dsh-log-slicer 0.1.3 → 0.1.5

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # 变更记录
 
+## 0.1.5
+
+（0.1.4 被 registry 的一次半途 PUT 占了号，发不出去；内容原样并入本版。）
+
+这一版防的不是"宿主换个界面"，是"宿主换套日志"。不加新功能，只把最容易断的几处改成认表；宿主真改了格式，第一眼能看见报错，而不是切出一节空文。
+
+- **加宽**：日志文件名认 `session[.v<N>].jsonl.zstd` 与 `.zst`，不再写死 `session.v4`；`--list` 与两个工具的输出都带上认到的 schema 号。
+- **加宽**：事件名认一张表（`user/message`、`message/user`、`assistant/message`、`message/assistant`、`user`、`assistant`、`model/message`、`message/model`）；正文位置认 `data.content`、`data.message.content`、`data.message`、`content`、`message`、`text`、`data.text`，以及 `{content:[…]}`/`{parts:[…]}` 两种包装；文本块认 `{text}`、`{text:{value}}`、`input_text`、`output_text`。
+- **改成大声失败**：一条轮次都认不出来时报 `LogSchemaError`，消息里带"读了多少事件、见过的事件类型各几条"——宿主升级之后，这一句就是诊断书。
+- **加宽**：注册口先认 `ctx.tools.register`，认不到试 `ctx.tool()`，都没有才抛；`apply` 照旧吞异常，不带崩应用。
+- **修**：`dsh.plugin.json` 的版本号一直停在 0.1.0，补到与包一致。
+
 ## 0.1.3
 
 0.1.2 装进 profile 之后，重启没有任何反应——两个工具都没出现。查下来两处：

package/bin/log-slicer.mjs

@@ -85,7 +85,7 @@ function main() {
       return `${startNo + i}. [${who}] ${when}  ${body}`;
     });
     process.stdout.write(
-      `# 日志 ${all.log}\n# ${all.frames} 帧（${all.failedFrames} 帧解不开）/ ${all.totalTurns} 轮有效\n\n` +
+      `# 日志 ${all.log}\n# schema ${all.schema ?? '?'} / ${all.frames} 帧（${all.failedFrames} 帧解不开）/ ${all.totalTurns} 轮有效\n\n` +
         lines.join('\n') + '\n',
     );
     return;
@@ -113,7 +113,7 @@ function main() {
 
   const out = [];
   out.push(`# 日志 ${built.log}`);
-  out.push(`# ${(built.logSize / 1048576).toFixed(1)} MB / ${built.frames} 帧（${built.failedFrames} 帧解不开）/ ${built.totalTurns} 轮有效`);
+  out.push(`# schema ${built.schema ?? '?'} / ${(built.logSize / 1048576).toFixed(1)} MB / ${built.frames} 帧（${built.failedFrames} 帧解不开）/ ${built.totalTurns} 轮有效`);
   out.push(`# 切成 ${built.segment.length} 轮，其中你 ${built.counts.you} 轮、我 ${built.counts.me} 轮`);
   if (built.imageCount) {
     out.push(`# 图片 ${built.imageCount} 张：`);

package/dsh.plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "dsh-log-slicer",
   "description": "Slice one section out of a DSH session log: verbatim turns, catalog row, backup and verification hash.",
-  "version": "0.1.0",
+  "version": "0.1.5",
   "entry": {
     "name": "dsh-log-slicer",
     "inject": [

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "dsh-log-slicer",
-  "version": "0.1.3",
-  "description": "Slice one section out of a DeepSeek Harness session log: verbatim turns, images, catalog row, backups and a verification hash 鈥?one command, dry-run by default.",
+  "version": "0.1.5",
+  "description": "Slice one section out of a DeepSeek Harness session log: verbatim turns, images, catalog row, backups and a verification hash — one command, dry-run by default.",
   "keywords": [
     "dsh",
     "dsh-plugin",
@@ -39,6 +39,9 @@
   "engines": {
     "node": ">=22.15.0"
   },
+  "scripts": {
+    "test": "node --test"
+  },
   "peerDependencies": {
     "@deepseek-ai/dsh-tools": "*"
   },
@@ -51,8 +54,5 @@
     "bundle": {
       "patch": "./cordis.patch.yml"
     }
-  },
-  "scripts": {
-    "test": "node --test"
   }
-}
+}

package/src/clean.js

@@ -3,6 +3,11 @@
  *
  * 只留人和模型真正说的话：宿主注入的环境文本、图片占位行都扔掉。
  * 图片留下 sha256 与媒体类型，供上层落成文件。
+ *
+ * 这一层是"宿主换版本最先崩的地方"：事件名、正文容器的形状都可能变。
+ * 所以这里不认死一个名字，认一张表；表里没见过的形状先跳过，
+ * 最后如果一条轮次都没认出来，就报错并列出见过的事件类型——宁可当场停下，
+ * 也不要静悄悄切出一节空文。
  */
 
 const ENV_PREFIXES = [
@@ -16,6 +21,37 @@ const ENV_PREFIXES = [
 
 const IMAGE_PLACEHOLDER = /^Image "[^"]+" \(sha256:[0-9a-f]{4,}/i;
 
