@sideid/id-profanity-filter 1.9.5 → 1.10.2

package/src/core/filter.ts

@@ -1,12 +1,13 @@
 import { FilterOptions, FilterResult, ProfanityWord } from "../types";
 import { findProfanity, findProfanityWithMetadata } from "./matcher";
-import { censorWord, escapeRegExp, normalizeText } from "../utils/stringUtils";
+import { censorWord, escapeRegExp } from "../utils/stringUtils";
 import { createWordRegex } from "../utils/regexUtils";
-import {
-  DEFAULT_OPTIONS,
-  makeRandomGrawlixString,
-  getRandomGrawlix,
-} from "../config/options";
+import { DEFAULT_OPTIONS, makeRandomGrawlixString } from "../config/options";
+
+interface FindProfanityFunction {
+  (text: string, options?: FilterOptions): string[];
+  lastActualMatches?: Map<string, string[]>;
+}
 
 /**
  * Menyensor kata kotor dalam teks
@@ -28,6 +29,11 @@ export function filter(
     useRandomGrawlix = false,
     keepFirstAndLast = false,
     indonesianVariation = false,
+    detectSplit = false,
+    detectSimilarity = false,
+    useLevenshtein = false,
+    maxLevenshteinDistance = 2,
+    similarityThreshold = 0.8,
   } = { ...DEFAULT_OPTIONS, ...options };
 
   const matches = findProfanity(text, {
@@ -36,8 +42,16 @@ export function filter(
     whitelist,
     checkSubstring,
     indonesianVariation,
+    detectSplit,
+    detectSimilarity,
+    useLevenshtein,
+    maxLevenshteinDistance,
+    similarityThreshold,
   });
 
+  const actualMatches: Map<string, string[]> =
+    (findProfanity as FindProfanityFunction).lastActualMatches || new Map();
+
   const matchDetails = findProfanityWithMetadata(text, options);
 
   if (matches.length === 0) {
@@ -66,49 +80,176 @@ export function filter(
           )),
     );
 
-    const regex = createWordRegex(word, {
-      wholeWord: true,
-      caseSensitive: false,
-      leetSpeak: false,
-      detectSplit: false,
-      indonesianVariation: false,
-    });
+    const variants = actualMatches.get(word.toLowerCase()) || [];
+    variants.push(word);
+
+    const uniqueVariants = [...new Set(variants)];
 
-    let match;
-    const textToSearch = filteredText;
+    uniqueVariants.forEach((variant) => {
+      const regex = new RegExp(`\\b${escapeRegExp(variant)}\\b`, "gi");
 
-    regex.lastIndex = 0;
+      let match;
+      while ((match = regex.exec(filteredText)) !== null) {
+        const originalWord = match[0];
 
-    while ((match = regex.exec(textToSearch)) !== null) {
-      const originalWord = match[0];
+        if (whitelist.includes(originalWord.toLowerCase())) continue;
 
-      if (whitelist.includes(originalWord.toLowerCase())) continue;
+        let censoredWord;
+        if (useRandomGrawlix) {
+          censoredWord = makeRandomGrawlixString(originalWord.length);
+        } else {
+          censoredWord = censorWord(
+            originalWord,
+            replaceWith,
+            !fullWordCensor && keepFirstAndLast,
+          );
+        }
 
-      let censoredWord;
-      if (useRandomGrawlix) {
-        censoredWord = makeRandomGrawlixString(originalWord.length);
-      } else {
-        censoredWord = censorWord(
-          originalWord,
-          replaceWith,
-          !fullWordCensor && keepFirstAndLast,
+        replacements.push({
+          original: originalWord,
+          censored: censoredWord,
+          metadata,
+        });
+
+        filteredText = filteredText.replace(
+          new RegExp(`\\b${escapeRegExp(originalWord)}\\b`, "g"),
+          censoredWord,
         );
       }
+    });
 
