@huyooo/ai-chat-shared 0.2.14 → 0.2.16

package/src/parser.ts

@@ -1,235 +0,0 @@
-/**
- * 内容解析器
- * 将 AI 输出的原始文本解析为结构化的内容块
- */
-
-import type { ContentBlock, TextBlock, CodeBlock } from './types'
-
-/** 生成唯一 ID */
-function generateId(): string {
-  return Math.random().toString(36).slice(2, 11)
-}
-
-/** 代码块正则：匹配 ```language\ncode\n``` 或 ```\ncode\n``` */
-const CODE_BLOCK_REGEX = /```(\w*)\n([\s\S]*?)```/g
-
-/**
- * 解析原始内容为内容块列表
- * @param raw 原始文本内容
- * @returns 内容块列表
- */
-export function parseContent(raw: string): ContentBlock[] {
-  if (!raw) return []
-
-  const blocks: ContentBlock[] = []
-  let lastIndex = 0
-
-  // 使用正则匹配所有代码块
-  let match: RegExpExecArray | null
-  const regex = new RegExp(CODE_BLOCK_REGEX.source, 'g')
-
-  while ((match = regex.exec(raw)) !== null) {
-    // 添加代码块前的文本
-    if (match.index > lastIndex) {
-      const textContent = raw.slice(lastIndex, match.index)
-      if (textContent.trim()) {
-        blocks.push(createTextBlock(textContent, undefined, true))
-      }
-    }
-
-    // 添加代码块
-    const language = match[1] || undefined
-    const code = match[2]
-    blocks.push(createCodeBlock(code, language))
-
-    lastIndex = match.index + match[0].length
-  }
-
-  // 添加最后一段文本
-  if (lastIndex < raw.length) {
-    const textContent = raw.slice(lastIndex)
-    if (textContent.trim()) {
-      blocks.push(createTextBlock(textContent, undefined, true))
-    }
-  }
-
-  return blocks
-}
-
-/** 创建文本块 */
-function createTextBlock(content: string, id?: string, trimContent = true): TextBlock {
-  return {
-    id: id || generateId(),
-    type: 'text',
-    content: trimContent ? content.trim() : content,
-  }
-}
-
-/** 创建代码块 */
-function createCodeBlock(content: string, language?: string, filename?: string, id?: string): CodeBlock {
-  return {
-    id: id || generateId(),
-    type: 'code',
-    content,
-    language,
-    filename,
-  }
-}
-
-// ============ 流式解析支持 ============
-
-/** 流式解析状态 */
-export interface StreamParseState {
-  /** 未完成的缓冲区 */
-  buffer: string
-  /** 已解析的块 */
-  blocks: ContentBlock[]
-  /** 是否在代码块中 */
-  inCodeBlock: boolean
-  /** 当前代码块语言 */
-  codeLanguage?: string
-  /** 当前代码块内容 */
-  codeContent: string
-  /** 当前代码块 ID（用于稳定 key，避免流式时每次 render 都生成新 id） */
-  codeBlockId: string | null
-  /** 当前文本块内容（流式追加，避免把一句话拆成很多 text block） */
-  textContent: string
-  /** 当前文本块 ID（稳定 key） */
-  textBlockId: string | null
-}
-
-/** 创建初始流式解析状态 */
-export function createStreamParseState(): StreamParseState {
-  return {
-    buffer: '',
-    blocks: [],
-    inCodeBlock: false,
-    codeLanguage: undefined,
-    codeContent: '',
-    codeBlockId: null,
-    textContent: '',
-    textBlockId: null,
-  }
-}
-
-function appendText(state: StreamParseState, text: string): void {
-  if (!text) return
-  if (state.textBlockId === null) state.textBlockId = generateId()
-  state.textContent += text
-}
-
-function flushTextBlock(state: StreamParseState): void {
-  if (!state.textContent.trim()) return
-  state.blocks.push(createTextBlock(state.textContent, state.textBlockId || undefined, false))
-  state.textContent = ''
-  state.textBlockId = null
-}
-
-/**
- * 流式解析（增量更新）
- * @param chunk 新增的文本块
- * @param state 当前解析状态
- * @returns 更新后的状态
