fix: HWP 읽기 — LibreOffice 제거, olefile 순수 Python 파서로 교체

- requirements.txt: olefile==0.47 추가
- Dockerfile: libreoffice-writer 제거 (이미지 크기 대폭 감소)
- app.py: _ext_hwp()를 OLE2 직접 파싱 방식으로 재구현
(HWPTAG_PARA_TEXT 레코드에서 UTF-16LE 텍스트 추출)

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

Dockerfile

@@ -2,11 +2,6 @@ FROM python:3.11-slim
 
 WORKDIR /code
 
-# LibreOffice: .hwp 파일을 텍스트로 변환
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    libreoffice-writer \
-    && rm -rf /var/lib/apt/lists/*
-
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 

app.py

@@ -286,20 +286,58 @@ def _ext_hwpx(path: Path) -> str:
 
 
 def _ext_hwp(path: Path) -> str:
-    import subprocess
-    import tempfile
-    with tempfile.TemporaryDirectory() as tmpdir:
-        r = subprocess.run(
-            ["libreoffice", "--headless", "--norestore",
-             "--convert-to", "txt:Text",
-             "--outdir", tmpdir, str(path)],
-            capture_output=True, text=True, timeout=60,
-            env={**os.environ, "HOME": "/tmp"},
-        )
-        txt = Path(tmpdir) / (path.stem + ".txt")
-        if not txt.exists():
-            raise RuntimeError(f"HWP → 텍스트 변환 실패\n{r.stderr[:300]}")
-        return _ext_txt(txt)
+    import struct, zlib
+    try:
+        import olefile
+    except ImportError:
+        raise RuntimeError("olefile 라이브러리 없음 (pip install olefile)")
+
+    ole = olefile.OleFileIO(str(path))
+
+    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)
+
+    lines: list[str] = []
+    idx = 1
+    while True:
+        sname = f"BodyText/Section{idx:04d}"
+        if not ole.exists(sname):
+            break
+        raw = ole.openstream(sname).read()
+        data = zlib.decompress(raw, -15) if compressed else raw
+
+        pos = 0
+        while pos + 4 <= len(data):
+            hval = struct.unpack_from("<I", data, pos)[0]
+            tag_id = hval & 0x3FF
+            size   = (hval >> 20) & 0xFFF
+            pos += 4
+            if size == 0xFFF:
+                if pos + 4 > len(data): break
+                size = struct.unpack_from("<I", data, pos)[0]
+                pos += 4
+
+            if tag_id == 67 and size >= 2:  # HWPTAG_PARA_TEXT
+                chunk = data[pos:pos + size]
+                t, chars = 0, []
+                while t + 2 <= len(chunk):
+                    cp = struct.unpack_from("<H", chunk, t)[0]
+                    t += 2
+                    if cp == 0x0D:
+                        lines.append("".join(chars)); chars = []
+                    elif 0x01 <= cp <= 0x0C:
+                        t += 8  # inline object: skip 8-byte attribute block
+                    elif cp >= 0x20:
+                        chars.append(chr(cp))
+                if chars:
+                    lines.append("".join(chars))
+            pos += size
+        idx += 1
+
+    return re.sub(r"\n{3,}", "\n\n", "\n".join(l for l in lines if l)).strip()
 
 
 def _ext_docx(path: Path) -> str:

requirements.txt

@@ -10,3 +10,4 @@ aiofiles==23.2.1
 python-multipart==0.0.12
 python-docx==1.1.2
 pypdf==5.1.0
+olefile==0.47