-      replacements.push({
-        original: originalWord,
-        censored: censoredWord,
-        metadata,
-      });
+    if (detectSplit || detectLeetSpeak) {
+      if (detectLeetSpeak) {
+        const leetRegex = createWordRegex(word, {
+          wholeWord: true,
+          caseSensitive: false,
+          leetSpeak: true,
+          detectSplit: false,
+          indonesianVariation: false,
+        });
 
-      const replaceRegex = new RegExp(
-        `\\b${escapeRegExp(originalWord)}\\b`,
-        "g",
-      );
-      filteredText = filteredText.replace(replaceRegex, censoredWord);
+        let match;
+        while ((match = leetRegex.exec(filteredText)) !== null) {
+          const originalWord = match[0];
+
+          if (whitelist.includes(originalWord.toLowerCase())) continue;
+
+          let censoredWord;
+          if (useRandomGrawlix) {
+            censoredWord = makeRandomGrawlixString(originalWord.length);
+          } else {
+            censoredWord = censorWord(
+              originalWord,
+              replaceWith,
+              !fullWordCensor && keepFirstAndLast,
+            );
+          }
+
+          replacements.push({
+            original: originalWord,
+            censored: censoredWord,
+            metadata,
+          });
+
+          filteredText = filteredText.replace(
+            new RegExp(escapeRegExp(originalWord), "g"),
+            censoredWord,
+          );
+        }
+      }
+
+      if (detectSplit) {
+        const splitRegex = createWordRegex(word, {
+          wholeWord: false,
+          caseSensitive: false,
+          leetSpeak: false,
+          detectSplit: true,
+          indonesianVariation: false,
+        });
+
+        let match;
+        while ((match = splitRegex.exec(filteredText)) !== null) {
+          const originalWord = match[0];
+
+          if (whitelist.includes(originalWord.toLowerCase())) continue;
+
+          let censoredWord;
+          if (useRandomGrawlix) {
+            censoredWord = makeRandomGrawlixString(originalWord.length);
+          } else {
+            censoredWord = censorWord(
+              originalWord,
+              replaceWith,
+              !fullWordCensor && keepFirstAndLast,
+            );
+          }
+
+          replacements.push({
+            original: originalWord,
+            censored: censoredWord,
+            metadata,
+          });
+
+          filteredText = filteredText.replace(
+            new RegExp(escapeRegExp(originalWord), "g"),
+            censoredWord,
+          );
+        }
+      }
     }
   });
 
+  if (detectSimilarity && useLevenshtein) {
+    matches.forEach((word) => {
+      const metadata = matchDetails.find(
+        (m) =>
+          m.word.toLowerCase() === word.toLowerCase() ||
+          (m.aliases &&
+            m.aliases.some(
+              (alias) => alias.toLowerCase() === word.toLowerCase(),
+            )),
+      );
+
+      const variants = actualMatches.get(word.toLowerCase()) || [];
+
+      variants.forEach((variant) => {
+        const exactVariantRegex = new RegExp(
+          `\\b${escapeRegExp(variant)}\\b`,
+          "gi",
+        );
+
+        let match;
+        while ((match = exactVariantRegex.exec(filteredText)) !== null) {
+          const originalWord = match[0];
+
+          if (whitelist.includes(originalWord.toLowerCase())) continue;
+
+          let censoredWord;
+          if (useRandomGrawlix) {
+            censoredWord = makeRandomGrawlixString(originalWord.length);
+          } else {
+            censoredWord = censorWord(
+              originalWord,
+              replaceWith,
+              !fullWordCensor && keepFirstAndLast,
+            );
+          }
+
+          replacements.push({
+            original: originalWord,
+            censored: censoredWord,
+            metadata,
+          });
+
+          filteredText = filteredText.replace(
+            new RegExp(`\\b${escapeRegExp(originalWord)}\\b`, "g"),
+            censoredWord,
+          );
+        }
+      });
+    });
+  }
+
   return {
     filtered: filteredText,
     censored: replacements.length,

package/src/core/matcher.ts

@@ -5,24 +5,34 @@ import {
   FilterOptions,
 } from "../types";
 
