acquire-lead/collectors/douyin/video.py

"""步骤：视频/作者信息采集、预检评论数、AI总结后入库。"""

from __future__ import annotations

import re
from urllib.parse import parse_qs, urlparse

from playwright.sync_api import Page

from config import settings
from collectors.douyin.ai import collect_ai_summary
from collectors.douyin.common import SkipVideoError, human_pause
from collectors.douyin.panel import (
    _click_side_panel_tab,
    _ensure_video_page_focus,
    _read_feed_comment_count,
    open_comment_panel,
)
from collectors.douyin.profile import (
    _extract_douyin_id,
    _open_author_profile,
    _open_profile_by_url,
    _to_abs_url,
)
from db.repository import save_video_only
from db.session import session_scope

def _extract_video_meta(page: Page, keyword: str) -> dict:
    """从当前活跃视频弹层提取视频与作者信息。"""
    raw = page.evaluate(
        """() => {
            const modalId = new URLSearchParams(location.search).get('modal_id') || '';
            // 评论打开后 DOM 里可能有多个视频卡片，必须锁定当前活跃视频
            const root = document.querySelector('[data-e2e="feed-active-video"]')
                || document.querySelector('[data-e2e="modal-video-container"]')
                || document;

            const infoEl = root.querySelector('[data-e2e="video-info"][data-e2e-aweme-id]')
                || root.querySelector('[data-e2e="video-info"]')
                || document.querySelector(`[data-e2e="video-info"][data-e2e-aweme-id="${modalId}"]`);
            const awemeId = infoEl ? (infoEl.getAttribute('data-e2e-aweme-id') || '') : '';

            const titleEl = root.querySelector('[data-e2e="video-desc"]');
            const nicknameEl = root.querySelector('[data-e2e="feed-video-nickname"]');
            const avatarEl = root.querySelector('[data-e2e="video-avatar"]');

            const title = titleEl ? (titleEl.innerText || '').trim() : '';
            let authorNickname = nicknameEl ? (nicknameEl.innerText || '').trim() : '';
            if (authorNickname.startsWith('@')) {
                authorNickname = authorNickname.slice(1);
            }
            const authorLink = avatarEl ? (avatarEl.getAttribute('href') || '') : '';
            return {
                videoId: awemeId || modalId || '',
                title,
                authorNickname,
                authorLink,
            };
        }"""
    )

    video_id = raw.get("videoId") or ""
    if not video_id:
        parsed = urlparse(page.url)
        video_id = (parse_qs(parsed.query).get("modal_id") or [""])[0]
        if not video_id:
            match = re.search(r"/video/(\d+)", page.url)
            video_id = match.group(1) if match else ""

    author_link = _to_abs_url(raw.get("authorLink") or "") if raw.get("authorLink") else ""
    author_douyin_id = ""

    if author_link:
        print(f"[视频作者] 准备进入主页: {author_link}")
        human_pause(1.0, 2.0)
        profile = None
        try:
            profile = _open_author_profile(page, author_link)
            human_pause(1.5, 2.5)
            author_douyin_id = _extract_douyin_id(profile)
            print(f"[视频作者] 抖音号: {author_douyin_id}")
        except Exception as exc:
            print(f"[视频作者] 读取抖音号失败: {exc}")
            # 再兜底一次：直接 goto 主页重试
            try:
                if profile is not None:
                    try:
                        profile.close()
                    except Exception:
                        pass
                profile = _open_profile_by_url(page, author_link)
                human_pause(2.0, 3.5)
                author_douyin_id = _extract_douyin_id(profile)
                print(f"[视频作者] 重试成功，抖音号: {author_douyin_id}")
            except Exception as exc2:
                print(f"[视频作者] 重试仍失败: {exc2}")
        finally:
            if profile is not None:
                try:
                    profile.close()
                except Exception:
                    pass
            _ensure_video_page_focus(page)
    else:
        print("[视频作者] 未找到作者主页链接")

    return {
        "keyword": keyword,
        "video_id": video_id,
        "title": raw.get("title") or "",
        "author_nickname": raw.get("authorNickname") or "",
        "author_douyin_id": author_douyin_id,
        "author_link": author_link,
        "video_url": page.url,
    }


def _print_video_meta(meta: dict) -> None:
    print("=" * 48)
    print("[视频信息]")
    print(f"  搜索关键词: {meta.get('keyword', '')}")
    print(f"  视频ID:     {meta.get('video_id', '')}")
    print(f"  标题:       {meta.get('title', '')}")
    print(f"  作者昵称:   {meta.get('author_nickname', '')}")
    print(f"  作者抖音号: {meta.get('author_douyin_id', '') or '(未获取)'}")
    print(f"  作者主页:   {meta.get('author_link', '')}")
    print(f"  视频链接:   {meta.get('video_url', '')}")
    print(f"  AI总结:     {meta.get('summary', '') or '(未获取)'}")
    print("=" * 48)

def collect_and_save_video_meta(
    page: Page,
    keyword: str,
    comment_limit: int = settings.COMMENT_LIMIT,
) -> dict:
    """
    1) 读取评论数（不打开评论区），不足则 SkipVideoError
    2) 采集视频/作者信息
    3) 点评论图标 → AI抖音总结 → 切回评论
    4) 入库，返回 meta（含 db_id=videos.id）
    """
    _ensure_video_page_focus(page)

    # 打开评论区之前：用侧栏数字判断是否值得采
    # 打开评论区之前：只读评论图标下方文案（勿与点赞数混淆）
    comment_total, comment_label = _read_feed_comment_count(page)
    print(f"[预检] 评论图标文案={comment_label or '(空)'} → 解析数={comment_total}")
    if comment_label and "抢首评" in comment_label:
        raise SkipVideoError("评论区显示「抢首评」，打开评论区前跳过")
    if comment_total >= 0 and comment_total < comment_limit:
        raise SkipVideoError(
            f"评论数 {comment_total} < {comment_limit}，打开评论区前跳过"
        )

    video_meta = _extract_video_meta(page, keyword)
    if not video_meta.get("video_id"):
        raise RuntimeError("未获取到 video_id，无法入库")

    _ensure_video_page_focus(page)

    # 固定流程：评论图标 → AI抖音 → 回评论
    print("[流程] 点击评论图标打开右侧栏...")
    open_comment_panel(page)

    try:
        video_meta["summary"] = collect_ai_summary(page)
    except Exception as exc:
        print(f"[AI抖音] 获取总结失败: {exc}")
        video_meta["summary"] = ""

    print("[流程] 切回评论 tab...")
    if not _click_side_panel_tab(page, "评论"):
        raise RuntimeError("AI总结后无法切回评论栏")
    try:
        page.wait_for_selector(
            '[data-e2e="comment-list"], [data-e2e="comment-item"]',
            timeout=10_000,
        )
    except Exception:
        print("[流程] 警告：切回评论后未见评论列表，后续采集可能失败")

    _print_video_meta(video_meta)
    with session_scope() as session:
        result = save_video_only(session, video_meta)
    video_meta["db_id"] = result["id"]
    print(
        f"[视频入库完成] platform_video_id={result['video_id']} "
        f"db_id={result['id']} summary={'有' if video_meta.get('summary') else '无'}"
    )
    return video_meta