@adguard/agtree 2.1.2 → 2.1.4

package/dist/agtree.d.ts

@@ -1,5 +1,5 @@
 /*
- * AGTree v2.1.2 (build date: Thu, 19 Sep 2024 15:44:40 GMT)
+ * AGTree v2.1.4 (build date: Mon, 25 Nov 2024 16:56:07 GMT)
  * (c) 2024 Adguard Software Ltd.
  * Released under the MIT license
  * https://github.com/AdguardTeam/tsurlfilter/tree/master/packages/agtree#readme
@@ -2357,9 +2357,11 @@ declare class FilterListParser extends ParserBase {
      * @param ast AST to generate
      * @param preferRaw If `true`, then the parser will use `raws.text` property of each rule
      * if it is available. Default is `false`.
+     * @param tolerant If `true`, errors during rule generation will be logged to the console and invalid rules
+     * will be skipped. If `false`, an error will be thrown on the first invalid rule. Default is `true`.
      * @returns Serialized filter list
      */
-    static generate(ast: FilterList, preferRaw?: boolean): string;
+    static generate(ast: FilterList, preferRaw?: boolean, tolerant?: boolean): string;
     /**
      * Serializes a filter list node to binary format.
      *

package/dist/agtree.js

@@ -1,5 +1,5 @@
 /*
- * AGTree v2.1.2 (build date: Thu, 19 Sep 2024 15:44:40 GMT)
+ * AGTree v2.1.4 (build date: Mon, 25 Nov 2024 16:56:07 GMT)
  * (c) 2024 Adguard Software Ltd.
  * Released under the MIT license
  * https://github.com/AdguardTeam/tsurlfilter/tree/master/packages/agtree#readme
@@ -343,10 +343,11 @@ class StringUtils {
      * @param searchedCharacter - Searched character
      * @param start - Start index
      * @param escapeCharacter - Escape character, \ by default
+     * @param end - End index (excluded)
      * @returns Index or -1 if the character not found
      */
-    static findNextUnescapedCharacter(pattern, searchedCharacter, start = 0, escapeCharacter = ESCAPE_CHARACTER) {
-        for (let i = start; i < pattern.length; i += 1) {
+    static findNextUnescapedCharacter(pattern, searchedCharacter, start = 0, escapeCharacter = ESCAPE_CHARACTER, end = pattern.length) {
+        for (let i = start; i < end; i += 1) {
             // The searched character cannot be preceded by an escape
             if (pattern[i] === searchedCharacter && pattern[i - 1] !== escapeCharacter) {
                 return i;
@@ -361,10 +362,11 @@ class StringUtils {
      * @param searchedCharacter - Searched character
      * @param start - Start index
      * @param escapeCharacter - Escape character, \ by default
+     * @param end - End index (Included)
      * @returns Index or -1 if the character not found
      */
-    static findNextUnescapedCharacterBackwards(pattern, searchedCharacter, start = pattern.length - 1, escapeCharacter = ESCAPE_CHARACTER) {
-        for (let i = start; i >= 0; i -= 1) {
+    static findNextUnescapedCharacterBackwards(pattern, searchedCharacter, start = pattern.length - 1, escapeCharacter = ESCAPE_CHARACTER, end = 0) {
+        for (let i = start; i >= end; i -= 1) {
             // The searched character cannot be preceded by an escape
             if (pattern[i] === searchedCharacter && pattern[i - 1] !== escapeCharacter) {
                 return i;
@@ -1911,6 +1913,9 @@ class AgentCommentRuleParser extends ParserBase {
     static serialize(node, buffer) {
         buffer.writeUint8(BinaryTypeMap.AgentRuleNode);
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(AgentRuleSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -1965,6 +1970,11 @@ class AgentCommentRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -2860,6 +2870,9 @@ class HintCommentRuleParser extends ParserBase {
             buffer.writeUint8(SYNTAX_SERIALIZATION_MAP.get(exports.AdblockSyntax.Adg) ?? 0);
         }
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(HintRuleSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -2913,6 +2926,11 @@ class HintCommentRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -5346,6 +5364,9 @@ class ModifierListParser extends ParserBase {
     static serialize(node, buffer) {
         buffer.writeUint8(BinaryTypeMap.ModifierListNode);
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(ModifierListNodeSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -5397,6 +5418,11 @@ class ModifierListParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -7052,9 +7078,6 @@ class UboParameterListParser extends ParameterListParser {
                     // Next non-whitespace character after the closing quote should be the separator
                     const nextSeparatorIndex = StringUtils.skipWS(raw, possibleClosingQuoteIndex + 1);
                     if (nextSeparatorIndex === length) {
-                        if (requireQuotes) {
-                            throw new AdblockSyntaxError('Expected separator, got end of string', baseOffset + nextSeparatorIndex, baseOffset + length);
-                        }
                         // If the separator is not found, the param end is the end of the string
                         paramEnd = StringUtils.skipWSBack(raw, length - 1) + 1;
                         offset = length;
@@ -7068,9 +7091,41 @@ class UboParameterListParser extends ParameterListParser {
                         if (requireQuotes) {
                             throw new AdblockSyntaxError(`Expected separator, got: '${raw[nextSeparatorIndex]}'`, baseOffset + nextSeparatorIndex, baseOffset + length);
                         }
-                        // Param end should be the last separator before the quote
-                        offset = StringUtils.findNextUnescapedCharacterBackwards(raw, separator, possibleClosingQuoteIndex) + 1;
-                        paramEnd = StringUtils.skipWSBack(raw, offset - 2) + 1;
+                        /**
