Add code/cube3d/training/copy_obj_from_json.py

code/cube3d/training/copy_obj_from_json.py

@@ -0,0 +1,171 @@
+import os
+import shutil
+import json
+import re
+from collections import defaultdict
+
+
+def extract_leading_digits(prefix):
+    """提取前缀中最前面的连续数字，无数字则返回None"""
+    match = re.match(r'^(\d+)', prefix)
+    return match.group(1) if match else None
+
+
+def compare_json_and_prefixes(json_file_path):
+    """对比JSON中的.dat键名和提取的前缀，找出差异"""
+    try:
+        print("\n" + "=" * 80)
+        print("第一步：JSON键名与提取前缀对比分析")
+        print("=" * 80)
+        
+        # 1. 读取JSON所有.dat键名
+        json_keys = []
+        with open(json_file_path, 'r', encoding='utf-8') as f:
+            dat_mapping = json.load(f)
+            for key in dat_mapping.keys():
+                key_stripped = key.strip()
+                if key_stripped.endswith('.dat'):
+                    json_keys.append(key_stripped)
+        
+        total_json_keys = len(json_keys)
+        print(f"1. 从JSON中读取到 {total_json_keys} 个.dat键名")
+
+        # 2. 提取前缀并分析
+        extracted_prefixes = {}
+        prefix_to_keys = defaultdict(list)  # 记录前缀对应的所有键名
+        for key in json_keys:
+            original_prefix = os.path.splitext(key)[0].strip()
+            lower_prefix = original_prefix.lower()
+            extracted_prefixes[lower_prefix] = original_prefix
+            prefix_to_keys[lower_prefix].append(key)
+        
+        total_extracted = len(extracted_prefixes)
+        print(f"2. 提取到 {total_extracted} 个唯一前缀")
+
+        # 3. 分析差异原因
+        if total_extracted < total_json_keys:
+            diff_count = total_json_keys - total_extracted
+            print(f"\n⚠️  发现差异：前缀数比JSON键名少 {diff_count} 个")
+            print("   原因：存在不同键名对应相同前缀的情况（如下所示）")
+            
+            # 显示前5个存在重复前缀的例子
+            duplicate_prefixes = {p: keys for p, keys in prefix_to_keys.items() if len(keys) > 1}
+            print(f"\n   存在重复前缀的键名数量：{len(duplicate_prefixes)} 组")
+            for i, (prefix, keys) in enumerate(list(duplicate_prefixes.items())[:5], 1):
+                print(f"   示例 {i}: 前缀 '{prefix}' 对应 {len(keys)} 个键名 → {keys[:2]}...")
+        else:
+            print("\n✅ 前缀数与JSON键名数一致，无差异")
+
+        return json_keys, prefix_to_keys
+
+    except Exception as e:
+        print(f"❌ 对比分析失败: {str(e)}")
+        return [], {}
+
+
+def copy_obj_files(json_file_path, obj_source_dir, dest_dir, json_keys):
+    """复制OBJ文件，基于JSON中的.dat键名"""
+    try:
+        print("\n" + "=" * 80)
+        print("第二步：OBJ文件复制操作")
+        print("=" * 80)
+        
+        # 1. 初始化目录
+        os.makedirs(dest_dir, exist_ok=True)
+        print(f"📂 目标目录已准备：{dest_dir}")
+
+        # 2. 预处理OBJ文件
+        obj_lower_to_info = {}
+        for filename in os.listdir(obj_source_dir):
+            if filename.lower().endswith('.obj'):
+                obj_original_prefix = os.path.splitext(filename)[0].strip()
+                obj_lower_prefix = obj_original_prefix.lower()
+                obj_full_path = os.path.join(obj_source_dir, filename)
+                obj_lower_to_info[obj_lower_prefix] = (obj_original_prefix, obj_full_path)
+        print(f"🔍 已预处理 {len(obj_lower_to_info)} 个OBJ文件")
+
+        # 3. 提取所有dat前缀（保留原始顺序）
