Update app.py

app.py

@@ -118,16 +118,78 @@ def extract_text_with_olefile(input_path):
         return None, f"OLE 파일 처리 오류: {str(e)}"
 
 
+def get_hwp5_command_paths(command_name):
+    """hwp5 명령어 경로 목록 반환"""
+    paths = [
+        command_name,
+        os.path.join(os.path.dirname(sys.executable), command_name),
+        os.path.join(SCRIPT_DIR, 'bin', command_name),
+        os.path.join(PYHWP_PATH, 'bin', command_name),
+    ]
+    
+    # Windows의 경우 Scripts 폴더
+    if sys.platform == 'win32':
+        paths.append(os.path.join(os.path.dirname(sys.executable), 'Scripts', command_name))
+        paths.append(os.path.join(os.path.dirname(sys.executable), 'Scripts', f'{command_name}.exe'))
+    
+    return paths
+
+
+def run_hwp5_module(module_args, capture_output=True):
+    """python -m hwp5 명령 실행"""
+    try:
+        cmd = [sys.executable, '-m', 'hwp5'] + module_args
+        result = subprocess.run(
+            cmd,
+            capture_output=capture_output,
+            timeout=120
+        )
+        return result
+    except Exception as e:
+        print(f"python -m hwp5 실행 오류: {e}")
+        return None
+
+
 def convert_to_html_subprocess(input_path, output_dir):
     """subprocess로 hwp5html 실행"""
     output_name = "output.html"
     output_path = os.path.join(output_dir, output_name)
     
-    hwp5html_paths = [
-        'hwp5html',
-        os.path.join(SCRIPT_DIR, 'bin', 'hwp5html'),
-        os.path.join(PYHWP_PATH, 'bin', 'hwp5html'),
-    ]
+    # 방법 1: python -m hwp5 html 실행
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'hwp5', 'html', '--output', output_path, input_path],
+            capture_output=True,
+            text=True,
+            timeout=120
+        )
+        
+        print(f"hwp5 html stdout: {result.stdout}")
+        print(f"hwp5 html stderr: {result.stderr}")
+        
+        if result.returncode == 0:
+            # 결과 파일/디렉토리 찾기
+            if os.path.isfile(output_path):
+                return output_path, None
+            if os.path.isdir(output_path):
+                return output_path, None
+            
+            # output_dir에서 HTML 찾기
+            for item in os.listdir(output_dir):
+                item_path = os.path.join(output_dir, item)
+                if item.lower().endswith(('.html', '.htm', '.xhtml')) and os.path.isfile(item_path):
+                    return item_path, None
+                if os.path.isdir(item_path):
+                    for sub_item in os.listdir(item_path):
+                        if sub_item.lower().endswith(('.html', '.htm', '.xhtml')):
+                            return item_path, None
+            
+            return output_dir, None
+    except Exception as e:
+        print(f"python -m hwp5 html 오류: {e}")
+    
+    # 방법 2: hwp5html 명령어 직접 실행
+    hwp5html_paths = get_hwp5_command_paths('hwp5html')
     
     for cmd_path in hwp5html_paths:
         try:
@@ -142,28 +204,20 @@ def convert_to_html_subprocess(input_path, output_dir):
             print(f"hwp5html stderr: {result.stderr}")
             
             if result.returncode == 0:
-                # 실제 생성된 파일/디렉토리 찾기
-                # 1. 지정한 output_path가 파일로 존재하는지
                 if os.path.isfile(output_path):
                     return output_path, None
-                
-                # 2. output_path가 디렉토리로 존재하는지
                 if os.path.isdir(output_path):
                     return output_path, None
                 
-                # 3. output_dir 내에서 HTML 파일 찾기
                 for item in os.listdir(output_dir):
                     item_path = os.path.join(output_dir, item)
                     if item.lower().endswith(('.html', '.htm', '.xhtml')) and os.path.isfile(item_path):
                         return item_path, None
-                    # hwp5html이 디렉토리를 생성했을 수 있음
-                    if os.path.isdir(item_path) and item != os.path.basename(input_path).replace('.hwp', ''):
-                        # 디렉토리 내에 HTML이 있는지 확인
+                    if os.path.isdir(item_path):
                         for sub_item in os.listdir(item_path):
                             if sub_item.lower().endswith(('.html', '.htm', '.xhtml')):
                                 return item_path, None
                 