+                         * At that point found `possibleClosingQuoteIndex` is wrong
+                         * | is `offset`
+                         * ~ is `possibleClosingQuoteIndex`
+                         * ^ is `nextSeparatorIndex`
+                         *
+                         * Example 1: "abc, ').cba='1'"
+                         *                  |      ~^
+                         * Example 2: "abc, ').cba, '1'"
+                         *                  |       ~^
+                         * Example 3: "abc, ').cba='1', cba"
+                         *                  |      ~^
+                         *
+                         * Search for separator before `possibleClosingQuoteIndex`
+                         */
+                        const separatorIndexBeforeQuote = StringUtils.findNextUnescapedCharacterBackwards(raw, separator, possibleClosingQuoteIndex, ESCAPE_CHARACTER, offset + 1);
+                        if (separatorIndexBeforeQuote !== -1) {
+                            // Found separator before (Example 2)
+                            paramEnd = StringUtils.skipWSBack(raw, separatorIndexBeforeQuote - 1) + 1;
+                            offset = separatorIndexBeforeQuote + 1;
+                        }
+                        else {
+                            // Didn't found separator before, search after
+                            const separatorIndexAfterQuote = StringUtils.findNextUnescapedCharacter(raw, separator, possibleClosingQuoteIndex);
+                            if (separatorIndexAfterQuote !== -1) {
+                                // We found separator after (Example 3)
+                                paramEnd = StringUtils.skipWSBack(raw, separatorIndexAfterQuote - 1) + 1;
+                                offset = separatorIndexAfterQuote + 1;
+                            }
+                            else {
+                                // If the separator is not found, the param end is the end of the string (Example 1)
+                                paramEnd = StringUtils.skipWSBack(raw, length - 1) + 1;
+                                offset = length;
+                            }
+                        }
                     }
                 }
                 else {
@@ -8580,10 +8635,16 @@ class NetworkRuleParser extends ParserBase {
         const pattern = ValueParser.parse(raw.slice(patternStart, patternEnd), options, baseOffset + patternStart);
         // Parse modifiers (if any)
         let modifiers;
+        // Get a last non-whitespace index
+        const lastNonWsIndex = StringUtils.skipWSBack(raw);
         // Find start and end index of the modifiers
         const modifiersStart = separatorIndex + 1;
-        const modifiersEnd = StringUtils.skipWSBack(raw) + 1;
+        const modifiersEnd = lastNonWsIndex + 1;
         if (separatorIndex !== -1) {
+            // Check for empty modifiers
+            if (separatorIndex === lastNonWsIndex) {
+                throw new AdblockSyntaxError('Empty modifiers are not allowed', baseOffset + separatorIndex, baseOffset + raw.length);
+            }
             modifiers = ModifierListParser.parse(raw.slice(modifiersStart, modifiersEnd), options, baseOffset + modifiersStart);
         }
         // Throw error if there is no pattern and no modifiers
@@ -8902,6 +8963,9 @@ class HostRuleParser extends ParserBase {
             buffer.writeUint32(node.end);
         }
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             // note: we store the count, because re-construction of the array is faster if we know the length
             if (count > UINT16_MAX) {
@@ -8945,6 +9009,11 @@ class HostRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
     /**
      * Serializes a host rule node to binary format.
@@ -9712,9 +9781,11 @@ class FilterListParser extends ParserBase {
      * @param ast AST to generate
      * @param preferRaw If `true`, then the parser will use `raws.text` property of each rule
      * if it is available. Default is `false`.
+     * @param tolerant If `true`, errors during rule generation will be logged to the console and invalid rules
+     * will be skipped. If `false`, an error will be thrown on the first invalid rule. Default is `true`.
      * @returns Serialized filter list
      */
-    static generate(ast, preferRaw = false) {
+    static generate(ast, preferRaw = false, tolerant = true) {
         let result = EMPTY;
         for (let i = 0; i < ast.children.length; i += 1) {
             const rule = ast.children[i];
@@ -9722,7 +9793,18 @@ class FilterListParser extends ParserBase {
                 result += rule.raws.text;
             }
             else {
-                result += RuleParser.generate(rule);
+                try {
+                    result += RuleParser.generate(rule);
+                }
+                catch (error) {
+                    if (tolerant) {
+                        // eslint-disable-next-line no-console
+                        console.error(`Error when generating: ${error}`);
+                    }
+                    else {
+                        throw new Error(String(error));
+                    }
+                }
             }
             switch (rule.raws?.nl) {
                 case 'crlf':
@@ -14669,7 +14751,7 @@ class RawFilterListConverter extends ConverterBase {
             return createConversionResult(rawFilterList, false);
         }
         // Otherwise, serialize the filter list and return the result
-        return createConversionResult(FilterListParser.generate(conversionResult.result), true);
+        return createConversionResult(FilterListParser.generate(conversionResult.result, false, tolerant), true);
     }
 }
 
