codex-work-receipt 0.7.2 → 0.7.3

package/CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## 0.7.3
+
+- Codex 会话日志改为分块逐行解析，避免超大 `.jsonl` 在转成单个 JavaScript 字符串时触发 V8 长度上限。
+- 会话行解析后立即丢弃 Prompt、回复、图片和工具输出，只在内存中保留统计所需的时间、事件、模型与 Token 字段。
+- 单行异常过大时带文件路径和行号跳过；文件读取失败时补充路径与体积，便于定位具体日志。
+- 本地 `history.jsonl` 改为逐条写入，避免历史增长后形成同类的大字符串拼接风险。
+- 小票生成或票仔安装成功后在终端展示开源仓库地址，并用对应的中英文文案邀请用户为项目点 Star。
+- 修正 npm 包的 repository、homepage 和 issues 地址，使其统一指向 `a-bai-2026/codex-work-receipt`。
+
 ## 0.7.2
 
 - 修复同一 Codex 会话存在多份 append-only 日志修订时生成重复 canonical factId、导致小程序云端入库拒绝的问题。

package/docs/privacy.en.md

@@ -10,6 +10,8 @@ AI Work Receipt reads Codex sessions, computes metrics, and renders receipts loc
 - Event types, timestamps, model names, and numerical metadata
 - Metrics such as turns, tool calls, Tokens, duration, and interruptions
 
+Session logs are parsed locally in bounded chunks, one JSONL row at a time. After each row is parsed, only fields needed for metrics remain in memory; prompts, responses, images, and tool output are discarded from the in-memory statistics representation.
+
 ## What receipts exclude
 
 - Prompts and response text

package/docs/privacy.md

@@ -10,6 +10,8 @@ AI 打工小票在本机读取 Codex 会话，并在本机完成统计和渲染
 - 提取事件类型、时间、模型名称和数字统计
 - 计算轮次、工具调用、Token、时长和中断次数等指标
 
+会话日志会在本机分块逐行解析。每行解析后只保留统计所需字段，Prompt、回复、图片和工具输出不会保留在内存统计对象中。
+
 ## 不会写入小票什么
 
 - Prompt 和回复正文

package/package.json

@@ -1,16 +1,16 @@
 {
   "name": "codex-work-receipt",
-  "version": "0.7.2",
+  "version": "0.7.3",
   "type": "module",
   "description": "Generate a privacy-first Codex work receipt from local session metadata.",
   "author": "a-bai-2026",
   "repository": {
     "type": "git",
-    "url": "git+https://github.com/a-bai-2026/ai-work-receipt.git"
+    "url": "git+https://github.com/a-bai-2026/codex-work-receipt.git"
   },
-  "homepage": "https://github.com/a-bai-2026/ai-work-receipt#readme",
+  "homepage": "https://github.com/a-bai-2026/codex-work-receipt#readme",
   "bugs": {
-    "url": "https://github.com/a-bai-2026/ai-work-receipt/issues"
+    "url": "https://github.com/a-bai-2026/codex-work-receipt/issues"
   },
   "license": "GPL-3.0-only",
   "keywords": [

package/src/cli.mjs

@@ -9,6 +9,7 @@ import QRCode from "qrcode";
 import { parseArgs, printHelp } from "./core/args.mjs";
 import { buildCanonicalFacts } from "./core/fact-buckets.mjs";
 import { collectMetrics } from "./core/metrics.mjs";
+import { printOpenSourcePrompt } from "./core/open-source.mjs";
 import { getRollingSummaryNotice, getScopeLabel } from "./core/presentation.mjs";
 import { encodeReceiptPayloads } from "./core/qr-payload.mjs";
 import { outputSlugForRange, resolveRange } from "./core/range.mjs";
@@ -81,6 +82,7 @@ async function main() {
       console.log("请重启 Codex，打开 Settings > Pets，点击 Refresh 后选择“票仔 · AI 小票工”。");
       console.log("输入 /pet 唤醒票仔；以后可以说：票仔，开今天的票。");
     }
+    printOpenSourcePrompt("pet", options.locale);
     return;
   }
   if (options.uninstallPet) {
@@ -163,6 +165,7 @@ async function main() {
     }
     if (!miniProgramCodeDataUrl) console.log("小程序码：尚未配置，页面使用明确占位符");
   }
+  printOpenSourcePrompt("receipt", options.locale);
   if (options.open) openFile(outputFile);
 }
 

