Update irpr/main.py

irpr/main.py

@@ -9,7 +9,7 @@ import os, traceback
 from irpr.models import IngestRequest, GenerateRequest
 from irpr.config import settings
 
-app = FastAPI(title="IR/PR Co-Pilot Pro", version="0.4.1 (OpenAI)")
+app = FastAPI(title="IR/PR Co-Pilot Pro", version="0.4.2 (OpenAI)")
 
 app.add_middleware(
     CORSMiddleware, allow_origins=["*"], allow_credentials=True,
@@ -19,7 +19,6 @@ app.add_middleware(
 def _ensure_dir_writable(path: str) -> bool:
     try:
         os.makedirs(path, exist_ok=True)
-        # 可能なら権限を広めに
         try:
             os.chmod(path, 0o777)
         except Exception:
@@ -36,25 +35,26 @@ def _pick_writable_dir() -> str:
     """
     優先順位:
     1) DATA_DIR (env で指示)
-    2) /mnt/data （この環境では最も安全）
+    2) /mnt/data （Spaces等で最も安全）
     3) /data
     4) ./var
     5) /tmp/irpr
-    6) カレント
+    6) .
     """
     candidates = []
-    if settings.DATA_DIR: candidates.append(settings.DATA_DIR)
+    if settings.DATA_DIR:
+        candidates.append(settings.DATA_DIR)
     candidates += ["/mnt/data", "/data", "./var", "/tmp/irpr", "."]
     for base in candidates:
-        if not base: 
-            continue
         if _ensure_dir_writable(base):
             return base
     return "."
 
 BASE_DIR = _pick_writable_dir()
 UPLOAD_DIR = os.path.join(BASE_DIR, "uploads")
+INDEX_DIR  = os.path.join(BASE_DIR, "simple_index")
 _ensure_dir_writable(UPLOAD_DIR)
+_ensure_dir_writable(INDEX_DIR)
 
 # 静的
 os.makedirs("static", exist_ok=True)
@@ -82,6 +82,13 @@ def ingest_edinet(req: IngestRequest):
     except Exception as e:
         return {"ok": False, "error": repr(e)}
 
+def _safe_filename(name: str) -> str:
+    # パス区切り無効化 + 先頭末尾空白削除
+    name = os.path.basename(name or "").strip()
+    if not name:
+        name = "upload.pdf"
+    return name
+
 @app.post("/ingest/upload")
 async def ingest_upload(files: list[UploadFile] = File(...)):
     from rag.ingest import ingest_pdf_bytes
@@ -90,21 +97,42 @@ async def ingest_upload(files: list[UploadFile] = File(...)):
         # 念のため再確認
         if not _ensure_dir_writable(UPLOAD_DIR):
             raise PermissionError(f"UPLOAD_DIR not writable: {UPLOAD_DIR}")
+
+        if not files:
+            return {"ok": False, "error": "no files", "ingested_chunks": 0, "saved": [], "base_dir": BASE_DIR}
+
         for f in files:
-            if not f.filename.lower().endswith(".pdf"):
+            fname = _safe_filename(f.filename)
+            if not fname.lower().endswith(".pdf"):
+                # PDF以外はスキップ
                 continue
+
             blob = await f.read()
-            save_path = os.path.join(UPLOAD_DIR, f.filename)
+            if not blob:
+                return {"ok": False, "error": f"empty file: {fname}", "ingested_chunks": 0, "saved": [], "base_dir": BASE_DIR}
+
+            # 親ディレクトリを都度保証
+            os.makedirs(UPLOAD_DIR, exist_ok=True)
             try:
+                save_path = os.path.join(UPLOAD_DIR, fname)
                 with open(save_path, "wb") as w:
                     w.write(blob)
             except Exception as e:
-                # 失敗したパスも返す
-                return {"ok": False, "error": f"{repr(e)} at path={save_path}", "base_dir": BASE_DIR}
-            source_url = f"/files/uploads/{f.filename}"
-            total += ingest_pdf_bytes(title=f.filename, source_url=source_url, pdf_bytes=blob)
+                return {"ok": False, "error": f"{repr(e)} at path={os.path.join(UPLOAD_DIR, fname)}", "ingested_chunks": total, "saved": saved, "base_dir": BASE_DIR}
+
+            source_url = f"/files/uploads/{fname}"
+
+            try:
+                added = ingest_pdf_bytes(title=fname, source_url=source_url, pdf_bytes=blob)
+            except Exception as e:
+                # 解析側での例外も捕捉し、どのファイルで落ちたか返す
+                return {"ok": False, "error": f"{repr(e)} while ingesting file={fname}", "ingested_chunks": total, "saved": saved, "base_dir": BASE_DIR}
+
+            total += added
             saved.append(source_url)
+
         return {"ok": True, "ingested_chunks": total, "saved": saved, "base_dir": BASE_DIR}
+
     except Exception as e:
         return {"ok": False, "error": repr(e), "ingested_chunks": total, "saved": saved, "base_dir": BASE_DIR}
 
@@ -162,4 +190,5 @@ from starlette.requests import Request as SRequest
 @app.exception_handler(Exception)
 async def _all_exception_handler(request: SRequest, exc: Exception):
     traceback.print_exc()
-    return JSONResponse(status_code=200, content={"ok": False, "error": repr(exc), "base_dir": BASE_DIR})
+    tb = "".join(traceback.format_exception(type(exc), exc, exc.__traceback__))[-2000:]
+    return JSONResponse(status_code=200, content={"ok": False, "error": repr(exc), "trace": tb, "base_dir": BASE_DIR})