@@ -15617,7 +15699,7 @@ class RuleCategorizer {
     }
 }
 
-const version = "2.1.2";
+const version = "2.1.4";
 
 /**
  * @file AGTree version

package/dist/agtree.mjs

@@ -1,5 +1,5 @@
 /*
- * AGTree v2.1.2 (build date: Thu, 19 Sep 2024 15:44:40 GMT)
+ * AGTree v2.1.4 (build date: Mon, 25 Nov 2024 16:56:07 GMT)
  * (c) 2024 Adguard Software Ltd.
  * Released under the MIT license
  * https://github.com/AdguardTeam/tsurlfilter/tree/master/packages/agtree#readme
@@ -323,10 +323,11 @@ class StringUtils {
      * @param searchedCharacter - Searched character
      * @param start - Start index
      * @param escapeCharacter - Escape character, \ by default
+     * @param end - End index (excluded)
      * @returns Index or -1 if the character not found
      */
-    static findNextUnescapedCharacter(pattern, searchedCharacter, start = 0, escapeCharacter = ESCAPE_CHARACTER) {
-        for (let i = start; i < pattern.length; i += 1) {
+    static findNextUnescapedCharacter(pattern, searchedCharacter, start = 0, escapeCharacter = ESCAPE_CHARACTER, end = pattern.length) {
+        for (let i = start; i < end; i += 1) {
             // The searched character cannot be preceded by an escape
             if (pattern[i] === searchedCharacter && pattern[i - 1] !== escapeCharacter) {
                 return i;
@@ -341,10 +342,11 @@ class StringUtils {
      * @param searchedCharacter - Searched character
      * @param start - Start index
      * @param escapeCharacter - Escape character, \ by default
+     * @param end - End index (Included)
      * @returns Index or -1 if the character not found
      */
-    static findNextUnescapedCharacterBackwards(pattern, searchedCharacter, start = pattern.length - 1, escapeCharacter = ESCAPE_CHARACTER) {
-        for (let i = start; i >= 0; i -= 1) {
+    static findNextUnescapedCharacterBackwards(pattern, searchedCharacter, start = pattern.length - 1, escapeCharacter = ESCAPE_CHARACTER, end = 0) {
+        for (let i = start; i >= end; i -= 1) {
             // The searched character cannot be preceded by an escape
             if (pattern[i] === searchedCharacter && pattern[i - 1] !== escapeCharacter) {
                 return i;
@@ -1891,6 +1893,9 @@ class AgentCommentRuleParser extends ParserBase {
     static serialize(node, buffer) {
         buffer.writeUint8(BinaryTypeMap.AgentRuleNode);
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(AgentRuleSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -1945,6 +1950,11 @@ class AgentCommentRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -2840,6 +2850,9 @@ class HintCommentRuleParser extends ParserBase {
             buffer.writeUint8(SYNTAX_SERIALIZATION_MAP.get(AdblockSyntax.Adg) ?? 0);
         }
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(HintRuleSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -2893,6 +2906,11 @@ class HintCommentRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -5326,6 +5344,9 @@ class ModifierListParser extends ParserBase {
     static serialize(node, buffer) {
         buffer.writeUint8(BinaryTypeMap.ModifierListNode);
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             buffer.writeUint8(ModifierListNodeSerializationMap.Children);
             // note: we store the count, because re-construction of the array is faster if we know the length
@@ -5377,6 +5398,11 @@ class ModifierListParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
 }
 
