xiaoxia-saas/apps/api/app/api/routes/scripts_ai.py

"""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")

_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） ────────────────────────────


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
    if not data.get("success", True) and data.get("error"):
        err = 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):
    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")
    return task_id


def _mediakit_asr_poll(self, task_id, poll_interval=2.0, max_attempts=90):
    for _ in range(max_attempts):
        time.sleep(poll_interval)
        data = self._get_json("/tasks/" + task_id)
        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") or {}
            raise MediaKitError(
                "MediaKit ASR 任务失败: %s" % err.get("message", "unknown"),
                code=err.get("code", "TaskFailed"),
            )
    raise MediaKitError(
        "MediaKit ASR 超时（%ss 未完成）" % int(poll_interval * max_attempts),
        code="Timeout",
    )


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 _ytdlp_extract_video_url(page_url, cookiefile=None):
    try:
        import yt_dlp
    except ImportError:
        logger.warning("yt-dlp 未安装，无法解析抖音直链")
        return None, 0.0
    opts = {
        "quiet": True,
        "no_warnings": True,
        "noplaylist": True,
        "skip_download": True,
        "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/",
        },
    }
    if cookiefile:
        opts["cookiefile"] = cookiefile
    try:
        with yt_dlp.YoutubeDL(opts) as ydl:
            info = ydl.extract_info(page_url, download=False)
    except Exception as exc:
        _dbg("ytdlp_err", str(exc)[:300])
        logger.info("yt-dlp 解析抖音直链失败（将降级）: %s", str(exc)[:200])
        return None, 0.0
    if not info:
        return None, 0.0
    video_url = info.get("url")
    if not video_url:
        for f in (info.get("requested_formats") or info.get("formats") or []):
            if f.get("url"):
                video_url = f["url"]
                break
    try:
        duration = float(info.get("duration") or 0)
    except (TypeError, ValueError):
        duration = 0.0
    return video_url, duration


def _ytdlp_download_and_local_asr(page_url, temp_dir, cookiefile=None):
    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
    opts = {
        "format": "best[ext=mp4]/best",
        "outtmpl": temp_dir + "/%(id)s.%(ext)s",
        "quiet": True,
        "no_warnings": True,
        "noplaylist": True,
        "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/",
        },
    }
    if cookiefile:
        opts["cookiefile"] = cookiefile

    info = None
    try:
        with yt_dlp.YoutubeDL(opts) as ydl:
            info = ydl.extract_info(page_url, download=True)
    except yt_dlp.utils.DownloadError as exc:
        msg = str(exc)
        logger.warning("抖音下载失败: url=%s error=%s", page_url, msg)
        if _is_cookies_related_error(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) from exc
        is_bad_url = any(
            kw in msg.lower()
            for kw in (
                "404", "not found", "unable to download webpage",
                "unsupported url", "no video formats", "video unavailable",
                "this video isn't available",
            )
        )
        _detail = "无法解析该抖音链接，请确认链接有效且视频未被下架" if is_bad_url else "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_400_BAD_REQUEST if is_bad_url else status.HTTP_502_BAD_GATEWAY,
            detail=_detail,
        ) from exc
    except Exception as exc:
        msg = str(exc)
        logger.exception("抖音视频下载异常: url=%s error=%s", page_url, msg)
        if _is_cookies_related_error(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) from exc
        _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) from exc

    if info is None:
        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="无法解析该抖音链接")
    duration = 0.0
    video_path = ""
    try:
        with yt_dlp.YoutubeDL(opts) as ydl:
            video_path = ydl.prepare_filename(info)
        try:
            duration = float(info.get("duration") or 0)
        except (TypeError, ValueError):
            duration = 0.0
        if not os.path.isfile(video_path) or os.path.getsize(video_path) == 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")
    except HTTPException:
        raise
    except Exception as exc:
        msg = str(exc)
        logger.exception("抖音视频后处理异常: url=%s error=%s", page_url, msg)
        if _is_cookies_related_error(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) from exc
        _detail = "视频处理失败，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) from exc

    try:
        text = transcribe_to_text(video_path)
    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", video_path)
        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


# ── 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()

    # 路径 A：yt-dlp 拿直链 + MediaKit 云端 ASR
    direct_url, meta_duration = _ytdlp_extract_video_url(page_url, cookiefile=cookiefile)
    if meta_duration:
        duration = meta_duration
    _dbg("direct_url", direct_url or "<none>")

    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: url=%s text_len=%d duration=%.1f",
                page_url, len(text), duration,
            )
        except MediaKitError as exc:
            logger.warning("MediaKit ASR 失败，回退本地 ASR: %s", exc)
            text = ""

    # 路径 B：回退下载 + 本地 ASR
    if not text:
        _dbg("fallback", "download+local_asr")
        with tempfile.TemporaryDirectory(prefix="douyin_extract_") as temp_dir:
            text, dl_duration = _ytdlp_download_and_local_asr(
                page_url, temp_dir, cookiefile=cookiefile
            )
            if dl_duration and not duration:
                duration = dl_duration

    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)