Files
MediaCrawler/mac-agent-os-main/05_tools/09_ave/scripts/lib/cost_tracker.py
T
butubb 2112c1a870
Deploy VitePress site to Pages / build (push) Canceled after 0s
Deploy VitePress site to Pages / Deploy (push) Canceled after 0s
feat(monitor): 抖音 Web 接口客户端 —— 绕开爬虫子进程,直接发 HTTP
移植自 mac-agent-os 的 mediacrawler_adapter:不起子进程、不开页面,用浏览器里那份
登录态直接调抖音 Web 接口。产物键名照抄 store/douyin,所以 ingest 那条链路一个字不用改。

**目前能用的(真环境实测,非推断)**:

    profile/other : 200, 7075 字节  —— 博主主页指标(粉丝/获赞/作品数/昵称)
    aweme/detail  : 200, 45425 字节 —— 单条作品详情(含点赞/评论/收藏/分享)

**目前不能用的:作品列表 `aweme/post`。** 两个互相独立的原因:
  1. 这个接口被抖音单独升级成了真校验:不带 x-tt-argus 回 403「Uifid Not Found」,
     带上 dummy 值回 200 + **空 body**。也就是说「头在不在」骗得过,「真校验」过不了。
     同一套头打 profile/other 和 aweme/detail 都是通的 —— 抖音是挑着接口加保护的,
     挑中的恰好是「批量拉作品列表」这个最敏感的动作。
  2. 改走页面截获也不行:CDP 浏览器打开博主主页会落到「验证码中间页」(当天大量探测的
     代价,过几小时要重测)。

  所以现在的边界是:**已知作品的指标刷新能做,自动发现新作品做不了**。

**排查中控住变量后得到的两条事实**(都写进注释了):
  · `Accept` / `Accept-Language` / `Referer` 才是主页接口能返回真数据的原因 —— 只有
    UA+client hints+Cookie 时是 200 但仅 121 字节的空壳,补上这三个头变 7074 字节。
    (我先前猜的 sec-ch-ua 不是关键。)
  · 因此 UA 与 client hints 必须**成套地取自同一个浏览器**,所以 BrowserIdentity 一次
    从 CDP 取齐 cookie + UA + hints,而不是各自写死。

「200 + 空 body 必须当场报错」也是刻意写死的:放过去它会在下游变成「这个博主没作品」,
把一次失败伪装成一条正常结果 —— 爬虫那条路正是这么栽的,还被翻译成「账号被封」。

测试 +11:cookie 解析、请求头成套性(含 uifid 缺失/回退)、产物键名与 store 对齐、
以及 _get 的三条失败路径(空 body / 403 带网关原话 / 正常返回)。
2026-10-10 17:12:25 +08:00

