tuntu 0.1.0__py3-none-any.whl

tuntu/__init__.py

@@ -0,0 +1,3 @@
+"""吞吐 (tuntu) — 本机 Claude Code token 用量看板。"""
+
+__version__ = "0.1.0"

tuntu/__main__.py

@@ -0,0 +1,3 @@
+from .cli import main
+
+main()

tuntu/cli.py

@@ -0,0 +1,496 @@
+#!/usr/bin/env python3
+"""吞吐 (tuntu) — 本机 Claude Code token 用量看板。
+
+只用 Python 标准库。读取本机的 Claude Code 会话日志，在 localhost 上
+打开一个看板页面。数据不会离开这台电脑。
+
+    tuntu            # 扫描并打开看板
+    tuntu --demo     # 用虚构数据演示（适合截图）
+
+从源码目录运行时用 python3 -m tuntu。
+"""
+
+import argparse
+import collections
+import datetime as dt
+import glob
+import json
+import os
+import random
+import re
+import sys
+import threading
+import webbrowser
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from urllib.parse import parse_qs, urlparse
+
+from . import __version__ as VERSION
+
+HERE = os.path.dirname(os.path.abspath(__file__))
+CTX_BUCKETS = 96          # 会话详情图最多画多少根柱子
+SCRATCH_KEY = "\x00scratch"
+
+# 一条请求记录的字段下标
+T, CWD, MODEL, SIDE, IN, OUT, CW, CR, H1, EFFORT = range(10)
+
+
+# ---------------------------------------------------------------- 读取日志
+
+def find_roots(custom=None):
+    """返回存在的日志根目录列表。"""
+    if custom:
+        cands = [custom]
+    else:
+        cands = []
+        env = os.environ.get("CLAUDE_CONFIG_DIR")
+        if env:
+            cands += [os.path.join(p.strip(), "projects") for p in env.split(",") if p.strip()]
+        cands += ["~/.claude/projects", "~/.config/claude/projects"]
+    roots = []
+    for c in cands:
+        p = os.path.abspath(os.path.expanduser(c))
+        if os.path.isdir(p) and p not in roots:
+            roots.append(p)
+    return roots
+
+
+def tidy_home(path):
+    """把家目录换成 ~，页面上不显示用户名。"""
+    home = os.path.expanduser("~")
+    if path == home or path.startswith(home + os.sep):
+        return "~" + path[len(home):]
+    return path
+
+
+def project_key(cwd):
+    """把工作目录归并成项目：去掉 worktree 后缀，临时会话合并成一项。"""
+    if not cwd:
+        return "?"
+    p = cwd.replace("\\", "/").rstrip("/")
+    if "/scratch-workspaces/" in p:
+        return SCRATCH_KEY
+    for marker in ("/.claude/worktrees/", "/.claude-worktrees/"):
+        i = p.find(marker)
+        if i > 0:
+            p = p[:i]
+    return p or "/"
+
+
+def encode_path(path):
+    """Claude Code 给项目目录起名的方式：非字母数字的字符都换成 -。"""
+    return re.sub(r"[^A-Za-z0-9]", "-", path)
+
+
+def resolve_dirs(cwd_seen):
+    """把日志目录名还原成项目路径。
+
+    会话中途 cd 到子目录时 cwd 会变，所以项目按日志目录归，而不是按每条
+    记录的 cwd 归；目录名里的 - 有歧义，用记录里出现过的 cwd 反查。
+    """
+    resolved = {}
+    for dirname, cwds in cwd_seen.items():
+        for c in cwds:
+            if encode_path(c) == dirname:
+                resolved[dirname] = c
+                break
+    by_enc = {encode_path(project_key(c)): project_key(c) for c in resolved.values()}
+    home_enc = encode_path(os.path.expanduser("~"))
+    out = {}
+    for dirname in cwd_seen:
+        if dirname in resolved:
+            out[dirname] = project_key(resolved[dirname])
+            continue
+        parent = re.split(r"--claude-worktrees-|-claude-worktrees-", dirname)[0]
+        if parent in by_enc:
+            out[dirname] = by_enc[parent]
+        elif "-scratch-workspaces-" in dirname:
+            out[dirname] = SCRATCH_KEY
+        else:
+            name = parent[len(home_enc) + 1:] if parent.startswith(home_enc + "-") else parent
+            out[dirname] = "/" + (name or parent)
+    return out
+
+
+def parse_file(path, force_side, recs, tools, proj, cwds):
+    """解析一个 jsonl 文件，把请求写进 recs，把工具调用写进 tools。"""
+    try:
+        fh = open(path, "r", encoding="utf-8", errors="ignore")
+    except OSError:
+        return
+    with fh:
+        for line in fh:
+            if '"usage"' not in line:
+                continue
+            try:
+                d = json.loads(line)
+            except ValueError:
+                continue
+            m = d.get("message")
+            if not isinstance(m, dict):
+                continue
+            u = m.get("usage")
+            if not isinstance(u, dict):
+                continue
+            model = m.get("model") or "unknown"
+            if model == "<synthetic>":
+                continue
+            ts = d.get("timestamp")
+            if not isinstance(ts, str):
+                continue
+            try:
+                t = dt.datetime.fromisoformat(ts.replace("Z", "+00:00")).astimezone()
+            except ValueError:
+                continue
+            cwd = d.get("cwd")
+            if isinstance(cwd, str) and len(cwds) < 500:
+                cwds.add(cwd)
+            key = (m.get("id") or d.get("uuid"), d.get("requestId"))
+            cc = u.get("cache_creation")
+            h1 = cc.get("ephemeral_1h_input_tokens") if isinstance(cc, dict) else 0
+            rec = [
+                t,
+                proj,
+                model,
+                bool(force_side or d.get("isSidechain")),
+                u.get("input_tokens") or 0,
+                u.get("output_tokens") or 0,
+                u.get("cache_creation_input_tokens") or 0,
+                u.get("cache_read_input_tokens") or 0,
+                h1 or 0,
+                d.get("perTurnEffort") or d.get("effort") or None,
+            ]
+            old = recs.get(key)
+            if old is not None:
+                # 同一条消息会按内容块写多行，用量取最大的那次
+                for i in (IN, OUT, CW, CR, H1):
+                    rec[i] = max(rec[i], old[i])
+                rec[T] = min(rec[T], old[T])
+            recs[key] = rec
+            content = m.get("content")
+            if isinstance(content, list):
+                for c in content:
+                    if isinstance(c, dict) and c.get("type") == "tool_use" and c.get("id"):
+                        name = str(c.get("name") or "?")
+                        if name.startswith("mcp__"):
+                            name = name.split("__")[-1] or name
+                        tools[c["id"]] = (t.strftime("%Y-%m-%d"), name)
+
+
+def collect(roots):
+    """扫描所有日志，按会话归组。返回 (sessions, 文件数, 目录名到项目的映射)。"""
+    sessions = {}
+    cwd_seen = {}
+    nfiles = 0
+    for root in roots:
+        for path in sorted(glob.glob(os.path.join(root, "**", "*.jsonl"), recursive=True)):
+            rel = os.path.relpath(path, root).split(os.sep)
+            if len(rel) < 2:
+                continue
+            nfiles += 1
+            is_main = len(rel) == 2
+            sid = rel[1][:-len(".jsonl")] if is_main else rel[1]
+            s = sessions.setdefault(sid, {"recs": {}, "tools": {}})
+            parse_file(path, not is_main, s["recs"], s["tools"], rel[0], cwd_seen.setdefault(rel[0], set()))
+    return sessions, nfiles, resolve_dirs(cwd_seen)
+
+
+# ---------------------------------------------------------------- 聚合
+
+def aggregate(sessions, meta, dirmap=None):
+    """把会话记录聚合成页面要用的数据。返回 (data, details)。"""
+    seen = set()
+    tool_seen = set()
+    buckets = collections.defaultdict(lambda: [0, 0, 0, 0, 0, 0])
+    tools = collections.Counter()
+    proj_ids, model_ids = {}, {}
+    sess_out, details = [], {}
+
+    dirmap = dirmap or {}
+
+    def pid(key):
+        key = dirmap.get(key, key)
+        return proj_ids.setdefault(key, len(proj_ids))
+
+    def mid(key):
+        return model_ids.setdefault(key, len(model_ids))
+
+    for sid in sorted(sessions):
+        s = sessions[sid]
+        recs = s["recs"]
+        if not recs:
+            continue
+        tok = [0, 0, 0, 0]
+        sub_req = sub_tok = 0
+        by_proj, by_model = collections.Counter(), collections.Counter()
+        main = []
+        for key, r in recs.items():
+            total = r[IN] + r[OUT] + r[CW] + r[CR]
+            for i, f in enumerate((IN, OUT, CW, CR)):
+                tok[i] += r[f]
+            by_proj[r[CWD]] += total
+            by_model[r[MODEL]] += total
+            if r[SIDE]:
+                sub_req += 1
+                sub_tok += total
+            else:
+                main.append(r)
+            # 续接或分叉的会话会带着前半段记录；总量只算一次
+            if key in seen:
+                continue
+            seen.add(key)
+            b = buckets[(r[T].strftime("%Y-%m-%d"), r[T].hour, pid(r[CWD]), mid(r[MODEL]),
+                         1 if r[SIDE] else 0, r[EFFORT])]
+            b[0] += r[IN]
+            b[1] += r[OUT]
+            b[2] += r[CW]
+            b[3] += r[CR]
+            b[4] += r[H1]
+            b[5] += 1
+        for tid, (date, name) in s["tools"].items():
+            if tid in tool_seen:
+                continue
+            tool_seen.add(tid)
+            tools[(date, name)] += 1
+
+        main.sort(key=lambda r: r[T])
+        ctx = [r[IN] + r[CW] + r[CR] for r in main]
+        drops = [i for i in range(1, len(ctx)) if ctx[i - 1] > 20000 and ctx[i] < ctx[i - 1] * 0.5]
+        times = [r[T] for r in recs.values()]
+        sess_out.append({
+            "id": sid,
+            "p": pid(by_proj.most_common(1)[0][0]),
+            "m": mid(by_model.most_common(1)[0][0]),
+            "start": min(times).strftime("%Y-%m-%dT%H:%M"),
+            "end": max(times).strftime("%Y-%m-%dT%H:%M"),
+            "tok": tok,
+            "req": len(recs),
+            "subReq": sub_req,
+            "subTok": sub_tok,
+            "peak": max(ctx) if ctx else 0,
+            "drops": len(drops),
+        })
+        details[sid] = downsample(ctx, drops)
+
+    data = {
+        "meta": meta,
+        "projects": project_names(proj_ids),
