Sidney Zhang
9020752eed
- 为 build_studio_urls.py 和 lookup_groups.py 添加 PEP 723 依赖声明 - 修复 HTTP/2 伪头导致 requests 抛 InvalidHeader 的问题 - 添加 CDP 端口连通性预检查及启动指引
327 lines
14 KiB
Python
327 lines
14 KiB
Python
# -*- coding: utf-8 -*-
|
||
# /// script
|
||
# requires-python = ">=3.10"
|
||
# dependencies = [
|
||
# "playwright>=1.40",
|
||
# ]
|
||
# ///
|
||
r"""
|
||
YouTube Studio 内容管理器「导出当前视图 → 逗号分隔值 (.csv)」下载脚本
|
||
（拦截响应 + 解码 zip + 完整性校验 + 自动保存到下载目录 + 重名去重）。
|
||
|
||
机制（已实证）：前端点击导出后向后端
|
||
POST https://studio.youtube.com/youtubei/v1/yta_web/csv_export?alt=json
|
||
后端把打好的 zip 以 base64 内联在响应 `zippedData` 字段里（开头 `UEsDBBQ` 即 ZIP 文件头 `PK`）。
|
||
注意两个字段细节（均为实证结论）：
|
||
- `zippedData` 是 **URL-safe base64**（-/_ 代替 +//，且省略 = 填充）；
|
||
- 当前界面导出入口是右上角**下载图标按钮**：只有 `aria-label="导出当前视图"`，没有可见文本。
|
||
本脚本拦截该响应，base64 解码并校验 zip 完整性后按 `<维度标签> <起始日>_<结束日> <账号名>.zip`
|
||
写盘，重名自动加 ` (n)` 后缀，n 从 1 起；未捕获到有效响应时自动重试一次触发导出。
|
||
|
||
登录态：脚本不负责登录，必须复用已登录 YouTube Studio 的浏览器会话，二选一：
|
||
- --user-data-dir + --channel chrome|msedge ：用已登录的用户数据目录启动（需先关闭该浏览器）
|
||
- --connect http://localhost:9222 ：附加到已在调试端口运行的浏览器
|
||
|
||
用法：
|
||
python youtube_export_download.py --selftest
|
||
python youtube_export_download.py --url "<explore URL>" --channel chrome --user-data-dir "C:/Users/<you>/AppData/Local/Google/Chrome/User Data"
|
||
python youtube_export_download.py --url "<explore URL>" --connect http://localhost:9222
|
||
"""
|
||
|
||
import argparse
|
||
import base64
|
||
import io
|
||
import json
|
||
import os
|
||
import re
|
||
import sys
|
||
import time
|
||
import zipfile
|
||
|
||
DOWNLOAD_DIR = r"D:\Downloads" # 默认下载目录，可用 --download-dir 覆盖
|
||
|
||
EXPORT_RESPONSE_TIMEOUT = 30 # 每次触发导出后等待 csv_export 响应的秒数
|
||
MAX_EXPORT_ATTEMPTS = 2 # 未捕获到有效响应时的最大触发次数（1 次重试）
|
||
|
||
# 维度类型 -> 文件名前缀标签（生产环境建议从页面「维度」按钮文本读取，这里兜底映射）。
|
||
DIMENSION_LABEL = {
|
||
"VIDEO": "内容",
|
||
"USER": "频道",
|
||
"CONTENT_OWNER": "内容",
|
||
}
|
||
|
||
CSV_EXPORT_PATH = "/youtubei/v1/yta_web/csv_export"
|
||
|
||
|
||
def dedup_path(directory, filename):
|
||
"""返回不冲突的落盘路径；重名按 `名称 (n).后缀` 递增，n 从 1 起。"""
|
||
directory = os.path.abspath(directory)
|
||
base, ext = os.path.splitext(filename)
|
||
candidate = os.path.join(directory, filename)
|
||
n = 1
|
||
while os.path.exists(candidate):
|
||
candidate = os.path.join(directory, f"{base} ({n}){ext}")
|
||
n += 1
|
||
return candidate
|
||
|
||
|
||
def build_export_filename(export_query, account_name, dimension_label=None):
|
||
"""从 csv_export 请求体 `exportQuery` 反推文件名。"""
|
||
def fmt_dateid(yyyymmdd):
|
||
s = str(yyyymmdd)
|
||
return f"{s[0:4]}-{s[4:6]}-{s[6:8]}"
|
||
|
||
date_range = None
|
||
dimension = None
|
||
nodes = export_query.get("joinRequest", {}).get("nodes") or []
|
||
for node in nodes:
|
||
q = node.get("value", {}).get("query") or {}
|
||
if not date_range:
|
||
tr = q.get("timeRange", {}).get("dateIdRange")
