feat(third-party): add handling of Angular CLI schematics, and rework…

… registry subset (#169)

* feat(third-party): add handling of Angular CLI schematics, and rework registry subset

This way we don't change the keywords anymore, and can do actual tests on the package contents.

cc @manekinekko 🙌

* avoid temp variable

* chore: rename to computedKeywords and computedMetadata

* chore: simplify code

src/__tests__/__snapshots__/config.test.js.snap

@@ -28,6 +28,7 @@ Object {
     "attributesForFaceting": Array [
       "filterOnly(concatenatedName)",
       "searchable(keywords)",
+      "searchable(computedKeywords)",
       "searchable(owner.name)",
       "deprecated",
     ],

src/__tests__/formatPkg.test.js

@@ -24,28 +24,49 @@ it('truncates long readmes', () => {
     formatted.readme.length - truncatedEnding.length
   );
 
-  expect(formatted.readme).toHaveLength(451070);
+  expect(formatted.readme).toHaveLength(451134);
   expect(ending).toBe(truncatedEnding);
 
   formatted.lastCrawl = '<!-- date replaced -->';
   expect(formatted).toMatchSnapshot();
 });
 
+describe('adds angular cli schematics', () => {
+  const angularSchema = {
+    name: 'angular-cli-schema-1',
+    schematics: 'bli-blo',
+    keywords: ['hi'],
+    lastPublisher: { name: 'angular-god' },
+  };
+
+  const formatted = formatPkg(angularSchema);
+  expect(formatted.keywords).toEqual(['hi', 'angular-cli-schematic']);
+  expect(formatted.computedKeywords).toEqual(['angular-cli-schematic']);
+  expect(formatted.computedMetadata).toEqual({
+    schematics: 'bli-blo',
+  });
+});
+
 describe('adds babel plugins', () => {
   const dogs = {
     name: '@babel/plugin-dogs',
+    keywords: 'babel',
     lastPublisher: { name: 'xtuc' },
   };
   const unofficialDogs = {
     name: 'babel-plugin-dogs',
+    keywords: ['dogs'],
     lastPublisher: { name: 'unknown' },
   };
 
   const formattedDogs = formatPkg(dogs);
   const formattedUnofficialDogs = formatPkg(unofficialDogs);
 
-  expect(formattedDogs.keywords).toEqual(['babel-plugin']);
-  expect(formattedUnofficialDogs.keywords).toEqual(['babel-plugin']);
+  expect(formattedDogs.keywords).toEqual(['babel', 'babel-plugin']);
+  expect(formattedUnofficialDogs.keywords).toEqual(['dogs', 'babel-plugin']);
+
+  expect(formattedDogs.computedKeywords).toEqual(['babel-plugin']);
+  expect(formattedUnofficialDogs.computedKeywords).toEqual(['babel-plugin']);
 });
 
 describe('adds vue-cli plugins', () => {
@@ -69,6 +90,10 @@ describe('adds vue-cli plugins', () => {
   expect(formattedDogs.keywords).toEqual(['vue-cli-plugin']);
   expect(formattedUnofficialDogs.keywords).toEqual(['vue-cli-plugin']);
   expect(formattedScopedDogs.keywords).toEqual(['vue-cli-plugin']);
+
+  expect(formattedDogs.computedKeywords).toEqual(['vue-cli-plugin']);
+  expect(formattedUnofficialDogs.computedKeywords).toEqual(['vue-cli-plugin']);
+  expect(formattedScopedDogs.computedKeywords).toEqual(['vue-cli-plugin']);
 });
 
 describe('test getRepositoryInfo', () => {

src/config.js

@@ -27,6 +27,7 @@ const defaultConfig = {
     attributesForFaceting: [
       'filterOnly(concatenatedName)' /* optionalFacetFilters to boost the name */,
       'searchable(keywords)',
+      'searchable(computedKeywords)',
       'searchable(owner.name)',
       'deprecated',
     ],

src/formatPkg.js

@@ -53,7 +53,8 @@ export default function formatPkg(pkg) {
 
   const owner = getOwner(repository, lastPublisher, author); // always favor the repository owner
   const badPackage = isBadPackage(owner);
-  const keywords = getKeywords(cleaned);
+  const { computedKeywords, computedMetadata } = getComputedData(cleaned, pkg);
+  const keywords = [...getKeywords(cleaned), ...computedKeywords]; // concat with the subset for backward compat
 
   const dependencies = cleaned.dependencies || {};
   const devDependencies = cleaned.devDependencies || {};
@@ -86,6 +87,8 @@ export default function formatPkg(pkg) {
     homepage: getHomePage(cleaned.homepage, cleaned.repository),
     license,
     keywords,
+    computedKeywords,
+    computedMetadata,
     created: Date.parse(cleaned.created),
     modified: Date.parse(cleaned.modified),
     lastPublisher,
@@ -209,31 +212,54 @@ function getVersions(cleaned) {
   return {};
 }
 
-const forcedKeywords = {
-  'babel-plugin': ({ name }) =>
-    name.startsWith('@babel/plugin') || name.startsWith('babel-plugin-'),
-  'vue-cli-plugin': ({ name }) =>
-    /^(@vue\/|vue-|@[\w-]+\/vue-)cli-plugin-/.test(name),
-};
+const registrySubsetRules = [
+  ({ name }) => ({
+    name: 'babel-plugin',
+    include:
+      name.startsWith('@babel/plugin') || name.startsWith('babel-plugin-'),
+  }),
+
+  ({ name }) => ({
+    name: 'vue-cli-plugin',
+    include: /^(@vue\/|vue-|@[\w-]+\/vue-)cli-plugin-/.test(name),
+  }),
+
+  (_, { schematics = '' }) => ({
+    name: 'angular-cli-schematic',
+    include: schematics.length > 0,
+    metadata: { schematics },
+  }),
+];
+
+function getComputedData(cleaned, original) {
+  const registrySubsets = registrySubsetRules.reduce(
+    (acc, matcher) => {
+      const { include, metadata, name } = matcher(cleaned, original);
+      return include
+        ? {
+            computedKeywords: [...acc.computedKeywords, name],
+            computedMetadata: {
+              ...acc.computedMetadata,
+              ...metadata,
+            },
+          }
+        : acc;
+    },
+    { computedKeywords: [], computedMetadata: {} }
+  );
+  return registrySubsets;
+}
 
 function getKeywords(cleaned) {
-  // Forced keywords
-  const keywords = [];
-  for (const keyword in forcedKeywords) {
-    if (forcedKeywords[keyword](cleaned)) {
-      keywords.push(keyword);
-    }
-  }
-
   if (cleaned.keywords) {
     if (Array.isArray(cleaned.keywords)) {
-      return [...cleaned.keywords, ...keywords];
+      return [...cleaned.keywords];
     }
     if (typeof cleaned.keywords === 'string') {
-      return [cleaned.keywords, ...keywords];
+      return [cleaned.keywords];
     }
   }
-  return [...keywords];
+  return [];
 }
 
 function getGitHubRepoInfo({ repository, gitHead = 'master' }) {