fix(algolia): chunk and validate size of data (GoogleChrome#220)

* fix(algolia): chunk and validate size of data

* refactor(algolia): truncate post content instead of skipping it

* refactor(algolia): small tweaks to indexing script

* refactor(algolia): give all objects uuid and date when indexed so we can update them and remove old pages easily

algolia.js

@@ -15,9 +15,40 @@
  */
 require('dotenv').config();
 const algoliasearch = require('algoliasearch');
+
 const fs = require('fs');
+const {sizeof} = require('sizeof');
+
+const maxChunkSizeInBytes = 10000000; // 10,000,000
+
+/**
+ * Chunks array of AlgoliaCollectionItem into array of array of AlgoliaCollectionItem smaller than 10 MB.
+ *
+ * @param {AlgoliaCollectionItem[]} arr
+ * @return {AlgoliaCollectionItem[][]}
+ */
+const chunkAlgolia = arr => {
+  const chunked = [];
+  let tempSizeInBytes = 0;
+  let temp = [];
+  for (const current of arr) {
+    const currentSizeInBytes = sizeof(current);
+    if (tempSizeInBytes + currentSizeInBytes < maxChunkSizeInBytes) {
+      temp.push(current);
+      tempSizeInBytes += currentSizeInBytes;
+    } else {
+      chunked.push(temp);
+      temp = [current];
+      tempSizeInBytes = currentSizeInBytes;
+    }
+  }
+  chunked.push(temp);
+  return chunked;
+};
 
 async function index() {
+  const indexedOn = new Date();
+
   if (!process.env.ALGOLIA_APP_ID || !process.env.ALGOLIA_API_KEY) {
     console.warn('Missing Algolia environment variables, skipping indexing.');
     return;
@@ -26,17 +57,45 @@ async function index() {
   const raw = fs.readFileSync('dist/algolia.json', 'utf-8');
   const algoliaData = JSON.parse(raw);
 
+  // Set date of when object is being added to algolia
+  algoliaData.map(e => {
+    e.indexedOn = indexedOn.getTime();
+    return e;
+  });
+
+  const chunkedAlgoliaData = chunkAlgolia(algoliaData);
+  const postsCount = algoliaData.length;
+
   const client = algoliasearch(
     process.env.ALGOLIA_APP_ID,
     process.env.ALGOLIA_API_KEY
   );
   const index = client.initIndex('prod_developer_chrome');
 
-  console.log(`Indexing ${algoliaData.length} articles`);
+  console.log(
+    `Indexing ${postsCount} articles amongst ${chunkedAlgoliaData.length} chunk(s).`
+  );
+
+  // When indexing data we mark these two fields as fields that can be filtered by.
+  await index.setSettings({
+    attributesForFaceting: ['locale', 'tags'],
+  });
+
+  // Update algolia index with new data
+  for (let i = 0; i < chunkedAlgoliaData.length; i++) {
+    await index.saveObjects(chunkedAlgoliaData[i], {
+      autoGenerateObjectIDIfNotExist: true,
+    });
+  }
+
+  console.log('Updated algolia data.');
 
-  await index.replaceAllObjects(algoliaData, {
-    autoGenerateObjectIDIfNotExist: true,
+  console.log('Deleting old data no longer in algolia.json.');
+  await index.deleteBy({
+    filters: `indexedOn < ${indexedOn.getTime()}`,
   });
+  console.log('Deleted old data.');
+
   console.log('Done!');
 }
 

package.json

@@ -72,6 +72,7 @@
     "ava": "^3.12.1",
     "chalk": "^4.1.0",
     "cheerio": "^1.0.0-rc.3",
+    "crypto": "^1.0.1",
     "csso": "^4.0.3",
     "dotenv": "^8.2.0",
     "eslint-plugin-ava": "^11.0.0",
@@ -103,6 +104,7 @@
     "rimraf": "^3.0.2",
     "rollup-plugin-copy": "^3.3.0",
     "rollup-plugin-terser": "^7.0.2",
+    "sizeof": "^1.0.0",
     "stylelint": "^13.7.0",
     "stylelint-config-sass-guidelines": "^7.1.0",
     "typescript": "^3.8.3",

site/_collections/algolia.js

@@ -15,8 +15,31 @@
  */
 
 const removeMarkdown = require('remove-markdown');
+const {createHash} = require('crypto');
+
 const {generateSrc} = require('../_shortcodes/img');
 
+/**
+ * Shrink the size of the given fulltext to fit within a certain limit, at the
+ * nearest found newline character.
+ *
+ * @param {string} content
+ * @param {number} [limit]
+ * @return {string}
+ */
+function limitText(content, limit = 7500) {
+  if (content.length <= limit) {
+    return content;
+  }
+
+  // Find the nearest prior newline to the 10k limit.
+  let newlineIndex = content.lastIndexOf('\n', limit);
+  if (newlineIndex === -1) {
+    newlineIndex = limit;
+  }
+  return content.slice(0, newlineIndex);
+}
+
 /**
  * @param {EleventyCollectionObject} collections
  * @returns {AlgoliaCollectionItem[]}
@@ -38,11 +61,12 @@ module.exports = collections => {
     algoliaCollectionItems.push({
       title: item.data.title,
       description: item.data.description,
-      content: removeMarkdown(item.template.frontMatter.content),
+      content: limitText(removeMarkdown(item.template.frontMatter.content)),
       url: item.url,
       tags: item.data.tags || [],
       locale: item.data.locale,
       photo: item.data.hero && generateSrc(item.data.hero),
+      objectID: createHash('md5').update(item.url).digest('hex'),
     });
   }
   return algoliaCollectionItems;

types/site/_collections/algolia.d.ts

@@ -23,6 +23,7 @@ declare global {
     tags: string[];
     locale: string;
     photo?: string;
+    objectID: string;
   }
 }