package/src/core/open-source.mjs

@@ -0,0 +1,43 @@
+export const OPEN_SOURCE_REPOSITORY_URL = "https://github.com/a-bai-2026/codex-work-receipt";
+
+const PROMPTS = {
+  "zh-CN": {
+    receipt: {
+      label: "开源项目",
+      message: "如果你也喜欢这个 AI 小票工具，欢迎来 GitHub 给我点个 Star ⭐",
+    },
+    pet: {
+      label: "票仔开源地址",
+      message: "如果你也喜欢票仔，欢迎来 GitHub 给我点个 Star ⭐",
+    },
+  },
+  en: {
+    receipt: {
+      label: "Open-source project",
+      message: "If you enjoy AI Work Receipt, a GitHub Star would mean a lot ⭐",
+    },
+    pet: {
+      label: "Ticket Buddy source",
+      message: "If you enjoy Ticket Buddy, a GitHub Star would mean a lot ⭐",
+    },
+  },
+};
+
+export function getOpenSourcePrompt(kind = "receipt", locale = "zh-CN") {
+  const normalizedLocale = locale === "en" ? "en" : "zh-CN";
+  const localized = PROMPTS[normalizedLocale];
+  const copy = localized[kind] || localized.receipt;
+  return {
+    url: OPEN_SOURCE_REPOSITORY_URL,
+    label: copy.label,
+    message: copy.message,
+    separator: normalizedLocale === "en" ? ": " : "：",
+  };
+}
+
+export function printOpenSourcePrompt(kind, locale, writeLine = console.log) {
+  const prompt = getOpenSourcePrompt(kind, locale);
+  writeLine("");
+  writeLine(`${prompt.label}${prompt.separator}${prompt.url}`);
+  writeLine(prompt.message);
+}

package/src/core/receipt-record.mjs

@@ -155,11 +155,13 @@ export function persistReceiptRecord(record, outputHtmlPath, requestedDataDir =
   }
   const history = [...deduplicated.values()]
     .sort((left, right) => String(left.period?.end_at).localeCompare(String(right.period?.end_at)));
-  fs.writeFileSync(
-    path.join(dataDir, "history.jsonl"),
-    history.length ? `${history.map((item) => JSON.stringify(item)).join("\n")}\n` : "",
-    "utf8",
-  );
+  const historyPath = path.join(dataDir, "history.jsonl");
+  const historyDescriptor = fs.openSync(historyPath, "w");
+  try {
+    for (const item of history) fs.writeSync(historyDescriptor, `${JSON.stringify(item)}\n`, null, "utf8");
+  } finally {
+    fs.closeSync(historyDescriptor);
+  }
 
   const companionPath = /\.html?$/i.test(outputHtmlPath)
     ? outputHtmlPath.replace(/\.html?$/i, ".json")

package/src/parsers/codex.mjs

@@ -5,6 +5,17 @@ import path from "node:path";
 
 import { rowDate } from "../lib/time.mjs";
 
+const READ_CHUNK_BYTES = 256 * 1024;
+const MAX_JSONL_ROW_BYTES = 64 * 1024 * 1024;
+const TAIL_HASH_BYTES = 4096;
+const TOKEN_KEYS = [
+  "input_tokens",
+  "cached_input_tokens",
+  "output_tokens",
+  "reasoning_output_tokens",
+  "total_tokens",
+];
+
 function walkJsonlFiles(directory, accumulator = []) {
   if (!fs.existsSync(directory)) return accumulator;
   for (const entry of fs.readdirSync(directory, { withFileTypes: true })) {
@@ -15,27 +26,183 @@ function walkJsonlFiles(directory, accumulator = []) {
   return accumulator;
 }
 
