Update app.py

app.py

@@ -3,147 +3,154 @@ import sqlite3
 import pandas as pd
 import os
 import re
-import base64
+import base64 # 추가됨
 
 # 1️⃣ 저장소 설정
 UPLOAD_DIR = "uploaded_dbs"
 if not os.path.exists(UPLOAD_DIR):
-    os.makedirs(UPLOAD_DIR)
+    os.makedirs(UPLOAD_DIR)
 
 db_registry = []
 
-# BLOB 데이터를 Base64 HTML 이미지 태그로 변환
+# BLOB 데이터를 Base64 HTML 이미지 태그로 변환하는 함수
 def blob_to_base64_html(blob_data):
-    if blob_data is None or pd.isna(blob_data):
-        return ""
-    try:
-        if isinstance(blob_data, (bytes, bytearray)):
-            encoded_string = base64.b64encode(blob_data).decode('utf-8')
-            return f'<img src="data:image/png;base64,{encoded_string}" width="200" height="auto" />'
-        return str(blob_data)
-    except:
-        return "[이미지 변환 에러]"
+    if blob_data is None:
+        return ""
+    try:
+        # 데이터가 이미 바이트 형태인지 확인 (b'\x89PNG' 등)
+        if isinstance(blob_data, bytes):
+            encoded_string = base64.b64encode(blob_data).decode('utf-8')
+            # 너비(width)는 원하시는 대로 조절하세요.
+            return f'<img src="data:image/png;base64,{encoded_string}" width="150" />'
+        return str(blob_data) # 바이트가 아니면 그냥 문자열로 반환
+    except:
+        return "[이미지 변환 에러]"
 
 def natural_sort_key(s):
-    if s is None: return []
-    return [int(text) if text.isdigit() else text.lower()
-            for text in re.split(r'(\d+)', str(s))]
+    if s is None: return []
+    return [int(text) if text.isdigit() else text.lower()
+            for text in re.split(r'(\d+)', str(s))]
 
 def refresh_registry_data():
-    global db_registry
-    db_registry = []
-    if not os.path.exists(UPLOAD_DIR): return []
-    db_files = [f for f in os.listdir(UPLOAD_DIR) if f.endswith(".db")]
-    for filename in db_files:
-        name_only = filename.replace(".db", "")
-        parts = name_only.split("_")
-        if len(parts) >= 2:
-            db_registry.append({
-                "path": os.path.join(UPLOAD_DIR, filename),
-                "standard": parts[0],
-                "version": parts[1]
-            })
-    return sorted(list(set([db["standard"] for db in db_registry])))
+    global db_registry
+    db_registry = []
+    if not os.path.exists(UPLOAD_DIR): return []
+    db_files = [f for f in os.listdir(UPLOAD_DIR) if f.endswith(".db")]
+    for filename in db_files:
+        name_only = filename.replace(".db", "")
+        parts = name_only.split("_")
+        if len(parts) >= 2:
+            db_registry.append({
+                "path": os.path.join(UPLOAD_DIR, filename),
+                "standard": parts[0],
+                "version": parts[1]
+            })
+    return sorted(list(set([db["standard"] for db in db_registry])))
 
 # 2️⃣ UI 업데이트 함수
 def on_load():
-    standards = refresh_registry_data()
-    return gr.Dropdown(choices=standards, value=None)
+    standards = refresh_registry_data()
+    return gr.Dropdown(choices=standards, value=None)
 
 def update_version_dd(standard):
-    if not standard: return gr.Dropdown(choices=[], value=None)
-    versions = sorted(list(set([db["version"] for db in db_registry if db["standard"] == standard])))
-    return gr.Dropdown(choices=versions, value=None)
+    if not standard: return gr.Dropdown(choices=[], value=None)
+    versions = sorted(list(set([db["version"] for db in db_registry if db["standard"] == standard])))
+    return gr.Dropdown(choices=versions, value=None)
 