-                # 4. output_dir 자체를 반환 (내부에서 HTML 검색하도록)
                 return output_dir, None
                 
         except (FileNotFoundError, subprocess.TimeoutExpired):
@@ -179,11 +233,22 @@ def convert_to_odt_subprocess(input_path, output_dir):
     """subprocess로 hwp5odt 실행"""
     output_path = os.path.join(output_dir, "output.odt")
     
-    hwp5odt_paths = [
-        'hwp5odt',
-        os.path.join(SCRIPT_DIR, 'bin', 'hwp5odt'),
-        os.path.join(PYHWP_PATH, 'bin', 'hwp5odt'),
-    ]
+    # 방법 1: python -m hwp5 odt 실행
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'hwp5', 'odt', '--output', output_path, input_path],
+            capture_output=True,
+            text=True,
+            timeout=120
+        )
+        
+        if result.returncode == 0 and os.path.exists(output_path):
+            return output_path, None
+    except Exception as e:
+        print(f"python -m hwp5 odt 오류: {e}")
+    
+    # 방법 2: hwp5odt 명령어 직접 실행
+    hwp5odt_paths = get_hwp5_command_paths('hwp5odt')
     
     for cmd_path in hwp5odt_paths:
         try:
@@ -208,11 +273,24 @@ def convert_to_txt_subprocess(input_path, output_dir):
     """subprocess로 hwp5txt 실행"""
     output_path = os.path.join(output_dir, "output.txt")
     
-    hwp5txt_paths = [
-        'hwp5txt',
-        os.path.join(SCRIPT_DIR, 'bin', 'hwp5txt'),
-        os.path.join(PYHWP_PATH, 'bin', 'hwp5txt'),
-    ]
+    # 방법 1: python -m hwp5 txt 실행
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'hwp5', 'txt', input_path],
+            capture_output=True,
+            text=True,
+            timeout=120
+        )
+        
+        if result.returncode == 0 and result.stdout:
+            with open(output_path, 'w', encoding='utf-8') as f:
+                f.write(result.stdout)
+            return output_path, None
+    except Exception as e:
+        print(f"python -m hwp5 txt 오류: {e}")
+    
+    # 방법 2: hwp5txt 명령어 직접 실행
+    hwp5txt_paths = get_hwp5_command_paths('hwp5txt')
     
     for cmd_path in hwp5txt_paths:
         try:
@@ -236,30 +314,99 @@ def convert_to_txt_subprocess(input_path, output_dir):
 
 
 def convert_to_xml_subprocess(input_path, output_dir):
-    """subprocess로 hwp5xml 실행"""
+    """hwp5 Python API 또는 subprocess로 XML 변환"""
     output_path = os.path.join(output_dir, "output.xml")
     
