fix: HWP 추출 오류 진단 강화 + openstream 문자열 경로로 변경

- _ext_hwp: list 대신 문자열 경로 사용 (BodyText/SectionN)
- _ext_hwp: OLE 항목 대소문자 무시 비교
- _ext_hwp: 각 단계별 명시적 오류 메시지 + print 로그
- _ext_hwp: zlib 실패 시 일반 zlib 재시도 후 raw 폴백
- /extract: str(e) 빈 문자열이면 type name 대신 사용

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

app.py

@@ -290,32 +290,55 @@ def _ext_hwp(path: Path) -> str:
     try:
         import olefile
     except ImportError:
-        raise RuntimeError("olefile 라이브러리 없음 (pip install olefile)")
+        raise RuntimeError("olefile 미설치 (pip install olefile)")
 
-    ole = olefile.OleFileIO(str(path))
+    try:
+        ole = olefile.OleFileIO(str(path))
+    except Exception as exc:
+        raise RuntimeError(f"HWP OLE 열기 실패 [{type(exc).__name__}]: {exc or 'no detail'}")
 
     compressed = True
-    if ole.exists("FileHeader"):
-        hdr = ole.openstream("FileHeader").read()
-        if len(hdr) >= 40:
-            compressed = bool(struct.unpack_from("<I", hdr, 36)[0] & 0x1)
-
-    # Discover section names dynamically — handles Section0, Section0001, etc.
-    sections = sorted(
-        (e[1] for e in ole.listdir()
-         if len(e) == 2 and e[0] == "BodyText" and e[1].startswith("Section")),
-        key=lambda s: int(re.sub(r"\D", "", s) or 0),
-    )
+    try:
+        if ole.exists("FileHeader"):
+            hdr = ole.openstream("FileHeader").read()
+            if len(hdr) >= 40:
+                compressed = bool(struct.unpack_from("<I", hdr, 36)[0] & 0x1)
+    except Exception:
+        pass
+
+    # Discover sections: case-insensitive, any naming convention
+    try:
+        all_entries = ole.listdir()
+        print(f"[HWP] OLE entries: {all_entries[:20]}", flush=True)
+        sections = sorted(
+            (e[1] for e in all_entries
+             if isinstance(e, (list, tuple)) and len(e) >= 2
+             and str(e[0]).lower() == "bodytext"
+             and str(e[1]).lower().startswith("section")),
+            key=lambda s: int(re.sub(r"\D", "", s) or "0"),
+        )
+    except Exception as exc:
+        raise RuntimeError(f"섹션 목록 오류 [{type(exc).__name__}]: {exc or 'no detail'}")
+
     if not sections:
-        raise RuntimeError("BodyText 섹션을 찾을 수 없습니다. 지원되지 않는 HWP 버전일 수 있습니다.")
+        entry_dump = str([str(e) for e in all_entries[:20]])
+        raise RuntimeError(f"BodyText/Section* 없음. OLE 항목: {entry_dump}")
 
     lines: list[str] = []
     for sec_name in sections:
-        raw = ole.openstream(["BodyText", sec_name]).read()
+        try:
+            raw = ole.openstream(f"BodyText/{sec_name}").read()
+        except Exception as exc:
+            print(f"[HWP] openstream {sec_name} 오류: {exc}", flush=True)
+            continue
+
         try:
             data = zlib.decompress(raw, -15) if compressed else raw
         except zlib.error:
-            data = raw
+            try:
+                data = zlib.decompress(raw)
+            except zlib.error:
+                data = raw
 
         pos = 0
         while pos + 4 <= len(data):
@@ -337,7 +360,7 @@ def _ext_hwp(path: Path) -> str:
                     if cp == 0x0D:
                         lines.append("".join(chars)); chars = []
                     elif 0x01 <= cp <= 0x0C:
-                        t += 8  # inline object: skip 8-byte attribute block
+                        t += 8
                     elif cp >= 0x20:
                         chars.append(chr(cp))
                 if chars:
@@ -428,7 +451,8 @@ async def extract_file(file: UploadFile = File(...)):
         else:
             text = _ext_txt(tmp_path)
     except Exception as e:
-        raise HTTPException(500, str(e))
+        detail = str(e).strip() or type(e).__name__
+        raise HTTPException(500, detail)
     finally:
         tmp_path.unlink(missing_ok=True)