AG-37643 Convert ABP's rules to AdGuard syntax.

Squashed commit of the following: commit a709d71 Merge: 9502b10 e64bdd1 Author: jellizaveta <[email protected]> Date: Wed Dec 18 15:40:51 2024 +0300 Merge branch 'fix/AG-37643' of ssh://bit.int.agrd.dev:7999/adguard-filters/tsurlfilter into fix/AG-37643 commit 9502b10 Author: jellizaveta <[email protected]> Date: Wed Dec 18 15:40:27 2024 +0300 add jsDoc comment commit e64bdd1 Author: Dávid Tóta <[email protected]> Date: Wed Dec 18 14:35:57 2024 +0300 Applied suggestion commit 89715a4 Author: jellizaveta <[email protected]> Date: Wed Dec 18 12:51:02 2024 +0300 add test, fix chngelog commit 6c57008 Author: Slava Leleka <[email protected]> Date: Wed Dec 18 12:50:03 2024 +0300 Applied suggestion commit 07b2ad4 Author: Slava Leleka <[email protected]> Date: Wed Dec 18 12:49:56 2024 +0300 Applied suggestion commit 00342b6 Author: jellizaveta <[email protected]> Date: Mon Dec 16 12:44:47 2024 +0300 fix changelog, styles commit 8496d6a Author: Dávid Tóta <[email protected]> Date: Mon Dec 16 12:46:49 2024 +0300 Applied suggestion commit 9fe7dcf Author: jellizaveta <[email protected]> Date: Mon Dec 16 11:39:23 2024 +0300 fix jsDoc commit 06ae9b8 Author: Slava Leleka <[email protected]> Date: Mon Dec 16 11:24:48 2024 +0300 Applied suggestion commit 959f8dd Author: jellizaveta <[email protected]> Date: Fri Dec 13 19:18:11 2024 +0300 fix naming in tests commit 429c699 Author: jellizaveta <[email protected]> Date: Fri Dec 13 19:13:01 2024 +0300 update version to 2.3.0 commit 3c7b8e2 Author: jellizaveta <[email protected]> Date: Fri Dec 13 19:06:54 2024 +0300 eslint commit a38ad21 Author: jellizaveta <[email protected]> Date: Fri Dec 13 19:06:01 2024 +0300 add jsDocs commit 1dab9d0 Author: jellizaveta <[email protected]> Date: Fri Dec 13 19:04:19 2024 +0300 fix the latest tests commit 1a93eca Author: scripthunter7 <[email protected]> Date: Fri Dec 13 17:01:00 2024 +0100 fix one test commit 004f3f4 Author: scripthunter7 <[email protected]> Date: Fri Dec 13 17:00:38 2024 +0100 add util method for last slot commit 21ab05e Author: scripthunter7 <[email protected]> Date: Fri Dec 13 17:00:16 2024 +0100 fix error offsets in css token stream constructor commit 071fca7 Merge: c6f6196 2db9ade Author: jellizaveta <[email protected]> Date: Fri Dec 13 18:31:24 2024 +0300 resolve conflict commit c6f6196 Author: jellizaveta <[email protected]> Date: Fri Dec 13 18:30:06 2024 +0300 add check for curly breckets ... and 11 more commits
AdguardTeam · Dec 18, 2024 · 45c8052 · 45c8052
1 parent 014a3df
commit 45c8052
Show file tree

Hide file tree

Showing 8 changed files with 728 additions and 14 deletions.
diff --git a/packages/agtree/CHANGELOG.md b/packages/agtree/CHANGELOG.md
@@ -8,6 +8,15 @@ The format is based on [Keep a Changelog][keepachangelog], and this project adhe
 [keepachangelog]: https://keepachangelog.com/en/1.0.0/
 [semver]: https://semver.org/spec/v2.0.0.html
 
+## [2.3.0] - 2024-12-18
+
+### Added
+
+- Support for ABP-syntax CSS injection rules [tsurlfilter#143].
+
+[2.3.0]: https://github.com/AdguardTeam/tsurlfilter/releases/tag/agtree-v2.3.0
+[tsurlfilter#143]: https://github.com/AdguardTeam/tsurlfilter/issues/143
+
 ## [2.2.0] - 2024-11-27
 
