add fetch-based checks to CI for new data entries

.github/workflows/test-and-validate.yml

@@ -1,6 +1,11 @@
 name: Test and Validate
 
-on: [push, pull_request]
+on:
+  push:
+    branches: [main]
+  pull_request:
+    paths:
+      - react-native-libraries.json
 
 jobs:
   test:
@@ -19,3 +24,7 @@ jobs:
       run: yarn data:validate
     - name: Test react-native-libraries.json
       run: yarn data:test
+    - name: Check new entries in react-native-libraries.json
+      run: yarn ci:validate
+      env:
+        CI_CHECKS_TOKEN: ${{ secrets.CI_CHECKS_TOKEN }}

package.json

@@ -12,6 +12,7 @@
     "libraries:recalculate": "node --no-warnings scripts/recalculate-popularity.js",
     "libraries:format": "prettier --write react-native-libraries.json",
     "libraries:check": "node --no-warnings scripts/check-resources.js",
+    "ci:validate": "node --no-warnings scripts/validate-new-entries.js",
     "precommit": "simple-git-hooks && lint-staged"
   },
   "dependencies": {

scripts/build-and-score-data.js

@@ -6,7 +6,7 @@ import { calculateDirectoryScore, calculatePopularityScore } from './calculate-s
 import { fetchGithubData, fetchGithubRateLimit, loadGitHubLicenses } from './fetch-github-data.js';
 import { fetchNpmData, fetchNpmDataBulk } from './fetch-npm-data.js';
 import fetchReadmeImages from './fetch-readme-images.js';
-import { fillNpmName, sleep } from './helpers.js';
+import { fillNpmName, hasMismatchedPackageData, sleep } from './helpers.js';
 import debugGithubRepos from '../debug-github-repos.json' assert { type: 'json' };
 import githubRepos from '../react-native-libraries.json' assert { type: 'json' };
 import { isLaterThan, TimeRange } from '../util/datetime.js';
@@ -45,9 +45,7 @@ const buildAndScoreData = async () => {
 
   // Detect mismatched package and package.json content
   data.forEach(project => {
-    if (
-      (project.npmPkg ?? project.githubUrl.split('/').pop()).toLowerCase() !== project.github.name
-    ) {
+    if (hasMismatchedPackageData(project)) {
       mismatchedRepos.push(project);
     }
   });
@@ -210,7 +208,7 @@ const buildAndScoreData = async () => {
   }
 };
 
-async function fetchGithubDataThrottled({ data, chunkSize, staggerMs }) {
+export async function fetchGithubDataThrottled({ data, chunkSize, staggerMs }) {
   let results = [];
   const chunks = chunk(data, chunkSize);
   for (const c of chunks) {
@@ -233,8 +231,8 @@ async function fetchGithubDataThrottled({ data, chunkSize, staggerMs }) {
   return results;
 }
 
-function getDataForFetch() {
-  if (wantedPackageName) {
+function getDataForFetch(wantedPackage) {
+  if (wantedPackage) {
     const match = DATASET.find(
       entry =>
         entry?.npmPkg === wantedPackageName || entry.githubUrl.endsWith(`/${wantedPackageName}`)
@@ -249,7 +247,7 @@ function getDataForFetch() {
 }
 
 async function loadRepositoryDataAsync() {
-  const data = getDataForFetch();
+  const data = getDataForFetch(wantedPackageName);
 
   let githubResultsFileExists = false;
   try {

scripts/fetch-github-data.js

@@ -9,7 +9,7 @@ config();
 
 const GRAPHQL_API = 'https://api.github.com/graphql';
 
-const AUTHORIZATION = `bearer ${process.env.GITHUB_TOKEN}`;
+const AUTHORIZATION = `bearer ${process.env.GITHUB_TOKEN ?? process.env.CI_CHECKS_TOKEN}`;
 
 const licenses = {};
 
@@ -87,7 +87,7 @@ const parseUrl = url => {
 
 export const fetchGithubData = async (data, retries = 2) => {
   if (retries < 0) {
-    console.warn(`[GH] ERROR fetching ${data.githubUrl} - OUT OF RETRIES`);
+    console.error(`[GH] ERROR fetching ${data.githubUrl} - OUT OF RETRIES`);
     return data;
   }
   try {
@@ -151,6 +151,7 @@ const createRepoDataWithResponse = (json, monorepo) => {
 
       json.newArchitecture = Boolean(packageJson.codegenConfig);
       json.name = packageJson.name;
+      json.isPackagePrivate = packageJson.private ?? false;
 