+/** 事件名 → 说话的人。见过的名字都收在这里；宿主改名时往这里加一行就行。 */
+const EVENT_ROLES = {
+  'user/message': 'user',
+  'message/user': 'user',
+  'user': 'user',
+  'assistant/message': 'me',
+  'message/assistant': 'me',
+  'assistant': 'me',
+  'model/message': 'me',
+  'message/model': 'me',
+};
+
+/** 正文可能待在的几个位置，按先后试。 */
+const CONTENT_PATHS = [
+  (ev) => ev.data && ev.data.content,
+  (ev) => ev.data && ev.data.message && ev.data.message.content,
+  (ev) => ev.data && ev.data.message,
+  (ev) => ev.content,
+  (ev) => ev.message,
+  (ev) => ev.text,
+  (ev) => ev.data && ev.data.text,
+];
+
+export class LogSchemaError extends Error {
+  constructor(message, eventTypes = []) {
+    super(message);
+    this.name = 'LogSchemaError';
+    this.eventTypes = eventTypes;
+  }
+}
+
 export function isEnvironmentText(text) {
   if (typeof text !== 'string') return true;
   const t = text.trim();
@@ -24,11 +60,43 @@ export function isEnvironmentText(text) {
   return ENV_PREFIXES.some((p) => t.startsWith(p));
 }
 
+/** 正文块可能是字符串、`{text}`、`{text:{value}}`；都收。 */
+function blockText(x) {
+  if (typeof x === 'string') return x;
+  if (!x || typeof x !== 'object') return '';
+  if (typeof x.text === 'string') return x.text;
+  if (x.text && typeof x.text === 'object' && typeof x.text.value === 'string') return x.text.value;
+  if (typeof x.value === 'string') return x.value;
+  return '';
+}
+
+function itemsOf(ev) {
+  for (const pick of CONTENT_PATHS) {
+    let v;
+    try {
+      v = pick(ev);
+    } catch {
+      continue;
+    }
+    if (v == null) continue;
+    if (typeof v === 'string') return v ? [{ type: 'text', text: v }] : [];
+    if (Array.isArray(v)) return v;
+    if (typeof v === 'object') {
+      // 有些版本把正文包成 {content:[...]} 或 {parts:[...]}
+      if (Array.isArray(v.content)) return v.content;
+      if (Array.isArray(v.parts)) return v.parts;
+      if (typeof v.text === 'string') return [{ type: 'text', text: v.text }];
+    }
+  }
+  return [];
+}
+
 function textOf(items) {
   if (!Array.isArray(items)) return '';
   return items
-    .filter((x) => x && x.type === 'text' && typeof x.text === 'string')
-    .map((x) => x.text)
+    .filter((x) => x && (x.type === 'text' || x.type === 'input_text' || x.type === 'output_text'))
+    .map(blockText)
+    .filter(Boolean)
     .join('\n');
 }
 
@@ -37,19 +105,21 @@ function imagesOf(items) {
   const out = [];
   for (const x of items) {
     if (!x || x.type !== 'image') continue;
-    const a = x.attachment || {};
-    const id = String(a.attachmentId || a.id || '').replace(/^sha256:sha256:/, 'sha256:');
-    out.push({ id, sha256: id.replace(/^sha256:/, ''), mediaType: a.mediaType || '' });
+    const a = x.attachment || x.source || {};
+    const id = String(a.attachmentId || a.id || a.sha256 || '').replace(/^sha256:sha256:/, 'sha256:');
+    out.push({ id, sha256: id.replace(/^sha256:/, ''), mediaType: a.mediaType || a.media_type || '' });
   }
   return out;
 }
 