 def update_category_dd(standard, version):
-    if not standard or not version: return gr.Dropdown(choices=[], value=None)
-    choices = ["ALL"]
-    try:
-        target_db = next(db for db in db_registry if db["standard"] == standard and db["version"] == version)
-        conn = sqlite3.connect(target_db["path"])
-        all_tables = pd.read_sql("SELECT name FROM sqlite_master WHERE type='table';", conn)['name'].tolist()
-        
-        main_t = f"{standard}_{version}"
-        if main_t not in all_tables: main_t = all_tables[0]
-        
-        # 컬럼명 대소문자 구분 없이 chapter/category 찾기
-        cols_info = pd.read_sql(f"PRAGMA table_info([{main_t}])", conn)
-        cols_lower = [c.lower() for c in cols_info['name'].tolist()]
-        
-        if 'chapter' in cols_lower and 'category' in cols_lower:
-            real_ch = cols_info['name'].tolist()[cols_lower.index('chapter')]
-            real_cat = cols_info['name'].tolist()[cols_lower.index('category')]
-            df_cat = pd.read_sql(f"SELECT DISTINCT [{real_ch}], [{real_cat}] FROM [{main_t}]", conn)
-            cats = [f"{row[real_ch]}.{row[real_cat]}" for _, row in df_cat.iterrows() if row[real_ch]]
-            cats.sort(key=natural_sort_key)
-            choices.extend(cats)
-        
-        prefix = f"{standard}_{version}_"
-        sub_tables = [t.replace(prefix, "") for t in all_tables if t != main_t]
-        choices.extend(sorted(sub_tables))
-        conn.close()
-    except: pass
-    return gr.Dropdown(choices=choices, value=None)
-
-# 3️⃣ 데이터 조회 로직 (정렬 지우고 이미지+DB 순서 유지)
+    if not standard or not version: return gr.Dropdown(choices=[], value=None)
+    choices = ["ALL"]
+    try:
+        target_db = next(db for db in db_registry if db["standard"] == standard and db["version"] == version)
+        conn = sqlite3.connect(target_db["path"])
+        all_tables = pd.read_sql("SELECT name FROM sqlite_master WHERE type='table';", conn)['name'].tolist()
+        main_t = f"{standard}_{version}"
+        if main_t not in all_tables: main_t = all_tables[0]
+        
+        cols = pd.read_sql(f"PRAGMA table_info([{main_t}])", conn)['name'].tolist()
+        if 'chapter' in cols and 'category' in cols:
+            df_cat = pd.read_sql(f"SELECT DISTINCT chapter, category FROM [{main_t}]", conn)
+            cats = [f"{row['chapter']}.{row['category']}" for _, row in df_cat.iterrows()]
+            cats.sort(key=natural_sort_key)
+            choices.extend(cats)
+        
+        prefix = f"{standard}_{version}_"
+        sub_tables = [t.replace(prefix, "") for t in all_tables if t != main_t]
+        choices.extend(sorted(sub_tables))
+        conn.close()
+    except: pass
+    return gr.Dropdown(choices=choices, value=None)
+
+# 3️⃣ 데이터 조회 로직 (BLOB 처리 추가)
 def display_data(standard, version, selection):
     if not all([standard, version, selection]): return None
     try:
         target_db = next(db for db in db_registry if db["standard"] == standard and db["version"] == version)
         conn = sqlite3.connect(target_db["path"])
         all_tables = pd.read_sql("SELECT name FROM sqlite_master WHERE type='table';", conn)['name'].tolist()
+        
         main_t = f"{standard}_{version}"
         if main_t not in all_tables: main_t = all_tables[0]
 
         is_sub_table = False 
-        # ORDER BY rowid로 DB 저장 순서 그대로 가져오기
         if selection == "ALL":
-            df = pd.read_sql(f"SELECT * FROM [{main_t}] ORDER BY rowid", conn)
+            df = pd.read_sql(f"SELECT * FROM [{main_t}]", conn)
         elif "." in selection:
             ch, ca = selection.split('.', 1)
-            # 대소문자 대응을 위해 필터링 쿼리 작성
-            cols_info = pd.read_sql(f"PRAGMA table_info([{main_t}])", conn)
-            cols_lower = [c.lower() for c in cols_info['name'].tolist()]
-            real_ch = cols_info['name'].tolist()[cols_lower.index('chapter')]
-            real_cat = cols_info['name'].tolist()[cols_lower.index('category')]
-            df = pd.read_sql(f"SELECT * FROM [{main_t}] WHERE [{real_ch}]=? AND [{real_cat}]=? ORDER BY rowid", conn, params=[ch, ca])
+            df = pd.read_sql(f"SELECT * FROM [{main_t}] WHERE chapter=? AND category=?", conn, params=[ch, ca])
         else:
             is_sub_table = True
             actual_table = next((t for t in all_tables if t.endswith(selection)), selection)
-            df = pd.read_sql(f"SELECT * FROM [{actual_table}] ORDER BY rowid", conn)
+            df = pd.read_sql(f"SELECT * FROM [{actual_table}]", conn)
         conn.close()
 
