Add EasyList filter support

2026-05-16 22:07:39 -07:00
parent f4f1a7a78d
commit 46444b193b
12 changed files with 171818 additions and 228 deletions
--- a/test/archiver.test.mjs
+++ b/test/archiver.test.mjs
@@ -1,6 +1,82 @@
 import assert from "node:assert/strict";
 import test from "node:test";
-import { renderPage } from "../src/archiver.mjs";
+import {
+  getCosmeticCssForHostname,
+  parseFilterRules,
+  renderPage,
+  shouldBlockRequestWithRules
+} from "../src/archiver.mjs";
+
+test("parses EasyList-style network rules, exceptions, and badfilter entries", () => {
+  const rules = parseFilterRules(`
+    [Adblock Plus 2.0]
+    ||ads.example.com^$script,third-party
+    @@||ads.example.com/allowed.js$script,domain=publisher.test
+    banner$~image
+    ||disabled.example^$script
+    ||disabled.example^$script,badfilter
+  `);
+
+  assert.equal(
+    shouldBlockRequestWithRules(
+      rules,
+      "https://ads.example.com/banner.js",
+      "script",
+      "www.publisher.test"
+    ),
+    true
+  );
+  assert.equal(
+    shouldBlockRequestWithRules(
+      rules,
+      "https://ads.example.com/banner.png",
+      "image",
+      "www.publisher.test"
+    ),
+    false
+  );
+  assert.equal(
+    shouldBlockRequestWithRules(
+      rules,
+      "https://ads.example.com/allowed.js",
+      "script",
+      "www.publisher.test"
+    ),
+    false
+  );
+  assert.equal(
+    shouldBlockRequestWithRules(
+      rules,
+      "https://disabled.example/ad.js",
+      "script",
+      "www.publisher.test"
+    ),
+    false
+  );
+});
+
+test("applies cosmetic filters with domain exceptions and skips unsupported procedural selectors", () => {
+  const rules = parseFilterRules(`
+    ##.generic-ad
+    example.com##.site-ad
+    example.com#@#.generic-ad
+    ~news.example.com,example.com##.except-news
+    example.*##.entity-ad
+    bad.example##div:has-text(ad)
+    foo.com#$#.adguard { display: none !important; }
+  `);
+
+  const exampleCss = getCosmeticCssForHostname(rules, "www.example.com").join("\n");
+  assert.doesNotMatch(exampleCss, /\.generic-ad/);
+  assert.match(exampleCss, /\.site-ad/);
+  assert.match(exampleCss, /\.except-news/);
+  assert.match(exampleCss, /\.entity-ad/);
+
+  const newsCss = getCosmeticCssForHostname(rules, "news.example.com").join("\n");
+  assert.doesNotMatch(newsCss, /\.except-news/);
+  assert.equal(getCosmeticCssForHostname(rules, "bad.example").length, 1);
+  assert.match(getCosmeticCssForHostname(rules, "foo.com").join("\n"), /\.adguard/);
+});

 test("renderPage serializes CSSOM-inserted style rules", async () => {
  const html = `<!doctype html>
--- a/test/asset-inliner.test.mjs
+++ b/test/asset-inliner.test.mjs
@@ -81,3 +81,25 @@ test("external asset reporting parses srcset-like attributes without splitting U
    "https://media.example/photos/id/master/w_240,c_limit/photo.jpg"
  ]);
 });
+
+test("asset inliner skips URLs blocked by the filter hook", async () => {
+  const blocked = [];
+  const inliner = new AssetInliner({
+    shouldBlockAsset: (url, resourceType) => {
+      blocked.push([url, resourceType]);
+      return true;
+    }
+  });
+
+  const output = await inliner.inlineHtml(`
+    <link rel="stylesheet" href="https://ads.example/ad.css">
+    <img src="https://ads.example/ad.png">
+  `, "https://publisher.example/article");
+
+  assert.doesNotMatch(output, /ad\.css/);
+  assert.match(output, /data:image\/gif;base64/);
+  assert.deepEqual(blocked, [
+    ["https://ads.example/ad.css", "stylesheet"],
+    ["https://ads.example/ad.png", "image"]
+  ]);
+});