Simplify html filters usage (#12)

packages/adblocker-webextension/src/index.ts

@@ -12,7 +12,6 @@ import { parse } from 'tldts-experimental';
 import {
   FiltersEngine,
   HTMLSelector,
-  HTMLModifier,
   isUTF8,
   Request,
   StreamingHtmlFilter,
@@ -136,13 +135,12 @@ export function filterRequestHTML(
   filterResponseData: Browser['webRequest']['filterResponseData'],
   { id }: { id: string },
   rules: HTMLSelector[],
-  modifiers: HTMLModifier[],
 ): void {
   // Create filter to observe loading of resource
   const filter = filterResponseData(id) as StreamFilter;
   const decoder = new TextDecoder();
   const encoder = new TextEncoder();
-  const htmlFilter = new StreamingHtmlFilter(rules, modifiers);
+  const htmlFilter = new StreamingHtmlFilter(rules);
 
   const teardown = (event: { data?: ArrayBuffer }) => {
     // Before disconnecting our streaming filter, we need to be extra careful
@@ -408,14 +406,8 @@ export class WebExtensionBlocker extends FiltersEngine {
       typeof TextEncoder !== 'undefined'
     ) {
       const htmlFilters = this.getHtmlFilters(request);
-      const htmlModifiers = this.getHtmlModifiers(request);
-      if (htmlFilters.length !== 0 && htmlModifiers.length !== 0) {
-        filterRequestHTML(
-          browser.webRequest.filterResponseData,
-          request,
-          htmlFilters,
-          htmlModifiers,
-        );
+      if (htmlFilters.length !== 0) {
+        filterRequestHTML(browser.webRequest.filterResponseData, request, htmlFilters);
       }
     }
   }

packages/adblocker/src/engine/bucket/network.ts

@@ -187,7 +187,7 @@ export default class NetworkFilterBucket {
     return this.badFiltersIds.has(filter.getId());
   }
 
-  public getHtmlModifiers(
+  public getHTMLFilters(
     request: Request,
     isFilterExcluded?: (filter: NetworkFilter) => boolean,
   ): NetworkFilter[] {

packages/adblocker/src/engine/engine.ts

@@ -19,7 +19,7 @@ import {
   fetchResources,
   fullLists,
 } from '../fetch.js';
-import { HTMLModifier, HTMLSelector } from '../html-filtering.js';
+import { HTMLSelector } from '../html-filtering.js';
 import CosmeticFilter from '../filters/cosmetic.js';
 import NetworkFilter from '../filters/network.js';
 import { block } from '../filters/dsl.js';
@@ -95,11 +95,16 @@ type NetworkFilterMatchingContext = {
   filterType: FilterType.NETWORK;
 };
 
-type CosmeticFilterMatchingContext = {
-  url: string;
-  callerContext: any; // Additional context given from user
-  filterType: FilterType.COSMETIC;
-};
+type CosmeticFilterMatchingContext =
+  | {
+      url: string;
+      callerContext: any; // Additional context given from user
+      filterType: FilterType.COSMETIC;
+    }
+  | {
+      request: Request; // For HTML Filters
+      filterType: FilterType.COSMETIC;
+    };
 
 type NetworkFilterMatchEvent = (request: Request, result: BlockingResponse) => void;
 
@@ -835,81 +840,86 @@ export default class FilterEngine extends EventEmitter<EngineEventHandlers> {
   /**
    * Return a list of HTML filtering rules.
    */
-  public getHtmlFilters({
-    // Page information
-    url,
-    hostname,
-    domain,
-
-    callerContext,
-  }: {
-    url: string;
-    hostname: string;
-    domain: string | null | undefined;
-
-    callerContext?: any | undefined;
-  }): HTMLSelector[] {
+  public getHtmlFilters(request: Request): HTMLSelector[] {
     const htmlSelectors: HTMLSelector[] = [];
 
-    if (this.config.enableHtmlFiltering === false || this.config.loadCosmeticFilters === false) {
+    if (this.config.enableHtmlFiltering === false) {
       return htmlSelectors;
     }
 
-    domain ||= '';
+    if (this.config.loadCosmeticFilters === true) {
+      const domain = request.domain || '';
 
-    const { filters, unhides } = this.cosmetics.getHtmlFilters({
-      domain,
-      hostname,
-      isFilterExcluded: this.isFilterExcluded.bind(this),
-    });
-    const exceptions = new Map(unhides.map((unhide) => [unhide.getSelector(), unhide]));
+      const { filters, unhides } = this.cosmetics.getHtmlFilters({
+        domain,
+        hostname: request.hostname,
+        isFilterExcluded: this.isFilterExcluded.bind(this),
+      });
+      const exceptions = new Map(unhides.map((unhide) => [unhide.getSelector(), unhide]));
 
