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/cli.py

@@ -42,9 +42,16 @@ def crawl(
     try:
         # 检查是否已抓取过（非 force 模式）
         if not force and not date_str:
-            existing = db.scalar(select(func.count(Paper.id)).where(Paper.paper_date == target)) or 0
+            existing = (
+                db.scalar(
+                    select(func.count(Paper.id)).where(Paper.paper_date == target)
+                )
+                or 0
+            )
             if existing > 0:
-                typer.echo(f"⏭️  {target} 已有 {existing} 篇论文，跳过（用 --force 强制重抓）")
+                typer.echo(
+                    f"⏭️  {target} 已有 {existing} 篇论文，跳过（用 --force 强制重抓）"
+                )
                 return
 
         typer.echo(f"📡 开始抓取 {target} ...")
@@ -56,7 +63,12 @@ def crawl(
         )
         if need_fallback:
             fallback = yesterday_str()
-            existing = db.scalar(select(func.count(Paper.id)).where(Paper.paper_date == fallback)) or 0
+            existing = (
+                db.scalar(
+                    select(func.count(Paper.id)).where(Paper.paper_date == fallback)
+                )
+                or 0
+            )
             if existing > 0:
                 typer.echo(
                     f"⏭️  {fallback} 已有 {existing} 篇论文，跳过（用 --force 强制重抓）"
@@ -103,7 +115,9 @@ def summarize(
     import os
 
     if pdf_mode not in ("auto", "inject", "search"):
-        typer.echo(f"❌ 无效的 pdf_mode: {pdf_mode}，只支持 auto / inject / search", err=True)
+        typer.echo(
+            f"❌ 无效的 pdf_mode: {pdf_mode}，只支持 auto / inject / search", err=True
+        )
         raise typer.Exit(code=1)
 
     if backend:
@@ -122,6 +136,8 @@ def summarize(
         datefmt="%H:%M:%S",
     )
 
+    from app.exceptions import ConflictError, NotFoundError
+
     db = SessionLocal()
     try:
         if arxiv_id:
@@ -131,16 +147,13 @@ def summarize(
             typer.echo(f"🤖 开始批量总结 pending 论文 (mode={pdf_mode}) ...")
             result = asyncio.run(summarize_batch(db, pdf_mode=pdf_mode))
 
-        if result.get("status") in ("success", "done"):
-            typer.echo(f"✅ 总结完成：{result}")
-        elif result.get("status") == "conflict":
-            typer.echo("⚠️  已有批量总结任务在运行中", err=True)
-            raise typer.Exit(code=1)
-        elif result.get("status") == "not_found":
-            typer.echo(f"❌ 论文未找到：{arxiv_id}", err=True)
-            raise typer.Exit(code=1)
-        else:
-            typer.echo(f"⚠️  总结结果：{result}", err=True)
+        typer.echo(f"✅ 总结完成：{result}")
+    except NotFoundError as exc:
+        typer.echo(f"❌ {exc.message}", err=True)
+        raise typer.Exit(code=1) from exc
+    except ConflictError as exc:
+        typer.echo(f"⚠️  {exc.message}", err=True)
+        raise typer.Exit(code=1) from exc
     finally:
         db.close()