dsh-log-contract 0.2.0 → 0.2.1

package/bin/dsh-log-contract.mjs

@@ -161,13 +161,16 @@ function cmdContracts() {
 function cmdFix(args) {
   const json = args.includes('--json');
   const removeMarkers = args.includes('--remove-markers');
+  const dropFailedTurns = args.includes('--drop-failed-turns');
+  const trimIdx = args.indexOf('--trim-last');
+  const trimLast = trimIdx >= 0 && args[trimIdx + 1] ? Number(args[trimIdx + 1]) : undefined;
   const apply = args.includes('--apply');
   const backupDirIdx = args.indexOf('--backup-dir');
   const backupDir = backupDirIdx >= 0 && args[backupDirIdx + 1] ? args[backupDirIdx + 1] : undefined;
   const file = args.find((a) => !a.startsWith('-'));
   if (!file) fail(USAGE);
 
-  const result = repairSession(file, { removeMarkers, apply, backupDir });
+  const result = repairSession(file, { removeMarkers, dropFailedTurns, trimLast, apply, backupDir });
   if (json) {
     process.stdout.write(JSON.stringify(result, null, 2) + '\n');
     process.exit(result.ok ? 0 : 1);
@@ -176,7 +179,7 @@ function cmdFix(args) {
   process.stdout.write(`\n🔧 dsh-log-contract fix —— ${file}\n`);
   process.stdout.write(`   诊断：${result.issues.length === 0 ? '无问题' : result.issues.map((i) => `[${i.kind}] ${i.detail}`).join('\n         ')}\n`);
   if (result.applied) {
-    process.stdout.write(`   已应用修复：移除 ${result.removed} 个 marker，重编号 ${result.renumbered} 行\n`);
+    process.stdout.write(`   已应用修复：移除 ${result.removed} 项，重编号 ${result.renumbered} 行\n`);
     process.stdout.write(`   备份：${result.backupPath}\n`);
     process.stdout.write(`   修复后体检：error ${result.check.summary?.bySeverity?.error ?? '?'} ｜ surface ${result.check.summary?.surfaceNodes ?? '?'} 节点\n`);
   } else if (apply && !result.ok) {
@@ -184,7 +187,7 @@ function cmdFix(args) {
   } else if (apply) {
     process.stdout.write('   （--apply 且无问题——无内容可修）\n');
   } else {
-    process.stdout.write(`   （干跑模式：${result.removed} 个 marker 可移除、${result.renumbered} 行待重编号；加 --apply 落盘，--remove-markers 启用于移除）\n`);
+    process.stdout.write(`   （干跑模式：${result.removed} 项可移除、${result.renumbered} 行待重编号；加 --apply 落盘，--remove-markers / --drop-failed-turns / --trim-last N 启用于对应修复）\n`);
   }
   process.stdout.write('\n');
   process.exit(result.ok ? 0 : 1);

package/lib/index.js

@@ -7,5 +7,5 @@
 export { loadSessionLog } from './log-reader.js';
 export { validateSessionLog } from './validate.js';
 export { createPreWriter, preWriterFromLog } from './prewrite.js';
-export { repairSession, strictScanText, removeMarkersText, rebuildZstdText } from './repair.js';
+export { repairSession, strictScanText, removeMarkersText, dropFailedTurnsText, trimLastMessagesText, rebuildZstdText } from './repair.js';
 export { CONTRACT_RULES, LAYER, SEVERITY, ruleById } from './contracts.js';

package/lib/repair.js

@@ -208,6 +208,168 @@ export function removeMarkersText(text) {
   return { text: fixed, removed, renumbered, markerSeqs };
 }
 
