feat: pdf2zh + DeepSeek V4 Flash 翻译集成

api/server.py

@@ -482,6 +482,59 @@ def serve_translated(arxiv_id: str):
     return FileResponse(fp, media_type="application/pdf",
         headers={"Content-Disposition": "inline"})
 
+# ─── Routes: Trigger translation ───────────────────────
+from concurrent.futures import ThreadPoolExecutor
+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")
+    
+    out_path = TRANSLATED_DIR / f"{arxiv_id}.pdf"
+    if out_path.exists():
+        return {"arxiv_id": arxiv_id, "status": "already_translated"}
+    
+    if arxiv_id in _translating:
+        return {"arxiv_id": arxiv_id, "status": "in_progress"}
+    
+    def do_translate():
+        try:
+            _translating.add(arxiv_id)
+            from pdf2zh.doclayout import OnnxModel
+            from pdf2zh.high_level import translate
+            model = OnnxModel.from_pretrained()
+            translate(
+                [str(pdf_path)], output=str(TRANSLATED_DIR),
+                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"
+            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}")
+        except Exception as e:
+            log.error(f"Translation failed for {arxiv_id}: {e}")
+        finally:
+            _translating.discard(arxiv_id)
+    
+    ThreadPoolExecutor(max_workers=1).submit(do_translate)
+    return {"arxiv_id": arxiv_id, "status": "started"}
+
+@app.get("/api/translate/status")
+def translation_status():
+    return {"translating": list(_translating)}
+
 # ─── Health ─────────────────────────────────────────────
 @app.get("/api/health")
 def health():