+        "models": [k for k, _ in sorted(model_ids.items(), key=lambda kv: kv[1])],
+        "buckets": sorted([list(k) + v for k, v in buckets.items()], key=lambda b: (b[0], b[1])),
+        "tools": sorted([[d, n, c] for (d, n), c in tools.items()]),
+        "sessions": sess_out,
+    }
+    data["meta"]["sessions"] = len(sess_out)
+    return data, details
+
+
+def downsample(ctx, drops):
+    """把逐次请求的上下文长度压成最多 CTX_BUCKETS 根柱子（取每段最大值）。"""
+    n = len(ctx)
+    if n == 0:
+        return {"n": 0, "per": 1, "ctx": [], "drop": []}
+    per = max(1, -(-n // CTX_BUCKETS))
+    dropset = set(drops)
+    out, flags = [], []
+    for i in range(0, n, per):
+        out.append(max(ctx[i:i + per]))
+        flags.append(1 if any(j in dropset for j in range(i, min(n, i + per))) else 0)
+    return {"n": n, "per": per, "ctx": out, "drop": flags}
+
+
+def project_names(proj_ids):
+    """给每个项目起显示名：只用目录名，重名时带上上一级目录。"""
+    keys = [k for k, _ in sorted(proj_ids.items(), key=lambda kv: kv[1])]
+    base = collections.Counter(os.path.basename(k) for k in keys if k != SCRATCH_KEY)
+    home = os.path.expanduser("~").replace("\\", "/")
+    names = []
+    for k in keys:
+        if k == SCRATCH_KEY:
+            names.append("临时会话")
+        elif k == home:
+            names.append("~")
+        else:
+            b = os.path.basename(k) or k
+            if base[b] > 1:
+                b = os.path.basename(os.path.dirname(k)) + "/" + b
+            names.append(b)
+    return names
+
+
+# ---------------------------------------------------------------- 演示数据
+
+def demo_sessions(seed=7, days=45):
+    """生成一份虚构的用量记录，结构和真实日志解析后一样。"""
+    rnd = random.Random(seed)
+    now = dt.datetime.now().astimezone().replace(minute=0, second=0, microsecond=0)
+    projects = [("/demo/moon-base", 55), ("/demo/noodle-shop", 25), ("/demo/side-quest", 14), ("/demo/dotfiles", 6)]
+    models = [("claude-opus-5-5", 80), ("claude-sonnet-5-5", 16), ("claude-haiku-5-5", 4)]
+    efforts = [("high", 50), ("medium", 40), ("low", 10)]
+    tool_names = [("Bash", 50), ("Read", 18), ("Edit", 14), ("Grep", 8), ("Write", 5), ("Glob", 3), ("WebFetch", 2)]
+    hour_w = [3, 2, 1, 0, 0, 0, 0, 1, 3, 6, 8, 8, 5, 6, 9, 10, 9, 7, 5, 6, 9, 10, 8, 5]
+
+    def pick(pairs):
+        return rnd.choices([p[0] for p in pairs], weights=[p[1] for p in pairs])[0]
+
+    sessions = {}
+    n = 0
+    for back in range(days - 1, -1, -1):
+        if back and rnd.random() < 0.12:
+            continue
+        day = (now - dt.timedelta(days=back)).replace(hour=0)
+        for _ in range(rnd.randint(1, 4)):
+            start = day + dt.timedelta(hours=rnd.choices(range(24), weights=hour_w)[0], minutes=rnd.randint(0, 59))
+            if start > now:
+                continue
+            sid = "demo-%04d" % len(sessions)
+            cwd, model = pick(projects), pick(models)
+            recs, tools = {}, {}
+            t, ctx = start, rnd.randint(15000, 30000)
+            for i in range(int(rnd.lognormvariate(4.3, 0.9)) + 5):
+                t += dt.timedelta(seconds=rnd.randint(15, 140))
+                if t > now:
+                    break
+                n += 1
+                side = rnd.random() < 0.07
+                grow = rnd.randint(300, 4000)
+                if side:
+                    cr, cw = rnd.randint(8000, 30000), rnd.randint(500, 6000)
+                else:
+                    if ctx > rnd.randint(150000, 190000):
+                        ctx = rnd.randint(25000, 40000)      # 自动压缩
+                        cr, cw = 0, ctx
+                    else:
+                        cr, cw = ctx, grow
+                    ctx += grow
+                recs[("demo", n)] = [t, cwd, model, side, rnd.randint(1, 40), rnd.randint(40, 1800), cw, cr,
+                                     cw if rnd.random() < 0.85 else 0, pick(efforts)]
+                for k in range(rnd.choice([0, 1, 1, 1, 2])):
+                    tools[("demo", n, k)] = (t.strftime("%Y-%m-%d"), pick(tool_names))
+            if recs:
+                sessions[sid] = {"recs": recs, "tools": tools}
+    return sessions
+
+
+# ---------------------------------------------------------------- 服务
+
+class App:
+    def __init__(self, args):
+        self.args = args
+        self.lock = threading.Lock()
+        self.data = None
+        self.details = {}
+
+    def scan(self):
+        with self.lock:
+            now = dt.datetime.now().astimezone()
+            meta = {"version": VERSION, "generated": now.strftime("%Y-%m-%dT%H:%M"), "demo": bool(self.args.demo)}
+            if self.args.demo:
+                sessions, nfiles, dirmap = demo_sessions(), 0, None
+                meta["root"] = "演示数据"
+            else:
+                roots = find_roots(self.args.dir)
+                sessions, nfiles, dirmap = collect(roots)
+                meta["root"] = "、".join(tidy_home(r) for r in roots) or "~/.claude/projects"
+            meta["files"] = nfiles
+            self.data, self.details = aggregate(sessions, meta, dirmap)
+            return self.data
+
+
+def make_handler(app, page):
+    class Handler(BaseHTTPRequestHandler):
+        server_version = "tuntu/" + VERSION
+
+        def log_message(self, *a):
+            pass
+
+        def send(self, code, body, ctype):
+            self.send_response(code)
+            self.send_header("Content-Type", ctype)
+            self.send_header("Content-Length", str(len(body)))
+            self.send_header("Cache-Control", "no-store")
+            self.send_header("X-Content-Type-Options", "nosniff")
+            if ctype.startswith("text/html"):
+                self.send_header("Content-Security-Policy", page["csp"])
+            self.end_headers()
+            self.wfile.write(body)
+
+        def send_json(self, obj, code=200):
+            self.send(code, json.dumps(obj, ensure_ascii=False, separators=(",", ":")).encode("utf-8"),
+                      "application/json; charset=utf-8")
+
+        def do_GET(self):
+            # 只接受发往本机的请求，挡掉 DNS 重绑定
+            host = (self.headers.get("Host") or "").rsplit(":", 1)[0]
+            if host not in ("127.0.0.1", "localhost"):
+                return self.send(403, b"forbidden", "text/plain")
+            url = urlparse(self.path)
+            qs = parse_qs(url.query)
+            if url.path in ("/", "/index.html"):
+                return self.send(200, page["html"], "text/html; charset=utf-8")
+            if url.path == "/api/data":
+                if "rescan" in qs or app.data is None:
+                    app.scan()
+                return self.send_json(app.data)
+            if url.path == "/api/session":
+                detail = app.details.get((qs.get("id") or [""])[0])
+                if detail is None:
+                    return self.send_json({"error": "not found"}, 404)
+                return self.send_json(detail)
+            return self.send(404, b"not found", "text/plain")
+
+    return Handler
+
+
+def load_page(web_fonts):
+    with open(os.path.join(HERE, "web", "index.html"), "r", encoding="utf-8") as fh:
+        html = fh.read()
+    csp = "default-src 'none'; script-src 'unsafe-inline'; style-src 'unsafe-inline'; connect-src 'self'; img-src data:"
+    fonts = ""
+    if web_fonts:
+        fonts = ('<link rel="stylesheet" href="https://fonts.googleapis.com/css2?'
+                 'family=Bricolage+Grotesque:opsz,wght@12..96,500..800&family=ZCOOL+QingKe+HuangYou&display=swap">')
+        csp = csp.replace("style-src 'unsafe-inline'", "style-src 'unsafe-inline' https://fonts.googleapis.com")
+        csp += "; font-src https://fonts.gstatic.com"
+    return {"html": html.replace("<!--FONTS-->", fonts).encode("utf-8"), "csp": csp}
+
+
+def main():
+    ap = argparse.ArgumentParser(prog="tuntu", description="吞吐 (tuntu)：本机 Claude Code token 用量看板")
+    ap.add_argument("--port", type=int, default=8787, help="端口，默认 8787（被占用时自动往后找）")
+    ap.add_argument("--dir", help="日志目录，默认 ~/.claude/projects")
+    ap.add_argument("--demo", action="store_true", help="用虚构数据演示，不读取任何日志")
+    ap.add_argument("--no-open", action="store_true", help="不自动打开浏览器")
+    ap.add_argument("--web-fonts", action="store_true", help="从 Google Fonts 加载展示字体（会产生外网请求）")
+    ap.add_argument("--json", action="store_true", help="把聚合结果打印成 JSON 后退出")
+    ap.add_argument("--version", action="version", version="tuntu " + VERSION)
+    args = ap.parse_args()
+
+    app = App(args)
+    data = app.scan()
+    if args.json:
+        json.dump(data, sys.stdout, ensure_ascii=False)
+        return
+    if not args.demo and not data["buckets"]:
+        print("没有找到用量记录（%s）。可以用 --dir 指定目录，或用 --demo 看演示。" % data["meta"]["root"])
+
+    handler = make_handler(app, load_page(args.web_fonts))
+    server = None
+    for port in range(args.port, args.port + 20):
+        try:
+            server = ThreadingHTTPServer(("127.0.0.1", port), handler)
+            break
+        except OSError:
+            continue
+    if server is None:
+        sys.exit("端口 %d 到 %d 都被占用了，请用 --port 换一个。" % (args.port, args.port + 19))
+    url = "http://127.0.0.1:%d/" % server.server_address[1]
+    print("吞吐已启动：%s  （Ctrl+C 退出）" % url)
+    if not args.no_open:
+        threading.Timer(0.3, webbrowser.open, args=(url,)).start()
+    try:
+        server.serve_forever()
+    except KeyboardInterrupt:
+        print("\n再见。")
+
+
+if __name__ == "__main__":
+    main()