 ### Removed

diff --git a/packages/agtree/src/parser/cosmetic/index.ts b/packages/agtree/src/parser/cosmetic/index.ts
@@ -1,12 +1,15 @@
 /* eslint-disable no-param-reassign */
 import { sprintf } from 'sprintf-js';
+import { TokenType } from '@adguard/css-tokenizer';
 
 import { CosmeticRuleSeparatorUtils } from '../../utils/cosmetic-rule-separator';
 import { AdblockSyntax } from '../../utils/adblockers';
 import { DomainListParser } from '../misc/domain-list';
 import { ModifierListParser } from '../misc/modifier-list';
 import {
     ADG_SCRIPTLET_MASK,
+    CSS_BLOCK_OPEN,
+    CSS_BLOCK_CLOSE,
     CLOSE_PARENTHESIS,
     CLOSE_SQUARE_BRACKET,
     COLON,
@@ -56,6 +59,7 @@ import { type InputByteBuffer } from '../../utils/input-byte-buffer';
 import { ValueParser } from '../misc/value';
 import { isUndefined } from '../../utils/type-guards';
 import { BINARY_SCHEMA_VERSION } from '../../utils/binary-schema-version';
+import { hasToken } from '../css/has-token';
 
 /**
  * Value map for binary serialization. This helps to reduce the size of the serialized data,
@@ -535,6 +539,37 @@ export class CosmeticRuleParser extends ParserBase {
             };
         };
 
+        /**
+         * Parses Adb CSS injection rules
+         * eg: example.com##.foo { display: none; }
+         *
+         * @returns parsed rule
+         */
+        const parseAbpCssInjection = (): Pick<CssInjectionRule, RestProps> | null => {
+            if (!options.parseAbpSpecificRules) {
+                return null;
+            }
+
+            // check if the rule contains both CSS block open and close characters
+            // if none of them is present we can stop parsing
+            if (rawBody.indexOf(CSS_BLOCK_OPEN) === -1 && rawBody.indexOf(CSS_BLOCK_CLOSE) === -1) {
+                return null;
+            }
+
+            if (!hasToken(rawBody, new Set([TokenType.OpenCurlyBracket, TokenType.CloseCurlyBracket]))) {
+                return null;
+            }
+
+            // try to parse the raw body as an AdGuard CSS injection rule
+            const body = AdgCssInjectionParser.parse(rawBody, options, baseOffset + bodyStart);
+            // if the parsed rule type is a 'CssInjectionRuleBody', return the parsed rule
+            return {
+                syntax: AdblockSyntax.Abp,
+                type: CosmeticRuleType.CssInjectionRule,
+                body,
+            };
+        };
+
         const parseAbpSnippetInjection = (): Pick<ScriptletInjectionRule, RestProps> | null => {
             if (!options.parseAbpSpecificRules) {
                 throw new AdblockSyntaxError(
@@ -687,11 +722,23 @@ export class CosmeticRuleParser extends ParserBase {
         // the next function is called, and so on.
         // If all functions return null, an error should be thrown.
         const separatorMap = {
-            '##': [parseUboHtmlFiltering, parseUboScriptletInjection, parseUboCssInjection, parseElementHiding],
-            '#@#': [parseUboHtmlFiltering, parseUboScriptletInjection, parseUboCssInjection, parseElementHiding],
-
-            '#?#': [parseUboCssInjection, parseElementHiding],
-            '#@?#': [parseUboCssInjection, parseElementHiding],
+            '##': [
+                parseUboHtmlFiltering,
+                parseUboScriptletInjection,
+                parseUboCssInjection,
+                parseAbpCssInjection,
+                parseElementHiding,
+            ],
+            '#@#': [
+                parseUboHtmlFiltering,
+                parseUboScriptletInjection,
+                parseUboCssInjection,
+                parseAbpCssInjection,
+                parseElementHiding,
+            ],
+
+            '#?#': [parseUboCssInjection, parseAbpCssInjection, parseElementHiding],
+            '#@?#': [parseUboCssInjection, parseAbpCssInjection, parseElementHiding],
 
