const searchIndex = buildIndex_1.buildIndex(allDocuments, config);
-            debug_1.debugInfo("writing index to disk");
+            const allDocuments = yield (0, scanDocuments_1.scanDocuments)(data);
+            (0, debug_1.debugInfo)("building index");
+            const searchIndex = (0, buildIndex_1.buildIndex)(allDocuments, config);
+            (0, debug_1.debugInfo)("writing index to disk");
             yield writeFileAsync(path_1.default.join(buildData.outDir, "search-index.json"), JSON.stringify(searchIndex), { encoding: "utf8" });
-            debug_1.debugInfo("index written to disk successfully!");
+            (0, debug_1.debugInfo)("index written to disk successfully!");
         });
     };
 }

package/dist/server/server/utils/processDocInfos.js

@@ -12,7 +12,9 @@ function processDocInfos({ routesPaths, outDir, baseUrl, siteConfig }, { indexDo
         }
         const route = url.substr(baseUrl.length).replace(/\/$/, "");
         // Do not index homepage, error page and search page.
-        if (route === "" || route === "404.html" || route === "search") {
+        if (((!docsRouteBasePath || docsRouteBasePath[0] !== "") && route === "") ||
+            route === "404.html" ||
+            route === "search") {
             return;
         }
         // ignore files

package/dist/server/server/utils/scanDocuments.js

@@ -19,9 +19,9 @@ function scanDocuments(DocInfoWithFilePathList) {
         const contentDocuments = [];
         const allDocuments = [titleDocuments, headingDocuments, contentDocuments];
         yield Promise.all(DocInfoWithFilePathList.map(({ filePath, url, type }) => tslib_1.__awaiter(this, void 0, void 0, function* () {
-            debug_1.debugVerbose(`parsing %s file %o of %o`, type, path_1.default.relative(process.cwd(), filePath), url);
+            (0, debug_1.debugVerbose)(`parsing %s file %o of %o`, type, path_1.default.relative(process.cwd(), filePath), url);
             const html = yield readFileAsync(filePath, { encoding: "utf8" });
-            const { pageTitle, sections, breadcrumb } = parse_1.parse(html, type, url);
+            const { pageTitle, sections, breadcrumb } = (0, parse_1.parse)(html, type, url);
             const titleId = getNextDocId();
             titleDocuments.push({
                 i: titleId,

package/dist/server/server/utils/tokenizer.js

@@ -30,7 +30,7 @@ function tokenizer(input, metadata) {
         if (/\w/.test(word[0])) {
             tokens.push(new lunr_1.default.Token(word, Object.assign(Object.assign({}, lunr_1.default.utils.clone(metadata)), { position: [start, word.length], index: tokens.length })));
             // Try to cut `api_gateway` to `api` and `gateway`.
-            const subWords = cutWordByUnderscore_1.cutWordByUnderscore(word);
+            const subWords = (0, cutWordByUnderscore_1.cutWordByUnderscore)(word);
             if (subWords.length > 1) {
                 let i = 0;
                 for (const subWord of subWords) {

package/dist/server/server/utils/validateOptions.js

@@ -15,6 +15,7 @@ const schema = utils_validation_1.Joi.object({
     docsDir: isStringOrArrayOfStrings.default(["docs"]),
     blogDir: isStringOrArrayOfStrings.default(["blog"]),
     removeDefaultStopWordFilter: utils_validation_1.Joi.boolean().default(false),
+    removeDefaultStemmer: utils_validation_1.Joi.boolean().default(false),
     highlightSearchTermsOnTargetPage: utils_validation_1.Joi.boolean().default(false),
     searchResultLimits: utils_validation_1.Joi.number().default(8),
     searchResultContextMaxLength: utils_validation_1.Joi.number().default(50),

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@easyops-cn/docusaurus-search-local",
-  "version": "0.23.2",
+  "version": "0.24.1",
   "description": "An offline/local search plugin for Docusaurus v2.",
   "repository": "https://github.com/easyops-cn/docusaurus-search-local",
   "homepage": "https://github.com/easyops-cn/docusaurus-search-local",
@@ -27,18 +27,19 @@
   "license": "MIT",
   "dependencies": {
     "@docusaurus/utils": "^2.0.0-beta.4",
+    "@docusaurus/utils-common": "^2.0.0-beta.20",
     "@docusaurus/utils-validation": "^2.0.0-beta.4",
     "@easyops-cn/autocomplete.js": "^0.38.1",
     "@node-rs/jieba": "^1.6.0",
     "cheerio": "^1.0.0-rc.3",
     "clsx": "^1.1.1",
     "debug": "^4.2.0",
-    "fs-extra": "^9.0.1",
+    "fs-extra": "^10.0.0",
     "klaw-sync": "^6.0.0",
     "lunr": "^2.3.9",
     "lunr-languages": "^1.4.0",
     "mark.js": "^8.11.1",
-    "tslib": "^2.2.0"
+    "tslib": "^2.4.0"
   },
   "devDependencies": {
     "@babel/core": "^7.12.3",
@@ -52,32 +53,33 @@
     "@types/enzyme": "^3.10.7",
     "@types/enzyme-adapter-react-16": "^1.0.6",
     "@types/fs-extra": "^9.0.2",
-    "@types/jest": "^26.0.14",
+    "@types/jest": "^27.0.0",
     "@types/klaw-sync": "^6.0.0",
     "@types/lunr": "^2.3.3",
     "@types/react": "^16.9.55",
     "@types/react-dom": "^16.9.8",
     "@types/react-helmet": "^6.1.0",
     "@types/react-router-dom": "^5.1.6",
-    "@typescript-eslint/eslint-plugin": "^4.4.1",
-    "@typescript-eslint/parser": "^4.4.1",
-    "babel-jest": "^26.5.2",
-    "concurrently": "^6.0.0",
+    "@typescript-eslint/eslint-plugin": "^5.0.0",
+    "@typescript-eslint/parser": "^5.0.0",
+    "babel-jest": "^28.0.0",
+    "concurrently": "^7.0.0",
     "copyfiles": "^2.4.0",
     "enzyme": "^3.11.0",
     "enzyme-adapter-react-16": "^1.15.5",
     "enzyme-to-json": "^3.6.1",
-    "eslint": "^7.11.0",
-    "eslint-config-prettier": "^7.0.0",
+    "eslint": "^8.0.0",
+    "eslint-config-prettier": "^8.0.0",
     "eslint-plugin-react": "^7.21.4",
     "eslint-plugin-react-hooks": "^4.1.2",
-    "husky": "^4.3.0",
+    "husky": "^8.0.0",
     "identity-obj-proxy": "^3.0.0",
-    "jest": "^26.5.3",
-    "lint-staged": "^10.4.1",
+    "jest": "^28.0.0",
+    "jest-environment-jsdom": "^28.1.0",
+    "lint-staged": "^12.0.0",
     "prettier": "^2.1.2",
     "rimraf": "^3.0.2",
     "standard-version": "^9.0.0",
-    "typescript": "^4.0.3"
+    "typescript": "^4.6.4"
   }
 }