Upload 7 files

old/dataset/1_caption_remove_space.py

@@ -0,0 +1,24 @@
+import os
+
+# 获取当前工作目录
+current_dir = os.getcwd()
+
+# 遍历当前目录下的所有文件
+for filename in os.listdir(current_dir):
+  # 检查是否为 txt 文件
+  if filename.endswith('.txt'):
+    filepath = os.path.join(current_dir, filename)
+
+    # 读取文件内容并去掉换行符
+    try:
+      with open(filepath, 'r', encoding='utf-8') as f:
+        # 读取所有行，并用空字符连接，达到去掉换行符的效果
+        content = ''.join(line.strip() for line in f)
+
+      # 写回原文件
+      with open(filepath, 'w', encoding='utf-8') as f:
+        f.write(content)
+
+      print(f'已处理: {filename}')
+    except Exception as e:
+      print(f'处理 {filename} 失败: {e}')

old/dataset/2_img_compressor.bat

@@ -0,0 +1,24 @@
+@echo off
+setlocal enabledelayedexpansion
+
+echo ========================================================
+echo 警告：此操作将批量将当前目录及子目录下的所有 .png 图片
+echo 转换为 .jpg 格式，并【永久删除】原始的 .png 文件！
+echo ========================================================
+set /p "choice=确定要继续吗？(Y/N): "
+
+if /i "%choice%" NEQ "Y" (
+    echo 操作已取消。
+    pause
+    exit /b
+)
+
+for /R %%f in (*.png) do (
+    ffmpeg -i "%%f" -q:v 2 "%%~dpnf.jpg"
+    if exist "%%~dpnf.jpg" (
+        del "%%f"
+    )
+)
+
+echo Conversion complete.
+pause

old/dataset/3_mask_clean_metadata.py

@@ -0,0 +1,44 @@
+import os
+from PIL import Image
+
+
+def remove_all_png_metadata_in_current_dir():
+  # 获取当前脚本所在的目录
+  current_dir = os.getcwd()
+  print(f"正在处理目录: {current_dir}\n" + "-" * 40)
+
+  success_count = 0
+
+  # 遍历当前目录及所有子目录
+  for root, dirs, files in os.walk(current_dir):
+    for file in files:
+      if file.lower().endswith(".png"):
+        file_path = os.path.join(root, file)
+        try:
+          with Image.open(file_path) as img:
+            # 读取图片像素数据，彻底与原图的元数据对象断开连接
+            data = list(img.getdata())
+            clean_img = Image.new(img.mode, img.size)
+            clean_img.putdata(data)
+
+            # 覆盖保存为干净的 PNG
+            clean_img.save(file_path, "PNG")
+
+          print(f"成功清除: {file_path}")
+          success_count += 1
+        except Exception as e:
+          print(f"处理失败 [{file}]: {e}")
+
+  print("-" * 40)
+  print(f"处理完成！共清理了 {success_count} 张 PNG 图片的元数据。")
+
+
+if __name__ == "__main__":
+  # 提示用户，防止误操作覆盖重要数据
+  confirm = input(
+      "此脚本将直接【覆盖】当前目录下所有 PNG 文件并移除其元数据，是否继续？(y/n): "
+  )
+  if confirm.lower() == "y":
+    remove_all_png_metadata_in_current_dir()
+  else:
+    print("操作已取消。")

