diff --git a/src/config/CleaningConfig.ts b/src/config/CleaningConfig.ts index 3d0913d..4cbc51e 100644 --- a/src/config/CleaningConfig.ts +++ b/src/config/CleaningConfig.ts @@ -45,13 +45,6 @@ export interface RepetitionThresholds { export interface ContaminationPatterns { /** Instruction patterns to remove from text beginning */ instructionPatterns: string[]; - /** XML pattern groups for different contexts */ - xmlPatternGroups: { - completeXmlTags: string[]; - sentenceBoundedTags: string[]; - lineBoundedTags: string[]; - standaloneTags: string[]; - }; /** Context patterns for general cleanup */ contextPatterns: string[]; /** Prompt snippet lengths for partial matching */ @@ -113,27 +106,6 @@ export const CLEANING_CONFIG: CleaningConfig = { '화자의 발언 내용만 정확히 기록해주세요' ], - xmlPatternGroups: { - completeXmlTags: [ - '/]*>([\\s\\S]*?)<\\/TRANSCRIPT>/g', - '/]*>([\\s\\S]*?)<\\/transcript>/g', - '/]*>([\\s\\S]*?)<\\/TRANSCRIPTION>/g' - ], - sentenceBoundedTags: [ - '/<\\/?TRANSCRIPT[^>]*>/g', - '/<\\/?transcript[^>]*>/g', - '/<\\/?TRANSCRIPTION[^>]*>/g' - ], - lineBoundedTags: [ - '/^\\s*<[^>]*>\\s*$/gm', - '/^\\s*<\\/[^>]*>\\s*$/gm' - ], - standaloneTags: [ - '/<[^>]*\\/>/g', - '/<\\w+[^>]*>\\s*<\\/\\w+>/g' - ] - }, - contextPatterns: [ // Universal context patterns (structural) '/^\\s*\\([^)]*Speaker[^)]*only[^)]*\\)\\s*$/gmi',