tuntu/web/index.html

@@ -0,0 +1,611 @@
+<!doctype html>
+<html lang="zh-CN">
+<head>
+<meta charset="utf-8">
+<meta name="viewport" content="width=device-width, initial-scale=1">
+<title>吞吐 · 我的 Token 都去哪了</title>
+<!--FONTS-->
+<style>
+:root {
+  --ink: #111318; --bg: #EEF0F7; --muted: #3C4150; --white: #FFFFFF;
+  --yellow: #FFD43B; --blue: #3550E8; --orange: #FF7A3D; --mint: #7CE3B1; --pink: #FFB3DD; --sky: #C3CEFF;
+  --num: "Bricolage Grotesque", ui-rounded, "SF Pro Rounded", system-ui, -apple-system, "Segoe UI", "PingFang SC", "Microsoft YaHei", sans-serif;
+  --cn: "ZCOOL QingKe HuangYou", "Yuanti SC", "YouYuan", "PingFang SC", "Microsoft YaHei", sans-serif;
+}
+* { box-sizing: border-box; }
+body { margin: 0; background: var(--bg); color: var(--ink); font-family: var(--num); font-variant-numeric: tabular-nums; }
+a { color: inherit; }
+button { font: inherit; color: inherit; cursor: pointer; }
+.wrap { max-width: 1320px; margin: 0 auto; padding: 36px 24px 56px; display: flex; flex-direction: column; gap: 24px; }
+.row { display: flex; flex-wrap: wrap; gap: 24px; }
+.tile { min-width: 0; background: var(--white); border: 2.5px solid var(--ink); border-radius: 22px; box-shadow: 6px 6px 0 var(--ink); padding: 28px; display: flex; flex-direction: column; gap: 14px; }
+.h { font-size: 20px; font-weight: 700; }
+.sub { font-size: 14px; font-weight: 500; color: var(--muted); }
+.cap { font-size: 15px; font-weight: 500; line-height: 1.5; }
+.head { display: flex; flex-wrap: wrap; align-items: baseline; justify-content: space-between; gap: 8px; }
+.cn { font-family: var(--cn); font-weight: 700; }
+.scroll { overflow-x: auto; }
+
+header.top { display: flex; flex-wrap: wrap; align-items: center; justify-content: space-between; gap: 16px; }
+.brand { display: flex; align-items: baseline; flex-wrap: wrap; gap: 14px; }
+.brand .logo { font-size: 60px; line-height: 1; }
+.tag { font-size: 17px; font-weight: 700; background: var(--ink); color: var(--white); padding: 6px 12px; border-radius: 999px; transform: rotate(-2deg); }
+.tag.demo { background: var(--orange); color: var(--ink); border: 2px solid var(--ink); transform: rotate(2deg); }
+.controls, .pills { display: flex; flex-wrap: wrap; align-items: center; gap: 8px; }
+.pill { min-height: 44px; padding: 0 18px; border: 2.5px solid var(--ink); border-radius: 999px; background: var(--white); font-size: 15px; font-weight: 700; display: inline-flex; align-items: center; gap: 8px; text-decoration: none; }
+.pill[aria-pressed="true"] { background: var(--ink); color: var(--white); }
+.pill.small { padding: 0 14px; border-width: 2px; font-size: 14px; }
+.pill.go { background: var(--mint); box-shadow: 3px 3px 0 var(--ink); }
+.pill.back { box-shadow: 3px 3px 0 var(--ink); }
+.pill:focus-visible, a.srow:focus-visible { outline: 3px solid var(--blue); outline-offset: 3px; }
+
+.hero { flex: 3 1 520px; background: var(--yellow); padding: 28px 32px; gap: 10px; }
+.hero .line { display: flex; align-items: baseline; flex-wrap: wrap; gap: 14px; }
+.hero .n { font-size: clamp(84px, 12vw, 172px); font-weight: 800; line-height: 0.9; letter-spacing: -0.04em; }
+.hero .u { font-size: 72px; line-height: 1; }
+.hero .t { font-size: 22px; font-weight: 700; }
+.chips { display: flex; flex-wrap: wrap; gap: 10px; margin-top: 10px; }
+.chip { font-size: 15px; font-weight: 700; background: var(--white); border: 2px solid var(--ink); border-radius: 999px; padding: 7px 14px; }
+.chip.s { font-size: 14px; padding: 6px 14px; }
+.side { flex: 2 1 380px; padding: 28px 32px; gap: 12px; }
+.side .n { font-size: 84px; font-weight: 800; line-height: 0.95; letter-spacing: -0.03em; }
+.ratio { background: var(--blue); color: var(--white); }
+.dots { display: flex; flex-wrap: wrap; gap: 3px; }
+.dots i { width: 7px; height: 7px; border-radius: 2px; background: var(--sky); }
+.dots i.w { background: var(--orange); }
+
+.hit { flex: 1 1 250px; background: var(--mint); }
+.hit .n { font-size: 64px; font-weight: 800; line-height: 0.95; letter-spacing: -0.03em; }
+.waffle { display: grid; grid-template-columns: repeat(10, minmax(0, 1fr)); gap: 4px; max-width: 216px; }
+.waffle i { aspect-ratio: 1 / 1; border: 2px solid var(--ink); border-radius: 4px; }
+.waffle i.on { background: var(--ink); }
+.daily { flex: 3 1 460px; }
+.bars { display: flex; align-items: flex-end; height: 236px; }
+.bars .col { flex: 1 1 0; min-width: 0; display: flex; flex-direction: column; align-items: center; gap: 4px; }
+.bars .tg { font-size: 12px; font-weight: 800; white-space: nowrap; height: 16px; }
+.bars .b { width: 100%; background: var(--blue); border: 2px solid var(--ink); border-radius: 6px 6px 0 0; }
+.bars .b.pk { background: var(--orange); }
+.bars .b.z { border: 0; background: transparent; }
+.xl { display: flex; }
+.xl span { flex: 1 1 0; min-width: 0; text-align: center; font-size: 11px; font-weight: 500; color: var(--muted); white-space: nowrap; }
+.clockt { flex: 1 1 280px; background: var(--pink); align-items: center; gap: 12px; }
+.clock { position: relative; width: 240px; height: 240px; }
+.clock .hb { position: absolute; left: 50%; top: 50%; width: 12px; margin-left: -6px; background: var(--ink); border: 2px solid var(--ink); border-radius: 0 0 6px 6px; transform-origin: 50% 0; }
+.clock .hb.pk { background: var(--yellow); }
+.clock .mid { position: absolute; left: 50%; top: 50%; width: 64px; height: 64px; margin: -32px 0 0 -32px; border-radius: 50%; background: var(--ink); color: var(--white); display: flex; align-items: center; justify-content: center; font-size: 18px; font-weight: 800; }
+.clock .l { position: absolute; font-size: 12px; font-weight: 700; }
+
+.badges { display: flex; flex-wrap: wrap; gap: 20px; }
+.badge { flex: 1 1 190px; min-width: 0; border: 2.5px solid var(--ink); border-radius: 18px; box-shadow: 4px 4px 0 var(--ink); padding: 16px 18px; display: flex; flex-direction: column; gap: 4px; }
+.badge .k { font-size: 12px; font-weight: 700; letter-spacing: 0.08em; }
+.badge .ti { font-size: 28px; line-height: 1.15; }
+.badge .li { font-size: 14px; font-weight: 500; }
+.badge.locked { background: transparent !important; border-style: dashed; box-shadow: none; color: var(--muted); transform: none !important; }
+
+.stack { display: flex; height: 34px; border: 2.5px solid var(--ink); border-radius: 10px; overflow: hidden; background: var(--bg); }
+.legend { display: grid; grid-template-columns: repeat(auto-fit, minmax(min(200px, 100%), 1fr)); gap: 14px; }
+.legend .it { display: flex; gap: 10px; align-items: flex-start; }
+.sw { width: 16px; height: 16px; border: 2px solid var(--ink); border-radius: 5px; margin-top: 2px; flex: none; }
+.legend b { display: block; font-size: 15px; }
+.legend span { font-size: 13px; font-weight: 500; color: var(--muted); }
+.erow { display: flex; align-items: center; gap: 12px; font-size: 15px; font-weight: 700; }
+.erow .nm { flex: 1 1 auto; }
+.erow .dt { width: 160px; flex: none; text-align: right; font-size: 13px; font-weight: 500; color: var(--muted); }
+
+.rank { display: flex; flex-direction: column; gap: 8px; }
+.rank .top { display: flex; justify-content: space-between; align-items: baseline; gap: 12px; }
+.rank .nm { font-size: 17px; font-weight: 800; min-width: 0; overflow-wrap: anywhere; }
+.rank .vv { font-size: 15px; font-weight: 700; flex: none; }
+.meter { height: 16px; border: 2px solid var(--ink); border-radius: 8px; overflow: hidden; background: var(--bg); }
+.meter i { display: block; height: 100%; border-right: 2px solid var(--ink); }
+.meter.thin { height: 14px; border-radius: 7px; }
+.meter.thin i { border-right: 0; }
+
+.heat { min-width: 560px; display: flex; flex-direction: column; gap: 4px; }