+        dat_prefixes = [os.path.splitext(key)[0].strip() for key in json_keys]
+        print(f"📋 待匹配的dat前缀总数：{len(dat_prefixes)} 个")
+
+        # 4. 第一次匹配：完整前缀匹配
+        first_copied = []  # [(dat前缀, OBJ文件名)]
+        for prefix in dat_prefixes:
+            lower_prefix = prefix.lower()
+            if lower_prefix in obj_lower_to_info:
+                obj_original, obj_path = obj_lower_to_info[lower_prefix]
+                obj_filename = os.path.basename(obj_path)
+                dest_path = os.path.join(dest_dir, obj_filename)
+                
+                if not os.path.exists(dest_path):
+                    shutil.copy2(obj_path, dest_path)
+                first_copied.append((prefix, obj_filename))
+        
+        print(f"\n第一次匹配完成：{len(first_copied)} 个前缀匹配成功")
+
+        # 5. 第二次匹配：未匹配前缀提取前导数字
+        matched_prefixes = {p for p, _ in first_copied}
+        not_found_first = [p for p in dat_prefixes if p not in matched_prefixes]
+        print(f"初始未匹配的前缀：{len(not_found_first)} 个")
+
+        second_copied = []  # [(dat前缀, OBJ文件名, 提取的数字)]
+        if not_found_first:
+            for prefix in not_found_first:
+                leading_digits = extract_leading_digits(prefix)
+                if not leading_digits:
+                    continue
+                
+                target_lower = leading_digits.lower()
+                if target_lower in obj_lower_to_info:
+                    obj_original, obj_path = obj_lower_to_info[target_lower]
+                    obj_filename = os.path.basename(obj_path)
+                    dest_path = os.path.join(dest_dir, obj_filename)
+                    
+                    if not os.path.exists(dest_path):
+                        shutil.copy2(obj_path, dest_path)
+                    second_copied.append((prefix, obj_filename, leading_digits))
+        
+        print(f"第二次匹配完成：{len(second_copied)} 个前缀匹配成功")
+
+        # 6. 最终结果汇总
+        print("\n" + "=" * 60)
+        print("最终结果汇总")
+        print("=" * 60)
+        total_matched = len(first_copied) + len(second_copied)
+        print(f"📊 统计信息：")
+        print(f"   - JSON中.dat键名总数：{len(json_keys)}")
+        print(f"   - 第一次匹配成功：{len(first_copied)}")
+        print(f"   - 第二次匹配成功：{len(second_copied)}")
+        print(f"   - 总匹配成功：{total_matched}")
+        print(f"   - 最终未匹配：{len(dat_prefixes) - total_matched}")
+
+        print(f"\n🏁 所有操作完成！复制的文件保存至：{dest_dir}")
+
+    except Exception as e:
+        print(f"❌ 复制操作失败: {str(e)}")
+
+
+def main():
+    # 配置文件路径
+    JSON_FILE_PATH = "/public/home/wangshuo/gap/assembly/data/car_1k/subset_self/label_mapping_freq.json"
+    OBJ_SOURCE_DIR = "/public/home/wangshuo/gap/assembly/data/part_obj"
+    DEST_DIR = "/public/home/wangshuo/gap/assembly/data/part_obj_300"
+
+    # 执行完整流程
+    print("=" * 80)
+    print("JSON键名分析与OBJ文件复制工具")
+    print("=" * 80)
+    
+    # 第一步：分析JSON键名与前缀的关系
+    json_keys, _ = compare_json_and_prefixes(JSON_FILE_PATH)
+    if not json_keys:
+        print("❌ 未获取到有效的JSON键名，无法继续操作")
+        return
+    
+    # 第二步：基于分析结果进行OBJ文件复制
+    copy_obj_files(JSON_FILE_PATH, OBJ_SOURCE_DIR, DEST_DIR, json_keys)
+
+
+if __name__ == "__main__":
+    main()
+