ai-write/src/utils/textDiff.js

/**
 * 文本差异计算工具
 * 用于对比原文和重写内容，生成差异片段
 */

/**
 * 将文本按句子拆分，并记录每个句子的位置信息
 * @param {string} text - 输入文本
 * @returns {Array<{text: string, start: number, end: number}>} - 句子数组（含位置）
 */
export const splitIntoSentencesWithPosition = (text) => {
  if (!text) return []

  const sentences = []
  const regex = /[^。！？；\n]+[。！？；\n]*/g
  let match

  while ((match = regex.exec(text)) !== null) {
    const sentence = match[0].trim()
    if (sentence) {
      sentences.push({
        text: sentence,
        start: match.index,
        end: match.index + match[0].length
      })
    }
  }

  return sentences
}

/**
 * 将文本按句子拆分（简单版本，不含位置）
 * @param {string} text - 输入文本
 * @returns {string[]} - 句子数组
 */
export const splitIntoSentences = (text) => {
  return splitIntoSentencesWithPosition(text).map(s => s.text)
}

/**
 * 计算两个字符串的相似度（基于 Levenshtein 距离）
 * @param {string} s1 - 字符串1
 * @param {string} s2 - 字符串2
 * @returns {number} - 相似度 0-1
 */
export const similarity = (s1, s2) => {
  if (!s1 && !s2) return 1
  if (!s1 || !s2) return 0

  const longer = s1.length > s2.length ? s1 : s2
  const shorter = s1.length > s2.length ? s2 : s1

  if (longer.length === 0) return 1

  // 简化版：使用字符重叠率
  const set1 = new Set(s1)
  const set2 = new Set(s2)
  const intersection = [...set1].filter(c => set2.has(c)).length
  const union = new Set([...set1, ...set2]).size

  return intersection / union
}

/**
 * 计算两段文本的差异（含位置信息）
 * @param {string} original - 原文
 * @param {string} rewritten - 重写后的文本
 * @returns {Array<{type: 'unchanged'|'modified'|'added'|'removed', original: string, rewritten: string, idx: number, originalStart?: number, originalEnd?: number}>}
 */
export const computeDiff = (original, rewritten) => {
  const originalSentences = splitIntoSentencesWithPosition(original)
  const rewrittenSentences = splitIntoSentencesWithPosition(rewritten)

  const diff = []
  let oIdx = 0
  let rIdx = 0
  let diffIdx = 0

  // 使用简单的贪心匹配算法
  while (oIdx < originalSentences.length || rIdx < rewrittenSentences.length) {
    const oItem = originalSentences[oIdx]
    const rItem = rewrittenSentences[rIdx]
    const oSentence = oItem?.text
    const rSentence = rItem?.text

    if (oIdx >= originalSentences.length) {
      // 原文已结束，剩余都是新增
      diff.push({
        type: 'added',
        original: '',
        rewritten: rSentence,
        idx: diffIdx++,
        // 新增内容插入到最后一个原文句子之后
        insertAfterIdx: originalSentences.length > 0 ? originalSentences[originalSentences.length - 1].end : 0
      })
      rIdx++
      continue
    }

    if (rIdx >= rewrittenSentences.length) {
      // 重写已结束，剩余都是删除
      diff.push({
        type: 'removed',
        original: oSentence,
        rewritten: '',
        idx: diffIdx++,
        originalStart: oItem.start,
        originalEnd: oItem.end
      })
      oIdx++
      continue
    }

    // 计算相似度
    const sim = similarity(oSentence, rSentence)

    if (sim > 0.8) {
      // 高度相似，视为未修改
      diff.push({
        type: 'unchanged',
        original: oSentence,
        rewritten: rSentence,
        idx: diffIdx++,
        originalStart: oItem.start,
        originalEnd: oItem.end
      })
      oIdx++
      rIdx++
    } else if (sim > 0.4) {
      // 中等相似，视为修改
      diff.push({
        type: 'modified',
        original: oSentence,
        rewritten: rSentence,
        idx: diffIdx++,
        originalStart: oItem.start,
        originalEnd: oItem.end
      })
      oIdx++
      rIdx++
    } else {
      // 低相似度，尝试向前查找匹配
      let foundMatch = false