+.heat .hr { display: flex; align-items: center; gap: 4px; }
+.heat .wl { width: 36px; flex: none; font-size: 13px; font-weight: 700; }
+.heat .c { flex: 1 1 0; height: 26px; border-radius: 5px; background: var(--bg); }
+.heat .hl { flex: 1 1 0; font-size: 11px; font-weight: 500; color: var(--muted); text-align: center; }
+.trow { display: flex; align-items: center; gap: 12px; font-size: 14px; font-weight: 700; }
+.trow .nm { width: 118px; flex: none; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
+.trow .meter { flex: 1 1 auto; }
+.trow .ct { width: 60px; flex: none; text-align: right; }
+
+.slist { min-width: 860px; display: flex; flex-direction: column; gap: 10px; padding: 0 6px 6px 0; }
+a.srow { display: grid; grid-template-columns: 44px 1.1fr 1.7fr 0.8fr 2fr 0.9fr 0.8fr 0.7fr; gap: 14px; align-items: center; min-height: 56px; padding: 0 16px; border: 2px solid var(--ink); border-radius: 14px; box-shadow: 3px 3px 0 var(--ink); background: var(--white); text-decoration: none; font-size: 15px; font-weight: 700; }
+a.srow:hover { background: #F7F8FC; transform: translate(-1px, -1px); box-shadow: 4px 4px 0 var(--ink); }
+.medal { width: 36px; height: 36px; border: 2px solid var(--ink); border-radius: 50%; display: flex; align-items: center; justify-content: center; font-weight: 800; }
+.r { text-align: right; }
+.lt { font-weight: 500; }
+.ell { min-width: 0; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
+
+.kpis { display: grid; grid-template-columns: repeat(auto-fit, minmax(min(220px, 100%), 1fr)); gap: 20px; }
+.kpi { border: 2.5px solid var(--ink); border-radius: 18px; box-shadow: 4px 4px 0 var(--ink); padding: 18px 22px; display: flex; flex-direction: column; gap: 4px; background: var(--white); }
+.kpi .k { font-size: 15px; font-weight: 700; }
+.kpi .v { font-size: 40px; font-weight: 800; letter-spacing: -0.02em; line-height: 1.05; }
+.kpi .s { font-size: 13px; font-weight: 500; }
+.ctxwrap { min-width: 760px; display: flex; gap: 10px; }
+.yax { display: flex; flex-direction: column; justify-content: space-between; height: 276px; padding-top: 20px; font-size: 12px; font-weight: 700; text-align: right; width: 52px; flex: none; }
+.ctx { display: flex; align-items: flex-end; gap: 3px; height: 276px; border-bottom: 2.5px solid var(--ink); }
+.ctx .b { border-bottom: 0; border-radius: 5px 5px 0 0; }
+.ctx .tg { font-size: 11px; }
+.duo { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 14px; }
+.duo > div { border: 2px solid var(--ink); border-radius: 14px; padding: 16px; display: flex; flex-direction: column; gap: 4px; }
+.duo .k { font-size: 14px; font-weight: 700; }
+.duo .v { font-size: 30px; font-weight: 800; line-height: 1.1; }
+.duo .s { font-size: 13px; font-weight: 500; }
+.tip { margin-top: auto; align-self: flex-start; background: var(--yellow); border: 2.5px solid var(--ink); border-radius: 16px; box-shadow: 4px 4px 0 var(--ink); padding: 14px 18px; font-size: 15px; font-weight: 700; line-height: 1.5; transform: rotate(-1.5deg); }
+
+footer { display: flex; flex-wrap: wrap; justify-content: space-between; gap: 8px; font-size: 13px; font-weight: 500; color: var(--muted); }
+.empty { align-items: flex-start; }
+@media (max-width: 560px) {
+  .wrap { padding: 24px 16px 40px; }
+  .tile { padding: 20px; box-shadow: 4px 4px 0 var(--ink); }
+  .brand .logo { font-size: 48px; }
+  .side .n { font-size: 64px; }
+  .hero .u { font-size: 52px; }
+}
+</style>
+</head>
+<body>
+<div class="wrap" id="app"><div class="tile"><div class="h">正在扫描日志…</div></div></div>
+<script>
+'use strict';
+// bucket 行的字段下标：日期、小时、项目、模型、是否子代理、推理强度、新输入、输出、缓存写入、缓存读取、其中 1 小时档写入、请求数
+const D = 0, HOUR = 1, P = 2, M = 3, SIDE = 4, EFF = 5, IN = 6, OUT = 7, CW = 8, CR = 9, H1 = 10, REQ = 11;
+const WK = ['周一', '周二', '周三', '周四', '周五', '周六', '周日'];
+const COLORS = { cr: 'var(--blue)', cw: 'var(--sky)', out: 'var(--orange)', inp: 'var(--ink)' };
+const app = document.getElementById('app');
+const state = { range: 'all', metric: 'total', busy: false };
+let DATA = null;
+
+const esc = (s) => String(s).replace(/[&<>"']/g, (c) => ({ '&': '&amp;', '<': '&lt;', '>': '&gt;', '"': '&quot;', "'": '&#39;' }[c]));
+const num = (n) => Math.round(n).toLocaleString('zh-CN');
+const fmt = (n) => n >= 1e8 ? (n / 1e8).toFixed(2) + ' 亿' : n >= 1e4 ? (n / 1e4).toFixed(n >= 1e6 ? 0 : 1) + ' 万' : num(n);
+const pct = (a, b, d = 2) => b ? (100 * a / b).toFixed(d) + '%' : '—';
+const pad = (n) => String(n).padStart(2, '0');
+const dstr = (d) => d.getFullYear() + '-' + pad(d.getMonth() + 1) + '-' + pad(d.getDate());
+const parseD = (s) => new Date(+s.slice(0, 4), +s.slice(5, 7) - 1, +s.slice(8, 10));
+const addDays = (d, n) => new Date(d.getFullYear(), d.getMonth(), d.getDate() + n);
+const weekday = (s) => (parseD(s).getDay() + 6) % 7;
+const md = (s) => s.slice(5, 10);
+const sum4 = (a) => a[0] + a[1] + a[2] + a[3];
+const hitOf = (a) => 100 * a[3] / ((a[0] + a[2] + a[3]) || 1);
+
+function modelName(id) {
+  const parts = String(id).replace(/^claude-/, '').replace(/-\d{8}$/, '').split('-');
+  const out = [];
+  for (const p of parts) {
+    if (/^\d+$/.test(p) && out.length && /^[\d.]+$/.test(out[out.length - 1])) out[out.length - 1] += '.' + p;
+    else out.push(/^\d+$/.test(p) ? p : p.charAt(0).toUpperCase() + p.slice(1));
+  }
+  return out.join(' ');
+}
+function spanText(a, b) {
+  const mins = Math.max(0, (new Date(b) - new Date(a)) / 60000);
+  if (mins < 60) return [Math.max(1, Math.round(mins)), '分钟'];
+  if (mins < 72 * 60) return [Math.round(mins / 60), '小时'];
+  return [Math.round(mins / 1440), '天'];
+}
+const when = (s) => s.slice(5, 10) + ' ' + s.slice(11, 16);
+
+// ---------------------------------------------------------------- 计算
+
+function compute(range) {
+  const today = dstr(new Date());
+  const back = { '30': 29, '7': 6, '1': 0 }[range];
+  const cutoff = range === 'all' ? null : dstr(addDays(new Date(), -back));
+  const rows = cutoff ? DATA.buckets.filter((b) => b[D] >= cutoff) : DATA.buckets;
+  const S = [0, 0, 0, 0, 0];            // 新输入、输出、缓存写入、缓存读取、请求
+  let h1 = 0, sideTok = 0, sideReq = 0;
+  const perDay = new Map(), byM = new Map(), byP = new Map(), byE = new Map();
+  const hours = Array(24).fill(0);
+  const heat = WK.map(() => Array(24).fill(0));
+  const add = (map, key, b) => {
+    let a = map.get(key);
+    if (!a) map.set(key, a = [0, 0, 0, 0, 0]);
+    a[0] += b[IN]; a[1] += b[OUT]; a[2] += b[CW]; a[3] += b[CR]; a[4] += b[REQ];
+  };
+  for (const b of rows) {
+    const t = b[IN] + b[OUT] + b[CW] + b[CR];
+    S[0] += b[IN]; S[1] += b[OUT]; S[2] += b[CW]; S[3] += b[CR]; S[4] += b[REQ];
+    h1 += b[H1];
+    if (b[SIDE]) { sideTok += t; sideReq += b[REQ]; }
+    add(perDay, b[D], b); add(byM, b[M], b); add(byP, b[P], b); add(byE, b[EFF], b);
+    hours[b[HOUR]] += t;
+    heat[weekday(b[D])][b[HOUR]] += t;
+  }
+  const total = sum4(S);
+  const first = cutoff || (rows.length ? rows[0][D] : today);
+  const last = range === 'all' && rows.length ? rows[rows.length - 1][D] : today;
+  const days = [];
+  for (let d = parseD(first); dstr(d) <= last; d = addDays(d, 1)) days.push(dstr(d));
+  const tools = new Map();
+  for (const [date, name, c] of DATA.tools) if (!cutoff || date >= cutoff) tools.set(name, (tools.get(name) || 0) + c);
+  const sessions = DATA.sessions.filter((s) => !cutoff || s.end.slice(0, 10) >= cutoff)
+    .sort((a, b) => sum4(b.tok) - sum4(a.tok));
+  return { range, cutoff, S, total, h1, sideTok, sideReq, perDay, byM, byP, byE, hours, heat, days, first, last,
+    tools: [...tools.entries()].sort((a, b) => b[1] - a[1]), sessions };
+}
+
+function badges() {