old/dataset/4_gen_par_yaml_update_md.py

@@ -0,0 +1,175 @@
+import os
+import glob
+import pandas as pd
+from datasets import Dataset, Features, Image, Value
+from huggingface_hub import HfApi, DatasetCard
+import tempfile
+
+# ==========================================
+# 配置区：填写你的 Hugging Face 仓库 ID
+# ==========================================
+REPO_ID = "waveshaker/dataset_collections" 
+REPO_TYPE = "dataset"
+HF_TOKEN = "***" # 如果只传 README，依然需要 Write 权限的 Token
+# ==========================================
+
+def generate_parquet_for_folder(directory):
+    """扫描文件夹，直接生成包含图片的 train.parquet"""
+    jpg_files = glob.glob(os.path.join(directory, "*.jpg")) + glob.glob(os.path.join(directory, "*.jpeg"))
+    if not jpg_files:
+        return False
+
+    parquet_path = os.path.join(directory, "train.parquet")
+    data_rows = []
+
+    print(f"  📦 正在打包图片: {os.path.basename(directory)}")
+
+    for jpg_path in jpg_files:
+        file_name = os.path.basename(jpg_path)
+        base_name = os.path.splitext(file_name)[0]
+        
+        txt_path = os.path.join(directory, f"{base_name}.txt")
+        text_content = ""
+        if os.path.exists(txt_path):
+            with open(txt_path, "r", encoding="utf-8") as f:
+                text_content = f.read().strip()
+
+        try:
+            with open(jpg_path, "rb") as f:
+                image_bytes = {"bytes": f.read(), "path": None}
+        except Exception as e:
+            print(f"    ⚠️ 跳过损坏图片 {file_name}: {e}")
+            continue
+
+        data_rows.append({
+            "image": image_bytes,
+            "file_name": file_name,
+            "text": text_content
+        })
+
+    if not data_rows:
+        return False
+
+    df = pd.DataFrame(data_rows)
+    features = Features({
+        "image": Image(),
+        "file_name": Value("string"),
+        "text": Value("string")
+    })
+    ds = Dataset.from_pandas(df, features=features)
+    ds.to_parquet(parquet_path)
+    print(f"  ✅ 已生成: {os.path.basename(parquet_path)} ({len(data_rows)} 条数据)")
+    return True
+
+def get_region_priority(config_name, path_str):
+    priority = {"CN": 1, "KR": 2, "JP": 3, "FICTIONAL": 4, "WE": 5}
+    path_prefix = path_str.split('/')[0].upper() if path_str else ""
+    if not path_prefix:
+        for key in priority.keys():
+            if key.lower() in config_name.lower():
+                return priority[key]
+        return 99
+    return priority.get(path_prefix, 99)
+
+def main():
+    api = HfApi(token=HF_TOKEN)
+    current_dir = os.getcwd() 
+    
+    # 1. 交互输入地区前缀
+    print("\n" + "="*50)
+    print("请选择该数据集的地区/分类前缀：")
+    print("1. CN | 2. KR | 3. JP | 4. Fictional | 5. WE | 6. 自定义")
+    choice = input("\n👉 请输入数字或直接输入前缀 (默认回车为 KR): ").strip().upper()
+    if not choice: choice = "2"
+    mapping = {"1": "CN", "2": "KR", "3": "JP", "4": "FICTIONAL", "5": "WE"}
+    region_prefix = mapping.get(choice, choice).replace(" ", "_").replace("-", "_")
+    print(f"✅ 已选择前缀: {region_prefix}\n" + "="*50 + "\n")
+
+    local_folder_name = os.path.basename(current_dir).replace(" ", "_").replace("-", "_")
+
+    # 2. 拉取云端 YAML
+    print(f"☁️ 正在读取云端 {REPO_ID} 的现有配置...")
+    try:
+        remote_readme_path = api.hf_hub_download(repo_id=REPO_ID, filename="README.md", repo_type=REPO_TYPE)
+        with open(remote_readme_path, "r", encoding="utf-8") as f:
+            content = f.read()
+        card = DatasetCard(content)
+        print("  ✅ 成功读取云端 README。")
+    except Exception as e:
+        print(f"  ⚠️ 找不到云端 README，将新建: {e}")
+        card = DatasetCard("")
+
+    yaml_data = card.data if card.data else {}
+    if "configs" not in yaml_data: yaml_data["configs"] = []
+    
+    yaml_data["dataset_info"] = {
+        "features": [
+            {"name": "image", "dtype": "image"},
+            {"name": "file_name", "dtype": "string"},
+            {"name": "text", "dtype": "string"}
+        ]
+    }
+
+    existing_config_names = [c.get("config_name") for c in yaml_data["configs"]]
+
+    # 3. 本地打包并新增配置
+    new_configs_added = False
+
+    print(f"🔍 正在扫描本地目录: {current_dir}")
+    for sub_folder_name in os.listdir(current_dir):
+        sub_folder_path = os.path.join(current_dir, sub_folder_name)
+        
+        if os.path.isdir(sub_folder_path) and not sub_folder_name.startswith('.'):
+            if generate_parquet_for_folder(sub_folder_path):
+                
+                config_name = f"{local_folder_name}_{sub_folder_name}".lower()
+                path_str = f"{region_prefix}/{local_folder_name}/{sub_folder_name}/train.parquet"
+                
+                if config_name not in existing_config_names:
+                    yaml_data["configs"].append({
+                        "config_name": config_name,
+                        "data_files": [{"split": "train", "path": path_str}]
+                    })
+                    new_configs_added = True
+                    print(f"  ➕ 已添加配置: {config_name}")
+
+    # 4. 排序
+    def sort_key(config):
+        data_files = config.get("data_files", [])
+        path = data_files[0].get("path", "") if data_files else ""
+        config_name = config.get("config_name", "")
+        priority = get_region_priority(config_name, path)
+        return (priority, config_name)
+
+    yaml_data["configs"] = sorted(yaml_data["configs"], key=sort_key)
+    print("\n🔄 已根据 [CN > KR > JP > Fictional > WE] 的顺序对 YAML 进行了重排序。")
+
+    # 5. 保存并仅推送 README.md
+    if new_configs_added:
+        card.data = yaml_data
+        
+        with tempfile.NamedTemporaryFile(mode='w', encoding='utf-8', suffix='.md', delete=False) as temp_readme:
+            temp_readme.write(str(card))
+            temp_readme_path = temp_readme.name
+
+        print("\n🚀 开始推送最新的 README.md 到 Hugging Face...")
+        try:
+            api.upload_file(
+                path_or_fileobj=temp_readme_path,
+                path_in_repo="README.md",
+                repo_id=REPO_ID,
+                repo_type=REPO_TYPE,
+                commit_message="Auto-update README YAML configs"
+            )
+            print("\n🎉 README.md 已成功更新到云端！")
+            print("👉 别忘了用你的上传脚本把本地新生成的 train.parquet 和图片传上去哦。")
+        except Exception as e:
+            print(f"\n❌ README.md 上传失败: {e}")
+        finally:
+            os.remove(temp_readme_path)
+    else:
+        print("\n✅ 云端 YAML 已经包含了当前目录的配置，无需更新 README。")
+        print("👉 你只需要用你的上传脚本把本地文件传上去即可。")
+
+if __name__ == "__main__":
+    main()

