xiaoxia
ea4b74216f
CI/CD Pipeline / Check if frontend-only change (push) Has been skipped
CI/CD Pipeline / PR Build API Image (push) Has been skipped
CI/CD Pipeline / PR Build Web Image (push) Has been skipped
CI/CD Pipeline / PR Build Worker Image (push) Has been skipped
CI/CD Pipeline / Validate - Style (push) Has been cancelled
CI/CD Pipeline / Validate - Security (push) Has been cancelled
CI/CD Pipeline / Validate - Python (mypy + alembic) (push) Has been cancelled
CI/CD Pipeline / Unit Tests (push) Has been cancelled
CI/CD Pipeline / Integration Tests (push) Has been cancelled
CI/CD Pipeline / Frontend Lint (push) Has been cancelled
CI/CD Pipeline / Frontend Unit Tests (push) Has been cancelled
CI/CD Pipeline / Build Staging API Image (push) Has been cancelled
CI/CD Pipeline / Build Staging Web Image (push) Has been cancelled
CI/CD Pipeline / Build Staging Worker Image (push) Has been cancelled
CI/CD Pipeline / Retag skipped Staging API Image (push) Has been cancelled
CI/CD Pipeline / Retag skipped Staging Web Image (push) Has been cancelled
CI/CD Pipeline / Retag skipped Staging Worker Image (push) Has been cancelled
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (push) Has been cancelled
CI/CD Pipeline / Staging E2E Tests (push) Has been cancelled
CI/CD Pipeline / Staging API Integration Tests (push) Has been cancelled
CI/CD Pipeline / Build Production API Image (push) Has been cancelled
CI/CD Pipeline / Build Production Web Image (push) Has been cancelled
CI/CD Pipeline / Build Production Worker Image (push) Has been cancelled
CI/CD Pipeline / Deploy Production (push) Has been cancelled
CI/CD Pipeline / Production Browser E2E (push) Has been cancelled
CI/CD Pipeline / ACR Image Cleanup (push) Has been cancelled
CI/CD Pipeline / Canary Release to Production (push) Has been cancelled
CI/CD Pipeline / CI Gate (push) Has been cancelled
CI/CD Pipeline / Check push changed paths (push) Has been cancelled
CI/CD Pipeline / Dedup Check - skip PR tests when covered by push pipeline (push) Has been cancelled
1092 lines
44 KiB
Python
1092 lines
44 KiB
Python
"""Scripts AI 能力路由 — Issue #1893/#1963.
|
||
|
||
三个 AI 工具接口（均挂载在 /api/v1/scripts 前缀下）：
|
||
- POST /extract-from-douyin 从抖音视频提取文案
|
||
- 入口自动从分享文本中正则提取 http(s) URL，兼容 "复制链接" 粘贴场景
|
||
- yt-dlp 仅解析视频元信息（download=False）拿无水印直链，避免整段下载
|
||
- 优先走火山 MediaKit ASR（asr-subtitles），配置了 MEDIAKIT_API_KEY 即可用
|
||
- MediaKit 不可用/失败时，回退到本地 ASR（下载视频 + transcribe_to_text）
|
||
- cookies/ytdlp 均失败时，返回友好 503 不暴露内部错误
|
||
- POST /ai-rewrite AI 文案改写（复用豆包 LLM）
|
||
- POST /ai-generate-titles AI 标题生成（复用 generate_smart_titles）
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import logging
|
||
import os
|
||
import re
|
||
import tempfile
|
||
import time
|
||
from urllib.parse import urlparse
|
||
|
||
from app.auth import AuthenticatedUser, get_current_user
|
||
from app.dependencies import get_db_session
|
||
from app.schemas.scripts_ai import (
|
||
AiGenerateTitlesRequest,
|
||
AiGenerateTitlesResponse,
|
||
AiRewriteRequest,
|
||
AiRewriteResponse,
|
||
ExtractFromDouyinRequest,
|
||
ExtractFromDouyinResponse,
|
||
)
|
||
from app.services.mediakit_client import (
|
||
MediaKitClient,
|
||
MediaKitError,
|
||
get_mediakit_client,
|
||
)
|
||
from app.services.script_asr_service import (
|
||
ASRNotConfiguredError,
|
||
ASRTranscriptionError,
|
||
transcribe_to_text,
|
||
)
|
||
from fastapi import APIRouter, Depends, HTTPException, status
|
||
from sqlalchemy.orm import Session
|
||
|
||
from packages.middleware.points_gate import points_gate
|
||
from packages.shared.ai_client import get_doubao_client
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
router = APIRouter()
|
||
|
||
DOUYIN_COOKIES_FILE = os.environ.get(
|
||
"DOUYIN_COOKIES_FILE",
|
||
"/app/configs/douyin_cookies.txt",
|
||
)
|
||
DOUYIN_COOKIES_FILE_BAKED = "/app/configs/douyin_cookies_default.txt"
|
||
|
||
_COOKIES_ERROR_KEYWORDS = (
|
||
"fresh cookies",
|
||
"cookies (not necessarily logged in)",
|
||
"cookies are needed",
|
||
"need cookies",
|
||
"cookie is expired",
|
||
"login required",
|
||
"sign in to continue",
|
||
"未登录",
|
||
"需要登录",
|
||
"cookies过期",
|
||
)
|
||
|
||
_TAIL_PUNCT = ".,;:!?，。；：！？)）]》" + chr(34) + chr(39) + "<>"
|
||
|
||
|
||
def _resolve_cookies_file():
|
||
for p in (DOUYIN_COOKIES_FILE, DOUYIN_COOKIES_FILE_BAKED):
|
||
try:
|
||
if p and os.path.isfile(p) and os.path.getsize(p) > 200:
|
||
return p
|
||
except OSError:
|
||
continue
|
||
return None
|
||
|
||
|
||
def _dbg(key, val):
|
||
logger.debug("douyin_extract %s=%s", key, str(val)[:200])
