feat: HWP/HWPX 실제 추출 (rhwp WASM + Node 22 Docker) — getSectionCount/getParagraphCount fix

Dockerfile

@@ -1,14 +1,20 @@
 FROM python:3.12-slim
-WORKDIR /app
 
-# system deps
+# Node.js 22 (rhwp.js WASM HWP/HWPX 추출용 — HanSoo 이식)
 RUN apt-get update && apt-get install -y --no-install-recommends \
-    curl ca-certificates \
+    curl ca-certificates gnupg \
+    && curl -fsSL https://deb.nodesource.com/setup_22.x | bash - \
+    && apt-get install -y --no-install-recommends nodejs \
     && rm -rf /var/lib/apt/lists/*
 
+WORKDIR /app
+
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
+# HWP 추출 (rhwp WASM via Node)
+COPY pkg/ ./pkg/
+
 COPY app.py .
 COPY static/ ./static/
 

app.py

@@ -185,18 +185,48 @@ def extract_txt(data: bytes) -> str:
         except UnicodeDecodeError: continue
     return data.decode("utf-8", errors="replace")
 
+import tempfile, asyncio as _asyncio
+PKG_DIR = Path(__file__).parent / "pkg"
+RHWP_SCRIPT = PKG_DIR / "extract_hwp.mjs"
+
 async def extract_hwp(data: bytes, fname: str) -> str:
-    if not HWP_API_URL:
-        return f"[HWP/HWPX 파일 {fname} — 서버에 HanSoo 변환 API 설정 안 됨 (env JGOS_HWP_API_URL). 파일 내용 추출 불가.]"
+    """HanSoo rhwp WASM (Rust) — Node.js subprocess로 HWP/HWPX 한글 텍스트 직접 추출.
+    외부 API 의존성 0. JGOS Space 내부 처리.
+    """
+    if not RHWP_SCRIPT.exists():
+        return f"[HWP 처리기 미설치: {RHWP_SCRIPT}]"
+    suffix = ".hwpx" if fname.lower().endswith(".hwpx") else ".hwp"
+    tmp_path = None
     try:
-        async with httpx.AsyncClient(timeout=60) as client:
-            r = await client.post(HWP_API_URL, files={"file": (fname, data)})
-            if r.status_code == 200:
-                j = r.json()
-                return j.get("text", f"[HanSoo API 응답 text 없음]")
-            return f"[HanSoo API HTTP {r.status_code}]"
+        with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as tf:
+            tf.write(data); tf.flush(); tmp_path = tf.name
+        proc = await _asyncio.create_subprocess_exec(
+            "node", str(RHWP_SCRIPT), tmp_path,
+            stdout=_asyncio.subprocess.PIPE, stderr=_asyncio.subprocess.PIPE,
+        )
+        try:
+            stdout, stderr = await _asyncio.wait_for(proc.communicate(), timeout=60)
+        except _asyncio.TimeoutError:
+            proc.kill(); return f"[HWP 변환 타임아웃 60s: {fname}]"
+        if proc.returncode != 0:
+            err = (stderr.decode("utf-8", errors="ignore")[:300]) if stderr else "no stderr"
+            return f"[HWP 변환 실패 (exit {proc.returncode}): {err}]"
+        try:
+            result = json.loads(stdout.decode("utf-8", errors="ignore"))
+        except Exception as e:
+            return f"[HWP 결과 파싱 실패: {e}]"
+        if not result.get("ok"):
+            return f"[HWP 추출 실패: {result.get('error','unknown')}]"
+        text = result.get("text", "")
+        return text or f"[{fname}: 추출 텍스트 없음 (paragraphs={result.get('paragraphs',0)})]"
+    except FileNotFoundError as e:
+        return f"[Node.js 미설치 — Docker rebuild 필요: {e}]"
     except Exception as e:
-        return f"[HanSoo API 오류: {type(e).__name__}: {e}]"
+        return f"[HWP 처리 오류: {type(e).__name__}: {e}]"
+    finally:
+        if tmp_path:
+            try: os.unlink(tmp_path)
+            except: pass
 
 def extract_image_meta(data: bytes, fname: str) -> str:
     """이미지: 메타만 텍스트로. 실제 vision 분석은 base64로 LLM에 직접 전달."""

pkg/extract_hwp.mjs

@@ -0,0 +1,79 @@
+// extract_hwp.mjs — HWP/HWPX 파일을 텍스트로 추출 (rhwp WASM 사용)
+// 사용: node extract_hwp.mjs <input_hwp_path>
+// 출력: stdout JSON {ok, text, paragraphs, sections, error?}
+
+import init, { HwpDocument } from "./rhwp.js";
+import { readFile } from "node:fs/promises";
+
+async function main() {
+  const inputPath = process.argv[2];
+  if (!inputPath) {
+    console.error(JSON.stringify({ ok: false, error: "missing arg: input path" }));
+    process.exit(1);
+  }
+
+  try {
+    // WASM 초기화 — Node.js fs로 직접 로드
+    const { readFileSync } = await import("node:fs");
+    const wasmPath = new URL("./rhwp_bg.wasm", import.meta.url);
+    const wasmBytes = readFileSync(wasmPath);
+    await init({ module_or_path: wasmBytes });
+
+    // HWP 파일 로드
+    const fileBytes = await readFile(inputPath);
+    const doc = new HwpDocument(new Uint8Array(fileBytes));
+
+    // Method 1: getTextRange로 모든 section/paragraph 순회
+    const allParas = [];
+    let sectionCount = 0;
+    try {
+      sectionCount = doc.getSectionCount();
+    } catch (e) {
+      sectionCount = 1;
+    }
+
+    for (let s = 0; s < sectionCount; s++) {
+      let paraCount = 0;
+      try {
+        paraCount = doc.getParagraphCount(s);
+      } catch (e) { continue; }
+
+      for (let p = 0; p < paraCount; p++) {
+        let paraLen = 0;
+        try {
+          paraLen = doc.getParagraphLength(s, p);
+        } catch (e) { continue; }
+        if (paraLen === 0) { allParas.push(""); continue; }
+        try {
+          const text = doc.getTextRange(s, p, 0, paraLen);
+          allParas.push(text || "");
+        } catch (e) {
+          allParas.push(`[para ${s}.${p} 추출 실패]`);
+        }
+      }
+    }
+
+    const finalText = allParas.filter(p => p && p.trim()).join("\n");
+
+    // free
+    try { doc.free(); } catch (e) {}
+
+    console.log(JSON.stringify({
+      ok: true,
+      text: finalText,
+      paragraphs: allParas.length,
+      sections: sectionCount,
+      char_count: finalText.length,
+    }, null, 0));
+
+  } catch (e) {
+    console.error(JSON.stringify({
+      ok: false,
+      error: `${e.name || "Error"}: ${e.message || e}`,
+      stack: (e.stack || "").split("\n").slice(0, 5).join(" | "),
+    }));
+    process.exit(2);
+  }
+}
+
+main();

pkg/rhwp_bg.wasm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72dd73f7343948584bc7f8b0ffcd7c8f5d05489a63b41b1f657ea41b24095bea
+size 3703777