-import { wordObjects, getWordsByFilter } from "../constants/wordList";
-import {
-  normalizeText,
-  escapeRegExp,
-  containsAnyWord,
-  detectSplitWords,
-} from "../utils/stringUtils";
-import {
-  createWordRegex,
-  addLeetSpeakVariations,
-  addIndonesianVariations,
-  addSplitVariations,
-} from "../utils/regexUtils";
+import { wordObjects } from "../constants/wordList";
+import { normalizeText } from "../utils/stringUtils";
+import { createWordRegex } from "../utils/regexUtils";
 import {
   findPossibleProfanityBySimiliarity,
-  stringSimilarity,
+  findProfanityByLevenshteinDistance,
 } from "../utils/similarityUtils";
 import { DEFAULT_OPTIONS } from "../config/options";
+import { AhoCorasick } from "../utils/ahoCorasick";
+
+const globalAhoCorasick = new AhoCorasick();
+let ahoCorasickInitialized = false;
+
+function initializeAhoCorasick(words: string[]) {
+  if (ahoCorasickInitialized) return;
+
+  for (const word of words) {
+    globalAhoCorasick.addPattern(word);
+  }
+
+  globalAhoCorasick.build();
+  ahoCorasickInitialized = true;
+}
+
+interface FindProfanityFunction {
+  (text: string, options?: FilterOptions): string[];
+  lastActualMatches?: Map<string, string[]>;
+}
 
 export function findProfanity(
   text: string,
@@ -40,53 +50,69 @@ export function findProfanity(
     detectSimilarity = false,
     similarityThreshold = 0.8,
     detectSplit = false,
+    useLevenshtein = false,
+    maxLevenshteinDistance = 2,
   } = { ...DEFAULT_OPTIONS, ...options };
 
   const normalizedText = normalizeText(text);
 
-  let wordsToCheck: string[] = wordList.length > 0 ? wordList : [];
+  let baseWordsToCheck: string[] = wordList.length > 0 ? wordList : [];
 
-  if (wordsToCheck.length === 0) {
-    if (categories || regions || severityThreshold > 0) {
-      wordsToCheck = wordObjects
-        .filter((word) => {
-          const matchCategory = categories
-            ? categories.includes(word.category)
-            : true;
-          const matchRegion = regions ? regions.includes(word.region) : true;
-          const matchSeverity = word.severity >= severityThreshold;
-          return matchCategory && matchRegion && matchSeverity;
-        })
-        .map((word) => word.word);
-    } else {
-      wordsToCheck = wordObjects.map((word) => word.word);
-    }
+  if (baseWordsToCheck.length === 0) {
+    const filteredWords = wordObjects.filter((word) => {
+      const matchCategory = categories
+        ? categories.includes(word.category)
+        : true;
+      const matchRegion = regions ? regions.includes(word.region) : true;
+      const matchSeverity = word.severity >= severityThreshold;
+      return matchCategory && matchRegion && matchSeverity;
+    });
+
+    baseWordsToCheck = filteredWords.map((word) => word.word);
   }
 
-  wordsToCheck = wordsToCheck.filter(
-    (word) => !whitelist.includes(word.toLocaleLowerCase()),
-  );
+  const aliasMap = new Map<string, string>();
+  wordObjects.forEach((wordObj) => {
+    if (wordObj.aliases && wordObj.aliases.length > 0) {
+      const matchCategory = categories
+        ? categories.includes(wordObj.category)
+        : true;
+      const matchRegion = regions ? regions.includes(wordObj.region) : true;
+      const matchSeverity = wordObj.severity >= severityThreshold;
+
+      if (matchCategory && matchRegion && matchSeverity) {
+        wordObj.aliases.forEach((alias) => {
+          aliasMap.set(alias.toLowerCase(), wordObj.word.toLowerCase());
+        });
+      }
+    }
+  });
+
+  const wordsToCheck = [
+    ...baseWordsToCheck,
+    ...Array.from(aliasMap.keys()),
+  ].filter((word) => !whitelist.includes(word.toLowerCase()));
 
   if (wordsToCheck.length === 0) {
     return [];
   }
 
   const matches = new Set<string>();
+  const actualMatches = new Map<string, string[]>();
 
