core: refactor out entity-classification lookup tables (#14744)

core/runner.js

@@ -154,37 +154,26 @@ class Runner {
 
     /** @type {Array<LH.Result.LhrEntity>} */
     const entities = [];
-    /** @type {Record<string, number>} */
-    const entityIndexByOrigin = {};
-    /** @type {Record<string, number>} */
-    const entityIndexByName = {};
-
     for (const [entity, entityUrls] of classifiedEntities.urlsByEntity) {
+      const uniqueOrigins = new Set();
+      for (const url of entityUrls) {
+        const origin = UrlUtils.getOrigin(url);
+        if (origin) uniqueOrigins.add(origin);
+      }
+
       /** @type {LH.Result.LhrEntity} */
       const shortEntity = {
         name: entity.name,
         homepage: entity.homepage,
+        origins: [...uniqueOrigins],
       };
-
       // Reduce payload size in LHR JSON by omitting whats falsy.
       if (entity === classifiedEntities.firstParty) shortEntity.isFirstParty = true;
       if (entity.isUnrecognized) shortEntity.isUnrecognized = true;
-
-      const id = entities.push(shortEntity) - 1;
-      for (const url of entityUrls) {
-        const origin = UrlUtils.getOrigin(url);
-        if (!origin) continue;
-        entityIndexByOrigin[origin] = id;
-      }
-      entityIndexByName[shortEntity.name] = id;
+      entities.push(shortEntity);
     }
 
-    return {
-      list: entities,
-      firstParty: classifiedEntities.firstParty?.name,
-      entityIndexByOrigin,
-      entityIndexByName,
-    };
+    return entities;
   }
 
   /**

proto/lighthouse-result.proto

@@ -234,7 +234,7 @@ message LighthouseResult {
   google.protobuf.Value full_page_screenshot = 19;
 
   // Entity classification data.
-  Entities entities = 20;
+  repeated LhrEntity entities = 20;
 }
 
 // Message containing a category
@@ -614,26 +614,11 @@ message StackPack {
   map<string, string> descriptions = 4;
 }
 
-// Message containing Entity Classification.
-message Entities {
-  // List of entities.
-  repeated LhrEntity list = 1;
-
-  // Name of the first-party entity.
-  string firstParty = 2;
-
-  // Lookup table that maps URL origins to corresponding index in entities.
-  map<string, uint32> entityIndexByOrigin = 3;
-
-  // Lookup table that maps entity names to corresponding index in entities.
-  map<string, uint32> entityIndexByName = 4;
-}
-
 // Message containing an Entity.
 message LhrEntity {
   // Name of the entity.
   string name = 1;
-  
+
   // An optional homepage URL of the entity.
   optional string homepage = 2;
 
@@ -645,4 +630,7 @@ message LhrEntity {
 
   // An optional flag indicating if the entity is not recognized.
   optional bool isUnrecognized = 5;
+
+  // A list of URL origin strings that belong to this entity.
+  repeated string origins = 6;
 }

report/renderer/report-ui-features.js

@@ -322,7 +322,7 @@ export class ReportUIFeatures {
    */
   _getThirdPartyRows(rowEls, finalDisplayedUrl) {
     const finalDisplayedUrlRootDomain = Util.getRootDomain(finalDisplayedUrl);
-    const firstPartyEntityName = this.json.entities?.firstParty;
+    const firstPartyEntityName = this.json.entities?.find(e => e.isFirstParty === true)?.name;
 
     /** @type {Array<HTMLElement>} */
     const thirdPartyRowEls = [];

report/renderer/report-utils.js

@@ -109,11 +109,11 @@ class ReportUtils {
 
   /**
    * Mark TableItems/OpportunityItems with entity names.
-   * @param {LH.Result.Entities|undefined} entityClassification
+   * @param {LH.Result.Entities|undefined} entities
    * @param {import('../../types/lhr/audit-result').Result} audit
    */
-  static classifyEntities(entityClassification, audit) {
-    if (!entityClassification) return;
+  static classifyEntities(entities, audit) {
+    if (!entities) return;
     if (audit.details?.type !== 'opportunity' && audit.details?.type !== 'table') {
       return;
     }
@@ -130,17 +130,15 @@ class ReportUtils {
       const url = urlLocatorFn(item);
       if (!url) continue;
 
-      let origin;
+      let origin = '';
       try {
         // Non-URLs can appear in valueType: url columns, like 'Unattributable'
         origin = Util.parseURL(url).origin;
       } catch {}
       if (!origin) continue;
 
-      const entityIndex = entityClassification.entityIndexByOrigin[origin];
-      if (entityIndex === undefined) return;
-      const entity = entityClassification.list[entityIndex];
-      item.entity = entity.name;
+      const entity = entities.find(e => e.origins.includes(origin));
+      if (entity) item.entity = entity.name;
     }
   }
 

report/test/renderer/report-ui-features-test.js

@@ -240,39 +240,28 @@ describe('ReportUIFeatures', () => {
 
         before(() => {
           // Setup entity-classification with recognized entities first.
-          lhrJson.entities = {
-            firstParty: 'example.com',
-            list: [
-              {
-                name: 'example.com',
-                isFirstParty: true,
-                isUnrecognized: true,
-              },
-              {
-                name: 'cdn.com',
-                isUnrecognized: true,
-              },
-              {
-                name: 'notexample.com',
-                isUnrecognized: true,
-              },
-              {
-                name: 'google.com',
-              },
-            ],
-            entityIndexByOrigin: {
-              'http://www.example.com': 0,
-              'http://www.cdn.com': 1,
-              'http://www.notexample.com': 2,
-              'http://www.google.com': 3,
+          lhrJson.entities = [
+            {
+              name: 'example.com',
+              isFirstParty: true,
+              isUnrecognized: true,
+              origins: ['http://www.example.com'],
             },
-            entityIndexByName: {
-              'example.com': 0,
-              'cdn.com': 1,
-              'notexample.com': 2,
-              'google.com': 3,
+            {
+              name: 'cdn.com',
+              isUnrecognized: true,
+              origins: ['http://www.cdn.com'],
             },
-          };
+            {
+              name: 'notexample.com',
+              isUnrecognized: true,
+              origins: ['http://www.notexample.com'],
+            },
+            {
+              name: 'google.com',
+              origins: ['http://www.google.com'],
+            },
+          ];
 
           // Entity resolution is done during prepareReportResult
           const result = ReportUtils.prepareReportResult(lhrJson);

types/lhr/lhr.d.ts

@@ -155,18 +155,8 @@ declare module Result {
 
   /**
    * Entity classification for the run, for resolving URLs/items to entities in report.
-   * The two lookup tables (LUT) below provide space-optimized, O(1) index lookup into entities.list.
    */
-  interface Entities {
-    /** All entities (1st and 3rd party) discovered during the run */
-    list: Array<LhrEntity>;
-    /** Name of the first-party entity */
-    firstParty?: string;
-    /** Entity-name to entity index lookup table  */
-    entityIndexByName: Record<string, number>;
-    /** URL origin to entity index lookup table */
-    entityIndexByOrigin: Record<string, number>;
-  }
+  interface Entities extends Array<LhrEntity> {}
 
   /**
    * An entity that's either recognized by third-party-web or made up by Lighthouse.
@@ -182,6 +172,8 @@ declare module Result {
     isFirstParty?: boolean;
     /** Is this entity recognized by third-party-web? */
     isUnrecognized?: boolean;
+    /** List of origin strings that belong to this entity found in network records. */
+    origins: Array<string>;
   }
 
   /**