bookworm-smart-assistant/scripts/patches/patch-r2-precompact-tier-output.js

#!/usr/bin/env node
/**
 * patch-r2-precompact-tier-output.js · 2026-04-26
 *
 * R2: pre-compact-handoff.js 增加工具输出分级摘要
 * 在 compact 前扫描 transcript_path JSONL, 识别 TOP-N 大工具结果,
 * 按工具类型差异化保留, 写入 handoff.json.tool_output_tiers
 *
 * 分级规则:
 *  - Bash 输出 >2000B: 保留头 1500B + 尾 500B + 截断行数
 *  - Read 结果 >3000B: 保留路径+行范围+前 200B 摘要
 *  - Write/Edit 结果 >500B: 仅保留路径+行数确认
 *  - Agent/Task 结果 >2000B: 保留头 1000B + "(Agent 完整结果已 dump)"
 *  - 其他工具 >5000B: 头 2000B + 尾 500B
 *
 * 幂等: sentinel "TOOL_OUTPUT_TIER_V1"
 */
'use strict';
const fs = require('fs');
const path = require('path');

const TARGET = path.join(__dirname, '..', '..', 'hooks', 'pre-compact-handoff.js');
const SENTINEL = 'TOOL_OUTPUT_TIER_V1';

const OLD_BLOCK = `    // 构造 handoff 数据
    const handoff = {
      timestamp: new Date().toISOString(),
      session_id: hookData.session_id || \`session-\${Date.now()}\`,
      context_hint: '会话因上下文压缩中断，以下是压缩前的状态摘要',
      conversation_summary: hookData.transcript_summary || '(由 PreCompact hook 自动捕获)',
      tool_call_count: hookData.tool_call_count || 'unknown',
      working_directory: process.cwd(),
      note: '此文件由 pre-compact-handoff.js 自动生成，SessionStart 时自动读取并注入恢复上下文'
    };`;

const NEW_BLOCK = `    // TOOL_OUTPUT_TIER_V1 - 扫描 transcript 提取大工具输出分级摘要
    const toolOutputTiers = scanToolOutputTiers(hookData.transcript_path);

    // 构造 handoff 数据
    const handoff = {
      timestamp: new Date().toISOString(),
      session_id: hookData.session_id || \`session-\${Date.now()}\`,
      context_hint: '会话因上下文压缩中断，以下是压缩前的状态摘要',
      conversation_summary: hookData.transcript_summary || '(由 PreCompact hook 自动捕获)',
      tool_call_count: hookData.tool_call_count || 'unknown',
      working_directory: process.cwd(),
      tool_output_tiers: toolOutputTiers,
      note: '此文件由 pre-compact-handoff.js 自动生成，SessionStart 时自动读取并注入恢复上下文'
    };`;

const HELPER_FN = `
// === TOOL_OUTPUT_TIER_V1 ===
// 扫描 transcript JSONL, 按工具类型分级保留大输出, 输出 TOP-10 摘要
function scanToolOutputTiers(transcriptPath) {
  if (!transcriptPath || !fs.existsSync(transcriptPath)) {
    return { applied: false, reason: 'no transcript_path' };
  }
  try {
    const raw = fs.readFileSync(transcriptPath, 'utf8');
    const lines = raw.split('\\n').filter(Boolean);
    const items = [];
    for (const line of lines) {
      let obj;
      try { obj = JSON.parse(line); } catch { continue; }
      // 只关注 tool_result 类型 (含工具调用响应)
      const content = obj?.message?.content || obj?.content;
      if (!Array.isArray(content)) continue;
      for (const part of content) {
        if (part?.type !== 'tool_result') continue;
        const text = typeof part.content === 'string'
          ? part.content
          : Array.isArray(part.content) ? part.content.map(c => c?.text || '').join('') : '';
        const size = Buffer.byteLength(text, 'utf8');
        if (size < 500) continue;
        items.push({ size, text, tool_use_id: part.tool_use_id });
      }
    }
    items.sort((a, b) => b.size - a.size);
    const top = items.slice(0, 10).map(it => tierize(it));
    const totalBytes = items.reduce((s, it) => s + it.size, 0);
    return {
      applied: true,
      total_tool_results_scanned: items.length,
      total_bytes: totalBytes,
      top_offenders: top
    };
  } catch (e) {
    return { applied: false, reason: 'scan_error: ' + (e.message || e) };
  }
}