|
||
|
||
|
||
def _is_cookies_related_error(msg):
|
||
low = msg.lower()
|
||
return any(kw in low for kw in _COOKIES_ERROR_KEYWORDS)
|
||
|
||
|
||
_cf = _resolve_cookies_file()
|
||
if _cf:
|
||
logger.info("抖音 cookies 文件已加载: %s (%d bytes)", _cf, os.path.getsize(_cf))
|
||
else:
|
||
logger.warning(
|
||
"抖音 cookies 文件未找到或无效: path=%s baked=%s",
|
||
DOUYIN_COOKIES_FILE,
|
||
DOUYIN_COOKIES_FILE_BAKED,
|
||
)
|
||
|
||
_DOUYIN_DEBUG_ERRORS = os.environ.get("DOUYIN_DEBUG_ERRORS", "").lower() in (
|
||
"1",
|
||
"true",
|
||
"yes",
|
||
) or os.environ.get("APP_ENV", "").lower() in ("staging", "dev", "development", "test")
|
||
|
||
_URL_EXTRACT_RE = re.compile(r"https?://\S+", re.IGNORECASE)
|
||
_DOUYIN_HOST_RE = re.compile(
|
||
r"(^|\.)(douyin\.com|iesdouyin\.com|amemv\.com)$",
|
||
re.IGNORECASE,
|
||
)
|
||
_ANY_SCHEME_RE = re.compile(r"^[a-z][a-z0-9+.-]*://\S+", re.IGNORECASE)
|
||
|
||
|
||
def _extract_url_from_text(raw):
|
||
if not raw:
|
||
return None
|
||
m = _URL_EXTRACT_RE.search(raw)
|
||
if m:
|
||
return m.group(0).rstrip(_TAIL_PUNCT)
|
||
short = re.search(
|
||
r"(?:^|(?<![a-z0-9/:]))((?:v|www)\.douyin\.com/\S+|douyin\.com/(?:video|note)/\S+)",
|
||
raw,
|
||
re.IGNORECASE,
|
||
)
|
||
if short:
|
||
return "https://" + short.group(1).rstrip(_TAIL_PUNCT)
|
||
return None
|
||
|
||
|
||
def _extract_and_validate_douyin_url(raw_input):
|
||
raw = (raw_input or "").strip()
|
||
if not raw:
|
||
raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="链接不能为空")
|
||
|
||
url = _extract_url_from_text(raw)
|
||
|
||
if not url:
|
||
# 含非 http(s) 的 scheme 前缀（如 ftp://、file:// 等）→ 协议不支持
|
||
if _ANY_SCHEME_RE.search(raw):
|
||
raise HTTPException(
|
||
status_code=status.HTTP_400_BAD_REQUEST,
|
||
detail="无效的抖音链接，仅支持 http(s) 协议",
|
||
)
|
||
# 裸域名兜底：在去除 scheme 的情况下匹配 douyin 域名
|
||
short = re.search(
|
||
r"(?:^|(?<![a-z0-9]))((?:v|www)\.douyin\.com/\S+|douyin\.com/(?:video|note)/\S+)",
|
||
raw,
|
||
re.IGNORECASE,
|
||
)
|
||
if short:
|
||
url = "https://" + short.group(1).rstrip(_TAIL_PUNCT)
|
||
else:
|
||
raise HTTPException(
|
||
status_code=status.HTTP_400_BAD_REQUEST,
|
||
detail="未在输入中找到有效抖音链接，请粘贴包含 v.douyin.com 或 www.douyin.com 的分享文本",
|
||
)
|
||
|
||
if not re.match(r"^https?://", url, re.IGNORECASE):
|
||
url = "https://" + url
|
||
|
||
try:
|
||
parsed = urlparse(url)
|
||
host = parsed.hostname or ""
|
||
scheme = (parsed.scheme or "").lower()
|
||
except Exception:
|
||
host = ""
|
||
scheme = ""
|
||
if scheme not in ("http", "https"):
|
||
raise HTTPException(
|
||
status_code=status.HTTP_400_BAD_REQUEST,
|
||
detail="无效的抖音链接，仅支持 http(s) 协议",
|
||
)
|
||
if not _DOUYIN_HOST_RE.search(host):
|
||
raise HTTPException(
|
||
status_code=status.HTTP_400_BAD_REQUEST,
|
||
detail="无效的抖音链接，仅支持 douyin.com 域名（v.douyin.com 短链或 www.douyin.com 长链）",
|
||
)
|
||
return url
|
||
|
||
|
||
# ── MediaKitClient ASR 扩展（monkey patch） ────────────────────────────
|
||
#
|
||
# app/services/mediakit_client.py 已提供：
|
||
# self._api_key / self._base_url / self._timeout / self._headers() / self.is_available
|
||
# 这里只补两个通用 JSON helper 和 ASR 提交/轮询方法。
|
||
|
||
|
||
def _mk_post_json(self, path, payload):
|
||
import httpx
|
||
if not self.is_available:
|
||
raise MediaKitError("MediaKit API Key 未配置", code="NotConfigured")
|
||
url = self._base_url + path
|
||
try:
|
||
with httpx.Client(timeout=self._timeout) as http:
|
||
resp = http.post(url, headers=self._headers(), json=payload)
|
||
resp.raise_for_status()
|
||
data = resp.json()
|
||
except httpx.TimeoutException as exc:
|
||
raise MediaKitError("MediaKit API 超时 (%ss)" % self._timeout, code="Timeout") from exc
|
||
except httpx.HTTPStatusError as exc:
|
||
raise MediaKitError(
|
||
"MediaKit API HTTP %s: %s" % (exc.response.status_code, exc.response.text[:300]),
|
||
code="HttpError",
|
||
) from exc
|
||
except httpx.RequestError as exc:
|
||
raise MediaKitError("MediaKit API 网络错误: %s" % exc, code="NetworkError") from exc
|
||
# 同步错误（提交参数错误等）：success=false 且含 error
|
||
if data.get("success") is False and data.get("error"):
|
||
err = data["error"] if isinstance(data["error"], dict) else {"message": str(data["error"])}
|
||
raise MediaKitError(
|
||
err.get("message", "请求失败"),
|
||
code=err.get("code", "RequestFailed"),
|
||
)
|
||
return data
|
||
|
||
|
||
def _mk_get_json(self, path):
|
||
import httpx
|
||
if not self.is_available:
|
||
raise MediaKitError("MediaKit API Key 未配置", code="NotConfigured")
|
||
url = self._base_url + path
|
||
try:
|
||
with httpx.Client(timeout=self._timeout) as http:
|
||
resp = http.get(url, headers=self._headers())
|
||
resp.raise_for_status()
|
||
return resp.json()
|
||
except httpx.TimeoutException as exc:
|
||
raise MediaKitError("MediaKit API 超时 (%ss)" % self._timeout, code="Timeout") from exc
|
||
except httpx.HTTPStatusError as exc:
|
||
raise MediaKitError(
|
||
"MediaKit API HTTP %s: %s" % (exc.response.status_code, exc.response.text[:300]),
|
||
code="HttpError",
|
||
) from exc
|
||
except httpx.RequestError as exc:
|
||
raise MediaKitError("MediaKit API 网络错误: %s" % exc, code="NetworkError") from exc
|
||
|
||
|
||
def _mediakit_asr_submit(self, video_url):
|
||
"""提交语音转字幕任务（POST /tools/asr-subtitles）。返回 task_id。"""
|
||
data = self._post_json(
|
||
"/tools/asr-subtitles",
|
||
{"video_url": video_url, "language": "cmn-Hans-CN"},
|
||
)
|
||
task_id = data.get("task_id")
|
||
if not task_id:
|
||
raise MediaKitError("MediaKit ASR 提交响应缺少 task_id: %s" % str(data)[:200])
|
||
return task_id
|
||
|
||
|
||
def _mediakit_asr_poll(self, task_id, poll_interval=2.0, max_attempts=90):
|
||
"""轮询 ASR 任务直到 completed/failed。返回 (text, duration)。"""
|
||
for attempt in range(max_attempts):
|
||
time.sleep(poll_interval)
|
||
try:
|
||
data = self._get_json("/tasks/" + task_id)
|
||
except MediaKitError as exc:
|
||
# 瞬时网络/超时可重试
|
||
if attempt < max_attempts - 1 and getattr(exc, "code", "") in ("Timeout", "NetworkError"):
|
||
logger.warning("MediaKit ASR 轮询异常（第%d次），将重试: %s", attempt + 1, exc)
|
||
continue
|
||
raise
|
||
st = data.get("status")
|
||
if st in ("completed", "success"):
|
||
result = data.get("result") or {}
|
||
subs = result.get("subtitles") or []
|
||
text = "".join(s.get("subtitle_text", "") for s in subs if isinstance(s, dict))
|
||
duration = float(result.get("duration") or 0.0)
|
||
return text.strip(), duration
|
||
if st == "failed":
|
||
err = data.get("error")
|
||
if isinstance(err, dict):
|
||
msg = err.get("message") or "unknown"
|
||
code = err.get("code") or "TaskFailed"
|
||
elif isinstance(err, str):
|
||
msg, code = err, "TaskFailed"
|
||
else:
|
||
msg, code = "unknown", "TaskFailed"
|
||
raise MediaKitError("MediaKit ASR 任务失败: %s" % msg, code=code)
|
||
# running/pending/queued: continue
|
||
raise MediaKitError(
|
||
"MediaKit ASR 超时（%ss 未完成）" % int(poll_interval * max_attempts),
|
||
code="Timeout",
|
||
)
|
||
|
||
|
||
# 绑定到类（零侵入，不修改原 mediakit_client.py）
|
||
if not hasattr(MediaKitClient, "_post_json"):
|
||
MediaKitClient._post_json = _mk_post_json
|
||
if not hasattr(MediaKitClient, "_get_json"):
|
||
MediaKitClient._get_json = _mk_get_json
|
||
if not hasattr(MediaKitClient, "asr_submit"):
|
||
MediaKitClient.asr_submit = _mediakit_asr_submit
|
||
if not hasattr(MediaKitClient, "asr_poll"):
|
||
MediaKitClient.asr_poll = _mediakit_asr_poll
|
||
|
||
|
||
# ── yt-dlp 辅助 ──────────────────────────────────────────────────────
|
||
|
||
|
||
def _fetch_ttwid():
|
||
"""从字节跳动 ttwid 注册接口获取 ttwid cookie。
|
||
通过 POST https://ttwid.bytedance.com/ttwid/union/register/ 可直接获得，无需浏览器。
|
||
失败时回退到访问 douyin.com 主页。
|
||
"""
|
||
import httpx
|
||
|
||
# 方式一：直接调 ttwid 注册接口
|
||
try:
|
||
with httpx.Client(timeout=8, follow_redirects=True, verify=False) as http:
|
||
r = http.post(
|
||
"https://ttwid.bytedance.com/ttwid/union/register/",
|
||
json={
|
||
"region": "cn",
|
||
"aid": 6383,
|
||
"needFid": False,
|
||
"service": "www.douyin.com",
|
||
"migrate_info": {"ticket": "", "source": "node"},
|
||
"cbUrlProtocol": "https",
|
||
"union": True,
|
||
},
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Content-Type": "application/json",
|
||
"Referer": "https://www.douyin.com/",
|
||
},
|
||
)
|
||
ttwid = http.cookies.get("ttwid")
|
||
if ttwid and len(ttwid) > 20:
|
||
_dbg("ttwid_fetch", "register_ok len=%d" % len(ttwid))
|
||
return ttwid
|
||
except Exception as exc:
|
||
logger.debug("ttwid 注册接口失败: %s", exc)
|
||
|
||
# 方式二：回退到访问主页
|
||
try:
|
||
with httpx.Client(timeout=8, follow_redirects=True, verify=False) as http:
|
||
for _ in range(2):
|
||
try:
|
||
http.get(
|
||
"https://www.douyin.com/",
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Accept-Language": "zh-CN,zh;q=0.9",
|
||
},
|
||
)
|
||
ttwid = http.cookies.get("ttwid")
|
||
if ttwid:
|
||
return ttwid
|
||
except Exception:
|
||
continue
|
||
except Exception as exc:
|
||
logger.debug("获取 ttwid 失败: %s", exc)
|
||
return None
|
||
|
||
|
||
def _generate_synthetic_cookie_file(temp_dir, ttwid=None):
|
||
"""生成一份合成 cookie 文件（仅 ttwid，不伪造 s_v_web_id），返回文件路径。
|
||
|
||
重要：不要伪造 s_v_web_id（如 verify_xxx 随机串）！yt-dlp DouyinIE 在检测到