+/**
+ * 通用"删除事件 + 全量重编号"引擎：按谓词删行，seq/seq0/sourceEventSeqs/
+ * surfaceOp 范围同步平移。
+ * @param {string[]} parts - text.split('\n')。
+ * @param {(event: object) => boolean} dropPredicate - 命中即删除该事件行。
+ * @returns {{ text: string, removed: number, renumbered: number }}
+ */
+function renumberWithDrops(parts, dropPredicate) {
+  const dropped = new Set();
+  for (const raw of parts) {
+    if (!raw.trim()) continue;
+    let v;
+    try {
+      v = JSON.parse(raw);
+    } catch {
+      continue;
+    }
+    if (typeof v.seq === 'number' && dropPredicate(v)) dropped.add(v.seq);
+  }
+  const sortedDrops = [...dropped].sort((a, b) => a - b);
+  const shiftFor = (seq) => {
+    let lo = 0;
+    let hi = sortedDrops.length;
+    while (lo < hi) {
+      const mid = (lo + hi) >> 1;
+      if (sortedDrops[mid] < seq) lo = mid + 1;
+      else hi = mid;
+    }
+    return lo;
+  };
+  const isDrop = (seq) => dropped.has(seq);
+  const out = [];
+  let removed = 0;
+  let renumbered = 0;
+  for (const raw of parts) {
+    if (!raw.trim()) {
+      out.push(raw);
+      continue;
+    }
+    let v;
+    try {
+      v = JSON.parse(raw);
+    } catch {
+      out.push(raw);
+      continue;
+    }
+    if (typeof v.seq === 'number' && dropPredicate(v)) {
+      removed++;
+      continue;
+    }
+    let touched = false;
+    if (typeof v.seq === 'number') {
+      const s = shiftFor(v.seq);
+      if (s) {
+        v.seq -= s;
+        touched = true;
+      }
+    }
+    if (CHUNK_ROW_TYPES.has(v.type) && typeof v.seq0 === 'number') {
+      const s = shiftFor(v.seq0);
+      if (s) {
+        v.seq0 -= s;
+        touched = true;
+      }
+    }
+    if (Array.isArray(v.sourceEventSeqs)) {
+      const next = v.sourceEventSeqs.filter((x) => !isDrop(x)).map((x) => x - shiftFor(x));
+      if (next.length !== v.sourceEventSeqs.length || next.some((x, k) => x !== v.sourceEventSeqs[k])) {
+        v.sourceEventSeqs = next;
+        touched = true;
+      }
+    }
+    if (v.surfaceOp && v.surfaceOp.op === 'replace') {
+      const ns = v.surfaceOp.start - shiftFor(v.surfaceOp.start);
+      const ne = v.surfaceOp.end - shiftFor(v.surfaceOp.end);
+      if (ns !== v.surfaceOp.start || ne !== v.surfaceOp.end) {
+        v.surfaceOp.start = ns;
+        v.surfaceOp.end = ne;
+        touched = true;
+      }
+    }
+    out.push(touched ? JSON.stringify(v) : raw);
+    if (touched) renumbered++;
+  }
+  let fixed = out.join('\n');
+  if (!fixed.endsWith('\n')) fixed += '\n';
+  return { text: fixed, removed, renumbered };
+}
+
+/**
+ * 删除"本轮运行失败"的轮次（turn/end 带 reason.kind==='error' 的完整轮次：
+ * [turnStart..turnEnd]，含其中的 user 消息与失败产出），并全量重编号。
+ * 用于清掉界面上的失败报错气泡。
+ * @param {string} text - JSONL 全文。
+ * @returns {{ text: string, removed: number, renumbered: number, failedTurns: number }}
+ */
+export function dropFailedTurnsText(text) {
+  const parts = text.split('\n');
+  const spans = [];
+  let curTurnStart = null;
+  for (const raw of parts) {
+    if (!raw.trim()) continue;
+    let v;
+    try {
+      v = JSON.parse(raw);
+    } catch {
+      continue;
+    }
+    if (v.type === 'turn/start') curTurnStart = v.seq;
+    else if (v.type === 'turn/end') {
+      if (v.data?.reason?.kind === 'error' && curTurnStart !== null) spans.push([curTurnStart, v.seq]);
+      curTurnStart = null;
+    }
+  }
+  if (spans.length === 0) return { text, removed: 0, renumbered: 0, failedTurns: 0 };
+  const inSpan = (seq) => spans.some(([s, e]) => seq >= s && seq <= e);
+  const r = renumberWithDrops(parts, (v) => inSpan(v.seq));
+  return { ...r, failedTurns: spans.length };
+}
+
+/**
+ * 裁剪到最近 keepMessages 条 append 消息（保留其所在 turn 的结构），
+ * 同时移除全部 retrace/message-editor marker，并全量重编号。
+ * 用于"完整历史超出模型 context 窗口"（MiMo 1M tokens 实锤）。
+ * @param {string} text - JSONL 全文。
+ * @param {number} keepMessages - 保留的 append 消息数。
+ * @returns {{ text: string, removed: number, renumbered: number, kept: number, cutoff: number }}
+ */
+export function trimLastMessagesText(text, keepMessages) {
+  const parts = text.split('\n');
+  const msgSeqs = [];
+  let lastTurnStart = null;
+  for (const raw of parts) {
+    if (!raw.trim()) continue;
+    let v;
+    try {
+      v = JSON.parse(raw);
+    } catch {
+      continue;
+    }