+  // 成就按全部记录算，不跟时间范围走
+  const all = compute('all');
+  const peakH = all.hours.indexOf(Math.max(...all.hours));
+  const habit = (peakH >= 20 || peakH <= 3) ? '夜猫子' : peakH <= 8 ? '早鸟' : peakH <= 12 ? '上午战士' : '下午茶选手';
+  const hit = hitOf(all.S);
+  let longest = null;
+  for (const s of DATA.sessions) {
+    const mins = (new Date(s.end) - new Date(s.start)) / 60000;
+    if (!longest || mins > longest.mins) longest = { mins, s };
+  }
+  const tool = all.tools[0];
+  const dates = [...all.perDay.keys()].sort();
+  let streak = 0, run = 0, prev = null;
+  for (const d of dates) {
+    run = prev && dstr(addDays(parseD(prev), 1)) === d ? run + 1 : 1;
+    streak = Math.max(streak, run);
+    prev = d;
+  }
+  const span = longest ? spanText(longest.s.start, longest.s.end) : null;
+  const club = all.total >= 1e9 ? '十亿俱乐部' : '亿元户';
+  return [
+    { ok: all.total > 0, title: habit, line: peakH + ' 点是一天里最旺的时候', bg: 'var(--white)', rot: -2 },
+    { ok: hit >= 95, title: '缓存大师', line: '命中率 ' + hit.toFixed(2) + '%', lock: '命中率到 95% 解锁', bg: 'var(--yellow)', rot: 1.5 },
+    { ok: !!longest && longest.mins >= 720, title: '马拉松选手', line: span ? '单个会话前后跨了 ' + span[0] + ' ' + span[1] : '', lock: '单个会话跨满 12 小时解锁', bg: 'var(--mint)', rot: -1 },
+    { ok: !!tool && tool[1] >= 1000, title: tool ? tool[0] + ' 狂魔' : '工具狂魔', line: tool ? '用了 ' + num(tool[1]) + ' 次' : '', lock: '同一个工具用满 1,000 次解锁', bg: 'var(--pink)', rot: 2 },
+    { ok: streak >= 7, title: '全勤王', line: streak === dates.length ? '连续 ' + streak + ' 天，一天没断' : '最长连续 ' + streak + ' 天', lock: '连续用满 7 天解锁', bg: 'var(--sky)', rot: -1.5 },
+    { ok: all.total >= 1e8, title: club, line: '累计吞了 ' + fmt(all.total), lock: '累计吞满 1 亿解锁', bg: 'var(--white)', rot: 1 },
+  ];
+}
+
+// ---------------------------------------------------------------- 页面片段
+
+function header(extra) {
+  const meta = DATA.meta;
+  return `<header class="top">
+    <div class="brand">
+      <div class="logo cn">吞吐</div>
+      <div class="tag">我的 Token 都去哪了</div>
+      ${meta.demo ? '<div class="tag demo">演示数据</div>' : ''}
+    </div>
+    <div class="controls">${extra}</div>
+  </header>`;
+}
+function rangePills() {
+  const opts = [['all', '全部'], ['30', '近 30 天'], ['7', '近 7 天'], ['1', '今天']];
+  return `<div class="pills" role="group" aria-label="时间范围">${opts.map(([id, label]) =>
+    `<button type="button" class="pill" data-range="${id}" aria-pressed="${state.range === id}">${label}</button>`).join('')}</div>
+    <button type="button" class="pill go" data-act="rescan">${state.busy ? '扫描中…' : '重新扫描'}</button>`;
+}
+function footer() {
+  const meta = DATA.meta;
+  return `<footer><div>数据只在本机，不上传 · 读取 ${esc(meta.root)} · ${num(meta.sessions)} 个会话</div>
+    <div>扫描于 ${esc(when(meta.generated))} · tuntu ${esc(meta.version)}</div></footer>`;
+}
+function segs(parts, base) {
+  return parts.map(([name, v, color]) =>
+    `<div title="${esc(name)} ${fmt(v)}（${pct(v, base)}）" style="width:${(100 * v / (base || 1)).toFixed(3)}%;background:${color}"></div>`).join('');
+}
+function rankList(map, names, color, limit) {
+  const rows = [...map.entries()].map(([k, a]) => ({ name: names(k), a, t: sum4(a) })).sort((x, y) => y.t - x.t);
+  const all = rows.reduce((s, r) => s + r.t, 0);
+  let shown = rows.slice(0, limit);
+  if (rows.length > limit) {
+    const rest = rows.slice(limit).reduce((acc, r) => acc.map((v, i) => v + r.a[i]), [0, 0, 0, 0, 0]);
+    shown = shown.concat({ name: '其他 ' + (rows.length - limit) + ' 个', a: rest, t: sum4(rest) });
+  }
+  const max = shown.length ? Math.max(...shown.map((r) => r.t)) : 1;
+  return shown.map((r) => `<div class="rank">
+      <div class="top"><span class="nm">${esc(r.name)}</span><span class="vv">${fmt(r.t)} · ${pct(r.t, all, 1)}</span></div>
+      <div class="meter"><i style="width:${Math.max(1.5, 100 * r.t / max).toFixed(2)}%;background:${color}"></i></div>
+      <div class="sub">输出 ${fmt(r.a[1])} · 命中率 ${hitOf(r.a).toFixed(2)}% · ${num(r.a[4])} 次请求</div>
+    </div>`).join('') || '<div class="sub">这段时间没有记录</div>';
+}
+
+function dailyTile(c) {
+  const metric = state.metric;
+  const zero = [0, 0, 0, 0, 0];
+  const step = Math.max(1, Math.ceil(c.days.length / 92));
+  const groups = [];
+  for (let i = 0; i < c.days.length; i += step) {
+    const a = [0, 0, 0, 0, 0];
+    for (const d of c.days.slice(i, i + step)) (c.perDay.get(d) || zero).forEach((v, k) => { a[k] += v; });
+    groups.push({ date: c.days[i], a });
+  }
+  const val = (a) => metric === 'total' ? sum4(a) : metric === 'output' ? a[1] : metric === 'req' ? a[4] : (a[4] ? hitOf(a) : null);
+  const show = (v) => metric === 'req' ? num(v) + ' 次' : metric === 'hit' ? v.toFixed(1) + '%' : fmt(v);
+  const vals = groups.map((g) => val(g.a));
+  const real = vals.filter((v) => v !== null && v > 0);
+  const max = real.length ? Math.max(...real) : 1;
+  const floor = metric === 'hit' && real.length ? Math.max(0, Math.floor(Math.min(...real) / 5) * 5 - 5) : 0;
+  const peak = real.length ? vals.indexOf(max) : -1;
+  const every = Math.max(1, Math.ceil(groups.length / 8));
+  const gap = groups.length > 45 ? 2 : groups.length > 24 ? 4 : 6;
+  const cols = groups.map((g, i) => {
+    const v = vals[i];
+    const h = v ? Math.max(6, Math.round(200 * (v - floor) / ((max - floor) || 1))) : 0;
+    const tip = `${g.date}${step > 1 ? ' 起 ' + step + ' 天' : ''} · 全部 ${fmt(sum4(g.a))} · 输出 ${fmt(g.a[1])} · ${num(g.a[4])} 次请求`;
+    return `<div class="col"><div class="tg">${i === peak ? show(v) : ''}</div><div class="b${i === peak ? ' pk' : ''}${h ? '' : ' z'}" title="${esc(tip)}" style="height:${h}px"></div></div>`;
+  }).join('');
+  const labels = groups.map((g, i) => `<span>${i % every === 0 ? md(g.date) : ''}</span>`).join('');
+  const what = { total: '吞得最多', output: '写得最多', req: '请求最多', hit: '命中率最高' }[metric];
+  const caption = peak < 0 ? '这段时间没有记录。'
+    : `所选范围里${what}的${step > 1 ? '一段' : '一天'}是 ${md(groups[peak].date)}：${show(max)}。${metric === 'hit' ? '纵轴从 ' + floor + '% 起算。' : ''}`;
+  const opts = [['total', '全部'], ['output', '输出'], ['req', '请求'], ['hit', '命中率']];
+  return `<section class="tile daily">
+    <div class="head" style="align-items:center"><div class="h">每日吞吐</div>
+      <div class="pills" role="group" aria-label="指标">${opts.map(([id, label]) =>
+        `<button type="button" class="pill small" data-metric="${id}" aria-pressed="${metric === id}">${label}</button>`).join('')}</div></div>
+    <div class="scroll"><div style="min-width:${Math.min(520, 60 + groups.length * 22)}px;display:flex;flex-direction:column;gap:6px">
+      <div class="bars" style="gap:${gap}px">${cols}</div><div class="xl" style="gap:${gap}px">${labels}</div></div></div>
+    <div class="cap">${caption}</div>
+  </section>`;
+}
+
+function clockTile(c) {
+  const max = Math.max(...c.hours);
+  const peak = c.hours.indexOf(max);
+  const night = [20, 21, 22, 23, 0, 1, 2].reduce((s, h) => s + c.hours[h], 0);
+  const bars = c.hours.map((v, h) => {
+    const len = max ? Math.round(8 + 74 * v / max) : 8;
+    return `<div class="hb${max && h === peak ? ' pk' : ''}" title="${h}:00 · ${fmt(v)}" style="height:${len}px;transform:rotate(${180 + h * 15}deg) translateY(36px)"></div>`;
+  }).join('');
+  return `<section class="tile clockt">
+    <div class="h" style="align-self:flex-start">你的 24 小时</div>
+    <div class="clock" role="img" aria-label="24 小时用量钟面，${peak} 点最高">${bars}
+      <div class="mid">${max ? peak + ' 点' : '—'}</div>
+      <div class="l" style="left:50%;top:-4px;transform:translateX(-50%)">0</div>
+      <div class="l" style="right:-4px;top:50%;transform:translateY(-50%)">6</div>