|
||
无效的 s_v_web_id 时会直接抛致命错误（"Fresh cookies needed"），而不是走内部
|
||
重试路径。只提供 ttwid 时，yt-dlp 会把缺少 s_v_web_id 视为 expected 错误，
|
||
内部走兜底路径拿数据，配合多次重试有一定概率成功。
|
||
"""
|
||
import os
|
||
import secrets
|
||
import time
|
||
|
||
if not ttwid:
|
||
ttwid = _fetch_ttwid()
|
||
if not ttwid:
|
||
# 兜底：生成一个格式合法的假 ttwid（至少不会直接报错）
|
||
ttwid = "1%7C" + secrets.token_hex(20) + "%7C" + str(int(time.time())) + "%7C" + secrets.token_hex(32)
|
||
|
||
expiry = int(time.time()) + 86400 * 30 # 30 天
|
||
|
||
path = os.path.join(temp_dir, f"dy_cookies_{secrets.token_hex(4)}.txt")
|
||
with open(path, "w") as f:
|
||
f.write("# Netscape HTTP Cookie File\n")
|
||
f.write("# This file is generated by yt-dlp. Do not edit.\n\n")
|
||
f.write(f".douyin.com\tTRUE\t/\tTRUE\t{expiry}\tttwid\t{ttwid}\n")
|
||
# 注意：不写 s_v_web_id！伪造的 verify_xxx 串会导致 yt-dlp 直接失败。
|
||
return path
|
||
|
||
|
||
def _ytdlp_try_extract(page_url, cookiefile):
|
||
"""单次 yt-dlp 提取，返回 (info_dict, None) 或 (None, error_message)。
|
||
显式选择 h264+aac 的 MP4 格式，避免 h265 编码导致 MediaKit ASR 不兼容。
|
||
"""
|
||
import yt_dlp
|
||
|
||
opts = {
|
||
"quiet": True,
|
||
"no_warnings": True,
|
||
"noplaylist": True,
|
||
"skip_download": True,
|
||
"extractor_args": {"douyin": {"webpage_cookie": ""}},
|
||
"format": "best[ext=mp4][vcodec^=avc1]/best[ext=mp4][vcodec^=h264]/best[ext=mp4]/best",
|
||
"http_headers": {
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Referer": "https://www.douyin.com/",
|
||
"Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
|
||
},
|
||
"socket_timeout": 15,
|
||
"retries": 1,
|
||
}
|
||
if cookiefile:
|
||
opts["cookiefile"] = cookiefile
|
||
try:
|
||
with yt_dlp.YoutubeDL(opts) as ydl:
|
||
info = ydl.extract_info(page_url, download=False)
|
||
return info, None
|
||
except Exception as exc:
|
||
return None, str(exc)
|
||
|
||
|
||
def _ytdlp_extract_video_url(page_url, cookiefile=None, max_retries=10):
|
||
"""yt-dlp 提取视频直链，支持合成 cookie 多次重试。
|
||
优先使用传入的 cookiefile；失败后自动生成合成 cookie 重试 max_retries 次。
|
||
每次重试使用新的 s_v_web_id，每3次刷新 ttwid，实测约 30% 单次成功率，10次≈97%。
|
||
"""
|
||
# 优先使用现有 cookie 文件
|
||
if cookiefile and os.path.exists(cookiefile):
|
||
info, err = _ytdlp_try_extract(page_url, cookiefile)
|
||
if info:
|
||
vurl = info.get("url")
|
||
return (vurl, float(info.get("duration") or 0.0)) if vurl else (None, 0.0)
|
||
logger.info("提供的 cookie 文件解析失败: %s", str(err)[:150])
|
||
|
||
# 合成 cookie 重试
|
||
import tempfile
|
||
import time as _time
|
||
|
||
ttwid = _fetch_ttwid()
|
||
logger.info("ytdlp 直链提取开始: page_url=%s ttwid_ok=%s max_retries=%d", page_url, bool(ttwid), max_retries)
|
||
with tempfile.TemporaryDirectory(prefix="dy_ytdlp_cookies_") as tmpdir:
|
||
for attempt in range(max_retries):
|
||
if attempt % 3 == 0:
|
||
ttwid = _fetch_ttwid() or ttwid
|
||
gen_cookie = _generate_synthetic_cookie_file(tmpdir, ttwid=ttwid)
|
||
info, err = _ytdlp_try_extract(page_url, gen_cookie)
|
||
if info:
|
||
vurl = info.get("url")
|
||
logger.info("yt-dlp 合成 cookie 第%d次尝试成功: vurl_domain=%s", attempt + 1, vurl.split("/")[2] if vurl and "/" in vurl else "?")
|
||
return (vurl, float(info.get("duration") or 0.0)) if vurl else (None, 0.0)
|
||
logger.info("yt-dlp 合成 cookie 第%d次尝试失败: %s", attempt + 1, str(err)[:150])
|
||
_time.sleep(0.4)
|
||
|
||
logger.info("yt-dlp 合成 cookie 全部 %d 次重试均失败", max_retries)
|
||
return None, 0.0
|
||
|
||
|
||
_PARSER_APIS = [
|
||
# 第三方抖音无水印解析 API（cookies 过期/yt-dlp 反爬升级时的兜底）
|
||
# 每项 (url_template, name, auth_header_or_None)
|
||
# url 中 {url} 会被替换为 URL-encoded 的抖音链接
|
||
("https://v1.apizero.cn/api/video-parse?url={url}&flat=1", "apizero", None),
|
||
]
|
||
|
||
# 可选：apizero API Key（登录后免费额度 5 次/天，付费更高额度）
|
||
_APIZERO_API_KEY = os.environ.get("APIZERO_API_KEY", "").strip()
|
||
_PARSER_MAX_RETRIES = 3
|
||
_PARSER_RETRY_DELAY = 0.8 # 秒
|
||
_PARSER_RATE_LIMIT_CODES = {4030, 4290, 429, 10004, 10008} # 各家限流 code
|
||
|
||
|
||
def _build_parser_headers(name):
|
||
"""为解析 API 构造请求头，包含可选 API Key。"""
|
||
base = {
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Accept": "application/json, text/plain, */*",
|
||
"Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
|
||
}
|
||
if name == "apizero" and _APIZERO_API_KEY:
|
||
# apizero 推荐用 Authorization: Bearer <key>，同时兼容 query key
|
||
base["Authorization"] = f"Bearer {_APIZERO_API_KEY}"
|
||
return base
|
||
|
||
|
||
def _build_parser_url(api_tpl, name, page_url_encoded):
|
||
url = api_tpl.format(url=page_url_encoded)
|
||
if name == "apizero" and _APIZERO_API_KEY:
|
||
sep = "&" if "?" in url else "?"