@@ -7032,9 +7058,6 @@ class UboParameterListParser extends ParameterListParser {
                     // Next non-whitespace character after the closing quote should be the separator
                     const nextSeparatorIndex = StringUtils.skipWS(raw, possibleClosingQuoteIndex + 1);
                     if (nextSeparatorIndex === length) {
-                        if (requireQuotes) {
-                            throw new AdblockSyntaxError('Expected separator, got end of string', baseOffset + nextSeparatorIndex, baseOffset + length);
-                        }
                         // If the separator is not found, the param end is the end of the string
                         paramEnd = StringUtils.skipWSBack(raw, length - 1) + 1;
                         offset = length;
@@ -7048,9 +7071,41 @@ class UboParameterListParser extends ParameterListParser {
                         if (requireQuotes) {
                             throw new AdblockSyntaxError(`Expected separator, got: '${raw[nextSeparatorIndex]}'`, baseOffset + nextSeparatorIndex, baseOffset + length);
                         }
-                        // Param end should be the last separator before the quote
-                        offset = StringUtils.findNextUnescapedCharacterBackwards(raw, separator, possibleClosingQuoteIndex) + 1;
-                        paramEnd = StringUtils.skipWSBack(raw, offset - 2) + 1;
+                        /**
+                         * At that point found `possibleClosingQuoteIndex` is wrong
+                         * | is `offset`
+                         * ~ is `possibleClosingQuoteIndex`
+                         * ^ is `nextSeparatorIndex`
+                         *
+                         * Example 1: "abc, ').cba='1'"
+                         *                  |      ~^
+                         * Example 2: "abc, ').cba, '1'"
+                         *                  |       ~^
+                         * Example 3: "abc, ').cba='1', cba"
+                         *                  |      ~^
+                         *
+                         * Search for separator before `possibleClosingQuoteIndex`
+                         */
+                        const separatorIndexBeforeQuote = StringUtils.findNextUnescapedCharacterBackwards(raw, separator, possibleClosingQuoteIndex, ESCAPE_CHARACTER, offset + 1);
+                        if (separatorIndexBeforeQuote !== -1) {
+                            // Found separator before (Example 2)
+                            paramEnd = StringUtils.skipWSBack(raw, separatorIndexBeforeQuote - 1) + 1;
+                            offset = separatorIndexBeforeQuote + 1;
+                        }
+                        else {
+                            // Didn't found separator before, search after
+                            const separatorIndexAfterQuote = StringUtils.findNextUnescapedCharacter(raw, separator, possibleClosingQuoteIndex);
+                            if (separatorIndexAfterQuote !== -1) {
+                                // We found separator after (Example 3)
+                                paramEnd = StringUtils.skipWSBack(raw, separatorIndexAfterQuote - 1) + 1;
+                                offset = separatorIndexAfterQuote + 1;
+                            }
+                            else {
+                                // If the separator is not found, the param end is the end of the string (Example 1)
+                                paramEnd = StringUtils.skipWSBack(raw, length - 1) + 1;
+                                offset = length;
+                            }
+                        }
                     }
                 }
                 else {
@@ -8560,10 +8615,16 @@ class NetworkRuleParser extends ParserBase {
         const pattern = ValueParser.parse(raw.slice(patternStart, patternEnd), options, baseOffset + patternStart);
         // Parse modifiers (if any)
         let modifiers;
+        // Get a last non-whitespace index
+        const lastNonWsIndex = StringUtils.skipWSBack(raw);