214 lines
7.5 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""
AVE 成本追踪器
记录每次 API 调用的费用估算, 防止不知不觉扣费过多。
用法:
from lib.cost_tracker import CostTracker
tracker = CostTracker()
tracker.log("OmniHuman", duration=4.04, resolution="480P")
tracker.summary()
"""
import json
import os
from datetime import datetime
from pathlib import Path
# ── 已知价格 (元/秒) ───────────────────────────────────────
# 来源: 火山引擎即梦AI官方定价 (截至2026-05-09)
PRICING = {
# OmniHuman 1.5
"OmniHuman_480P": 0.5, # ¥0.5/秒
"OmniHuman_720P": 0.9, # ¥0.9/秒
# Wan2.2 (备用)
"Wan2.2_480P": 0.5,
"Wan2.2_720P": 0.9,
# DreamActor M1 (估算, 与OmniHuman同级)
"DreamActor_480P": 0.5,
"DreamActor_720P": 0.9,
# CosyVoice TTS (阿里云百炼, 按字符计费约 ¥0.002/字符)
"CosyVoice_per_char": 0.002,
# Pexels (免费)
"Pexels": 0,
# Kling LipSync (fal.ai, $0.014/5s ≈ ¥0.10/s)
"KlingLipSync_per_s": 0.10, # ¥0.10/秒 ≈ $0.014/5s
}
# ── 日志路径 ───────────────────────────────────────────────
LOG_DIR = Path(os.environ.get("AVE_LOCAL_DIR",
str(Path.home() / "workbuddy-agent-os/agent-local"))) / "tools" / "ave" / "logs"
LOG_DIR.mkdir(parents=True, exist_ok=True)
LOG_FILE = LOG_DIR / "cost_log.jsonl"
class CostTracker:
"""API 成本追踪器"""
def __init__(self):
self._session_log: list[dict] = []
def log(self, service: str, duration: float = 0, resolution: str = "480P",
chars: int = 0, status: str = "success", note: str = ""):
"""
记录一次 API 调用
参数:
service: API 服务名 (OmniHuman, DreamActor, CosyVoice, Pexels)
duration: 视频/音频时长(秒)
resolution: 分辨率 (480P/720P)
chars: CosyVoice 字符数
status: success/failed
note: 备注
"""
# 计算费用
if service == "CosyVoice":
unit_price = PRICING.get("CosyVoice_per_char", 0.002)
cost = round(chars * unit_price, 4)
elif service == "KlingLipSync":
unit_price = PRICING.get("KlingLipSync_per_s", 0.10)
cost = round(duration * unit_price, 4) if status == "success" else 0.0
elif service == "Pexels":
cost = 0.0
else:
key = f"{service}_{resolution}"
unit_price = PRICING.get(key, 0.5)
cost = round(duration * unit_price, 4) if status == "success" else 0.0
record = {
"time": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
"service": service,
"duration_s": duration,
"resolution": resolution,
"chars": chars,
"status": status,
"cost_yuan": cost,
"note": note,
}
self._session_log.append(record)
# 追加到日志文件
with open(LOG_FILE, "a", encoding="utf-8") as f:
f.write(json.dumps(record, ensure_ascii=False) + "\n")
# 同步写入 Dashboard DB (如果当前有 production)
if _current_production_id > 0:
try:
from lib.dashboard import log_cost as _dash_log_cost
_dash_log_cost(
production_id=_current_production_id,
service=service,
step_name=note.split(" ")[0] if " " in note else "",
duration_s=duration,
resolution=resolution,
chars=chars,
cost_yuan=cost,
status=status,
note=note,
)
except ImportError:
pass # dashboard 模块不可用时不报错
# 实时输出
if cost > 0:
print(f" 💰 {service}: {duration:.1f}s × ¥{unit_price}/s = ¥{cost:.2f}")
elif status == "failed":
print(f" 💰 {service}: 失败, 不计费")
else:
print(f" 💰 {service}: 免费")
return cost
def summary(self) -> dict:
"""输出本次会话的费用汇总"""
total = sum(r["cost_yuan"] for r in self._session_log)
success_calls = sum(1 for r in self._session_log if r["status"] == "success")
failed_calls = sum(1 for r in self._session_log if r["status"] == "failed")
print("\n" + "=" * 40)
print("💰 本次会话费用汇总")
print("=" * 40)
print(f" 成功调用: {success_calls} 次")
print(f" 失败调用: {failed_calls} 次")
print(f" 总费用: ¥{total:.2f}")
print(f" 账户余额: 未知 (请登录火山引擎控制台查看)")
print("=" * 40)
# 分服务统计
by_service = {}
for r in self._session_log:
s = r["service"]
if s not in by_service:
by_service[s] = {"count": 0, "total_s": 0, "cost": 0.0}
by_service[s]["count"] += 1
by_service[s]["total_s"] += r["duration_s"] if r["status"] == "success" else 0
by_service[s]["cost"] += r["cost_yuan"]
if by_service:
print("\n 按服务:")
for s, info in sorted(by_service.items()):
print(f" {s:15s} {info['count']}次, {info['total_s']:.1f}s, ¥{info['cost']:.2f}")
return {"total": total, "calls": success_calls + failed_calls}
def get_session_log(self) -> list[dict]:
return self._session_log
def _total_cost(self) -> float:
"""当前会话累计费用"""
return sum(r["cost_yuan"] for r in self._session_log)
# ── Dashboard 集成 ─────────────────────────────────────────
# 当前 production_id, 由 video_factory / main.py 设置
# CostTracker.log() 检测到该值 > 0 时自动写 dashboard DB
_current_production_id: int = 0
def set_current_production_id(pid: int):
"""设置当前 production ID, 后续 cost log 自动关联"""
global _current_production_id
_current_production_id = pid
def get_current_production_id() -> int:
return _current_production_id
# ── 全局单例 (在 main.py 中使用) ────────────────────────────
_tracker: CostTracker | None = None
def get_tracker() -> CostTracker:
global _tracker
if _tracker is None:
_tracker = CostTracker()
return _tracker
def print_history(days: int = 7):
"""打印最近 N 天的费用历史"""
if not LOG_FILE.exists():
print("暂无费用记录")
return
from datetime import timedelta
cutoff = datetime.now() - timedelta(days=days)
total = 0.0
count = 0
with open(LOG_FILE, encoding="utf-8") as f:
for line in f:
line = line.strip()
if not line:
continue
try:
r = json.loads(line)
t = datetime.strptime(r["time"], "%Y-%m-%d %H:%M:%S")
if t >= cutoff:
total += r["cost_yuan"]
count += 1
print(f" [{r['time']}] {r['service']:15s} {r['duration_s']:5.1f}s ¥{r['cost_yuan']:.2f} {r['note']}")
except (json.JSONDecodeError, KeyError, ValueError):
pass
print(f"\n 近 {days} 天: {count} 次调用, 总费用 ¥{total:.2f}")