-function readJsonl(filePath) {
+function shortString(value, maximumLength = 512) {
+  return typeof value === "string" && value.length <= maximumLength ? value : null;
+}
+
+function finiteNumber(value) {
+  return typeof value === "number" && Number.isFinite(value) ? value : null;
+}
+
+function compactTokenUsage(value) {
+  if (!value || typeof value !== "object") return null;
+  const usage = {};
+  for (const key of TOKEN_KEYS) {
+    const amount = finiteNumber(value[key]);
+    if (amount !== null) usage[key] = amount;
+  }
+  return Object.keys(usage).length ? usage : null;
+}
+
+function compactPayload(rowType, value) {
+  if (!value || typeof value !== "object") return null;
+  const payload = {};
+  for (const key of ["completed_at", "started_at", "timestamp"]) {
+    const timestamp = shortString(value[key], 128);
+    if (timestamp !== null) payload[key] = timestamp;
+  }
+
+  const payloadType = shortString(value.type, 128);
+  if (payloadType !== null) payload.type = payloadType;
+
+  if (rowType === "session_meta") {
+    const sessionId = shortString(value.session_id);
+    const fallbackId = shortString(value.id);
+    if (sessionId !== null) payload.session_id = sessionId;
+    if (fallbackId !== null) payload.id = fallbackId;
+  } else if (rowType === "turn_context") {
+    const model = shortString(value.model);
+    if (model !== null) payload.model = model;
+  } else if (rowType === "event_msg") {
+    const durationMs = finiteNumber(value.duration_ms);
+    const firstTokenMs = finiteNumber(value.time_to_first_token_ms);
+    if (durationMs !== null) payload.duration_ms = durationMs;
+    if (firstTokenMs !== null) payload.time_to_first_token_ms = firstTokenMs;
+    const totalTokenUsage = compactTokenUsage(value.info?.total_token_usage);
+    if (totalTokenUsage) payload.info = { total_token_usage: totalTokenUsage };
+  }
+
+  return Object.keys(payload).length ? payload : null;
+}
+
+function compactRow(value, sourceLine) {
+  if (!value || typeof value !== "object" || Array.isArray(value)) return null;
+  const row = { __sourceLine: sourceLine };
+  const timestamp = shortString(value.timestamp, 128);
+  const type = shortString(value.type, 128);
+  if (timestamp !== null) row.timestamp = timestamp;
+  if (type !== null) row.type = type;
+  const payload = compactPayload(type, value.payload);
+  if (payload) row.payload = payload;
+  return row;
+}
+
+function formatByteSize(value) {
+  const bytes = Math.max(0, Number(value || 0));
+  if (bytes >= 1024 * 1024) return `${(bytes / (1024 * 1024)).toFixed(1)} MiB`;
+  if (bytes >= 1024) return `${(bytes / 1024).toFixed(1)} KiB`;
+  return `${bytes} B`;
+}
+
+function appendTail(current, chunk) {
+  if (chunk.length >= TAIL_HASH_BYTES) return Buffer.from(chunk.subarray(chunk.length - TAIL_HASH_BYTES));
+  const combined = Buffer.concat([current, chunk]);
+  return combined.length > TAIL_HASH_BYTES
+    ? Buffer.from(combined.subarray(combined.length - TAIL_HASH_BYTES))
+    : combined;
+}
+
+function sessionReadError(file, error) {
+  return new Error(
+    `无法读取 Codex 会话：${file.filePath}（${formatByteSize(file.size)}）：${error.message}`,
+    { cause: error },
+  );
+}
+
+function readJsonl(file) {
   const rows = [];
-  const source = fs.readFileSync(filePath, "utf8");
-  for (const [lineIndex, line] of source.split("\n").entries()) {
-    if (!line.trim()) continue;
-    try {
-      rows.push({ ...JSON.parse(line), __sourceLine: lineIndex + 1 });
-    } catch {
-      console.warn(`跳过无法解析的记录：${path.basename(filePath)}:${lineIndex + 1}`);
+  const buffer = Buffer.allocUnsafe(READ_CHUNK_BYTES);
+  const lineParts = [];
+  let lineBytes = 0;
+  let lineIndex = 0;
+  let offset = 0;
+  let tail = Buffer.alloc(0);
+  let skippingOversizedLine = false;
+  let fileDescriptor = null;
+
+  const resetLine = () => {
+    lineParts.length = 0;
+    lineBytes = 0;
+    skippingOversizedLine = false;
+  };
+
+  const appendLinePart = (part) => {
+    lineBytes += part.length;
+    if (skippingOversizedLine) return;
+    if (lineBytes > MAX_JSONL_ROW_BYTES) {
+      skippingOversizedLine = true;
+      lineParts.length = 0;
+      return;
     }
+    if (part.length) lineParts.push(Buffer.from(part));
+  };
+
+  const finishLine = (finalPart = Buffer.alloc(0)) => {
+    lineIndex += 1;
+    const totalBytes = lineBytes + finalPart.length;
+    if (skippingOversizedLine || totalBytes > MAX_JSONL_ROW_BYTES) {
+      console.warn(
+        `跳过过大的 Codex 记录：${file.filePath}:${lineIndex}（${formatByteSize(totalBytes)}）`,
+      );
+      resetLine();
+      return;
+    }
+
+    const lineBuffer = lineParts.length
+      ? Buffer.concat([...lineParts, finalPart], totalBytes)
+      : finalPart;
+    const line = lineBuffer.toString("utf8").replace(/\r$/, "");
+    if (line.trim()) {
+      try {
+        const compact = compactRow(JSON.parse(line), lineIndex);
+        if (compact) rows.push(compact);
+      } catch {
+        console.warn(`跳过无法解析的记录：${file.filePath}:${lineIndex}`);
+      }
+    }
+    resetLine();
+  };
+
+  try {
+    fileDescriptor = fs.openSync(file.filePath, "r");
+    const snapshotBytes = Math.max(0, Number(file.size || 0));
+    while (offset < snapshotBytes) {
+      const requestedBytes = Math.min(buffer.length, snapshotBytes - offset);
+      const bytesRead = fs.readSync(fileDescriptor, buffer, 0, requestedBytes, offset);
+      if (!bytesRead) break;
+      const chunk = buffer.subarray(0, bytesRead);
+      tail = appendTail(tail, chunk);
+      offset += bytesRead;
+
+      let start = 0;
+      let newlineIndex = chunk.indexOf(10, start);
+      while (newlineIndex !== -1) {
+        finishLine(chunk.subarray(start, newlineIndex));
+        start = newlineIndex + 1;
+        newlineIndex = chunk.indexOf(10, start);
+      }
+      appendLinePart(chunk.subarray(start));
+    }
+    if (lineBytes || skippingOversizedLine) finishLine();
+  } catch (error) {
+    throw sessionReadError(file, error);
+  } finally {
+    if (fileDescriptor !== null) fs.closeSync(fileDescriptor);
   }
-  return { rows, source };
+
+  return {
+    rows,
+    byteLength: offset,
+    tailHash: crypto.createHash("sha256").update(tail).digest("hex"),
+  };
 }
 