         if not df.empty:
-            # 모든 셀에 대해 이미지 변환 처리
+            # 1. 먼저 BLOB(이미지) 데이터를 HTML로 변환
             for col in df.columns:
                 df[col] = df[col].apply(blob_to_base64_html)
 
-            # 열 필터링: 본문 조회 시 주요 열 + 이미지가 포함된 열만 표시
+            # 2. 정렬 로직 적용
+            sort_col = 'section' if 'section' in df.columns else df.columns[0]
+            df['sort_key'] = df[sort_col].apply(natural_sort_key)
+            df = df.sort_values(by='sort_key').drop(columns=['sort_key'])
+            
+            # 3. 💡 [열 필터링 로직]
             if not is_sub_table:
+                # 기본적으로 보여줄 열 정의
                 base_cols = [c for c in ['section', 'description'] if c in df.columns]
-                # 이미지가 들어있는 열(HTML 태그가 포함된 열)을 찾음
-                image_cols = [c for c in df.columns if c not in base_cols and '<img' in str(df[col].iloc[0] if len(df)>0 else "")]
+                
+                # 만약 'image'나 'blob' 같은 다른 열에 이미지가 변환되어 들어갔다면 그 열도 포함
+                image_cols = [
+                    c for c in df.columns 
+                    if c not in base_cols and df[c].astype(str).str.contains('<img', na=False).any()
+                ]
+                
+                # 최종적으로 section, description + 이미지 열만 표시
                 df = df[base_cols + image_cols]
+            else:
+                # 부속 테이블(Table A 등)은 모든 열을 보여주되, 정렬용 임시 열만 제거
+                df = df[[c for c in df.columns if not c.startswith('sort_')]]
                 
         return df
     except Exception as e:
         return pd.DataFrame({"Error": [f"조회 실패: {str(e)}"]})
-
+        
 # 4️⃣ UI 구성
 with gr.Blocks() as demo:
-    gr.Markdown("# 📜 Regulation Viewer")
+    gr.Markdown("# 📜 Regulation Viewer (Image Support)")
 
-    with gr.Row():
-        standard_dd = gr.Dropdown(label="1. 법규 선택")
-        version_dd = gr.Dropdown(label="2. Version 선택")
-        category_dd = gr.Dropdown(label="3. Category / Table 선택")
+    with gr.Row():
+        standard_dd = gr.Dropdown(label="1. 법규 선택")
+        version_dd = gr.Dropdown(label="2. Version 선택")
+        category_dd = gr.Dropdown(label="3. Category / Table 선택")
 
-    output_df = gr.Dataframe(wrap=True, interactive=False, datatype="html")
+    # 💡 [중요] datatype="html"을 추가하여 <img> 태그가 이미지로 보이게 함
+    output_df = gr.Dataframe(wrap=True, interactive=False, datatype="html")
 
-    demo.load(on_load, None, standard_dd)
-    standard_dd.change(update_version_dd, standard_dd, version_dd)
-    version_dd.change(update_category_dd, [standard_dd, version_dd], category_dd)
-    category_dd.change(display_data, [standard_dd, version_dd, category_dd], output_df)
+    demo.load(on_load, None, standard_dd)
+    standard_dd.change(update_version_dd, standard_dd, version_dd)
+    version_dd.change(update_category_dd, [standard_dd, version_dd], category_dd)
+    category_dd.change(display_data, [standard_dd, version_dd, category_dd], output_df)
 
 if __name__ == "__main__":
-    demo.launch(theme=gr.themes.Soft())
+    demo.launch(theme=gr.themes.Soft())