old/dataset/5_upload2data.py

@@ -0,0 +1,68 @@
+import os
+from huggingface_hub import HfApi
+
+# ==========================================
+# 配置区
+# ==========================================
+api = HfApi(token="***")
+
+repo_id = "waveshaker/character_loras" # 你的仓库名
+repo_type = "dataset" # 如果你的仓库是模型仓库，请改为 "model"
+# ==========================================
+
+def main():
+    current_dir = os.getcwd()
+    
+    # 1. 交互输入地区前缀 (已按最新要求重排序)
+    print("="*50)
+    print("请选择这批数据上传到哪个地区前缀下：")
+    print("1. CN | 2. KR | 3. JP | 4. Fictional | 5. WE | 6. 自定义")
+    choice = input("\n👉 请输入数字或直接输入前缀 (默认回车为 KR): ").strip().upper()
+    
+    # 默认回车选 KR (2)
+    if not choice: 
+        choice = "2"
+    
+    # 映射字典同步更新
+    mapping = {"1": "CN", "2": "KR", "3": "JP", "4": "Fictional", "5": "WE"}
+    region_prefix = mapping.get(choice, choice).replace(" ", "_")
+    
+    # 2. 自动获取当前所在的母文件夹名字
+    local_folder_name = os.path.basename(current_dir).replace(" ", "_")
+    
+    print(f"\n✅ 地区前缀: {region_prefix}")
+    print(f"✅ 母文件夹名: {local_folder_name}")
+    print("="*50 + "\n")
+
+    # 3. 遍历并上传所有子文件夹
+    for sub_folder_name in os.listdir(current_dir):
+        sub_folder_path = os.path.join(current_dir, sub_folder_name)
+        
+        # 过滤掉非文件夹和隐藏文件
+        if os.path.isdir(sub_folder_path) and not sub_folder_name.startswith('.'):
+            
+            # 动态构建云端路径: 例如 KR/YuJinNyang_KR2/1_clothing
+            path_in_repo = f"{region_prefix}/{local_folder_name}/{sub_folder_name}"
+            
+            print(f"📦 开始上传: {sub_folder_name}")
+            print(f"📂 目标路径: {path_in_repo}")
+            
+            try:
+                api.upload_folder(
+                    folder_path=sub_folder_path,
+                    repo_id=repo_id,
+                    repo_type=repo_type,
+                    path_in_repo=path_in_repo,
+                    # 动态 commit 信息
+                    commit_message=f"Update files for {path_in_repo}",
+                    # 👇 核心白名单：只上传这四种格式，完美过滤掉 .py 等垃圾文件
+                    allow_patterns=["*.jpg", "*.jpeg", "*.txt", "*.png", "*.parquet"] 
+                )
+                print(f"🎉 {sub_folder_name} 上传完成！\n")
+            except Exception as e:
+                print(f"❌ {sub_folder_name} 上传失败: {e}\n")
+
+    print("✅ 所有子文件夹处理完毕！")
+
+if __name__ == "__main__":
+    main()

old/dataset/aio.bat

@@ -0,0 +1,27 @@
+@echo off
+chcp 65001 >nul
+title 5合1 数据集自动化流水线
+color 07
+
+echo ====================================================
+echo    欢迎使用 LoRA 数据集 5合1 自动化流水线
+echo    正在启动...
+echo ====================================================
+echo.
+
+:: 检查 Python 是否安装
+python --version >nul 2>&1
+if %errorlevel% neq 0 (
+    echo [错误] 未检测到 Python！请先安装 Python 并配置环境变量。
+    pause
+    exit /b
+)
+
+:: 核心修改：调用 C 盘的绝对路径脚本，但工作目录保持在当前文件夹
+python "F:\tool\aio.py"
+
+echo.
+echo ====================================================
+echo    任务执行结束！
+echo ====================================================
+pause