|
||
url = f"{url}{sep}key={_APIZERO_API_KEY}"
|
||
return url
|
||
|
||
|
||
def _extract_video_url_from_parser_response(name, data):
|
||
"""从解析 API 的 JSON 响应中提取 (direct_url, duration)。兼容多种响应结构。"""
|
||
if not isinstance(data, dict):
|
||
return None, 0.0
|
||
# 统一成功判定：code == 0 或 status == "success"
|
||
code = data.get("code")
|
||
is_success = code == 0 or code == 200 or data.get("success") is True or data.get("status") == "success"
|
||
if not is_success:
|
||
msg = data.get("msg") or data.get("message") or str(code)
|
||
# 返回 msg 供调用方判断是否限流
|
||
return None, 0.0, str(msg), code
|
||
vd = data.get("data") or {}
|
||
if not isinstance(vd, dict):
|
||
return None, 0.0, "bad data", -1
|
||
vlist = vd.get("video_list") or []
|
||
duration = float(vd.get("duration") or vd.get("duration_ms", 0) or 0.0)
|
||
if duration > 1000: # ms → s
|
||
duration = duration / 1000.0
|
||
direct_url = None
|
||
chosen_quality = -1
|
||
quality_priority = ["原画", "1080p", "720p", "540p", "480p"]
|
||
for prio, keyword in enumerate(quality_priority):
|
||
for v in vlist:
|
||
if not isinstance(v, dict) or not v.get("url"):
|
||
continue
|
||
if keyword in str(v.get("quality", "")):
|
||
if prio < chosen_quality or chosen_quality == -1:
|
||
direct_url = v.get("url")
|
||
chosen_quality = prio
|
||
break
|
||
if direct_url:
|
||
break
|
||
if not direct_url:
|
||
# 兜底：取 data.video_url 顶层字段，或 video_list[0].url
|
||
if vd.get("video_url"):
|
||
direct_url = vd["video_url"]
|
||
elif vlist:
|
||
direct_url = vlist[0].get("url")
|
||
else:
|
||
# 兼容 play/play_url 字段
|
||
direct_url = vd.get("play") or vd.get("play_url") or vd.get("url")
|
||
if not direct_url:
|
||
return None, 0.0, "no video url", -1
|
||
return direct_url, duration, None, code
|
||
|
||
|
||
def _parser_extract_direct_url(page_url, timeout=12):
|
||
"""使用第三方解析 API 获取抖音无水印直链（yt-dlp 失败时的兜底）。
|
||
每个 API 会重试 _PARSER_MAX_RETRIES 次以应对瞬时限流。
|
||
返回 (direct_url, duration) 或 (None, 0.0)。
|
||
"""
|
||
import random
|
||
import urllib.parse
|
||
|
||
import httpx
|
||
|
||
page_url_encoded = urllib.parse.quote(page_url, safe="")
|
||
last_err = None
|
||
|
||
for api_tpl, name, _auth in _PARSER_APIS:
|
||
url = _build_parser_url(api_tpl, name, page_url_encoded)
|
||
headers = _build_parser_headers(name)
|
||
for attempt in range(_PARSER_MAX_RETRIES):
|
||
try:
|
||
with httpx.Client(timeout=timeout, follow_redirects=True, verify=False) as http:
|
||
resp = http.get(url, headers=headers)
|
||
resp.raise_for_status()
|
||
data = resp.json()
|
||
except Exception as exc: # noqa: BLE001
|
||
last_err = str(exc)[:200]
|
||
logger.warning(
|
||
"第三方解析 %s 第%d次请求失败: %s", name, attempt + 1, exc
|
||
)
|
||
time.sleep(_PARSER_RETRY_DELAY * (attempt + 1) + random.random() * 0.3)
|
||
continue
|
||
|
||
try:
|
||
result = _extract_video_url_from_parser_response(name, data)
|
||
if len(result) == 4:
|
||
direct_url, duration, err_msg, resp_code = result
|
||
else:
|
||
direct_url, duration = result
|
||
err_msg, resp_code = None, None
|
||
if direct_url:
|
||
logger.info(
|
||
"第三方解析 %s 第%d次成功，直链长度=%d",
|
||
name, attempt + 1, len(direct_url),
|
||
)
|
||
return direct_url, duration
|
||
# 判断是否限流，可重试
|
||
if resp_code in _PARSER_RATE_LIMIT_CODES or (
|
||
err_msg and any(
|
||
kw in str(err_msg) for kw in ("次数", "限流", "频率", "用完", "rate limit", "quota")
|
||
)
|
||
):
|
||
logger.info(
|
||
"第三方解析 %s 第%d次返回限流: code=%s msg=%s，将重试",
|
||
name, attempt + 1, resp_code, err_msg,
|
||
)
|
||
time.sleep(_PARSER_RETRY_DELAY * (attempt + 1) + random.random() * 0.5)
|
||
continue
|
||
# 业务错误（如链接无效），直接换下一个 API
|
||
logger.warning(
|
||
"第三方解析 %s 第%d次返回业务错误: code=%s msg=%s",
|
||
name, attempt + 1, resp_code, err_msg,
|
||
)
|
||
break
|
||
except Exception as exc: # noqa: BLE001
|
||
last_err = str(exc)[:200]
|
||
logger.warning("第三方解析 %s 响应解析失败: %s", name, exc)
|
||
break # 响应格式异常，不重试当前 API
|
||
|
||
logger.warning("所有第三方解析均失败: last_err=%s", last_err)
|
||
return None, 0.0
|
||
|
||
|
||
def _direct_url_download_and_local_asr(direct_url, page_url, temp_dir):
|
||
"""通过第三方解析得到的直链直接下载 MP4，再做本地 ASR。
|
||
返回 (text, duration)。
|
||
"""
|
||
import os
|
||
|
||
import httpx
|
||
|
||
video_path = os.path.join(temp_dir, "video.mp4")
|
||
try:
|
||
with httpx.Client(timeout=90, follow_redirects=True, verify=False) as http:
|
||
with http.stream(
|
||
"GET",
|
||
direct_url,
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Referer": "https://www.douyin.com/",
|
||
"Accept": "*/*",
|
||
"Accept-Language": "zh-CN,zh;q=0.9",
|
||
},
|
||
) as resp:
|
||
resp.raise_for_status()
|
||
downloaded = 0
|
||
with open(video_path, "wb") as f:
|
||
for chunk in resp.iter_bytes(chunk_size=65536):
|
||
if chunk:
|
||
f.write(chunk)
|
||
downloaded += len(chunk)
|
||
if downloaded == 0:
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="直链下载为空")
