完成转录功能，中文脚本搜索，播放界面显示脚本

2025-10-25 19:22:41 +08:00
parent d1b4fb6be0
commit 0d18595b68
34 changed files with 1827 additions and 404 deletions
@@ -0,0 +1,164 @@
+import { json } from '@/lib/json';
+import { getFileUrl } from '@/lib/minio';
+import { prisma } from '@/lib/prisma'; // 你的 Prisma 客户端实例
+import { NextResponse } from 'next/server';
+
+export async function GET(req: Request) {
+  const { searchParams } = new URL(req.url);
+  const q = (searchParams.get('q') || '').trim();
+  const page = Math.max(1, Number(searchParams.get('page') || 1));
+  const limit = Math.min(50, Math.max(1, Number(searchParams.get('limit') || 20)));
+  const offset = (page - 1) * limit;
+
+  if (!q) {
+    return NextResponse.json({ results: [], total: 0 });
+  }
+
+  try {
+    // 核心查询：Postgres 中文全文检索（视频转写 + 视频desc + 图文desc）+ 相关度排序 + 摘要高亮
+    // 说明：
+    // - 合并搜索来源：VideoTranscript.transcript_zh_tsv、Video.desc 与 ImagePost.desc
+    // - 使用 UNION ALL 合并视频和图文的搜索结果
+    // - rank 取两者匹配得分的最大值，snippet 优先展示转写匹配，否则展示 desc 匹配
+    const rows = await prisma.$queryRaw<
+      {
+        id: string;
+        awemeId: string;
+        type: 'video' | 'image';
+        rank: number;
+        snippet: string;
+      }[]
+    >`
+      WITH tsq AS (
+        SELECT websearch_to_tsquery('zhcfg', ${q}) AS query
+      )
+      -- 视频搜索结果
+      SELECT
+        COALESCE(vt.id, 'desc:' || v.aweme_id) AS id,
+        v.aweme_id AS "awemeId",
+        'video'::text AS type,
+        GREATEST(
+          COALESCE(ts_rank(vt.transcript_zh_tsv, tsq.query), 0),
+          COALESCE(ts_rank(to_tsvector('zhcfg', v.desc), tsq.query), 0)
+        ) AS rank,
+        CASE
+          WHEN vt.transcript_zh_tsv @@ tsq.query THEN
+            ts_headline(
+              'zhcfg',
+              array_to_string(vt.transcript, ' '),
+              tsq.query,
+              'StartSel=<mark>,StopSel=</mark>,MaxFragments=2,MinWords=2,MaxWords=20'
+            )
+          ELSE
+            ts_headline(
+              'zhcfg',
+              v.desc,
+              tsq.query,
+              'StartSel=<mark>,StopSel=</mark>,MaxFragments=2,MinWords=2,MaxWords=20'
+            )
+        END AS snippet
+      FROM "Video" v
+      LEFT JOIN "VideoTranscript" vt ON vt."videoId" = v.aweme_id,
+           tsq
+      WHERE
+        (vt.transcript_zh_tsv @@ tsq.query OR to_tsvector('zhcfg', v.desc) @@ tsq.query)
+      
+      UNION ALL
+      
+      -- 图文搜索结果
+      SELECT
+        'img_desc:' || ip.aweme_id AS id,
+        ip.aweme_id AS "awemeId",
+        'image'::text AS type,
+        ts_rank(to_tsvector('zhcfg', ip.desc), tsq.query) AS rank,
+        ts_headline(
+          'zhcfg',
+          ip.desc,
+          tsq.query,
+          'StartSel=<mark>,StopSel=</mark>,MaxFragments=2,MinWords=2,MaxWords=20'
+        ) AS snippet
+      FROM "ImagePost" ip, tsq
+      WHERE to_tsvector('zhcfg', ip.desc) @@ tsq.query
+      
+      ORDER BY rank DESC
+      OFFSET ${offset}
+      LIMIT ${limit};
+    `;
+
+    // 查询总数（参数化，避免注入）
+    const totalRows = await prisma.$queryRaw<{
+      count: number;
+    }[]>`
+      WITH tsq AS (
+        SELECT websearch_to_tsquery('zhcfg', ${q}) AS query
+      )
+      SELECT (
+        (SELECT COUNT(*)::int FROM "Video" v
+         LEFT JOIN "VideoTranscript" vt ON vt."videoId" = v.aweme_id
+         WHERE (vt.transcript_zh_tsv @@ tsq.query OR to_tsvector('zhcfg', v.desc) @@ tsq.query))
+        +
+        (SELECT COUNT(*)::int FROM "ImagePost" ip
+         WHERE to_tsvector('zhcfg', ip.desc) @@ tsq.query)
+      ) AS count
+      FROM tsq;
+    `;
+
+    // 分离视频和图文ID
+    const videoIds = rows.filter(r => r.type === 'video').map(r => r.awemeId);
+    const imagePostIds = rows.filter(r => r.type === 'image').map(r => r.awemeId);
+
+    // 批量查询视频元信息
+    const videos = videoIds.length > 0 ? (await prisma.video.findMany({
+      where: { aweme_id: { in: videoIds } },
+      select: {
+        aweme_id: true,
+        desc: true,
+        cover_url: true,
+        video_url: true,
+        duration_ms: true,
+        author: true
+      },
+    })).map(v => (
+      { ...v, cover_url: getFileUrl(v.cover_url || ''), 
+        author: { ...v.author, avatar_url: getFileUrl(v.author.avatar_url || '') },
+        video_url: getFileUrl(v.video_url || '') })
+    ) : [];
+
+    // 批量查询图文元信息
+    const imagePosts = imagePostIds.length > 0 ? (await prisma.imagePost.findMany({
+      where: { aweme_id: { in: imagePostIds } },
+      select: {
+        aweme_id: true,
+        desc: true,
+        author: true,
+        images: {
+          orderBy: { order: 'asc' },
+          take: 1,
+          select: {
+            url: true,
+            width: true,
+            height: true,
+          }
+        }
+      },
+    })).map(ip => ({
+      ...ip,
+      author: { ...ip.author, avatar_url: getFileUrl(ip.author.avatar_url || '') },
+      cover_url: ip.images[0] ? getFileUrl(ip.images[0].url) : null,
+    })) : [];
+
+    return json({
+      results: rows.map(r => ({
+        ...r,
+        video: r.type === 'video' ? videos.find(v => v.aweme_id === r.awemeId) : undefined,
+        imagePost: r.type === 'image' ? imagePosts.find(ip => ip.aweme_id === r.awemeId) : undefined,
+      })),
+      total: totalRows?.[0]?.count ?? 0,
+      page,
+      limit,
+    });
+  } catch (err) {
+    console.error('Search error:', err);
+    return NextResponse.json({ error: 'Search failed' }, { status: 500 });
+  }
+}