feat: pdf2zh + DeepSeek V4 Flash 翻译集成

api/batch_translate.py

@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+"""批量翻译 arXiv 论文 — pdf2zh + DeepSeek V4 Flash"""
+import json, os, sys, time
+
+PAPERS_DIR = "/app/papers"
+TRANSLATED_DIR = os.path.join(PAPERS_DIR, "translated")
+DATA_FILE = "/app/data/papers.json"
+os.makedirs(TRANSLATED_DIR, exist_ok=True)
+
+# Set DeepSeek credentials
+os.environ["DEEPSEEK_API_KEY"] = os.environ.get("DEEPSEEK_API_KEY", "")
+os.environ["DEEPSEEK_MODEL"] = os.environ.get("DEEPSEEK_MODEL", "deepseek-chat")
+
+# Load paper list
+with open(DATA_FILE) as f:
+    data = json.load(f)
+
+papers = []
+for mod in data.values():
+    for area in mod.get('areas', []):
+        for s in ('mainline', 'branches', 'forward'):
+            for p in area.get(s, []):
+                if p.get('arxiv'):
+                    papers.append(p['arxiv'])
+
+if len(sys.argv) > 1:
+    target = set(sys.argv[1:])
+    papers = [p for p in papers if p in target]
+
+total = len(papers)
+done = 0; skipped = 0; failed = []
+
+from pdf2zh.doclayout import OnnxModel
+
+print(f"=== Batch translate {total} papers (DeepSeek V4 Flash) ===")
+model = OnnxModel.from_pretrained()
+
+for arxiv_id in papers:
+    out_path = os.path.join(TRANSLATED_DIR, f"{arxiv_id}.pdf")
+    if os.path.exists(out_path):
+        skipped += 1; done += 1
+        continue
+    
+    pdf_path = os.path.join(PAPERS_DIR, "arxiv", f"{arxiv_id}.pdf")
+    if not os.path.exists(pdf_path):
+        print(f"[{done+1}/{total}] {arxiv_id}: PDF not found")
+        done += 1
+        continue
+    
+    try:
+        from pdf2zh.high_level import translate
+        translate(
+            [pdf_path], output=TRANSLATED_DIR,
+            lang_in='en', lang_out='zh',
+            service='deepseek', thread=4, model=model,
+        )
+        mono = os.path.join(TRANSLATED_DIR, f"{arxiv_id}-mono.pdf")
+        dual = os.path.join(TRANSLATED_DIR, f"{arxiv_id}-dual.pdf")
+        if os.path.exists(mono):
+            os.rename(mono, out_path)
+        if os.path.exists(dual):
+            os.remove(dual)
+        done += 1
+        print(f"[{done}/{total}] {arxiv_id}: OK")
+    except Exception as e:
+        done += 1
+        failed.append(arxiv_id)
+        print(f"[{done}/{total}] {arxiv_id}: FAILED - {e}")
+        time.sleep(5)
+    
+    time.sleep(1)
+
+print(f"\n=== Done: {done-skipped} translated, {skipped} skipped, {len(failed)} failed ===")
+if failed:
+    print(f"Failed: {failed}")