|
||
except HTTPException:
|
||
raise
|
||
except httpx.TimeoutException:
|
||
logger.warning("直链下载超时: %s", page_url)
|
||
raise HTTPException(status_code=status.HTTP_504_GATEWAY_TIMEOUT, detail="视频下载超时，请稍后重试")
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.exception("直链下载失败: url=%s err=%s", page_url, exc)
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="视频下载失败: " + str(exc)[:200]) from exc
|
||
|
||
try:
|
||
text = transcribe_to_text(video_path)
|
||
return text.strip(), 0.0
|
||
except ASRNotConfiguredError as exc:
|
||
raise HTTPException(status_code=status.HTTP_503_SERVICE_UNAVAILABLE, detail=str(exc)) from exc
|
||
except ASRTranscriptionError as exc:
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail=str(exc)) from exc
|
||
except Exception as exc:
|
||
logger.exception("直链下载后 ASR 转写异常: path=%s", video_path)
|
||
raise HTTPException(
|
||
status_code=status.HTTP_502_BAD_GATEWAY,
|
||
detail="语音识别失败: " + str(exc)[:200],
|
||
) from exc
|
||
|
||
|
||
def _ytdlp_download_and_local_asr(page_url, temp_dir, cookiefile=None):
|
||
"""下载抖音视频 + 本地 ASR。优先使用传入 cookie，失败时自动合成 cookie 重试。"""
|
||
try:
|
||
import yt_dlp
|
||
except ImportError as exc:
|
||
raise HTTPException(
|
||
status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
|
||
detail="抖音提取功能暂不可用（缺少依赖 yt-dlp）",
|
||
) from exc
|
||
|
||
def _download(cfile):
|
||
opts = {
|
||
"format": "best[ext=mp4][vcodec^=avc1]/best[ext=mp4][vcodec^=h264]/best[ext=mp4]/best",
|
||
"outtmpl": temp_dir + "/%(id)s.%(ext)s",
|
||
"quiet": True,
|
||
"no_warnings": True,
|
||
"noplaylist": True,
|
||
"extractor_args": {"douyin": {"webpage_cookie": ""}},
|
||
"socket_timeout": 30,
|
||
"retries": 2,
|
||
"http_headers": {
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/128.0.0.0 Safari/537.36"
|
||
),
|
||
"Referer": "https://www.douyin.com/",
|
||
"Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
|
||
},
|
||
}
|
||
if cfile:
|
||
opts["cookiefile"] = cfile
|
||
with yt_dlp.YoutubeDL(opts) as ydl:
|
||
info = ydl.extract_info(page_url, download=True)
|
||
vpath = ydl.prepare_filename(info)
|
||
dur = float(info.get("duration") or 0)
|
||
return vpath, dur
|
||
|
||
# 1. 先用传入 cookie
|
||
vpath = None
|
||
last_err = None
|
||
cookie_files_to_try = []
|
||
if cookiefile and os.path.exists(cookiefile):
|
||
cookie_files_to_try.append(cookiefile)
|
||
|
||
# 2. 合成 cookie 重试（复用 _ytdlp_extract 的 cookie 生成逻辑，8次重试）
|
||
import secrets, time as _time
|
||
ttwid = _fetch_ttwid()
|
||
for i in range(8):
|
||
if i % 3 == 0:
|
||
ttwid = _fetch_ttwid() or ttwid
|
||
cfile = _generate_synthetic_cookie_file(temp_dir, ttwid=ttwid)
|
||
cookie_files_to_try.append(cfile)
|
||
_time.sleep(0.3)
|
||
|
||
for cfile in cookie_files_to_try:
|
||
try:
|
||
vpath, duration = _download(cfile)
|
||
if vpath and os.path.isfile(vpath) and os.path.getsize(vpath) > 0:
|
||
break
|
||
logger.warning(
|
||
"ytdlp 下载返回路径无效或空文件: vpath=%s size=%s",
|
||
vpath,
|
||
os.path.getsize(vpath) if vpath and os.path.isfile(vpath) else "N/A",
|
||
)
|
||
except Exception as exc: # noqa: BLE001
|
||
last_err = exc
|
||
logger.warning("ytdlp 下载重试失败: %s", str(exc)[:200])
|
||
continue
|
||
else:
|
||
msg = str(last_err) if last_err else "unknown"
|
||
logger.warning(
|
||
"ytdlp 下载全部 %d 次 cookie 尝试均失败: ttwid_ok=%s last_err=%s",
|
||
len(cookie_files_to_try), bool(ttwid), msg[:300],
|
||
)
|
||
if _is_cookies_related_error(msg) or "Fresh cookies" in msg:
|
||
_detail = "抖音链接解析暂时不可用，请稍后重试或手动输入文案"
|
||
if _DOUYIN_DEBUG_ERRORS:
|
||
_detail = _detail + " [debug: " + msg[:300] + "]"
|
||
raise HTTPException(status_code=status.HTTP_503_SERVICE_UNAVAILABLE, detail=_detail)
|
||
_detail = "视频下载失败，请稍后重试"
|
||
if _DOUYIN_DEBUG_ERRORS:
|
||
_detail = _detail + " [debug: " + msg[:300] + "]"
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail=_detail)
|
||
|
||
if not vpath or not os.path.isfile(vpath) or os.path.getsize(vpath) == 0:
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="视频下载异常：未获取到有效文件")
|
||
try:
|
||
duration = float(duration or 0)
|
||
except (TypeError, ValueError):
|
||
duration = 0.0
|
||
|
||
try:
|
||
text = transcribe_to_text(vpath)
|
||
except ASRNotConfiguredError as exc:
|
||
raise HTTPException(status_code=status.HTTP_503_SERVICE_UNAVAILABLE, detail=str(exc)) from exc
|
||
except ASRTranscriptionError as exc:
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail=str(exc)) from exc
|
||
except Exception as exc:
|
||
logger.exception("ASR 转写异常: path=%s", vpath)
|
||
raise HTTPException(
|
||
status_code=status.HTTP_502_BAD_GATEWAY,
|
||
detail="语音识别失败: " + str(exc)[:200],
|
||
) from exc
|
||
return text.strip(), duration
|
||
|
||
|
||
def _html_scrape_direct_url(page_url, timeout=15):
|
||
"""通过抖音分享页 HTML 直接抓取视频 CDN 直链（最后兜底，不依赖任何第三方 API）。