+      <div class="l" style="left:50%;bottom:-4px;transform:translateX(-50%)">12</div>
+      <div class="l" style="left:-8px;top:50%;transform:translateY(-50%)">18</div>
+    </div>
+    <div class="cap">${max ? `${pct(night, c.total, 0)} 的 token 花在晚 8 点到凌晨 3 点。` : '这段时间没有记录。'}</div>
+  </section>`;
+}
+
+function heatTile(c) {
+  const max = Math.max(...c.heat.flat());
+  let pw = 0, ph = 0;
+  c.heat.forEach((row, w) => row.forEach((v, h) => { if (v > c.heat[pw][ph]) { pw = w; ph = h; } }));
+  const rows = c.heat.map((row, w) => `<div class="hr"><div class="wl">${WK[w]}</div>${row.map((v, h) =>
+    `<div class="c" title="${WK[w]} ${h}:00 · ${fmt(v)}"${v ? ` style="background:rgba(53,80,232,${(0.14 + 0.86 * v / max).toFixed(2)})"` : ''}></div>`).join('')}</div>`).join('');
+  const hl = Array.from({ length: 24 }, (_, h) => `<div class="hl">${h % 3 === 0 ? h : ''}</div>`).join('');
+  return `<section class="tile" style="flex:3 1 520px">
+    <div class="head"><div class="h">一周作息表</div><div class="sub">星期 × 小时，越深吞得越多</div></div>
+    <div class="scroll"><div class="heat">${rows}<div class="hr" style="margin-top:2px"><div class="wl"></div>${hl}</div></div></div>
+    <div class="cap">${max ? `最猛的一小时是${WK[pw]} ${ph} 点。` : '这段时间没有记录。'}</div>
+  </section>`;
+}
+
+function effortTile(c) {
+  const order = ['max', 'xhigh', 'high', 'medium', 'low', null];
+  const names = { max: '最大', xhigh: '超高', high: '高', medium: '中', low: '低' };
+  const palette = ['var(--yellow)', 'var(--orange)', 'var(--ink)', 'var(--blue)', 'var(--sky)', 'var(--white)'];
+  const keys = [...c.byE.keys()];
+  if (!keys.some((k) => k !== null)) return '';
+  keys.sort((a, b) => (order.indexOf(a) < 0 ? 9 : order.indexOf(a)) - (order.indexOf(b) < 0 ? 9 : order.indexOf(b)));
+  const items = keys.map((k) => ({ name: k === null ? '未记录' : (names[k] || k), a: c.byE.get(k), color: palette[Math.max(0, order.indexOf(k))] }));
+  return `<section class="tile" style="flex:2 1 360px;gap:16px">
+    <div class="h">用了多大的劲</div><div class="sub">按推理强度分的 token</div>
+    <div class="stack">${segs(items.map((e) => [e.name, sum4(e.a), e.color]), c.total)}</div>
+    ${items.map((e) => `<div class="erow"><span class="sw" style="margin:0;background:${e.color}"></span><div class="nm">${esc(e.name)}</div>
+      <div>${pct(sum4(e.a), c.total, 1)}</div><div class="dt">${fmt(sum4(e.a))} · ${num(e.a[4])} 次</div></div>`).join('')}
+  </section>`;
+}
+
+// ---------------------------------------------------------------- 总览页
+
+function renderHome() {
+  const c = compute(state.range);
+  const S = c.S, total = c.total;
+  const allIn = S[0] + S[2] + S[3], fresh = S[0] + S[1] + S[2];
+  if (!DATA.buckets.length) {
+    app.innerHTML = header(rangePills()) + `<section class="tile empty"><div class="h">还没找到用量记录</div>
+      <div class="cap">在 ${esc(DATA.meta.root)} 里没有读到 Claude Code 的会话日志。可以用 <b>--dir</b> 指定目录，或者用 <b>--demo</b> 先看看演示。</div></section>` + footer();
+    return;
+  }
+  const big = total >= 1e8 ? [(total / 1e8).toFixed(2), '亿'] : total >= 1e4 ? [(total / 1e4).toFixed(total >= 1e6 ? 0 : 1), '万'] : [num(total), ''];
+  const lead = { all: `自 ${+c.first.slice(5, 7)} 月 ${+c.first.slice(8, 10)} 日起，你一共吞了`, '30': '近 30 天，你一共吞了', '7': '近 7 天，你一共吞了', '1': '今天，你一共吞了' }[c.range];
+  const nd = c.days.length || 1;
+  const chips = [`平均每秒 ${num(total / (nd * 86400))} 个，昼夜不停`, `日均 ${fmt(total / nd)}`, `${num(S[4])} 次请求`, `写出 ${fmt(S[1])}`];
+  const ratio = S[1] ? Math.round(allIn / S[1]) : 0;
+  const per = Math.max(1, Math.ceil(ratio / 480));
+  const dots = '<i></i>'.repeat(Math.round(ratio / per)) + '<i class="w"></i>';
+  const hit = allIn ? 100 * S[3] / allIn : 0;
+  const cells = Math.floor(hit);
+  const waffle = Array.from({ length: 100 }, (_, i) => `<i${i < cells ? ' class="on"' : ''}></i>`).join('');
+  const parts = [['缓存读取', S[3], COLORS.cr, '每轮重读的历史上下文'], ['缓存写入', S[2], COLORS.cw, '第一次写进缓存的内容'],
+    ['输出', S[1], COLORS.out, '模型写出的回复和代码'], ['新输入', S[0], COLORS.inp, '没走缓存的输入']];
+  const facts = [];
+  if (c.h1 && S[2]) facts.push(`<div class="chip s" style="background:var(--sky)">缓存写入里 ${pct(c.h1, S[2], 1)} 是 1 小时档</div>`);
+  if (c.sideTok) facts.push(`<div class="chip s" style="background:var(--pink)">子代理吞了 ${pct(c.sideTok, total, 1)}（${fmt(c.sideTok)}，${num(c.sideReq)} 次请求）</div>`);
+  const toolMax = c.tools.length ? c.tools[0][1] : 1;
+  const medals = ['var(--yellow)', 'var(--sky)', 'var(--pink)'];
+  const top = c.sessions.slice(0, 10);
+  const sMax = top.length ? sum4(top[0].tok) : 1;
+  const effort = effortTile(c);
+
+  app.innerHTML = header(rangePills()) + `
+  <div class="row">
+    <section class="tile hero">
+      <div class="h">${lead}</div>
+      <div class="line"><div class="n">${big[0]}</div><div class="u cn">${big[1]}</div><div class="t">个 token</div></div>
+      <div class="chips">${chips.map((t) => `<div class="chip">${t}</div>`).join('')}</div>
+    </section>
+    <section class="tile side ratio">
+      <div class="h">读写比</div>
+      <div class="n">${ratio ? num(ratio) + ' : 1' : '—'}</div>
+      <div class="dots" role="img" aria-label="浅色方块代表读入，橙色方块代表写出">${dots}</div>
+      <div class="cap">${ratio ? `每写出 1 个 token（橙色那一格），先读了 ${num(ratio)} 个${per > 1 ? `（每个浅色格代表 ${per} 个）` : ''}。读的几乎全走缓存。` : '这段时间没有输出。'}</div>
+    </section>
+  </div>
+  <div class="row">
+    <section class="tile hit">
+      <div class="h">缓存命中率</div>
+      <div class="n">${allIn ? hit.toFixed(2) + '%' : '—'}</div>
+      <div class="waffle" role="img" aria-label="100 格华夫图，实心格为命中">${waffle}</div>
+      <div class="cap">100 格里中了 ${cells} 格</div>
+      <div class="sub" style="color:#1D3B2C">= 缓存读取 ÷ 全部输入</div>
+    </section>
+    ${dailyTile(c)}
+    ${clockTile(c)}
+  </div>
+  <section class="badges" aria-label="成就">${badges().map((b) => b.ok
+    ? `<div class="badge" style="background:${b.bg};transform:rotate(${b.rot}deg)"><div class="k">已解锁</div><div class="ti cn">${esc(b.title)}</div><div class="li">${esc(b.line)}</div></div>`
+    : `<div class="badge locked"><div class="k">未解锁</div><div class="ti cn">${esc(b.title)}</div><div class="li">${esc(b.lock || '')}</div></div>`).join('')}</section>
+  <div class="row">
+    <section class="tile" style="flex:3 1 520px;gap:16px">
+      <div class="h">吞下去的都是什么</div>
+      <div><div class="sub" style="margin-bottom:8px">全部 token</div><div class="stack">${segs(parts, total)}</div></div>
+      <div><div class="sub" style="margin-bottom:8px">去掉缓存读取，放大看剩下的 ${pct(fresh, total)}</div><div class="stack">${segs(parts.slice(1), fresh)}</div></div>
+      <div class="legend">${parts.map(([name, v, color, desc]) => `<div class="it"><span class="sw" style="background:${color}"></span>
+        <div><b>${name} · ${fmt(v)}</b><span>${pct(v, total)} · ${desc}</span></div></div>`).join('')}</div>
+      ${facts.length ? `<div class="chips" style="margin-top:0">${facts.join('')}</div>` : ''}
+    </section>
+    ${effort}
+  </div>
+  <div class="row">
+    <section class="tile" style="flex:1 1 420px;gap:18px"><div class="h">谁在干活（模型）</div>${rankList(c.byM, (k) => modelName(DATA.models[k]), 'var(--yellow)', 6)}</section>
+    <section class="tile" style="flex:1 1 420px;gap:18px"><div class="h">花在哪了（项目）</div>${rankList(c.byP, (k) => DATA.projects[k], 'var(--blue)', 8)}</section>
+  </div>
+  <div class="row">
+    ${heatTile(c)}
+    <section class="tile" style="flex:2 1 360px;gap:12px">
+      <div class="head"><div class="h">最爱用的工具</div><div class="sub">调用次数</div></div>
+      ${c.tools.slice(0, 10).map(([name, n]) => `<div class="trow"><div class="nm" title="${esc(name)}">${esc(name)}</div>