-  wordsToCheck.forEach((word) => {
-    const regex = createWordRegex(word, {
-      wholeWord: !checkSubstring,
-      caseSensitive: false,
-      leetSpeak: false,
-      detectSplit: false,
-      indonesianVariation: false,
-    });
+  initializeAhoCorasick(wordsToCheck);
 
-    let match;
-    while ((match = regex.exec(normalizedText)) !== null) {
-      matches.add(word.toLowerCase());
+  const basicMatches = globalAhoCorasick.searchUnique(normalizedText);
+  for (const match of basicMatches) {
+    const originalWord =
+      aliasMap.get(match.toLowerCase()) || match.toLowerCase();
+    matches.add(originalWord);
+
+    if (!actualMatches.has(originalWord)) {
+      actualMatches.set(originalWord, []);
     }
-  });
+    actualMatches.get(originalWord)?.push(match);
+  }
 
   if (detectLeetSpeak) {
     wordsToCheck.forEach((word) => {
@@ -100,7 +126,14 @@ export function findProfanity(
 
       let match;
       while ((match = leetRegex.exec(text)) !== null) {
-        matches.add(word.toLowerCase());
+        const originalWord =
+          aliasMap.get(word.toLowerCase()) || word.toLowerCase();
+        matches.add(originalWord);
+
+        if (!actualMatches.has(originalWord)) {
+          actualMatches.set(originalWord, []);
+        }
+        actualMatches.get(originalWord)?.push(match[0]);
       }
     });
   }
@@ -117,41 +150,85 @@ export function findProfanity(
 
       let match;
       while ((match = variantRegex.exec(text)) !== null) {
-        matches.add(word.toLowerCase());
+        const originalWord =
+          aliasMap.get(word.toLowerCase()) || word.toLowerCase();
+        matches.add(originalWord);
+
+        if (!actualMatches.has(originalWord)) {
+          actualMatches.set(originalWord, []);
+        }
+        actualMatches.get(originalWord)?.push(match[0]);
       }
     });
   }
 
   if (detectSplit) {
-    if (detectSplitWords(text, wordsToCheck)) {
-      wordsToCheck.forEach((word) => {
-        const splitRegex = createWordRegex(word, {
-          wholeWord: false,
-          caseSensitive: false,
-          leetSpeak: false,
-          detectSplit: true,
-          indonesianVariation: false,
-        });
+    wordsToCheck.forEach((word) => {
+      const splitRegex = createWordRegex(word, {
+        wholeWord: false,
+        caseSensitive: false,
+        leetSpeak: false,
+        detectSplit: true,
+        indonesianVariation: false,
+      });
 
-        if (splitRegex.test(text)) {
-          matches.add(word.toLowerCase());
+      let match;
+      while ((match = splitRegex.exec(text)) !== null) {
+        const originalWord =
+          aliasMap.get(word.toLowerCase()) || word.toLowerCase();
+        matches.add(originalWord);
+
+        if (!actualMatches.has(originalWord)) {
+          actualMatches.set(originalWord, []);
         }
-      });
-    }
+        actualMatches.get(originalWord)?.push(match[0]);
+      }
+    });
   }
 