             '#$#': [parseAdgCssInjection, parseAbpSnippetInjection],
             '#@$#': [parseAdgCssInjection, parseAbpSnippetInjection],
@@ -772,15 +819,14 @@ export class CosmeticRuleParser extends ParserBase {
      */
     public static generateBody(node: AnyCosmeticRule): string {
         let result = EMPTY;
-
         // Body
         switch (node.type) {
             case CosmeticRuleType.ElementHidingRule:
                 result = node.body.selectorList.value;
                 break;
 
             case CosmeticRuleType.CssInjectionRule:
-                if (node.syntax === AdblockSyntax.Adg) {
+                if (node.syntax === AdblockSyntax.Adg || node.syntax === AdblockSyntax.Abp) {
                     result = AdgCssInjectionParser.generate(node.body);
                 } else if (node.syntax === AdblockSyntax.Ubo) {
                     if (node.body.mediaQueryList) {

diff --git a/packages/agtree/src/parser/css/css-token-stream.ts b/packages/agtree/src/parser/css/css-token-stream.ts
@@ -112,14 +112,23 @@ export class CssTokenStream {
         // - end: end index of the token
         // - props: additional properties of the token, if any (we don't use it here, this is why we use underscore)
         // - balance: balance level of the token
-        tokenizeBalanced(source, (type, start, end, _, balance) => {
-            this.tokens.push({
-                type,
-                start,
-                end,
-                balance,
+        try {
+            tokenizeBalanced(source, (type, start, end, _, balance) => {
+                this.tokens.push({
+                    type,
+                    start,
+                    end,
+                    balance,
+                });
             });
-        });
+        } catch (error) {
+            // If the error is an AdblockSyntaxError, adjust the error positions to the base offset
+            if (error instanceof AdblockSyntaxError) {
+                error.start += baseOffset;
+                error.end += baseOffset;
+                throw error;
+            }
+        }
 
         this.index = 0;
 

diff --git a/packages/agtree/src/parser/css/has-token.ts b/packages/agtree/src/parser/css/has-token.ts
@@ -0,0 +1,38 @@
+import { type TokenType, tokenizeExtended } from '@adguard/css-tokenizer';
+
+/**
+ * Represents an error that occurs when an operation is aborted.
+ */
+class AbortError extends Error {
+    constructor() {
+        super('Aborted');
+    }
+}
+
+// TODO: AG-38480 add a stop function to the tokenizers callback and move `hasToken` to CSS Tokenizer as well
+/**
+ * Checks if the given raw string contains any of the specified tokens.
+ * This function uses error throwing inside the abort tokenization process.
+ *
+ * @param raw - The raw string to be tokenized and checked.
+ * @param tokens - A set of token types to check for in the raw string.
+ * @returns `true` if any of the specified tokens are found in the raw string, otherwise `false`.
+ */
+export const hasToken = (raw: string, tokens: Set<TokenType>): boolean => {
+    try {
+        tokenizeExtended(
+            raw,
+            (type: TokenType) => {
+                if (tokens.has(type)) {
+                    throw new AbortError();
+                }
+            },
+        );
+    } catch (e) {
+        if (e instanceof AbortError) {
+            return true;
+        }
+        throw e;
+    }
+    return false;
+};
diff --git a/packages/agtree/test/converter/converter.test.ts b/packages/agtree/test/converter/converter.test.ts
@@ -658,6 +658,32 @@ describe('Converter integration tests', () => {
                     expected: ['#?#.banner'],
                     shouldConvert: false,
                 },
+                {
+                    actual: '##.banner { display: none; }',
+                    expected: ['#$#.banner { display: none; }'],
+                    shouldConvert: true,
+                },
+                {
+                    actual: '##.banner { remove: true; }',
+                    expected: ['#$?#.banner { remove: true; }'],
+                    shouldConvert: true,
+                },
+                // case without spaces in css pseudo property
+                {
+                    actual: '##.banner {remove:true;}',
+                    expected: ['#$?#.banner { remove: true; }'],
+                    shouldConvert: true,
+                },
+                {
+                    actual: '##div[foo="yay{"]',
+                    expected: ['##div[foo="yay{"]'],
+                    shouldConvert: false,
+                },
+                {
+                    actual: '##div[foo="yay{"][href="yay}"]',
+                    expected: ['##div[foo="yay{"][href="yay}"]'],
+                    shouldConvert: false,
+                },
             ])('should convert \'$actual\' to \'$expected\'', (testData) => {
                 expect(testData).toBeConvertedProperly(RuleConverter, 'convertToAdg');
             });

diff --git a/packages/agtree/test/helpers/node-utils.ts b/packages/agtree/test/helpers/node-utils.ts
@@ -88,6 +88,18 @@ export class NodeExpectContext {
         };
     }
 
+    /**
+     * Gets the range for the last slot in the actual string.
+     *
+     * @returns Location range. See {@link Range}.
+     */
+    public getLastSlotRange(): Range {
+        return {
+            start: this.actual.length - 1,
+            end: this.actual.length,
+        };
+    }
+
     /**
      * Converts the given range to a tuple.
      *