+        <div class="meter thin"><i style="width:${Math.max(1.5, 100 * n / toolMax).toFixed(1)}%;background:var(--orange)"></i></div><div class="ct">${num(n)}</div></div>`).join('') || '<div class="sub">这段时间没有工具调用</div>'}
+    </section>
+  </div>
+  <section class="tile" style="gap:16px">
+    <div class="head"><div class="h">会话马拉松榜</div><div class="sub">点一行看这个会话的上下文过山车</div></div>
+    <div class="scroll"><div class="slist">${top.map((s, i) => {
+      const sp = spanText(s.start, s.end);
+      return `<a class="srow" href="#/s/${encodeURIComponent(s.id)}">
+        <div class="medal" style="background:${medals[i] || 'var(--white)'}">${i + 1}</div>
+        <div class="ell" title="${esc(DATA.projects[s.p])}">${esc(DATA.projects[s.p])}</div>
+        <div class="lt">${when(s.start)} → ${when(s.end)}</div>
+        <div>${sp[0]} ${sp[1]}</div>
+        <div class="meter thin"><i style="width:${(100 * sum4(s.tok) / sMax).toFixed(1)}%;background:var(--blue)"></i></div>
+        <div class="r" style="font-weight:800">${fmt(sum4(s.tok))}</div>
+        <div class="r lt">${hitOf(s.tok).toFixed(2)}%</div>
+        <div class="r lt">${num(s.req)} 次</div></a>`;
+    }).join('') || '<div class="sub">这段时间没有会话</div>'}</div></div>
+    <div class="sub">续接或分叉出来的会话会带着前半段记录，所以各会话相加会超过总量。</div>
+  </section>` + footer();
+}
+
+// ---------------------------------------------------------------- 会话详情页
+
+async function renderSession(id) {
+  const ranked = [...DATA.sessions].sort((a, b) => sum4(b.tok) - sum4(a.tok));
+  const idx = ranked.findIndex((s) => s.id === id);
+  const back = '<a class="pill back" href="#/">← 返回总览</a>';
+  if (idx < 0) {
+    app.innerHTML = header(back) + '<section class="tile empty"><div class="h">找不到这个会话</div><div class="cap">它可能在重新扫描后不见了。</div></section>' + footer();
+    return;
+  }
+  const s = ranked[idx];
+  let det = { n: 0, per: 1, ctx: [], drop: [] };
+  try {
+    const r = await fetch('/api/session?id=' + encodeURIComponent(id));
+    if (r.ok) det = await r.json();
+  } catch (e) { /* 详情图画不出来时，其余部分照常显示 */ }
+  if (location.hash !== '#/s/' + encodeURIComponent(id)) return;
+
+  const total = sum4(s.tok);
+  const allIn = s.tok[0] + s.tok[2] + s.tok[3];
+  const sp = spanText(s.start, s.end);
+  const all = compute('all');
+  const allRatio = all.S[1] ? Math.round((all.S[0] + all.S[2] + all.S[3]) / all.S[1]) : 0;
+  const ratio = s.tok[1] ? Math.round(allIn / s.tok[1]) : 0;
+  const mainReq = s.req - s.subReq, mainTok = total - s.subTok;
+  const top = Math.max(1, ...det.ctx);
+  const unit = top >= 2e5 ? 2e5 : top >= 4e4 ? 4e4 : 1e4;
+  const yMax = Math.ceil(top / unit) * unit;
+  const cols = det.ctx.map((v, i) => `<div class="col"><div class="tg">${det.drop[i] ? '压缩' : ''}</div>
+    <div class="b${det.drop[i] ? ' pk' : ''}" title="第 ${num(i * det.per + 1)} 次请求附近 · 上下文约 ${fmt(v)}${det.drop[i] ? ' · 这一段压缩过' : ''}" style="height:${Math.max(2, Math.round(250 * v / yMax))}px"></div></div>`).join('');
+  const parts = [['缓存读取', s.tok[3], COLORS.cr], ['缓存写入', s.tok[2], COLORS.cw], ['输出', s.tok[1], COLORS.out], ['新输入', s.tok[0], COLORS.inp]];
+  const kpis = [
+    ['缓存命中率', hitOf(s.tok).toFixed(2) + '%', '整体是 ' + hitOf(all.S).toFixed(2) + '%', 'var(--mint)'],
+    ['写出', fmt(s.tok[1]), '平均每次请求 ' + num(s.tok[1] / (s.req || 1)) + ' 个', 'var(--white)'],
+    ['每轮读取', fmt(s.tok[3] / (s.req || 1)), '缓存读取 ÷ 请求次数', 'var(--white)'],
+    ['读写比', ratio ? num(ratio) + ' : 1' : '—', '整体是 ' + (allRatio ? num(allRatio) + ' : 1' : '—'), 'var(--sky)'],
+  ];
+  app.innerHTML = header(back + `<div class="tag" style="transform:none;font-size:15px">马拉松榜 第 ${idx + 1} 名 · ${esc(DATA.projects[s.p])} · ${esc(modelName(DATA.models[s.m]))}</div>`) + `
+  <div class="row">
+    <section class="tile hero">
+      <div class="h">这个会话前后跨了</div>
+      <div class="line"><div class="n">${sp[0]}</div><div class="u cn">${sp[1]}</div><div class="t">吞了 ${fmt(total)}个 token</div></div>
+      <div class="chips"><div class="chip">${when(s.start)} → ${when(s.end)}</div><div class="chip">${num(s.req)} 次请求</div><div class="chip">占全部用量的 ${pct(total, all.total, 1)}</div></div>
+    </section>
+    <section class="tile side" style="background:var(--pink);gap:10px">
+      <div class="h">上下文最高涨到</div>
+      <div class="n">${fmt(s.peak)}</div>
+      <div class="cap">${s.drops ? `中途自动压缩了 ${s.drops} 次，每次回落后再重新往上爬。` : '中途没有触发过压缩。'}</div>
+    </section>
+  </div>
+  <section class="kpis" aria-label="会话指标">${kpis.map(([k, v, sub, bg]) =>
+    `<div class="kpi" style="background:${bg}"><div class="k">${k}</div><div class="v">${v}</div><div class="s">${sub}</div></div>`).join('')}</section>
+  <section class="tile">
+    <div class="head"><div class="h">上下文过山车</div><div class="sub">每根柱子约 ${num(det.per)} 次请求 · 橙色 = 这一段压缩过</div></div>
+    ${det.ctx.length ? `<div class="scroll"><div class="ctxwrap">
+      <div class="yax"><span>${fmt(yMax)}</span><span>${fmt(yMax / 2)}</span><span>0</span></div>
+      <div style="flex:1 1 auto;display:flex;flex-direction:column;gap:6px">
+        <div class="bars ctx">${cols}</div>
+        <div style="display:flex;justify-content:space-between" class="sub"><span>第 1 次请求</span><span>第 ${num(Math.ceil(det.n / 2))} 次</span><span>第 ${num(det.n)} 次</span></div>
+      </div></div></div>` : '<div class="sub">这个会话没有主线请求。</div>'}
+    <div class="cap">每次请求都要把当时的整段上下文读一遍，所以柱子越高，那一段每轮就越贵。</div>
+  </section>
+  <div class="row">
+    <section class="tile" style="flex:3 1 460px;gap:16px">
+      <div class="h">这一单吞了什么</div>
+      ${parts.map(([name, v, color]) => `<div class="rank"><div class="top" style="font-size:15px;font-weight:700">
+        <span style="display:flex;align-items:center;gap:10px"><span class="sw" style="margin:0;background:${color}"></span>${name}</span><span>${fmt(v)} · ${pct(v, total)}</span></div>
+        <div class="meter"><i style="width:${Math.max(1, 100 * v / (total || 1)).toFixed(2)}%;background:${color};border-right:0"></i></div></div>`).join('')}
+    </section>
+    <section class="tile" style="flex:2 1 340px;gap:16px">
+      <div class="h">主线 vs 子代理</div>
+      <div class="duo">
+        <div style="background:var(--sky)"><div class="k">主线对话</div><div class="v">${fmt(mainTok)}</div><div class="s">${num(mainReq)} 次${mainReq ? ' · 每次约 ' + fmt(mainTok / mainReq) : ''}</div></div>
+        <div style="background:var(--mint)"><div class="k">子代理</div><div class="v">${fmt(s.subTok)}</div><div class="s">${num(s.subReq)} 次${s.subReq ? ' · 每次约 ' + fmt(s.subTok / s.subReq) : ''}</div></div>
+      </div>
+      <div class="cap">子代理每次从干净的上下文起步，单次请求通常比主线便宜。</div>
+      <div class="tip">省 token 小贴士：换任务时开个新会话，比在越来越长的会话里接着聊更省。</div>
+    </section>
+  </div>` + footer();
+  window.scrollTo(0, 0);
+}
+
+// ---------------------------------------------------------------- 启动
+
+function route() {
+  if (!DATA) return;
+  const m = location.hash.match(/^#\/s\/(.+)$/);
+  if (m) renderSession(decodeURIComponent(m[1]));
+  else renderHome();
+}
+async function load(rescan) {
+  state.busy = true;
+  if (DATA) route();
+  try {
+    const r = await fetch('/api/data' + (rescan ? '?rescan=1' : ''));
+    DATA = await r.json();
+  } catch (e) {
+    if (!DATA) app.innerHTML = '<div class="tile empty"><div class="h">连不上本机服务</div><div class="cap">请确认 tuntu 还在运行，然后刷新页面。</div></div>';
+  }
+  state.busy = false;
+  route();
+}
+app.addEventListener('click', (e) => {
+  const b = e.target.closest('button');
+  if (!b) return;
+  if (b.dataset.range) { state.range = b.dataset.range; route(); }
+  else if (b.dataset.metric) { state.metric = b.dataset.metric; route(); }
+  else if (b.dataset.act === 'rescan' && !state.busy) load(true);
+});
+window.addEventListener('hashchange', route);
+load(false);
+</script>
+</body>
+</html>