old/dataset/aio.py

@@ -0,0 +1,302 @@
+import os
+import glob
+import subprocess
+import pandas as pd
+from datasets import Dataset, Features, Image as HFImage, Value
+from huggingface_hub import HfApi, DatasetCard
+import tempfile
+from PIL import Image
+
+# ==========================================
+# 配置区
+# ==========================================
+REPO_ID = "waveshaker/dataset_collections" 
+REPO_TYPE = "dataset"
+HF_TOKEN = ""  # 请填入你的 Token
+# ==========================================
+
+def clean_txt_newlines(current_dir):
+    """1. 递归清理所有 txt 文件中的换行符"""
+    print("\n" + "="*50)
+    print("🧹 [任务 1/5] 正在清理所有 TXT 文件中的换行符...")
+    print("="*50)
+    
+    txt_files = glob.glob(os.path.join(current_dir, "**", "*.txt"), recursive=True)
+    if not txt_files:
+        print("  ✅ 没有找到需要处理的 TXT 文件。")
+        return
+
+    cleaned_count = 0
+    for txt_path in txt_files:
+        # 跳过根目录下的 txt（可能是说明文档）
+        if os.path.dirname(txt_path) == current_dir:
+            continue
+            
+        try:
+            with open(txt_path, 'r', encoding='utf-8') as f:
+                content = ''.join(line.strip() for line in f)
+            
+            with open(txt_path, 'w', encoding='utf-8') as f:
+                f.write(content)
+            cleaned_count += 1
+        except Exception as e:
+            print(f"  ⚠️ 处理 {os.path.basename(txt_path)} 失败: {e}")
+            
+    print(f"  🎉 TXT 清理完成！共处理了 {cleaned_count} 个文件。")
+
+def remove_mask_metadata(current_dir):
+    """2. 递归剥离 masklabel.png 的元数据，保护隐私"""
+    print("\n" + "="*50)
+    print("🛡️ [任务 2/5] 正在剥离 Mask 蒙版图 (masklabel.png) 的元数据...")
+    print("="*50)
+    
+    png_files = glob.glob(os.path.join(current_dir, "**", "*.png"), recursive=True)
+    mask_files = [f for f in png_files if "masklabel" in os.path.basename(f).lower()]
+    
+    if not mask_files:
+        print("  ✅ 没有找到 masklabel.png 文件，跳过此步骤。")
+        return
+
+    success_count = 0
+    for file_path in mask_files:
+        try:
+            with Image.open(file_path) as img:
+                # 读取像素数据，彻底断开原元数据
+                data = list(img.getdata())
+                clean_img = Image.new(img.mode, img.size)
+                clean_img.putdata(data)
+                # 覆盖保存
+                clean_img.save(file_path, "PNG")
+            success_count += 1
+            print(f"  🔒 已清理元数据: {os.path.basename(file_path)}")
+        except Exception as e:
+            print(f"  ⚠️ 处理失败 [{os.path.basename(file_path)}]: {e}")
+
+    print(f"  🎉 元数据清理完成！共处理了 {success_count} 张 Mask 图。")
+
+def convert_png_to_jpg(current_dir):
+    """3. 批量转换常规 png 为 jpg，并删除原 png（绝对跳过 masklabel）"""
+    print("\n" + "="*50)
+    print("🎬 [任务 3/5] 正在批量转换常规 PNG 为 JPG...")
+    print("="*50)
+    
+    png_files = glob.glob(os.path.join(current_dir, "**", "*.png"), recursive=True)
+    if not png_files:
+        print("  ✅ 没有找到需要转换的 PNG 文件。")
+        return
+
+    converted_count = 0
+    for png_path in png_files:
+        # 🚨 核心过滤：跳过 masklabel 文件，保证蒙版依然是 png
+        if "masklabel" in png_path.lower():
+            continue
+            
+        jpg_path = os.path.splitext(png_path)[0] + ".jpg"
+        
+        try:
+            subprocess.run(
+                ["ffmpeg", "-y", "-i", png_path, "-q:v", "2", jpg_path],
+                stdout=subprocess.DEVNULL, 
+                stderr=subprocess.DEVNULL,
+                check=True
+            )
+            os.remove(png_path)
+            converted_count += 1
+            print(f"  🔄 转换: {os.path.basename(png_path)} -> {os.path.basename(jpg_path)}")
+        except subprocess.CalledProcessError:
+            print(f"  ⚠️ 转换失败: {png_path}")
+        except FileNotFoundError:
+            print("  ❌ 错误: 找不到 ffmpeg！请确保已安装并配置了环境变量。")
+            return
+