 /**
  * @param {string} text 解压后的 JSONL 文本
- * @returns {Array<{who: 'user'|'me', ts: number|null, text: string, images: Array}>}
+ * @returns {{turns: Array<{who: 'user'|'me', ts: number|null, text: string, images: Array}>, eventTypes: Record<string, number>}}
  */
-export function turnsFromText(text) {
+export function parseTurnLog(text) {
   const turns = [];
+  const eventTypes = {};
+  let events = 0;
   for (const line of String(text).split('\n')) {
     if (!line.trim()) continue;
     let ev;
@@ -59,19 +129,35 @@ export function turnsFromText(text) {
       continue;
     }
     if (!ev || typeof ev !== 'object') continue;
-    const ts = ev.time ?? ev.timestamp ?? null;
-    if (ev.type === 'user/message') {
-      const items = (ev.data && ev.data.content) || [];
-      const body = textOf(items);
-      const images = imagesOf(items);
-      if (isEnvironmentText(body) && images.length === 0) continue;
-      turns.push({ who: 'user', ts, text: body.trim(), images });
-    } else if (ev.type === 'assistant/message') {
-      const items = (ev.data && ev.data.message && ev.data.message.content) || [];
-      const body = textOf(items);
-      if (isEnvironmentText(body)) continue;
-      turns.push({ who: 'me', ts, text: body.trim(), images: [] });
-    }
+    events++;
+    const type = String(ev.type ?? '');
+    eventTypes[type] = (eventTypes[type] || 0) + 1;
+    const role = EVENT_ROLES[type];
+    if (!role) continue;
+    const ts = ev.time ?? ev.timestamp ?? (ev.data && ev.data.time) ?? null;
+    const items = itemsOf(ev);
+    const body = textOf(items);
+    const images = role === 'user' ? imagesOf(items) : [];
+    if (isEnvironmentText(body) && images.length === 0) continue;
+    turns.push({ who: role, ts, text: body.trim(), images });
+  }
+  if (turns.length === 0 && events > 0) {
+    const top = Object.entries(eventTypes)
+      .sort((a, b) => b[1] - a[1])
+      .slice(0, 8)
+      .map(([k, v]) => `${k || '(空)'}×${v}`)
+      .join('、');
+    throw new LogSchemaError(
+      `read ${events} events but recognised no turns (event types: ${top}). ` +
+        'The host log format probably changed — add the new names to EVENT_ROLES in src/clean.js, ' +
+        'or check CONTENT_PATHS if the names are the same but the body moved.',
+      Object.keys(eventTypes),
+    );
   }
-  return turns;
+  return { turns, eventTypes };
+}
+
+/** 兼容旧调用：只要轮次。 */
+export function turnsFromText(text) {
+  return parseTurnLog(text).turns;
 }

package/src/decode.js

@@ -63,8 +63,21 @@ export function decodeFile(filePath, options) {
 }
 
 /**
- * 在 $DSH_HOME/sessions 下找 session.v4.jsonl.zstd，按修改时间倒序。
- * sessions/<workspace>/<sessionId>/session.v4.jsonl.zstd
+ * 日志文件的形状：session.v4.jsonl.zstd、session.v5.jsonl.zstd、session.jsonl.zst……
+ * 版本号只用来排序与记录；不拿它当"认不认这份日志"的门槛——宿主改版本时，
+ * 这里多认一个文件名，比让整条链路失明强。
+ */
+const SESSION_FILE = /^session(?:\.v(\d+))?\.jsonl\.(?:zstd|zst)$/i;
+
+export function schemaOf(fileName) {
+  const m = SESSION_FILE.exec(path.basename(fileName));
+  if (!m) return null;
+  return m[1] ? Number(m[1]) : 0;
+}
+
+/**
+ * 在 $DSH_HOME/sessions 下找会话日志，按修改时间倒序。
+ * sessions/<workspace>/<sessionId>/session.v<N>.jsonl.zstd
  */
 export function findSessionLogs(root = path.join(dshHome(), 'sessions')) {
   const out = [];
@@ -84,12 +97,24 @@ export function findSessionLogs(root = path.join(dshHome(), 'sessions')) {
       continue;
     }
     for (const s of subs) {
-      const file = path.join(wsDir, s.name, 'session.v4.jsonl.zstd');
+      if (!s.isDirectory()) continue;
+      let files;
       try {
-        const st = fs.statSync(file);
-        out.push({ path: file, workspace: ws.name, session: s.name, mtime: st.mtimeMs, size: st.size });
+        files = fs.readdirSync(path.join(wsDir, s.name), { withFileTypes: true });
       } catch {
-        /* 不是会话目录 */
+        continue;
+      }
+      for (const f of files) {
+        if (!f.isFile()) continue;
+        const schema = schemaOf(f.name);
+        if (schema == null) continue;
+        const file = path.join(wsDir, s.name, f.name);
+        try {
+          const st = fs.statSync(file);
+          out.push({ path: file, workspace: ws.name, session: s.name, schema, mtime: st.mtimeMs, size: st.size });
+        } catch {
+          /* 读不到就算了 */
+        }
       }
     }
   }
@@ -99,6 +124,10 @@ export function findSessionLogs(root = path.join(dshHome(), 'sessions')) {
 