tuntu-0.1.0.dist-info/METADATA

@@ -0,0 +1,134 @@
+Metadata-Version: 2.4
+Name: tuntu
+Version: 0.1.0
+Summary: 吞吐：本机 Claude Code token 用量看板。A local, zero-dependency dashboard for Claude Code token usage.
+License-Expression: MIT
+Keywords: claude,claude-code,tokens,usage,dashboard
+Classifier: Environment :: Console
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Developers
+Classifier: Natural Language :: Chinese (Simplified)
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Utilities
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# 吞吐 tuntu
+
+我的 Token 都去哪了？——一个本机运行的 Claude Code token 用量看板。
+
+A local, zero-dependency dashboard for your Claude Code token usage. It reads the session logs already on your machine and shows them in a playful one-page dashboard. Nothing leaves your computer. (The UI is in Chinese.)
+
+## 运行
+
+需要 Python 3.9 或更高版本，没有任何第三方依赖。
+
+不用安装，一行运行（需要 [uv](https://docs.astral.sh/uv/)）：
+
+```bash
+uvx tuntu
+```
+
+或者装成命令：
+
+```bash
+pipx install tuntu
+```
+
+```bash
+tuntu
+```
+
+也可以直接从源码运行，不用安装：
+
+```bash
+git clone <本仓库地址>
+cd <仓库目录>
+python3 -m tuntu
+```
+
+它会扫描 `~/.claude/projects` 下的会话日志，然后在浏览器里打开 `http://127.0.0.1:8787/`。
+
+没有日志，或者只是想先看看长什么样（也适合截图）：
+
+```bash
+tuntu --demo
+```
+
+| 参数 | 作用 |
+|---|---|
+| `--demo` | 用虚构数据演示，不读取任何日志 |
+| `--dir PATH` | 指定日志目录（默认 `~/.claude/projects`，也会识别 `CLAUDE_CONFIG_DIR`） |
+| `--port N` | 端口，默认 8787，被占用时自动往后找 |
+| `--no-open` | 不自动打开浏览器 |
+| `--web-fonts` | 从 Google Fonts 加载展示字体，观感更接近设计稿（会产生外网请求） |
+| `--json` | 把聚合结果打印成 JSON 后退出 |
+| `--version` | 显示版本号 |
+
+## 能看到什么
+
+- **总量**：一共吞了多少 token，平均每秒多少，日均多少。
+- **读写比**：每写出 1 个 token，先读了多少个。
+- **缓存命中率**：缓存读取 ÷ 全部输入。
+- **每日吞吐**：可以切换看全部、输出、请求次数或命中率。
+- **24 小时钟、一周作息表**：你什么时候用得最多。
+- **构成**：缓存读取、缓存写入、输出、新输入各占多少；缓存写入里 1 小时档的占比；子代理的占比。
+- **推理强度、模型、项目、工具**：各自的分布。
+- **会话马拉松榜**：最耗 token 的会话。点进去能看到上下文长度随对话增长、被自动压缩的过程。
+- **成就贴纸**：夜猫子、缓存大师、马拉松选手……
+
+时间范围可以切换：全部、近 30 天、近 7 天、今天。
+
+## 隐私
+
+- 服务只监听 `127.0.0.1`，并且拒绝 Host 不是本机的请求。
+- 默认不发出任何外网请求，页面带有禁止外部资源的 CSP。只有加了 `--web-fonts` 才会去请求 Google Fonts。
+- 只取用日志里的用量数字、时间、模型名、工具名和工作目录名；对话内容不会被保存、显示或发送到任何地方。
+- 页面上的项目只显示目录名，不显示完整路径；日志目录里的家目录显示为 `~`。
+- 不写任何文件，不做缓存，关掉就什么都不留。
+
+分享截图前请留意：项目名和工具名来自你自己的机器。想要一张不含个人信息的图，用 `--demo`。
+
+## 数字是怎么算的
+
+- 数据来自 Claude Code 写在本机的会话日志（`*.jsonl`）里每次请求的 `usage` 字段。
+- 同一条消息在日志里会出现多行，按消息去重后再累加。
+- 续接或分叉出来的会话会带着前半段记录。算总量时每条消息只算一次；但会话榜里每个会话按它自己的全部记录算，所以各会话相加会超过总量。
+- 项目按会话启动时的目录归类，worktree 里的会话归到原项目。
+- “会话跨度”是第一次请求到最后一次请求之间的时间，中间可能隔了很久没用。
+- 时间按本机时区显示。
+- Claude Code 默认只保留一段时间的日志（由 `cleanupPeriodDays` 设置决定），更早的用量已经不在本机，这里也就看不到。
+- 这里只统计本机的 Claude Code，不包含 claude.ai 网页或手机上的对话。
+- 这些数字是从本机日志估算的，不是账单。以官方的用量页面为准。
+
+## 目录结构
+
+```
+tuntu/cli.py           扫描日志、聚合、本机 HTTP 服务（只用标准库）
+tuntu/web/index.html   页面（单文件，原生 JS，无依赖）
+pyproject.toml         打包配置
+```
+
+## 发布新版本（维护者）
+
+1. 改 `tuntu/__init__.py` 里的 `__version__`。
+2. 构建并上传：
+
+```bash
+uv build
+```
+
+```bash
+uv publish
+```
+
+## 说明
+
+这是个人项目，与 Anthropic 无关。Claude 和 Claude Code 是 Anthropic 的商标。
+
+## 许可证
+
+MIT，见 [LICENSE](LICENSE)。

tuntu-0.1.0.dist-info/RECORD

@@ -0,0 +1,10 @@
+tuntu/__init__.py,sha256=vHtqaYlbdeMFwKlpAa-8kwkZ8oOVyoHGHhCzKW4Cc6M,89
+tuntu/__main__.py,sha256=bYt9eEaoRQWdejEHFD8REx9jxVEdZptECFsV7F49Ink,30
+tuntu/cli.py,sha256=SVf8Y9AsMb3o-Bvkbqy5Iysq0yCMjjM57BK1_Npxgto,19389
+tuntu/web/index.html,sha256=27y_6dZEp0WgFl9a82BZzoHV44NZjVjInLg8B6KYluo,40319
+tuntu-0.1.0.dist-info/licenses/LICENSE,sha256=K9VX4dr242WEsDy5R0lzfSUXItlhjeEMt_pUqeVKiH0,1075
+tuntu-0.1.0.dist-info/METADATA,sha256=6AAnfdCTnd9MakrANZ7FqLIMAV7gmme4a5HvpwL04fM,5124
+tuntu-0.1.0.dist-info/WHEEL,sha256=mOdXR9C3wMVov7cSY8PY-XCzo3SmmWCHvGQr5csJoao,91
+tuntu-0.1.0.dist-info/entry_points.txt,sha256=T3bSvUOjTnQzGQy0tkTZMi6hqvUdqqDlQxGA_6pWSm4,41
+tuntu-0.1.0.dist-info/top_level.txt,sha256=ZeiQNOImlf3NtyffiSBGx1Kv8dU-hHuNPG-g8LkgZw0,6
+tuntu-0.1.0.dist-info/RECORD,,

tuntu-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (84.2.1)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

tuntu-0.1.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+tuntu = tuntu.cli:main

tuntu-0.1.0.dist-info/licenses/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 tuntu contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

tuntu-0.1.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+tuntu