|
||
if tr and tr.get("inclusiveStart"):
|
||
date_range = (tr["inclusiveStart"], tr.get("exclusiveEnd"))
|
||
dims = q.get("dimensions") or []
|
||
if dimension is None and dims:
|
||
dimension = dims[0].get("type")
|
||
|
||
if not date_range:
|
||
raise ValueError("无法从 exportQuery 解析日期范围")
|
||
|
||
if dimension_label is None:
|
||
dimension_label = DIMENSION_LABEL.get(dimension or "", "")
|
||
|
||
start, end = date_range
|
||
return f"{dimension_label} {fmt_dateid(start)}_{fmt_dateid(end)} {account_name}.zip"
|
||
|
||
|
||
def decode_zipped_data(payload):
|
||
"""把 csv_export 响应 payload 里的 zippedData 解码为 zip 字节流。
|
||
|
||
该字段是 URL-safe base64（-/_ 代替 +//，且省略 = 填充）：标准 b64decode 遇
|
||
-/_ 或非 4 对齐长度会抛 Incorrect padding，或静默解出损坏字节流。先补齐填充，
|
||
再用 altchars 兼容 URL-safe 与标准两种字符表；解码后校验 PK 文件头。
|
||
"""
|
||
zipped = payload.get("zippedData")
|
||
if not zipped:
|
||
raise ValueError("响应中缺少 zippedData 字段")
|
||
z = str(zipped).strip()
|
||
data = base64.b64decode(z + "=" * (-len(z) % 4), altchars=b"-_")
|
||
if not data.startswith(b"PK"):
|
||
raise ValueError(
|
||
"zippedData 解码结果不是 zip 字节流（缺 PK 文件头），"
|
||
"响应可能被网关改写或导出接口已变动")
|
||
return data
|
||
|
||
|
||
def verify_zip(data):
|
||
"""校验内存 zip 完整性：结构可读、成员 CRC 全过；否则抛 ValueError。"""
|
||
try:
|
||
with zipfile.ZipFile(io.BytesIO(data)) as zf:
|
||
bad = zf.testzip()
|
||
except zipfile.BadZipFile as e:
|
||
raise ValueError(f"zip 结构损坏: {e}") from e
|
||
if bad is not None:
|
||
raise ValueError(f"zip 成员 CRC 校验失败: {bad}")
|
||
|
||
|
||
def trigger_export(page):
|
||
"""点「导出当前视图」→「逗号分隔值 (.csv)」。
|
||
|
||
导出入口是右上角的下载图标按钮：只有 aria-label、无可见文本，
|
||
优先按 label 定位（get_by_text 兜底旧版有可见文本的界面）。
|
||
"""
|
||
page.get_by_label("导出当前视图").or_(page.get_by_text("导出当前视图")).first.click()
|
||
page.get_by_text("逗号分隔值 (.csv)").click()
|
||
|
||
|
||
def intercept_and_save(page, account_name, download_dir):
|
||
"""给 page 绑定 response 拦截器：命中 csv_export 就把 zip 保存到下载目录。"""
|
||
import pathlib
|
||
|
||
saved = []
|
||
|
||
def on_response(response):
|
||
if CSV_EXPORT_PATH not in response.url:
|
||
return
|
||
try:
|
||
payload = response.json()
|
||
data = decode_zipped_data(payload)
|
||
verify_zip(data)
|
||
|
||
filename = None
|
||
try:
|
||
body = json.loads(response.request.post_data or "{}")
|
||
filename = build_export_filename(body.get("exportQuery", {}), account_name)
|
||
except Exception:
|
||
filename = "export.zip" # 反推失败兜底，避免丢内容
|
||
|
||
filename = re.sub(r"[\\/:*?\"<>|]", "_", filename) # Windows 非法字符
|
||
path = dedup_path(download_dir, filename)
|
||
pathlib.Path(path).write_bytes(data)
|
||
saved.append((filename, len(data), path))
|
||
except Exception as e: # noqa: BLE001
|
||
print(f"[interceptor] 处理 csv_export 响应失败: {e}", file=sys.stderr)
|
||
|
||
page.on("response", on_response)
|
||
return saved
|
||
|
||
|
||
def _default_user_data_dir(channel):