 function sessionFromFile(file) {
-  const { rows, source } = readJsonl(file.filePath);
+  const { rows, byteLength, tailHash } = readJsonl(file);
   const meta = rows.find((row) => row.type === "session_meta")?.payload || {};
   const metadataSessionId = meta.session_id || meta.id || null;
   const timestamps = rows.map(rowDate).filter(Boolean).sort((left, right) => left - right);
   const fallbackDate = new Date(file.modifiedAt);
-  const sourceBuffer = Buffer.from(source, "utf8");
   return {
     rows,
     filePath: file.filePath,
@@ -45,8 +212,8 @@ function sessionFromFile(file) {
     sourceRevision: {
       kind: "append-only-jsonl-v1",
       row_count: rows.length,
-      byte_length: sourceBuffer.byteLength,
-      tail_hash: crypto.createHash("sha256").update(sourceBuffer.subarray(-4096)).digest("hex"),
+      byte_length: byteLength,
+      tail_hash: tailHash,
     },
     startAt: timestamps[0] || fallbackDate,
     endAt: timestamps.at(-1) || fallbackDate,
@@ -57,7 +224,14 @@ function codexSessionFiles() {
   const codexHome = process.env.CODEX_HOME || path.join(os.homedir(), ".codex");
   const sessionsDirectory = path.join(codexHome, "sessions");
   const files = walkJsonlFiles(sessionsDirectory)
-    .map((filePath) => ({ filePath, modifiedAt: fs.statSync(filePath).mtimeMs }))
+    .map((filePath) => {
+      try {
+        const stats = fs.statSync(filePath);
+        return { filePath, modifiedAt: stats.mtimeMs, size: stats.size };
+      } catch (error) {
+        throw sessionReadError({ filePath, size: 0 }, error);
+      }
+    })
     .sort((left, right) => right.modifiedAt - left.modifiedAt);
   if (!files.length) throw new Error(`没有在 ${sessionsDirectory} 找到 Codex 会话记录`);
   return files;