@aim-packages/subtitle 0.1.3 → 0.1.4

package/dist/index.d.ts

@@ -222,15 +222,41 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
            *
            * @returns A new Parser parser, with `parse`, `end` and `reset` methods.
            * @public
+           *
+           * Whisper字幕解析器
+           *
+           * 用于解析Whisper模型输出的字幕格式，支持VAD（Voice Activity Detection）时间调整
+           * 输入格式示例：
+           * [00:00:00.000 --> 00:00:03.000] 这是第一句话
+           * [00:00:03.000 --> 00:00:06.000] 这是第二句话
+           * [--Single--]  // 单个文件结束标记
+           * [--Full--]    // 所有文件结束标记
+           * [--end--]     // 解析结束标记
+           *
+           * @param options - 解析器配置选项
+           * @param options.vad - VAD时间片段数组，用于调整字幕时间戳
+           * @param options.vadPadding - VAD时间调整的填充值 [前填充, 后填充]，默认 [0.4, 0]
+           * @returns 返回一个Parser实例，包含feed、reset、end方法
            */
-          declare function create(options?: ParserOptions<AimSegments[], AimSegments[], string> & {
+          declare function create(options?: ParserOptions<AimSegments[], AimSegments[], string, AimSegments[]> & {
               vad?: Segment[];
               vadPadding?: [number, number];
           }): Parser<string>;
 
-          declare function create_2(options?: ParserOptions<RequiredByKey<Partial<AimSegments>, "index">[], Partial<AimSegments>[], string>): Parser<string>;
+          declare function create_2(options?: ParserOptions<RequiredByKey<Partial<AimSegments>, "index">[], Partial<AimSegments>[], string, RequiredByKey<Partial<AimSegments>, "index">[]>): Parser<string>;
 
-          declare function create_3(options?: ParserOptions<AimSegments[], AimSegments[], AimSegments[]> & {
+          /**
+           * 创建字幕分段解析器
+           * 主要功能：
+           * 1. 将输入的字幕片段按句子长度进行分组
+           * 2. 检测和处理重复内容
+           * 3. 支持多语言分句
+           * 4. 流式处理字幕数据
+           *
+           * @param options 解析器配置选项
+           * @returns 解析器实例
+           */
+          declare function create_3(options?: ParserOptions<AimSegments[], AimSegments[], AimSegments[], AimSegments[]> & {
               sentenceLength?: number;
               repeatString?: string[];
           }): Parser<AimSegments[]>;
@@ -274,6 +300,18 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
               }
           }
 
+          /**
+           * 查找文本中重复子串的位置
+           * @param text 要搜索的文本
+           * @param substring 要查找的重复子串
+           * @returns 重复子串在文本中的位置数组
+           */
+          declare function findRepeatedSubstringPositions(text: string, substring: string): {
+              start: number;
+              end: number;
+              match: string;
+          }[];
+
           /**
            * 将秒数转换为 xx:xx:xx.xxx 格式
            *
@@ -285,6 +323,13 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
 
           export declare type ISegment = [string, string, string, string | undefined];
 
+          /**
+           * 将多个子片段合并成一个片段
+           * @param s 要合并的子片段数组
+           * @returns 合并后的片段
+           */
+          declare function joinAimSegmentItems(s: AimSegments[]): AimSegments | undefined;
+
           export declare type LanguageCode = "auto" | "none" | "zh" | "zh_cn" | "zh_tw" | "yue" | "en" | "ja" | "ko" | "fr" | "es" | "ru" | "de" | "it" | "tr" | "pt" | "vi" | "id" | "th" | "ms" | "ar" | "hi" | "ro" | "ug" | "uz" | "kk" | "az" | "ky" | "fa" | "tg";
 
           /**
@@ -375,12 +420,13 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
            *
            * @public
            */
-          declare type ParseCallback<T> = (event: ParsedEvent<T>) => void;
+          declare type ParseCallback<T, R = never> = (event: ParsedEvent<T, R>) => void;
 
-          declare interface ParsedEvent<T> {
+          declare interface ParsedEvent<T, R = never> {
               type: "event";
               event: "start" | "message" | "end";
               data?: T;
+              result?: R;
           }
 
           /**
@@ -429,17 +475,18 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
            * @template S - 解析开始时的状态类型
            * @template P - 解析过程中使用的状态类型
            * @template E - 解析结束时的状态类型
+           * @template R - 解析结果类型（用于 onEnd 回调）
            *
            * @property {ParseCallback<S>} [onStart] - 解析开始时的回调函数
            * @property {ParseCallback<P>} [onParse] - 解析过程中调用的回调函数
            * @property {ParseCallback<P>} [onProgress] - 解析进度更新时的回调函数
-           * @property {ParseCallback<E>} [onEnd] - 解析结束时的回调函数
+           * @property {ParseCallback<E, R>} [onEnd] - 解析结束时的回调函数
            */
-          declare type ParserOptions<S, P, E> = {
+          declare type ParserOptions<S, P, E, R = never> = {
               onStart?: ParseCallback<S>;
               onParse?: ParseCallback<P>;
               onProgress?: ParseCallback<P>;
-              onEnd?: ParseCallback<E>;
+              onEnd?: ParseCallback<E, R>;
           };
 
           export declare type PartialByKey<T, K extends keyof T> = Omit<T, K> & Partial<Pick<T, K>>;
@@ -497,6 +544,14 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
               end: number;
           };
 
+          /**
+           * 从片段中移除指定数量的子项
+           * @param segment 父片段
+           * @param count 要移除的子项数量
+           * @returns 被移除的子项合并后的新片段
+           */
+          declare function shiftAimSegmentItems(segment: AimSegments, count?: number): AimSegments | undefined;
+
           export declare interface SpeakerData {
               settings: Record<string, {
                   spk: string;
@@ -661,6 +716,7 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
                   formatTime,
                   convertToSeconds,
                   cleanTimeDisplay,
+                  findRepeatedSubstringPositions,
                   containsCJKCharacters,
                   languageCodeToName,
                   convertHexColorToAssFormat,
@@ -668,7 +724,9 @@ declare function chunkArrayStrings(strings: string[], characterLimit: number): s
                   convertHexColorToFFmpegFormat,
                   chunkArrayStrings,
                   chunkSegmentStringsWithIndex,
-                  consolidateSegments
+                  consolidateSegments,
+                  joinAimSegmentItems,
+                  shiftAimSegmentItems
               }
           }