- */
-export function parseContentStream(chunk: string, state: StreamParseState): StreamParseState {
-  const newState: StreamParseState = { ...state, blocks: [...state.blocks] }
-  newState.buffer += chunk
-
-  // 处理缓冲区
-  while (true) {
-    if (newState.inCodeBlock) {
-      // 在代码块中，查找结束标记
-      const endIndex = newState.buffer.indexOf('```')
-      if (endIndex !== -1) {
-        // 找到结束标记
-        newState.codeContent += newState.buffer.slice(0, endIndex)
-        newState.blocks.push(
-          createCodeBlock(
-            newState.codeContent.trim(),
-            newState.codeLanguage,
-            undefined,
-            newState.codeBlockId || undefined
-          )
-        )
-        newState.buffer = newState.buffer.slice(endIndex + 3)
-        newState.inCodeBlock = false
-        newState.codeLanguage = undefined
-        newState.codeContent = ''
-        newState.codeBlockId = null
-      } else {
-        // 没找到结束标记，继续累积
-        newState.codeContent += newState.buffer
-        newState.buffer = ''
-        break
-      }
-    } else {
-      // 不在代码块中，查找开始标记
-      const startIndex = newState.buffer.indexOf('```')
-      if (startIndex !== -1) {
-        // 找到开始标记
-        const beforeCode = newState.buffer.slice(0, startIndex)
-        appendText(newState, beforeCode)
-        flushTextBlock(newState)
-
-        // 解析语言标识
-        const afterStart = newState.buffer.slice(startIndex + 3)
-        const newlineIndex = afterStart.indexOf('\n')
-        if (newlineIndex !== -1) {
-          newState.codeLanguage = afterStart.slice(0, newlineIndex).trim() || undefined
-          newState.buffer = afterStart.slice(newlineIndex + 1)
-          newState.inCodeBlock = true
-          newState.codeContent = ''
-          newState.codeBlockId = generateId()
-        } else {
-          // 语言行还没完整，等待更多数据
-          break
-        }
-      } else {
-        // 没有代码块标记：追加到“当前文本块”里（不 flush），避免碎片化
-        // 同时检查末尾是否可能是不完整的 ```，如果是就保留在 buffer 等待后续补全
-        const lastBackticks = newState.buffer.lastIndexOf('`')
-        if (lastBackticks !== -1 && newState.buffer.length - lastBackticks < 3) {
-          // 可能是不完整的 ```，保留
-          const safeText = newState.buffer.slice(0, lastBackticks)
-          appendText(newState, safeText)
-          newState.buffer = newState.buffer.slice(lastBackticks)
-        } else {
-          appendText(newState, newState.buffer)
-          newState.buffer = ''
-        }
-        break
-      }
-    }
-  }
-
-  return newState
-}
-
-/**
- * 完成流式解析（处理剩余缓冲区）
- * @param state 当前解析状态
- * @returns 最终的内容块列表
- */
-export function finishStreamParse(state: StreamParseState): ContentBlock[] {
-  const blocks = [...state.blocks]
-
-  if (state.inCodeBlock) {
-    // 未闭合的代码块，作为代码块处理
-    blocks.push(
-      createCodeBlock(
-        (state.codeContent + state.buffer).trim(),
-        state.codeLanguage,
-        undefined,
-        state.codeBlockId || undefined
-      )
-    )
-  } else {
-    // 剩余文本 + 累积文本：合并成一个 text block（稳定 id），避免碎片化
-    const merged = state.textContent + state.buffer
-    if (merged.trim()) {
-      blocks.push(createTextBlock(merged, state.textBlockId || undefined, false))
-    }
-  }
-
-  return blocks
-}