       if (monorepo) {
         json.homepageUrl = packageJson.homepage;
@@ -215,6 +216,7 @@ const createRepoDataWithResponse = (json, monorepo) => {
     },
     name: json.name,
     fullName: json.nameWithOwner,
+    isPrivate: json.isPackagePrivate,
     description: json.description,
     topics: json.topics,
     license: json.licenseInfo,

scripts/helpers.js

@@ -23,3 +23,9 @@ export function processTopics(topics) {
     )
     .filter(topic => topic?.length);
 }
+
+export function hasMismatchedPackageData(project) {
+  return (
+    (project.npmPkg ?? project.githubUrl.split('/').at(-1)).toLowerCase() !== project.github?.name
+  );
+}

scripts/validate-new-entries.js

@@ -0,0 +1,80 @@
+import fetch from 'cross-fetch';
+import differenceWith from 'lodash/differenceWith.js';
+import isEqual from 'lodash/isEqual.js';
+
+import { fetchGithubData } from './fetch-github-data.js';
+import { fetchNpmData } from './fetch-npm-data.js';
+import { fillNpmName, hasMismatchedPackageData } from './helpers.js';
+import libraries from '../react-native-libraries.json' assert { type: 'json' };
+
+async function makeBaseFileQuery() {
+  const result = await fetch(
+    'https://raw.githubusercontent.com/react-native-community/directory/main/react-native-libraries.json'
+  );
+  return await result.json();
+}
+
+const mainData = await makeBaseFileQuery();
+
+if (libraries.length !== mainData.length) {
+  console.log('🔍️ Detected changes in data entries, checking!');
+
+  const modifiedEntries = differenceWith(libraries, mainData, isEqual);
+
+  const checkResults = await Promise.all(
+    modifiedEntries.map(async entry => {
+      const entryWithNpmData = await fetchNpmData(fillNpmName(entry));
+
+      if (!entryWithNpmData.npm) {
+        console.error(
+          `Unable to fetch npm package data for ${entryWithNpmData.npmPkg} package! Please make sure that the package exist in npm registry.`
+        );
+        return false;
+      }
+
+      const entryWithGitHubData = await fetchGithubData(entryWithNpmData);
+
+      if (!entryWithGitHubData.github) {
+        console.error(`Unable to fetch data from ${entryWithGitHubData.githubUrl} repository!`);
+        return false;
+      }
+
+      if (!entryWithGitHubData.github.isPackagePrivate) {
+        console.error(
+          `Extracted 'package.json' from ${entryWithGitHubData.githubUrl} is marked as private! You might be linking to the monorepo/workspace root, instead of wanted package directory.`
+        );
+        return false;
+      }
+
+      if (!entryWithGitHubData.github.name) {
+        console.error(
+          `Extracted 'package.json' from ${entryWithGitHubData.githubUrl} does not contains package name! You might be linking to the monorepo/workspace root, instead of wanted package directory.`
+        );
+        return false;
+      }
+
+      if (hasMismatchedPackageData(entryWithGitHubData)) {
+        console.error(
+          `Package name extracted from 'package.json' at given GitHub repository URL differs with package name in the directory data!`
+        );
+        console.error(
+          `- Supplied package name: ${entryWithGitHubData.npmPkg ?? entryWithGitHubData.githubUrl.split('/').at(-1)}`
+        );
+        console.error(
+          `- Extracted package name: ${entryWithGitHubData.github.name ?? entryWithGitHubData.github.fullName.split('/').at(-1)}`
+        );
+
+        return false;
+      }
+      return true;
+    })
+  );
+
+  if (checkResults.filter(result => !result).length > 0) {
+    console.error('\n❌ There were errors spotted during new entries check!');
+    process.exit(1);
+  }
+  console.log('✅ All checks have passed!');
+}
+
+console.log('✅ There was no data changes detected!');

types/index.ts

@@ -54,6 +54,14 @@ export type Library = {
   newArchitectureNote?: string;
   alternatives?: string[];
   github: {
+    name: string;
+    isPackagePrivate: boolean;
+    fullName: string;
+    description: string;
+    topics?: string[];
+    hasTypes?: boolean;
+    newArchitecture?: boolean;
+    isArchived?: boolean;
     urls: {
       repo: string;
       clone: string;
@@ -72,10 +80,6 @@ export type Library = {
       stars: number;
       forks: number;
     };
-    name: string;
-    fullName: string;
-    description: string;
-    topics?: string[];
     license: {
       key: string;
       name: string;
@@ -90,9 +94,6 @@ export type Library = {
       publishedAt: Date | string;
       isPrerelease: boolean;
     };
-    hasTypes?: boolean;
-    newArchitecture?: boolean;
-    isArchived?: boolean;
   };
   npm?: {
     downloads?: number;