+    print(f"  🎉 转换完成！共转换 {converted_count} 张图片。")
+
+def generate_parquet_and_update_yaml(api, current_dir, region_prefix):
+    """4. 本地打包生成 parquet，拉取云端 YAML，追加并重排"""
+    print("\n" + "="*50)
+    print("📦 [任务 4/5] 正在本地打包 Parquet 并更新 YAML 配置...")
+    print("="*50)
+    
+    local_folder_name = os.path.basename(current_dir).replace(" ", "_").replace("-", "_")
+    
+    print(f"  ☁️ 正在读取云端 {REPO_ID} 的现有配置...")
+    try:
+        remote_readme_path = api.hf_hub_download(repo_id=REPO_ID, filename="README.md", repo_type=REPO_TYPE)
+        with open(remote_readme_path, "r", encoding="utf-8") as f:
+            content = f.read()
+        card = DatasetCard(content)
+        print("  ✅ 成功读取云端 README。")
+    except Exception as e:
+        print(f"  ⚠️ 找不到云端 README，将新建: {e}")
+        card = DatasetCard("")
+
+    yaml_data = card.data if card.data else {}
+    if "configs" not in yaml_data: yaml_data["configs"] = []
+    
+    yaml_data["dataset_info"] = {
+        "features": [
+            {"name": "image", "dtype": "image"},
+            {"name": "file_name", "dtype": "string"},
+            {"name": "text", "dtype": "string"}
+        ]
+    }
+
+    existing_config_names = [c.get("config_name") for c in yaml_data["configs"]]
+    new_configs_added = False
+    valid_folders = []
+
+    for sub_folder_name in os.listdir(current_dir):
+        sub_folder_path = os.path.join(current_dir, sub_folder_name)
+        
+        if os.path.isdir(sub_folder_path) and not sub_folder_name.startswith('.'):
+            jpg_files = glob.glob(os.path.join(sub_folder_path, "*.jpg")) + glob.glob(os.path.join(sub_folder_path, "*.jpeg"))
+            if not jpg_files:
+                continue
+                
+            print(f"  📦 正在打包: {sub_folder_name}")
+            data_rows = []
+            
+            for jpg_path in jpg_files:
+                file_name = os.path.basename(jpg_path)
+                base_name = os.path.splitext(file_name)[0]
+                txt_path = os.path.join(sub_folder_path, f"{base_name}.txt")
+                
+                text_content = ""
+                if os.path.exists(txt_path):
+                    with open(txt_path, "r", encoding="utf-8") as f:
+                        text_content = f.read().strip()
+
+                try:
+                    with open(jpg_path, "rb") as f:
+                        image_bytes = {"bytes": f.read(), "path": None}
+                except Exception as e:
+                    print(f"    ⚠️ 跳过损坏图片 {file_name}: {e}")
+                    continue
+
+                data_rows.append({
+                    "image": image_bytes,
+                    "file_name": file_name,
+                    "text": text_content
+                })
+
+            if data_rows:
+                df = pd.DataFrame(data_rows)
+                features = Features({
+                    "image": HFImage(),  # 👈 这里加上了 HF
+                    "file_name": Value("string"),
+                    "text": Value("string")
+                })
+                ds = Dataset.from_pandas(df, features=features)
+                ds.to_parquet(os.path.join(sub_folder_path, "train.parquet"))
+                valid_folders.append(sub_folder_name)
+                print(f"    ✅ 生成 train.parquet ({len(data_rows)} 条数据)")
+
+                config_name = f"{local_folder_name}_{sub_folder_name}".lower()
+                path_str = f"{region_prefix}/{local_folder_name}/{sub_folder_name}/train.parquet"
+                
+                if config_name not in existing_config_names:
+                    yaml_data["configs"].append({
+                        "config_name": config_name,
+                        "data_files": [{"split": "train", "path": path_str}]
+                    })