      // 在重写文本中向前查找原句的匹配
      for (let i = rIdx + 1; i < Math.min(rIdx + 3, rewrittenSentences.length); i++) {
        if (similarity(oSentence, rewrittenSentences[i].text) > 0.6) {
          // 找到匹配，中间的都是新增
          for (let j = rIdx; j < i; j++) {
            diff.push({
              type: 'added',
              original: '',
              rewritten: rewrittenSentences[j].text,
              idx: diffIdx++,
              insertAfterIdx: oItem.start
            })
          }
          rIdx = i
          foundMatch = true
          break
        }
      }

      if (!foundMatch) {
        // 在原文中向前查找重写句的匹配
        for (let i = oIdx + 1; i < Math.min(oIdx + 3, originalSentences.length); i++) {
          if (similarity(originalSentences[i].text, rSentence) > 0.6) {
            // 找到匹配，中间的都是删除
            for (let j = oIdx; j < i; j++) {
              diff.push({
                type: 'removed',
                original: originalSentences[j].text,
                rewritten: '',
                idx: diffIdx++,
                originalStart: originalSentences[j].start,
                originalEnd: originalSentences[j].end
              })
            }
            oIdx = i
            foundMatch = true
            break
          }
        }
      }

      if (!foundMatch) {
        // 没找到匹配，视为修改
        diff.push({
          type: 'modified',
          original: oSentence,
          rewritten: rSentence,
          idx: diffIdx++,
          originalStart: oItem.start,
          originalEnd: oItem.end
        })
        oIdx++
        rIdx++
      }
    }
  }

  return diff
}

/**
 * 根据选中的修改应用差异，在原文中精确替换
 * @param {string} original - 原文
 * @param {Array} diffSegments - 差异片段（含位置信息）
 * @param {Set<number>} acceptedChanges - 接受的修改索引集合
 * @returns {string} - 精确替换后的文本
 */
export const applySelectedChanges = (original, diffSegments, acceptedChanges) => {
  // 收集所有需要执行的替换操作
  const operations = []

  for (const segment of diffSegments) {
    const isAccepted = acceptedChanges.has(segment.idx)

    if (segment.type === 'modified' && isAccepted) {
      // 修改：替换原文中的对应位置
      if (segment.originalStart !== undefined && segment.originalEnd !== undefined) {
        operations.push({
          type: 'replace',
          start: segment.originalStart,
          end: segment.originalEnd,
          content: segment.rewritten
        })
      }
    } else if (segment.type === 'removed' && isAccepted) {
      // 删除：移除原文中的对应位置
      if (segment.originalStart !== undefined && segment.originalEnd !== undefined) {
        operations.push({
          type: 'delete',
          start: segment.originalStart,
          end: segment.originalEnd
        })
      }
    } else if (segment.type === 'added' && isAccepted) {
      // 新增：在指定位置插入
      if (segment.insertAfterIdx !== undefined) {
        operations.push({
          type: 'insert',
          position: segment.insertAfterIdx,
          content: segment.rewritten
        })
      }
    }
  }

  // 按位置从后往前排序（避免位置偏移）
  operations.sort((a, b) => {
    const posA = a.start !== undefined ? a.start : a.position
    const posB = b.start !== undefined ? b.start : b.position
    return posB - posA
  })

  // 执行替换操作
  let result = original
  for (const op of operations) {
    if (op.type === 'replace') {
      result = result.slice(0, op.start) + op.content + result.slice(op.end)
    } else if (op.type === 'delete') {
      result = result.slice(0, op.start) + result.slice(op.end)
    } else if (op.type === 'insert') {
      result = result.slice(0, op.position) + op.content + result.slice(op.position)
    }
  }

  return result
}

/**
 * 获取差异统计
 * @param {Array} diffSegments - 差异片段
 * @returns {{total: number, modified: number, added: number, removed: number}}
 */
export const getDiffStats = (diffSegments) => {
  return {
    total: diffSegments.length,
    modified: diffSegments.filter(s => s.type === 'modified').length,
    added: diffSegments.filter(s => s.type === 'added').length,
    removed: diffSegments.filter(s => s.type === 'removed').length,
    unchanged: diffSegments.filter(s => s.type === 'unchanged').length
  }
}