-    hwp5xml_paths = [
-        'hwp5xml',
-        os.path.join(SCRIPT_DIR, 'bin', 'hwp5xml'),
-        os.path.join(PYHWP_PATH, 'bin', 'hwp5xml'),
-    ]
+    # 방법 1: hwp5.xmldump 모듈 직접 사용 (가장 권장)
+    if PYHWP_AVAILABLE:
+        try:
+            from hwp5.xmlmodel import Hwp5File as XmlHwp5File
+            from hwp5.xmldump import xmldump_flat
+            
+            hwp5file = XmlHwp5File(input_path)
+            
+            with open(output_path, 'wb') as f:
+                xmldump_flat(hwp5file, f)
+            
+            if os.path.exists(output_path) and os.path.getsize(output_path) > 0:
+                print(f"XML 변환 성공 (xmldump_flat): {output_path}")
+                return output_path, None
+        except ImportError as e:
+            print(f"xmldump 모듈 import 실패: {e}")
+        except Exception as e:
+            print(f"xmldump_flat 오류: {e}")
+    
+    # 방법 2: hwp5.proc 모듈 사용
+    try:
+        from hwp5.proc import xml_command
+        from io import BytesIO
+        import sys
+        from contextlib import redirect_stdout
+        
+        # XML 출력 캡처
+        output_buffer = BytesIO()
+        
+        # hwp5proc xml 실행
+        old_stdout = sys.stdout
+        sys.stdout = open(output_path, 'w', encoding='utf-8')
+        try:
+            from hwp5 import proc
+            # proc.main(['xml', input_path]) 대신 직접 호출
+            sys.argv = ['hwp5proc', 'xml', input_path]
+            proc.main()
+        except SystemExit:
+            pass
+        finally:
+            sys.stdout.close()
+            sys.stdout = old_stdout
+        
+        if os.path.exists(output_path) and os.path.getsize(output_path) > 0:
+            print(f"XML 변환 성공 (proc.main): {output_path}")
+            return output_path, None
+    except Exception as e:
+        print(f"hwp5.proc 오류: {e}")
+    
+    # 방법 3: python -m hwp5 xml 실행
+    try:
+        result = subprocess.run(
+            [sys.executable, '-m', 'hwp5', 'xml', input_path],
+            capture_output=True,
+            timeout=120
+        )
+        
+        print(f"python -m hwp5 xml returncode: {result.returncode}")
+        print(f"python -m hwp5 xml stdout length: {len(result.stdout)}")
+        print(f"python -m hwp5 xml stderr: {result.stderr.decode('utf-8', errors='ignore')[:500]}")
+        
+        if result.returncode == 0 and result.stdout:
+            with open(output_path, 'wb') as f:
+                f.write(result.stdout)
+            print(f"XML 변환 성공 (python -m): {output_path}")
+            return output_path, None
+    except subprocess.TimeoutExpired:
+        print("python -m hwp5 xml 타임아웃")
+    except Exception as e:
+        print(f"python -m hwp5 xml 오류: {e}")
+    
+    # 방법 4: hwp5xml 명령어 직접 실행
+    hwp5xml_paths = get_hwp5_command_paths('hwp5xml')
     
     for cmd_path in hwp5xml_paths:
         try:
-            # hwp5xml은 stdout으로 출력하거나 --output 옵션 사용
+            # stdout으로 출력
             result = subprocess.run(
                 [cmd_path, input_path],
                 capture_output=True,
                 timeout=120
             )
             
-            print(f"hwp5xml stdout length: {len(result.stdout)}")
-            print(f"hwp5xml stderr: {result.stderr.decode('utf-8', errors='ignore')[:500]}")
+            print(f"{cmd_path} returncode: {result.returncode}")
+            print(f"{cmd_path} stdout length: {len(result.stdout)}")
             
             if result.returncode == 0 and result.stdout:
                 with open(output_path, 'wb') as f:
                     f.write(result.stdout)
+                print(f"XML 변환 성공 ({cmd_path}): {output_path}")
                 return output_path, None
             
             # --output 옵션 시도
@@ -271,15 +418,41 @@ def convert_to_xml_subprocess(input_path, output_dir):
             )
             
             if result.returncode == 0 and os.path.exists(output_path):
+                print(f"XML 변환 성공 ({cmd_path} --output): {output_path}")
                 return output_path, None
                 
-        except (FileNotFoundError, subprocess.TimeoutExpired):
+        except FileNotFoundError:
+            continue
+        except subprocess.TimeoutExpired:
+            print(f"{cmd_path} 타임아웃")
+            continue
+        except Exception as e:
+            print(f"{cmd_path} 오류: {e}")
+            continue
+    
+    # 방법 5: hwp5proc xml 직접 실행
+    hwp5proc_paths = get_hwp5_command_paths('hwp5proc')
+    
+    for cmd_path in hwp5proc_paths:
+        try:
+            result = subprocess.run(
+                [cmd_path, 'xml', input_path],
+                capture_output=True,
+                timeout=120
+            )
+            
+            if result.returncode == 0 and result.stdout:
+                with open(output_path, 'wb') as f:
+                    f.write(result.stdout)
+                print(f"XML 변환 성공 ({cmd_path} xml): {output_path}")
+                return output_path, None
+        except FileNotFoundError:
             continue
         except Exception as e:
-            print(f"hwp5xml error: {e}")
+            print(f"{cmd_path} xml 오류: {e}")
             continue
     
-    return None, "hwp5xml 명령을 찾을 수 없습니다."
+    return None, "hwp5xml 변환을 수행할 수 없습니다. pyhwp 설치를 확인하세요."
 
 
 def html_to_markdown_with_markdownify(html_content):