Add Lunr.js index script

1 files changed, 106 insertions, 0 deletions

diff --git a/scripts/lunr/preindex.js b/scripts/lunr/preindex.js
new file mode 100644
index 00000000..ffae7d8c
--- /dev/null
+++ b/scripts/lunr/preindex.js
@@ -0,0 +1,106 @@
+#!/usr/bin/env node
+
+/**
+ * @file preindex.js
+ * Creates data files required for Lunr search.
+ *
+ * This script creates two JSON files:
+ *   - lunr-index.js: A serialized search index.
+ *   - lunr-map.js: Maps index item IDs to their human-readable titles.
+ *
+ * @see https://lunrjs.com/guides/index_prebuilding.html
+ */
+
+/* eslint-disable no-console */
+
+const fs = require('fs');
+const lunr = require('lunr');
+const cheerio = require('cheerio');
+const glob = require('glob');
+
+const htmlSrc = 'public/';
+const outputDir = `${htmlSrc}assets/javascripts`;
+
+/**
+ * Find all HTML files within a given path,
+ * then execute a callback function to build the index.
+ */
+const buildIndex = (path, callback) => {
+  glob(`${path}/**/*.html`, callback);
+};
+
+/**
+ * Extracts text from a given HTML element.
+ *
+ * @param {cheerio} $
+ *   A Cheerio page object
+ * @param {String} element
+ *   An HTML element to search for
+ *
+ * @return {String}
+ *   All text contained within the given element
+ */
+const getText = ($, element) => {
+  const headingText = [];
+  $(element)
+    .toArray()
+    .forEach((el) => {
+      headingText.push($(el).text().replace('\n', ''));
+    });
+  return headingText.join(' ');
+};
+
+/**
+ * Build the index and output files.
+ */
+buildIndex(htmlSrc, (err, filenames) => {
+  if (err) {
+    console.error(err);
+  }
+
+  // Create an array of objects containing each page's text content.
+  const pages = [];
+  Object.keys(filenames).forEach((key) => {
+    const filename = filenames[key];
+    const $ = cheerio.load(fs.readFileSync(filename));
+    const title = getText($, 'h1');
+
+    if (title.length) {
+      pages.push({
+        id: filename.slice(htmlSrc.length),
+        h1: title,
+        h2: getText($, 'h2'),
+        h3: getText($, 'h3'),
+      });
+    }
+  });
+
+  // Build the serialized Lunr search index.
+  const idx = lunr((e) => {
+    e.ref('id');
+    e.field('h1', { boost: 10 });
+    e.field('h2', { boost: 5 });
+    e.field('h3', { boost: 2 });
+    pages.forEach((doc) => {
+      e.add(doc);
+    }, e);
+  });
+
+  // Write the index file.
+  fs.writeFile(`${outputDir}/lunr-index.json`, JSON.stringify(idx), (fsErr) => {
+    if (fsErr) {
+      console.error(fsErr);
+    }
+  });
+
+  // Write the map file.
+  // We can drop h2s and h3s from this since we don't display those in results.
+  const pageMap = pages.map(({ h2, h3, ...rest }) => {
+    return rest;
+  });
+  fs.writeFile(`${outputDir}/lunr-map.json`, JSON.stringify(pageMap), (fsErr) => {
+    if (fsErr) {
+      console.error(fsErr);
+    }
+  });
+});