|
||
"""返回指定浏览器的默认用户数据目录（Windows），用于复用已登录会话。"""
|
||
_local = os.environ.get("LOCALAPPDATA") or os.path.expanduser(r"~\AppData\Local")
|
||
if channel == "msedge":
|
||
return os.path.join(_local, "Microsoft", "Edge", "User Data")
|
||
return os.path.join(_local, "Google", "Chrome", "User Data")
|
||
|
||
|
||
def run(url, user_data_dir=None, channel=None, cdp_url=None,
|
||
download_dir=None, account_name=None):
|
||
"""启动/连接浏览器并触发导出。必须复用已登录会话，否则跳 Google 登录页。"""
|
||
from playwright.sync_api import sync_playwright
|
||
|
||
download_dir = download_dir or DOWNLOAD_DIR
|
||
os.makedirs(download_dir, exist_ok=True)
|
||
|
||
with sync_playwright() as p:
|
||
browser = None
|
||
context = None
|
||
|
||
if cdp_url:
|
||
browser = p.chromium.connect_over_cdp(cdp_url)
|
||
context = browser.contexts[0] if browser.contexts else \
|
||
browser.new_context(accept_downloads=True)
|
||
page = context.new_page()
|
||
page.goto(url, wait_until="domcontentloaded")
|
||
elif user_data_dir:
|
||
context = p.chromium.launch_persistent_context(
|
||
user_data_dir=user_data_dir or _default_user_data_dir(channel),
|
||
channel=channel,
|
||
headless=False,
|
||
accept_downloads=True,
|
||
args=["--disable-blink-features=AutomationControlled"],
|
||
)
|
||
page = context.new_page()
|
||
page.goto(url, wait_until="domcontentloaded")
|
||
else:
|
||
browser = p.chromium.launch(headless=False, channel=channel)
|
||
context = browser.new_context(accept_downloads=True)
|
||
page = context.new_page()
|
||
page.goto(url, wait_until="domcontentloaded")
|
||
|
||
if "accounts.google" in page.url:
|
||
print("[!] 当前会话未登录，已跳转到 Google 登录页。", file=sys.stderr)
|
||
print(" 请用 --user-data-dir 或 --connect 复用已登录浏览器后重试。",
|
||
file=sys.stderr)
|
||
|
||
if not account_name:
|
||
account_name = "WL Media"
|
||
try:
|
||
account_name = page.locator(
|
||
"ytcp-account-item button, .account-switcher button"
|
||
).first.inner_text(timeout=5000).strip() or account_name
|
||
except Exception:
|
||
pass
|
||
|
||
saved = intercept_and_save(page, account_name, download_dir)
|
||
|
||
# 触发导出并等待响应；未捕获到有效 zip（未触发/响应无效）自动重试一次
|
||
for attempt in range(1, MAX_EXPORT_ATTEMPTS + 1):
|
||
trigger_export(page)
|
||
deadline = time.time() + EXPORT_RESPONSE_TIMEOUT
|
||
while not saved and time.time() < deadline:
|
||
page.wait_for_timeout(500)
|
||
if saved:
|
||
break
|
||
if attempt < MAX_EXPORT_ATTEMPTS:
|
||
print(f"[重试] 第 {attempt} 次未捕获到有效导出响应，自动重试……",
|
||
file=sys.stderr)
|
||
|
||
if not saved:
|
||
print("未捕获到 csv_export 响应，请确认已点击导出且登录态有效。")
|
||
else:
|
||
for name, size, path in saved:
|
||
print(f"已保存: {path} ({size} bytes)")
|
||
|
||
if browser is not None:
|
||
browser.close()
|
||
elif context is not None:
|
||
context.close()
|
||
|
||
|
||
def selftest():
|
||
import tempfile
|
||
from pathlib import Path
|
||
|
||
with tempfile.TemporaryDirectory() as td:
|
||
p0 = Path(dedup_path(td, "需求文件.zip"))
|
||
assert p0.name == "需求文件.zip", p0.name
|
||
p0.write_bytes(b"a")
|
||
|
||
p1 = Path(dedup_path(td, "需求文件.zip"))
|
||
assert p1.name == "需求文件 (1).zip", p1.name
|
||
p1.write_bytes(b"b")
|
||
|
||
p2 = Path(dedup_path(td, "需求文件.zip"))
|
||
assert p2.name == "需求文件 (2).zip", p2.name
|
||
p2.write_bytes(b"c")
|
||
|
||
assert Path(dedup_path(td, "需求文件.zip")).name == "需求文件 (3).zip"
|
||
assert Path(dedup_path(td, "其他.zip")).name == "其他.zip"
|
||
|
||
export_query = {
|
||
"joinRequest": {"nodes": [{"value": {"query": {
|
||
"dimensions": [{"type": "VIDEO"}],
|
||
"timeRange": {"dateIdRange": {
|
||
"inclusiveStart": 20260723, "exclusiveEnd": 20260820}},
|
||
}}}]},
|
||
}
|
||
name = build_export_filename(export_query, "WL Media")
|
||
assert name == "内容 2026-07-23_2026-08-20 WL Media.zip", name
|
||
|
||
# zip 解码：标准 base64 与 URL-safe 无填充变体都要解出同一字节流
|
||
buf = io.BytesIO()
|
||
with zipfile.ZipFile(buf, "w", zipfile.ZIP_DEFLATED) as zf:
|
||
zf.writestr("a.csv", "x,y\n1,2")
|
||
raw = buf.getvalue()
|
||
std = base64.b64encode(raw).decode("ascii")
|
||
urlsafe_nopad = std.replace("+", "-").replace("/", "_").rstrip("=")
|
||
assert decode_zipped_data({"zippedData": std}) == raw
|
||
assert decode_zipped_data({"zippedData": urlsafe_nopad}) == raw
|
||
verify_zip(raw)
|
||
|
||
# 非 zip 字节流要报 PK 文件头错误，而不是静默落盘损坏文件
|
||
try:
|
||
decode_zipped_data({"zippedData": base64.b64encode(b"not a zip").decode()})
|
||
raise AssertionError("非 zip 数据应抛 ValueError")
|
||
except ValueError as e:
|
||
assert "PK" in str(e), e
|
||
|
||
print("selftest OK：去重、文件名反推、zip 解码与完整性校验全部通过")
|
||
|
||
|
||
if __name__ == "__main__":
|
||
ap = argparse.ArgumentParser()
|
||
ap.add_argument("--selftest", action="store_true", help="仅跑去重/命名自测")
|
||
ap.add_argument("--url", help="explore URL")
|
||
ap.add_argument("--user-data-dir", help="浏览器用户数据目录（复用登录态，需先关闭该浏览器）")
|
||
ap.add_argument("--channel", choices=["chrome", "msedge"],
|
||
help="浏览器品牌，复用登录态时必填其一")
|
||
ap.add_argument("--connect", help="通过 CDP 附加到已打开浏览器，如 http://localhost:9222")
|
||
ap.add_argument("--download-dir", help=f"下载目录，默认 {DOWNLOAD_DIR}")
|
||
ap.add_argument("--account-name", help="账号名（用于文件名），默认从页面读取")
|
||
args = ap.parse_args()
|
||
|
||
if args.selftest:
|
||
selftest()
|
||
elif args.url:
|
||
run(args.url, user_data_dir=args.user_data_dir, channel=args.channel,
|
||
cdp_url=args.connect, download_dir=args.download_dir,
|
||
account_name=args.account_name)
|
||
else:
|
||
selftest()
|
||
print("\n实际运行需先准备依赖环境（uv sync 或直接 uv run，见 uv-env-setup 技能），"
|
||
"再复用登录态（二选一）：\n"
|
||
" 方式 A: python youtube_export_download.py --url \"<explore URL>\" "
|
||
"--channel chrome --user-data-dir <你的 Chrome User Data 目录>\n"
|
||
" 方式 B: python youtube_export_download.py --url \"<explore URL>\" "
|
||
"--connect http://localhost:9222") |