 export function latestSessionLog(root) {
   const logs = findSessionLogs(root);
-  if (!logs.length) throw new Error('no session.v4.jsonl.zstd found under ' + (root || path.join(dshHome(), 'sessions')));
+  if (!logs.length) {
+    throw new Error(
+      'no session log found under ' + (root || path.join(dshHome(), 'sessions')) + ' — looked for session[.v<N>].jsonl.zstd/.zst',
+    );
+  }
   return logs[0];
 }

package/src/index.js

@@ -162,7 +162,7 @@ async function probe({ logPath, limit, width } = {}) {
   });
   return [
     `# ${built.log}`,
-    `# ${built.frames} frames (${built.failedFrames} failed) / ${built.totalTurns} turns`,
+    `# schema ${built.schema ?? '?'} / ${built.frames} frames (${built.failedFrames} failed) / ${built.totalTurns} turns`,
     '',
     lines.join('\n'),
   ].join('\n');
@@ -178,7 +178,7 @@ async function slice(args = {}) {
   });
   const head = [
     `# ${built.log}`,
-    `# ${built.frames} frames (${built.failedFrames} failed) / ${built.totalTurns} turns`,
+    `# schema ${built.schema ?? '?'} / ${built.frames} frames (${built.failedFrames} failed) / ${built.totalTurns} turns`,
     `# sliced ${built.segment.length} turns (you ${built.counts.you} / me ${built.counts.me})`,
   ].join('\n');
 
@@ -226,9 +226,17 @@ async function slice(args = {}) {
 
 async function registerAll(ctx) {
   const defineTool = await loadDefineTool();
+  // 宿主的注册口可能在版本间挪位置：先认 ctx.tools.register，认不到就试 ctx.tool()，
+  // 都没有才抛——抛了会在 apply 里被记一行日志，不会带崩应用。
+  const register = typeof ctx?.tools?.register === 'function'
+    ? ctx.tools.register.bind(ctx.tools)
+    : typeof ctx?.tool === 'function'
+      ? ctx.tool.bind(ctx)
+      : null;
+  if (!register) throw new Error('host exposes neither ctx.tools.register nor ctx.tool');
   const disposers = [
-    ctx.tools.register(define(defineTool, probeSpec, probe)),
-    ctx.tools.register(define(defineTool, sliceSpec, slice)),
+    register(define(defineTool, probeSpec, probe)),
+    register(define(defineTool, sliceSpec, slice)),
   ];
   return () => {
     for (const d of disposers.reverse()) d?.();

package/src/pipeline.js

@@ -5,8 +5,9 @@
  */
 
 import fs from 'node:fs';
-import { decodeFile, latestSessionLog } from './decode.js';
-import { turnsFromText } from './clean.js';
+import path from 'node:path';
+import { decodeFile, latestSessionLog, schemaOf } from './decode.js';
+import { parseTurnLog } from './clean.js';
 import { sliceTurns, countTurns } from './slice.js';
 import { renderChat, renderWork, applyImageNotes, findPointerLines } from './render.js';
 
@@ -23,9 +24,9 @@ export function buildSection(options = {}) {
   const { kind = 'chat', title, from, to, logPath, workHeader } = options;
   if (!title) throw new Error('title is required');
 
-  const log = logPath ? { path: logPath, size: fsSize(logPath), mtime: 0 } : latestSessionLog();
+  const log = logPath ? { path: logPath, size: fsSize(logPath), mtime: 0, schema: schemaOf(logPath) } : latestSessionLog();
   const decoded = decodeFile(log.path);
-  const turns = turnsFromText(decoded.text);
+  const { turns, eventTypes } = parseTurnLog(decoded.text);
   const segment = sliceTurns(turns, { from, to });
   const counts = countTurns(segment);
 
@@ -37,9 +38,12 @@ export function buildSection(options = {}) {
 
   return {
     log: log.path,
+    logFile: path.basename(log.path),
+    schema: log.schema ?? null,
     logSize: decoded.frames ? log.size || 0 : 0,
     frames: decoded.frames,
     failedFrames: decoded.failed,
+    eventTypes,
     totalTurns: turns.length,
     segment,
     counts,