+    if (v.type === 'user/message' || (v.type === 'assistant/message' && v.surfaceOp === 'append')) msgSeqs.push(v.seq);
+  }
+  const total = msgSeqs.length;
+  if (total <= keepMessages) return { text, removed: 0, renumbered: 0, kept: total, cutoff: 0 };
+  const cutoffMsg = msgSeqs[total - keepMessages];
+  // 往前取到包含 cutoffMsg 的那个 turn 的 turn/start，保住轮次结构
+  let cutoff = cutoffMsg;
+  for (const raw of parts) {
+    if (!raw.trim()) continue;
+    let v;
+    try {
+      v = JSON.parse(raw);
+    } catch {
+      continue;
+    }
+    if (v.type === 'turn/start' && v.seq < cutoffMsg) cutoff = v.seq;
+  }
+  const drop = (v) => (typeof v.seq === 'number' && v.seq < cutoff) || isRetraceMarker(v);
+  const r = renumberWithDrops(parts, drop);
+  return { ...r, kept: keepMessages, cutoff };
+}
+
 /**
  * 按官方 writer 帧格式重建 .jsonl.zstd（帧1=header，帧2=其余，均带 checksum）。
  * @param {string} text - JSONL 全文（以 "\n" 结尾）。
@@ -232,7 +394,10 @@ export function rebuildZstdText(text) {
 /**
  * 对单个会话日志执行诊断 +（可选）修复。
  * @param {string} file - .jsonl 或 .jsonl.zstd 路径。
- * @param {{ removeMarkers?: boolean, apply?: boolean, backupDir?: string }} opts
+ * @param {{
+ *   removeMarkers?: boolean, dropFailedTurns?: boolean, trimLast?: number,
+ *   apply?: boolean, backupDir?: string
+ * }} opts
  * @returns {{
  *   file, ok, issues: Array<{kind:string, detail:string}>,
  *   removed, renumbered, backupPath, applied,
@@ -270,25 +435,38 @@ export function repairSession(file, opts = {}) {
   }
   let removed = 0;
   let renumbered = 0;
-  if (opts.removeMarkers) {
-    const r = removeMarkersText(plain);
-    removed = r.removed;
-    renumbered = r.renumbered;
+  const applyFix = (label, r, detail) => {
+    removed += r.removed;
+    renumbered += r.renumbered;
     if (r.removed > 0) {
-      issues.push({ kind: 'markers', detail: `移除 ${r.removed} 个 retrace/message-editor marker（重编号 ${r.renumbered} 行）` });
+      issues.push({ kind: label, detail });
       plain = r.text;
-      // 修复后复检
-      const scan2 = strictScanText(plain);
-      if (scan2.failures.length > 0) issues.push({ kind: 'post-scan', detail: `修复后仍有 ${scan2.failures.length} 处 seq 不连续` });
-      const check2 = validateSessionLog(loadSessionLogFromText(plain));
-      if (!check2.ok) {
-        const errs2 = check2.violations.filter((v) => v.severity === 'error');
-        issues.push({ kind: 'post-contract', detail: `修复后仍有 ${errs2.length} 个 error 级违规` });
-      }
     }
+  };
+  if (opts.dropFailedTurns) {
+    const r = dropFailedTurnsText(plain);
+    applyFix('failed-turns', r, `移除 ${r.failedTurns} 个失败轮次（"本轮运行失败"报错气泡，重编号 ${r.renumbered} 行）`);
+  }
+  if (opts.removeMarkers) {
+    const r = removeMarkersText(plain);
+    applyFix('markers', r, `移除 ${r.removed} 个 retrace/message-editor marker（重编号 ${r.renumbered} 行）`);
+  }
+  if (typeof opts.trimLast === 'number') {
+    const r = trimLastMessagesText(plain, opts.trimLast);
+    applyFix('trim', r, `裁剪到最近 ${r.kept} 条消息（丢弃 ${r.removed} 行，重编号 ${r.renumbered} 行）`);
+  }
+  // 全部修复完成后做一次终检（中间态的临时违规不阻塞——后续修复可能已消除）
+  const scanFinal = strictScanText(plain);
+  const checkFinal = validateSessionLog(loadSessionLogFromText(plain));
+  if (scanFinal.failures.length > 0) {
+    const f = scanFinal.failures[0];
+    issues.push({ kind: 'final-scan', detail: `修复后仍不连续：line ${f.line} 期望 ${f.expected} 实际 ${f.got}——${scanFinal.failures.length} 处` });
+  }
+  if (!checkFinal.ok) {
+    const errs = checkFinal.violations.filter((v) => v.severity === 'error');
+    issues.push({ kind: 'final-contract', detail: `修复后仍有 ${errs.length} 个 error 级违规（首条：${errs[0]?.id ?? '-'} ${errs[0]?.message?.slice(0, 90) ?? ''}）` });
   }
-  const postBroken = issues.some((i) => i.kind.startsWith('post-'));
-  const ok = scan.failures.length === 0 && check.ok && !postBroken;
+  const ok = scanFinal.failures.length === 0 && checkFinal.ok;
   let backupPath = null;
   let applied = false;
   if (opts.apply && ok) {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "dsh-log-contract",
   "description": "日志契约守护 — DSH session log contract guard: offline health check (CLI) + pre-write validation for DeepSeek Harness session logs",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "packageManager": "pnpm@11.7.0",
   "type": "module",
   "main": "lib/index.js",