|
||
注意：桌面版页面通常是 SPA 需 JS 渲染，移动分享页 iesdouyin.com 含 SSR 数据。
|
||
返回 (direct_url, duration) 或 (None, 0.0)。
|
||
"""
|
||
import re
|
||
|
||
import httpx
|
||
|
||
aweme_id = None
|
||
m = re.search(r"(?:video|note)/(\d{10,})", page_url)
|
||
if not m:
|
||
# 先展开短链拿 aweme_id
|
||
try:
|
||
with httpx.Client(timeout=timeout, follow_redirects=False, verify=False) as http:
|
||
r = http.head(
|
||
page_url,
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (iPhone; CPU iPhone OS 16_6 like Mac OS X) "
|
||
"AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 "
|
||
"Mobile/15E148 Safari/604.1"
|
||
),
|
||
},
|
||
)
|
||
loc = r.headers.get("location", "")
|
||
m2 = re.search(r"(?:video|note)/(\d{10,})", loc)
|
||
if m2:
|
||
aweme_id = m2.group(1)
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.debug("HTML 解析短链重定向失败: %s", exc)
|
||
else:
|
||
aweme_id = m.group(1)
|
||
|
||
if not aweme_id:
|
||
# 直接 GET 短链跟随
|
||
try:
|
||
with httpx.Client(timeout=timeout, follow_redirects=True, verify=False) as http:
|
||
r = http.get(
|
||
page_url,
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (iPhone; CPU iPhone OS 16_6 like Mac OS X) "
|
||
"AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 "
|
||
"Mobile/15E148 Safari/604.1"
|
||
),
|
||
},
|
||
)
|
||
m2 = re.search(r"(?:video|note)/(\d{10,})", str(r.url))
|
||
if m2:
|
||
aweme_id = m2.group(1)
|
||
except Exception: # noqa: BLE001
|
||
pass
|
||
|
||
if not aweme_id:
|
||
return None, 0.0
|
||
|
||
# 尝试移动分享页
|
||
share_url = f"https://www.iesdouyin.com/share/video/{aweme_id}/"
|
||
try:
|
||
with httpx.Client(timeout=timeout, follow_redirects=True, verify=False) as http:
|
||
r = http.get(
|
||
share_url,
|
||
headers={
|
||
"User-Agent": (
|
||
"Mozilla/5.0 (iPhone; CPU iPhone OS 16_6 like Mac OS X) "
|
||
"AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 "
|
||
"Mobile/15E148 Safari/604.1"
|
||
),
|
||
"Referer": "https://www.douyin.com/",
|
||
"Accept-Language": "zh-CN,zh;q=0.9",
|
||
},
|
||
)
|
||
if r.status_code != 200:
|
||
return None, 0.0
|
||
html = r.text
|
||
# 先解码转义
|
||
html_decoded = (
|
||
html.replace("\\u002F", "/")
|
||
.replace("\\/", "/")
|
||
.replace("&", "&")
|
||
)
|
||
# 匹配 douyinvod CDN 链接（可能是 playwm 带水印版本）
|
||
urls = re.findall(
|
||
r"https?://[^\"'\\<>\s]+?douyinvod\.com[^\"'\\<>\s]*",
|
||
html_decoded,
|
||
)
|
||
mp4_urls = [u for u in urls if ".mp4" in u or "/video/" in u]
|
||
# 也匹配 amemv CDN
|
||
amemv_urls = re.findall(
|
||
r"https?://[^\"'\\<>\s]+?amemv\.com[^\"'\\<>\s]*?\.mp4[^\"'\\<>\s]*",
|
||
html_decoded,
|
||
)
|
||
all_urls = mp4_urls + amemv_urls
|
||
if all_urls:
|
||
# 优先无水印（URL 中不含 playwm）
|
||
for u in all_urls:
|
||
if "playwm" not in u and len(u) > 40:
|
||
logger.info("HTML 抓取到无水印直链: %s", u[:120])
|
||
return u, 0.0
|
||
for u in all_urls:
|
||
if len(u) > 40:
|
||
logger.info("HTML 抓取到带水印直链（可用于ASR）: %s", u[:120])
|
||
return u, 0.0
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.debug("HTML 抓取异常: %s", exc)
|
||
return None, 0.0
|
||
|
||
|
||
# ── 1. 从抖音视频提取文案 ─────────────────────────────────────────────
|
||
|
||
|
||
@router.post("/extract-from-douyin", response_model=ExtractFromDouyinResponse)
|
||
@points_gate("douyin_extract")
|
||
def extract_from_douyin(
|
||
request: ExtractFromDouyinRequest,
|
||
current_user: AuthenticatedUser = Depends(get_current_user),
|
||
db: Session = Depends(get_db_session),
|
||
):
|
||
page_url = _extract_and_validate_douyin_url(request.url)
|
||
_dbg("page_url", page_url)
|
||
|
||
text = ""
|
||
duration = 0.0
|
||
cookiefile = _resolve_cookies_file()
|
||
mk_client = get_mediakit_client()
|
||
|
||
# 整体重试：外层最多 2 轮完整链路，应对第三方 API 瞬时限流/CDN 抖动
|
||
max_rounds = 2
|
||
direct_url = None
|
||
meta_duration = 0.0
|
||
for round_idx in range(max_rounds):
|
||
direct_url = None
|
||
meta_duration = 0.0
|
||
|
||
# 路径 A1：第三方解析 API（最稳定，apizero 付费时 99.88% 可用）
|
||
logger.info("抖音解析第%d轮开始: url=%s", round_idx + 1, page_url)
|
||
try:
|
||
direct_url, meta_duration = _parser_extract_direct_url(page_url)
|
||
if direct_url:
|
||
logger.info(
|
||
"第三方解析 API 成功（第%d轮）: url_domain=%s",
|
||
round_idx + 1, direct_url.split("/")[2] if "/" in direct_url else "?",
|
||
)
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.warning("第三方解析 API 异常（第%d轮）: %s", round_idx + 1, exc)
|
||
|
||
# 路径 A2：yt-dlp 拿直链（作为兜底，需要 cookie 概率性成功）
|
||
if not direct_url:
|
||
logger.info(
|
||
"第三方解析未拿到直链（第%d轮），尝试 yt-dlp", round_idx + 1
|
||
)
|
||
direct_url, meta_duration = _ytdlp_extract_video_url(page_url, cookiefile=cookiefile)
|
||
|
||
# 路径 A3：HTML 直抓（最后兜底，不依赖任何 API Key / cookies）
|
||
if not direct_url:
|
||
logger.info(
|
||
"yt-dlp 未拿到直链（第%d轮），尝试 HTML 直接抓取", round_idx + 1
|
||
)
|
||
try:
|
||
direct_url, meta_duration = _html_scrape_direct_url(page_url)