+                    new_configs_added = True
+                    print(f"    ➕ 已添加配置: {config_name}")
+
+    def get_priority(config):
+        data_files = config.get("data_files", [])
+        path = data_files[0].get("path", "") if data_files else ""
+        # 直接使用原始大小写，确保与云端路径完全一致
+        priority_dict = {"CN": 1, "KR": 2, "JP": 3, "Fictional": 4, "WE": 5}
+        path_prefix = path.split('/')[0] if path else ""
+        return priority_dict.get(path_prefix, 99)
+
+    yaml_data["configs"] = sorted(yaml_data["configs"], key=lambda c: (get_priority(c), c.get("config_name", "")))
+    print("  🔄 已根据 [CN > KR > JP > Fictional > WE] 的顺序重排 YAML。")
+
+    if new_configs_added:
+        card.data = yaml_data
+        with tempfile.NamedTemporaryFile(mode='w', encoding='utf-8', suffix='.md', delete=False) as temp_readme:
+            temp_readme.write(str(card))
+            return temp_readme.name, valid_folders
+    return None, valid_folders
+
+def upload_files_and_readme(api, current_dir, region_prefix, temp_readme_path, valid_folders):
+    """5. 上传文件及 README"""
+    print("\n" + "="*50)
+    print("🚀 [任务 5/5] 正在上传文件到 Hugging Face...")
+    print("="*50)
+    
+    local_folder_name = os.path.basename(current_dir).replace(" ", "_").replace("-", "_")
+    
+    for sub_folder_name in valid_folders:
+        sub_folder_path = os.path.join(current_dir, sub_folder_name)
+        path_in_repo = f"{region_prefix}/{local_folder_name}/{sub_folder_name}"
+        
+        print(f"  📦 开始上传: {sub_folder_name} -> {path_in_repo}")
+        try:
+            api.upload_folder(
+                folder_path=sub_folder_path,
+                repo_id=REPO_ID,
+                repo_type=REPO_TYPE,
+                path_in_repo=path_in_repo,
+                commit_message=f"Update files for {path_in_repo}",
+                # 上传过滤：保留 masklabel.png，jpg，txt，parquet
+                allow_patterns=["*.jpg", "*.jpeg", "*.txt", "*.png", "*.parquet"]
+            )
+            print(f"  🎉 {sub_folder_name} 上传完成！")
+        except Exception as e:
+            print(f"  ❌ {sub_folder_name} 上传失败: {e}")
+
+    if temp_readme_path:
+        print("\n  📝 正在推送最新的 README.md...")
+        try:
+            api.upload_file(
+                path_or_fileobj=temp_readme_path,
+                path_in_repo="README.md",
+                repo_id=REPO_ID,
+                repo_type=REPO_TYPE,
+                commit_message="Auto-update README YAML configs"
+            )
+            print("  🎉 README.md 已成功更新到云端！")
+        except Exception as e:
+            print(f"  ❌ README.md 上传失败: {e}")
+        finally:
+            os.remove(temp_readme_path)
+
+def main():
+    api = HfApi(token=HF_TOKEN)
+    current_dir = os.getcwd() 
+    
+    print("\n" + "="*50)
+    print("🌟 欢迎使用 5合1 数据集自动化流水线 🌟")
+    print("="*50)
+    
+    print("请选择该数据集的地区/分类前缀：")
+    print("1. CN | 2. KR | 3. JP | 4. Fictional | 5. WE | 6. 自定义")
+    choice = input("\n👉 请输入数字或直接输入前缀 (默认回车为 KR): ").strip().upper()
+    if not choice: choice = "2"
+    mapping = {"1": "CN", "2": "KR", "3": "JP", "4": "Fictional", "5": "WE"}
+    region_prefix = mapping.get(choice, choice).replace(" ", "_").replace("-", "_")
+    print(f"\n✅ 已选择前缀: {region_prefix}")
+
+    # 任务 1：清理 txt 换行符
+    clean_txt_newlines(current_dir)
+
+    # 任务 2：清理 Mask 元数据
+    remove_mask_metadata(current_dir)
+
+    # 任务 3：转换 PNG -> JPG
+    convert_png_to_jpg(current_dir)
+
+    # 任务 4：打包 Parquet 并更新 YAML
+    temp_readme_path, valid_folders = generate_parquet_and_update_yaml(api, current_dir, region_prefix)
+
+    # 任务 5：上传
+    upload_files_and_readme(api, current_dir, region_prefix, temp_readme_path, valid_folders)
+
+    print("\n" + "="*50)
+    print("🎉 5合1 全流程处理完毕！")
+    print("="*50)
+
+if __name__ == "__main__":
+    main()