fix: 翻译 API 支持 arXiv + HF 两种论文源

api/server.py

@@ -489,23 +489,42 @@ import subprocess, threading
 _translate_lock = threading.Lock()
 _translating = set()
 
-@app.post("/api/translate/{arxiv_id}")
-async def trigger_translation(arxiv_id: str):
-    """Trigger pdf2zh translation for a paper (DeepSeek V4 Flash)"""
-    pdf_path = PAPERS_DIR / "arxiv" / f"{arxiv_id}.pdf"
-    if not pdf_path.exists():
-        raise HTTPException(status_code=404, detail="PDF not found")
+@app.get("/api/translate/status")
+def translation_status():
+    return {"translating": list(_translating)}
+
+@app.post("/api/translate/{paper_id}")
+async def trigger_translation(paper_id: str):
+    """Trigger pdf2zh translation for a paper (DeepSeek V4 Flash).
+    paper_id can be arxiv ID (e.g. 1706.03762) or a HF filename."""
+    # Find the PDF
+    pdf_path = None
+    # Try arxiv
+    candidate = PAPERS_DIR / "arxiv" / f"{paper_id}.pdf"
+    if candidate.exists():
+        pdf_path = candidate
+    else:
+        # Try HF papers directory
+        hf_dir = PAPERS_DIR / "hf"
+        if hf_dir.exists():
+            for f in hf_dir.glob("*.pdf"):
+                if paper_id in f.stem:
+                    pdf_path = f
+                    break
     
-    out_path = TRANSLATED_DIR / f"{arxiv_id}.pdf"
+    if not pdf_path:
+        raise HTTPException(status_code=404, detail=f"PDF not found for {paper_id}")
+    
+    out_path = TRANSLATED_DIR / f"{paper_id}.pdf"
     if out_path.exists():
-        return {"arxiv_id": arxiv_id, "status": "already_translated"}
+        return {"paper_id": paper_id, "status": "already_translated"}
     
-    if arxiv_id in _translating:
-        return {"arxiv_id": arxiv_id, "status": "in_progress"}
+    if paper_id in _translating:
+        return {"paper_id": paper_id, "status": "in_progress"}
     
     def do_translate():
         try:
-            _translating.add(arxiv_id)
+            _translating.add(paper_id)
             from pdf2zh.doclayout import OnnxModel
             from pdf2zh.high_level import translate
             model = OnnxModel.from_pretrained()
@@ -514,22 +533,22 @@ async def trigger_translation(arxiv_id: str):
                 lang_in='en', lang_out='zh',
                 service='deepseek', thread=4, model=model,
             )
-            mono = TRANSLATED_DIR / f"{arxiv_id}-mono.pdf"
-            dual = TRANSLATED_DIR / f"{arxiv_id}-dual.pdf"
+            mono = TRANSLATED_DIR / f"{paper_id}-mono.pdf"
+            dual = TRANSLATED_DIR / f"{paper_id}-dual.pdf"
             if mono.exists():
                 if out_path.exists():
                     out_path.unlink()
                 mono.rename(out_path)
             if dual.exists():
                 dual.unlink()
-            log.info(f"Translated: {arxiv_id}")
+            log.info(f"Translated: {paper_id}")
         except Exception as e:
-            log.error(f"Translation failed for {arxiv_id}: {e}")
+            log.error(f"Translation failed for {paper_id}: {e}")
         finally:
-            _translating.discard(arxiv_id)
+            _translating.discard(paper_id)
     
     ThreadPoolExecutor(max_workers=1).submit(do_translate)
-    return {"arxiv_id": arxiv_id, "status": "started"}
+    return {"paper_id": paper_id, "status": "started"}
 
 @app.get("/api/translate/status")
 def translation_status():