function tierize(item) {
  const { size, text, tool_use_id } = item;
  // 启发式工具类型判定 (transcript 不直接含工具名, 用文本特征)
  let kind = 'other';
  if (/^(File created successfully|Wrote \\d+ lines|The file .* has been (created|updated))/m.test(text)) kind = 'write';
  else if (/^\\s*\\d+→/m.test(text) || text.startsWith('     1\\t')) kind = 'read';
  else if (/<tool_use_error>|^bash:|stderr:/m.test(text) || /\\$ /m.test(text.slice(0, 100))) kind = 'bash';
  else if (/^(Found \\d+ files?|^[a-zA-Z]:\\\\.*\\.(ts|js|md|json))/m.test(text)) kind = 'glob_grep';
  else if (size > 3000 && text.includes('agent')) kind = 'agent';

  let summary;
  switch (kind) {
    case 'write':
      summary = text.split('\\n').slice(0, 2).join(' | ').slice(0, 200);
      break;
    case 'read':
      summary = '[Read] ' + text.slice(0, 200) + ' ... [+' + (size - 200) + ' bytes]';
      break;
    case 'bash':
      summary = text.slice(0, 1500) + '\\n... [truncated ' + Math.max(0, size - 2000) + ' bytes] ...\\n' + text.slice(-500);
      break;
    case 'agent':
      summary = text.slice(0, 1000) + '\\n... [Agent 完整结果已截断 ' + (size - 1000) + ' bytes]';
      break;
    case 'glob_grep':
      summary = '[Glob/Grep] ' + text.split('\\n').slice(0, 8).join(' | ').slice(0, 400);
      break;
    default:
      summary = text.slice(0, 2000) + '\\n... [+' + Math.max(0, size - 2500) + ' bytes] ...\\n' + text.slice(-500);
  }
  return { tool_use_id, kind, original_bytes: size, summary };
}
`;

function main() {
  const srcRaw = fs.readFileSync(TARGET, 'utf8');
  if (srcRaw.includes(SENTINEL)) {
    console.log('[r2] already applied, skip');
    return;
  }
  // CRLF 容忍: 检测原文行尾, 把 OLD_BLOCK/NEW_BLOCK 转换成同制
  const eol = srcRaw.includes('\r\n') ? '\r\n' : '\n';
  const oldNorm = OLD_BLOCK.replace(/\r?\n/g, eol);
  const newNorm = NEW_BLOCK.replace(/\r?\n/g, eol);
  if (!srcRaw.includes(oldNorm)) {
    console.error('[r2] anchor block not found, manual review needed (eol=' + JSON.stringify(eol) + ')');
    process.exit(1);
  }
  let next = srcRaw.replace(oldNorm, newNorm);
  // 在 IIFE 结尾 })(); 之后追加 helper 函数
  const helperNorm = HELPER_FN.replace(/\r?\n/g, eol);
  next = next.replace(/(\}\)\(\);\s*)$/, `$1${eol}${helperNorm}`);

  const bak = TARGET + '.bak.r2.' + Date.now();
  fs.copyFileSync(TARGET, bak);
  const tmp = TARGET + '.tmp.' + process.pid;
  fs.writeFileSync(tmp, next, 'utf8');
  fs.renameSync(tmp, TARGET);
  console.log('[r2] OK, bak:', path.basename(bak), 'eol=', JSON.stringify(eol));
}

main();