         // Find start and end index of the modifiers
         const modifiersStart = separatorIndex + 1;
-        const modifiersEnd = StringUtils.skipWSBack(raw) + 1;
+        const modifiersEnd = lastNonWsIndex + 1;
         if (separatorIndex !== -1) {
+            // Check for empty modifiers
+            if (separatorIndex === lastNonWsIndex) {
+                throw new AdblockSyntaxError('Empty modifiers are not allowed', baseOffset + separatorIndex, baseOffset + raw.length);
+            }
             modifiers = ModifierListParser.parse(raw.slice(modifiersStart, modifiersEnd), options, baseOffset + modifiersStart);
         }
         // Throw error if there is no pattern and no modifiers
@@ -8882,6 +8943,9 @@ class HostRuleParser extends ParserBase {
             buffer.writeUint32(node.end);
         }
         const count = node.children.length;
+        // If there are no children, we do not write any data related to them, to avoid using unnecessary storage,
+        // but children is a required field, so during deserialization we should initialize it as an empty array,
+        // if there are no children in the binary data.
         if (count) {
             // note: we store the count, because re-construction of the array is faster if we know the length
             if (count > UINT16_MAX) {
@@ -8925,6 +8989,11 @@ class HostRuleParser extends ParserBase {
             }
             prop = buffer.readUint8();
         }
+        // Maybe children are not present in the binary data,
+        // in this case, we should initialize it as an empty array.
+        if (!node.children) {
+            node.children = [];
+        }
     }
     /**
      * Serializes a host rule node to binary format.
@@ -9692,9 +9761,11 @@ class FilterListParser extends ParserBase {
      * @param ast AST to generate
      * @param preferRaw If `true`, then the parser will use `raws.text` property of each rule
      * if it is available. Default is `false`.
+     * @param tolerant If `true`, errors during rule generation will be logged to the console and invalid rules
+     * will be skipped. If `false`, an error will be thrown on the first invalid rule. Default is `true`.
      * @returns Serialized filter list
      */
-    static generate(ast, preferRaw = false) {
+    static generate(ast, preferRaw = false, tolerant = true) {
         let result = EMPTY;
         for (let i = 0; i < ast.children.length; i += 1) {
             const rule = ast.children[i];
@@ -9702,7 +9773,18 @@ class FilterListParser extends ParserBase {
                 result += rule.raws.text;
             }
             else {
-                result += RuleParser.generate(rule);
+                try {
+                    result += RuleParser.generate(rule);
+                }
+                catch (error) {
+                    if (tolerant) {
+                        // eslint-disable-next-line no-console
+                        console.error(`Error when generating: ${error}`);
+                    }
+                    else {
+                        throw new Error(String(error));
+                    }
+                }
             }
             switch (rule.raws?.nl) {
                 case 'crlf':
@@ -14649,7 +14731,7 @@ class RawFilterListConverter extends ConverterBase {
             return createConversionResult(rawFilterList, false);
         }
         // Otherwise, serialize the filter list and return the result
-        return createConversionResult(FilterListParser.generate(conversionResult.result), true);
+        return createConversionResult(FilterListParser.generate(conversionResult.result, false, tolerant), true);
     }
 }
 
@@ -15597,7 +15679,7 @@ class RuleCategorizer {
     }
 }
 
-const version = "2.1.2";
+const version = "2.1.4";
 
 /**
  * @file AGTree version

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adguard/agtree",
-  "version": "2.1.2",
+  "version": "2.1.4",
   "description": "Tool set for working with adblock filter lists",
   "keywords": [
     "adblock",