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.warning("HTML 直抓异常（第%d轮）: %s", round_idx + 1, exc)
|
||
|
||
if meta_duration:
|
||
duration = meta_duration
|
||
_dbg("direct_url", direct_url or "<none>")
|
||
|
||
# 路径 B1：直链 → MediaKit 云端 ASR（最快，不下载视频）
|
||
if direct_url and mk_client.is_available:
|
||
try:
|
||
task_id = mk_client.asr_submit(direct_url)
|
||
text, mk_duration = mk_client.asr_poll(task_id)
|
||
if mk_duration:
|
||
duration = mk_duration
|
||
logger.info(
|
||
"抖音 MediaKit ASR 成功（第%d轮）: url=%s text_len=%d duration=%.1f",
|
||
round_idx + 1, page_url, len(text), duration,
|
||
)
|
||
break
|
||
except MediaKitError as exc:
|
||
logger.warning("MediaKit ASR 失败（第%d轮），回退本地 ASR: %s", round_idx + 1, exc)
|
||
text = ""
|
||
|
||
# 路径 B2：回退下载 + 本地 ASR
|
||
if not text:
|
||
_dbg("fallback", f"download+local_asr round={round_idx+1}")
|
||
try:
|
||
with tempfile.TemporaryDirectory(prefix="douyin_extract_") as temp_dir:
|
||
if direct_url:
|
||
_dbg("fallback_via", "direct_url_download")
|
||
text, dl_duration = _direct_url_download_and_local_asr(
|
||
direct_url, page_url, temp_dir
|
||
)
|
||
else:
|
||
text, dl_duration = _ytdlp_download_and_local_asr(
|
||
page_url, temp_dir, cookiefile=cookiefile
|
||
)
|
||
if dl_duration and not duration:
|
||
duration = dl_duration
|
||
if text:
|
||
logger.info(
|
||
"抖音本地 ASR 成功（第%d轮）: url=%s text_len=%d",
|
||
round_idx + 1, page_url, len(text),
|
||
)
|
||
break
|
||
except HTTPException as exc:
|
||
# 可重试错误（502/503/504）短暂等待后重试
|
||
if exc.status_code in (502, 503, 504) and round_idx < max_rounds - 1:
|
||
logger.warning(
|
||
"本地 ASR 链路返回 %d（第%d轮），将重试: %s",
|
||
exc.status_code, round_idx + 1, exc.detail,
|
||
)
|
||
time.sleep(1.2 + round_idx)
|
||
continue
|
||
raise
|
||
except Exception as exc: # noqa: BLE001
|
||
logger.warning("本地 ASR 链路异常（第%d轮）: %s", round_idx + 1, exc)
|
||
if round_idx < max_rounds - 1:
|
||
time.sleep(1.2 + round_idx)
|
||
continue
|
||
raise
|
||
|
||
# 如果拿到 text 了就 break
|
||
if text:
|
||
break
|
||
|
||
# 本轮全链路失败，等一下重试（直链未拿到可能是瞬时限流）
|
||
if round_idx < max_rounds - 1:
|
||
logger.info("第%d轮全链路失败，等待后重试", round_idx + 1)
|
||
time.sleep(1.5)
|
||
|
||
if not text:
|
||
logger.warning("抖音文案提取全部失败: url=%s", page_url)
|
||
raise HTTPException(
|
||
status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
|
||
detail="抖音链接解析暂时不可用，请稍后重试或手动输入文案",
|
||
)
|
||
|
||
return ExtractFromDouyinResponse(
|
||
text=text,
|
||
duration_seconds=duration,
|
||
source_url=page_url,
|
||
)
|
||
|
||
|
||
# ── 2. AI 文案改写 ────────────────────────────────────────────────────
|
||
|
||
|
||
@router.post("/ai-rewrite", response_model=AiRewriteResponse)
|
||
@points_gate("ai_rewrite")
|
||
def ai_rewrite(
|
||
request: AiRewriteRequest,
|
||
current_user: AuthenticatedUser = Depends(get_current_user),
|
||
db: Session = Depends(get_db_session),
|
||
):
|
||
content = (request.content or "").strip()
|
||
if not content:
|
||
raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="文案内容不能为空")
|
||
style = request.style or "口语化"
|
||
client = get_doubao_client()
|
||
if not client.is_available:
|
||
raise HTTPException(
|
||
status_code=status.HTTP_502_BAD_GATEWAY,
|
||
detail="AI 服务不可用，请联系管理员配置豆包大模型 API Key",
|
||
)
|
||
system_prompt = (
|
||
"你是一个专业的短视频文案改写专家。请对以下文案进行改写，"
|
||
"要求：保留原意、口语化、适合短视频口播、调整语序避免查重。"
|
||
)
|
||
if style:
|
||
system_prompt = system_prompt + "\n风格要求：" + style
|
||
messages = [
|
||
{"role": "system", "content": system_prompt},
|
||
{"role": "user", "content": "请改写以下文案：\n\n" + content},
|
||
]
|
||
try:
|
||
rewritten = client.chat_completion(messages=messages, temperature=0.8, max_tokens=2048)
|
||
except Exception as exc:
|
||
logger.error("AI 改写调用失败: %s", exc)
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="AI 改写失败: " + str(exc)) from exc
|
||
if not rewritten:
|
||
raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY, detail="AI 改写未返回有效结果")
|
||
return AiRewriteResponse(original=content, rewritten=rewritten.strip(), style=style)
|
||
|
||
|
||
# ── 3. AI 标题生成 ────────────────────────────────────────────────────
|
||
|
||
|
||
@router.post("/ai-generate-titles", response_model=AiGenerateTitlesResponse)
|
||
@points_gate("ai_title")
|
||
def ai_generate_titles(
|
||
request: AiGenerateTitlesRequest,
|
||
current_user: AuthenticatedUser = Depends(get_current_user),
|
||
db: Session = Depends(get_db_session),
|
||
):
|
||
content = (request.content or "").strip()
|
||
if not content:
|
||
raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="文案内容不能为空")
|
||
count = max(1, min(5, request.count))
|
||
from app.services.ai_service import generate_smart_titles
|
||
result = generate_smart_titles(description=content, style="viral", count=count)
|
||
titles = result.get("titles", [])[:count]
|
||
return AiGenerateTitlesResponse(titles=titles)
|