-    for (const filter of filters) {
-      const extended = filter.getExtendedSelector();
-      if (extended === undefined) {
-        continue;
-      }
-      const exception = exceptions.get(filter.getSelector());
-      if (exception !== undefined) {
-        htmlSelectors.push(extended);
+      for (const filter of filters) {
+        const extended = filter.getExtendedSelector();
+        if (extended === undefined) {
+          continue;
+        }
+        const exception = exceptions.get(filter.getSelector());
+        if (exception !== undefined) {
+          htmlSelectors.push(extended);
+        }
+        this.emit(
+          'filter-matched',
+          { filter, exception },
+          {
+            request,
+            filterType: FilterType.COSMETIC,
+          },
+        );
       }
-      this.emit(
-        'filter-matched',
-        { filter, exception },
-        {
-          url,
-          callerContext,
-          filterType: FilterType.COSMETIC,
-        },
-      );
     }
 
-    if (htmlSelectors.length !== 0) {
-      this.emit('html-filtered', htmlSelectors, url);
-    }
+    if (this.config.loadNetworkFilters === true) {
+      const replaceFilters = this.filters.getHTMLFilters(
+        request,
+        this.isFilterExcluded.bind(this),
+      );
 
-    return htmlSelectors;
-  }
+      if (replaceFilters.length !== 0) {
+        const exception = this.exceptions.match(request, this.isFilterExcluded.bind(this));
+        let modifiers = [];
+        for (const filter of replaceFilters) {
+          const modifier = filter.getHtmlModifier();
 
-  public getHtmlModifiers(request: Request): HTMLModifier[] {
-    const htmlModifiers: HTMLModifier[] = [];
+          if (modifier !== null) {
+            if (!exception) {
+              modifiers.push(['replace', modifier]);
+            }
 
-    if (this.config.enableHtmlFiltering === false || this.config.loadNetworkFilters === false) {
-      return htmlModifiers;
+            this.emit(
+              'filter-matched',
+              { filter, exception },
+              {
+                request,
+                filterType: FilterType.COSMETIC,
+              },
+            );
+          } else {
+            // Disable all replace modifiers if empty replace modifier found
+            modifiers = [];
+            break;
+          }
+        }
+      }
+
+      if (modifiers.length !== 0) {
+        htmlSelectors.push(...modifiers.map((modifier) => ['replace', modifier]));
+      }
     }
 
-    const filters = this.filters.getHtmlModifiers(request, this.isFilterExcluded.bind(this));
-
-    for (const filter of filters) {
-      const modifier = filter.getHtmlModifier();
-      // Disable all replace modifiers if empty replace modifier found
-      if (modifier === null) {
-        return [];
-      }
-      htmlModifiers.push(modifier);
+    if (htmlSelectors.length !== 0) {
+      this.emit('html-filtered', htmlSelectors, request.url);
     }
 
-    return htmlModifiers;
+    return htmlSelectors;
   }
 
   /**

packages/adblocker/src/html-filtering.ts

@@ -10,7 +10,11 @@
 // which is able to consume an HTML document over time and filter part of it
 // using adblocker selectors.
 
-export type HTMLSelector = readonly ['script', readonly string[]];
+export type HTMLModifier = readonly [RegExp, string];
+
+export type HTMLSelector =
+  | readonly ['script', readonly string[]]
+  | readonly ['replace', HTMLModifier];
 
 export function extractHTMLSelectorFromRule(rule: string): HTMLSelector | undefined {
   if (rule.startsWith('^script') === false) {
@@ -139,7 +143,10 @@ type Patterns = readonly [readonly string[], readonly RegExp[]][];
 export function extractSelectorsFromRules(filter: HTMLSelector[]): Patterns {
   const patterns: [string[], RegExp[]][] = [];
 
-  for (const [, selectors] of filter) {
+  for (const [type, selectors] of filter) {
+    if (type !== 'script') {
+      continue;
+    }
     const plainPatterns: string[] = [];
     const regexpPatterns: RegExp[] = [];
 
@@ -220,8 +227,6 @@ export function removeTagsFromHtml(html: string, toRemove: [number, string][]):
   return filteredHtml;
 }
 
-export type HTMLModifier = readonly [RegExp, string];
-
 function applyModifiersToHtml(html: string, modifiers: HTMLModifier[]): string {
   if (modifiers.length === 0) {
     return html;
@@ -239,9 +244,18 @@ export default class StreamingHtmlFilter {
   private readonly patterns: Patterns;
   private readonly modifiers: HTMLModifier[];
 
-  constructor(selectors: HTMLSelector[], modifiers: HTMLModifier[] = []) {
+  constructor(selectors: HTMLSelector[]) {
     this.buffer = '';
-    this.patterns = extractSelectorsFromRules(selectors);
+    const modifiers = [];
+    const rules = [];
+    for (const selector of selectors) {
+      if (selector[0] === 'replace') {
+        modifiers.push(selector[1]);
+      } else if (selector[0] === 'script') {
+        rules.push(selector);
+      }
+    }
+    this.patterns = extractSelectorsFromRules(rules);
     this.modifiers = modifiers;
   }
 

packages/adblocker/src/index.ts

@@ -40,4 +40,4 @@ export { isUTF8 } from './encoding.js';
 export { default as Config } from './config.js';
 export { default as Resources } from './resources.js';
 export { default as StreamingHtmlFilter } from './html-filtering.js';
-export type { HTMLModifier, HTMLSelector } from './html-filtering.js';
+export type { HTMLSelector } from './html-filtering.js';