   if (detectSimilarity) {
-    const possibleProfanity = findPossibleProfanityBySimiliarity(
-      text,
-      wordsToCheck,
-      similarityThreshold,
-    );
-
-    possibleProfanity.forEach((item) => {
-      matches.add(item.original.toLowerCase());
-    });
+    if (useLevenshtein) {
+      const possibleProfanity = findProfanityByLevenshteinDistance(
+        text,
+        wordsToCheck,
+        similarityThreshold,
+        maxLevenshteinDistance,
+      );
+
+      possibleProfanity.forEach((item) => {
+        const originalWord =
+          aliasMap.get(item.original.toLowerCase()) ||
+          item.original.toLowerCase();
+        matches.add(originalWord);
+
+        if (!actualMatches.has(originalWord)) {
+          actualMatches.set(originalWord, []);
+        }
+        actualMatches.get(originalWord)?.push(item.word);
+      });
+    } else {
+      const possibleProfanity = findPossibleProfanityBySimiliarity(
+        text,
+        wordsToCheck,
+        similarityThreshold,
+      );
+
+      possibleProfanity.forEach((item) => {
+        matches.add(item.original.toLowerCase());
+
+        const originalWord =
+          aliasMap.get(item.original.toLowerCase()) ||
+          item.original.toLowerCase();
+        if (!actualMatches.has(originalWord)) {
+          actualMatches.set(originalWord, []);
+        }
+        actualMatches.get(originalWord)?.push(item.word);
+      });
+    }
   }
 
+  (findProfanity as FindProfanityFunction).lastActualMatches = actualMatches;
+
   return Array.from(matches);
 }
 

package/src/index.ts

@@ -21,7 +21,6 @@ import {
   CATEGORY_PRESETS,
   REGION_PRESETS,
   getPresetOptions,
-  makeRandomGrawlixString,
 } from "./config/options";
 
 export class IDProfanityFilter {
@@ -161,10 +160,30 @@ export class IDProfanityFilter {
   /**
    * Mengaktifkan deteksi berdasarkan kesamaan
    * @param threshold Threshold kesamaan (0-1)
+   * @param useLevenshtein Gunakan algoritma Levenshtein untuk deteksi
+   * @param maxLevenshteinDistance Jarak maksimal Levenshtein (default: 2)
    */
-  enableSimilarityDetection(threshold: number = 0.8) {
+  enableSimilarityDetection(
+    threshold: number = 0.8,
+    useLevenshtein: boolean = false,
+    maxLevenshteinDistance: number = 2,
+  ) {
+    this.options.detectSimilarity = true;
+    this.options.similarityThreshold = threshold;
+    this.options.useLevenshtein = useLevenshtein;
+    this.options.maxLevenshteinDistance = maxLevenshteinDistance;
+  }
+
+  /**
+   * Mengaktifkan deteksi berbasis Levenshtein distance
+   * @param threshold Threshold kesamaan (0-1)
+   * @param maxDistance Jarak maksimal Levenshtein (default: 2)
+   */
+  enableLevenshteinDetection(threshold: number = 0.8, maxDistance: number = 2) {
     this.options.detectSimilarity = true;
+    this.options.useLevenshtein = true;
     this.options.similarityThreshold = threshold;
+    this.options.maxLevenshteinDistance = maxDistance;
   }
 }
 

package/src/types/index.ts

@@ -50,9 +50,11 @@ export interface FilterOptions {
   useRandomGrawlix?: boolean;
   keepFirstAndLast?: boolean;
   indonesianVariation?: boolean;
-  detectSimilarity?: boolean;
-  similarityThreshold?: number;
+  detectSimilarity?: boolean; // Enable/disable Levenshtein distance matching
+  similarityThreshold?: number; // Threshold for Levenshtein distance similarity (0-1)
   detectSplit?: boolean;
+  useLevenshtein?: boolean; // New option specifically for Levenshtein algorithm
+  maxLevenshteinDistance?: number; // Maximum allowed Levenshtein distance
 }
 
 export interface FilterResult {