feat: refactor summarizer and PDF extraction pipeline

- Split summarizer into summary_generator and summary_persister modules
- Refactor pdf_image_extractor to two-phase pipeline with PicoDet layout detection
- Add layout_detector service for PicoDet-S_layout_3cls integration
- Add exceptions module with ConflictError and NotFoundError
- Improve admin dashboard with better statistics and task management
- Add design review document with system optimization suggestions
- Add new tests for crawler, pdf_downloader, pipeline, and summary_utils
- Update dependencies and configuration
- Clean up dead code and improve error handling

app/services/admin.py

@@ -9,10 +9,18 @@ from sqlalchemy import func, select, text
 from sqlalchemy.orm import Session
 
 from app.config import settings
-from app.models import CrawlLog, Paper, SummaryState, TaskLock
+from app.models import CrawlLog, Paper, PaperTag, SummaryState, SummaryStatus, TaskLock
 from app.services.scheduler import get_scheduler
 from app.utils import PAPERS_DIR, TMP_DIR
 
+# admin_papers 排序映射
+SORT_MAP = {
+    "date_desc": Paper.paper_date.desc(),
+    "date_asc": Paper.paper_date.asc(),
+    "upvotes_desc": Paper.upvotes.desc(),
+    "title_asc": Paper.title_en.asc(),
+}
+
 
 def _dir_size(path: Path) -> int:
     """递归计算目录总字节数。"""
@@ -52,7 +60,11 @@ def get_admin_stats(db: Session) -> dict:
     status_counts = {row[0]: row[1] for row in summary_rows}
 
     # ── 存储概况 ──────────────────────────────────────────────────────
-    db_size = _fmt_size(settings.db_path.stat().st_size) if settings.db_path.exists() else "0 B"
+    db_size = (
+        _fmt_size(settings.db_path.stat().st_size)
+        if settings.db_path.exists()
+        else "0 B"
+    )
     papers_size = _fmt_size(_dir_size(PAPERS_DIR))
     tmp_size = _fmt_size(_dir_size(TMP_DIR))
 
@@ -68,22 +80,14 @@ def get_admin_stats(db: Session) -> dict:
 
     # ── 最近日志（5 条） ──────────────────────────────────────────────
     recent_logs = (
-        db.execute(
-            select(CrawlLog)
-            .order_by(CrawlLog.started_at.desc())
-            .limit(5)
-        )
+        db.execute(select(CrawlLog).order_by(CrawlLog.started_at.desc()).limit(5))
         .scalars()
         .all()
     )
 
     # ── 活跃锁 ────────────────────────────────────────────────────────
     active_locks = (
-        db.execute(
-            select(TaskLock).where(TaskLock.status == "running")
-        )
-        .scalars()
-        .all()
+        db.execute(select(TaskLock).where(TaskLock.status == "running")).scalars().all()
     )
 
     return {
@@ -108,3 +112,81 @@ def get_admin_stats(db: Session) -> dict:
         "active_locks": active_locks,
         "upvote_refresh_days": settings.UPVOTE_REFRESH_DAYS,
     }
+
+
+def query_papers(
+    db: Session,
+    *,
+    q: str = "",
+    date_from: str | None = None,
+    date_to: str | None = None,
+    tag: str = "",
+    summary_status: str = "all",
+    sort: str = "date_desc",
+    page: int = 1,
+    per_page: int = 20,
+) -> tuple[list[Paper], int, dict[str, str]]:
+    """论文管理查询 — 构建过滤、排序、分页。
+
+    Returns:
+        (papers, total, statuses) — 论文列表、总数、{arxiv_id: summary_status}
+    """
+    query = select(Paper)
+
+    # 搜索
+    if q.strip():
+        query = query.where(
+            Paper.title_en.ilike(f"%{q}%")
+            | Paper.title_zh.ilike(f"%{q}%")
+            | Paper.abstract.ilike(f"%{q}%")
+        )
+
+    # 日期范围
+    if date_from:
+        query = query.where(Paper.paper_date >= date_from)
+    if date_to:
+        query = query.where(Paper.paper_date <= date_to)
+
+    # 标签筛选
+    if tag:
+        query = query.join(PaperTag, PaperTag.paper_id == Paper.id).where(
+            PaperTag.tag == tag
+        )
+
+    # 总结状态筛选
+    if summary_status != "all":
+        if summary_status == "none":
+            query = query.outerjoin(
+                SummaryStatus, SummaryStatus.paper_id == Paper.id
+            ).where(SummaryStatus.paper_id == None)  # noqa: E711
+        else:
+            query = query.join(SummaryStatus, SummaryStatus.paper_id == Paper.id).where(
+                SummaryStatus.status == summary_status
+            )
+
+    # 排序
+    order = SORT_MAP.get(sort, Paper.paper_date.desc())
+    query = query.order_by(order)
+
+    # 计数
+    total = db.scalar(select(func.count()).select_from(query.subquery()))
+
+    # 分页
+    papers = (
+        db.execute(query.offset((page - 1) * per_page).limit(per_page)).scalars().all()
+    )
+
+    # 每篇论文的总结状态
+    paper_ids = [p.id for p in papers]
+    statuses: dict[str, str] = {}
+    if paper_ids:
+        rows = db.execute(
+            select(SummaryStatus.paper_id, SummaryStatus.status).where(
+                SummaryStatus.paper_id.in_(paper_ids)
+            )
+        ).all()
+        paper_id_to_arxiv = {p.id: p.arxiv_id for p in papers}
+        for pid, st in rows:
+            statuses[paper_id_to_arxiv.get(pid, "")] = st
+
+    return papers, total or 0, statuses