Add code/cube3d/training/compare_ldr.py

code/cube3d/training/compare_ldr.py

@@ -0,0 +1,152 @@
+import os
+from collections import defaultdict
+
+def count_part_frequencies(ldr_content):
+    """
+    精确统计.ldr文件中零件使用频率（基于1开头的行）
+    
+    参数:
+        ldr_content: .ldr文件的内容字符串
+    
+    返回:
+        defaultdict: 零件文件名 -> 出现次数
+    """
+    freq = defaultdict(int)
+    for line in ldr_content.splitlines():
+        line = line.strip()
+        if line.startswith('1 '):
+            parts = line.split()
+            if len(parts) >= 15:  # 确保是完整的零件定义行
+                part_filename = parts[-1].lower()  # 统一小写处理
+                freq[part_filename] += 1
+    return freq
+
+def build_frequency_index(folder_path):
+    """
+    为文件夹中的所有.ldr文件建立频率索引
+    
+    参数:
+        folder_path: 文件夹路径
+    
+    返回:
+        dict: {文件名: 频率字典}
+    """
+    freq_index = {}
+    for filename in os.listdir(folder_path):
+        if filename.lower().endswith('.ldr'):
+            file_path = os.path.join(folder_path, filename)
+            with open(file_path, 'r') as f:
+                freq_index[filename] = count_part_frequencies(f.read())
+    return freq_index
+
+def compare_folders(folder1, folder2):
+    """
+    比较两个文件夹中的.ldr文件，找出没有匹配的文件
+    
+    参数:
+        folder1: 第一个文件夹路径
+        folder2: 第二个文件夹路径
+    """
+    # 为两个文件夹建立频率索引
+    freq_index1 = build_frequency_index(folder1)
+    freq_index2 = build_frequency_index(folder2)
+    
+    # 转换第二个文件夹的索引为频率字典到文件名的映射
+    freq_to_files2 = defaultdict(list)
+    for filename, freq in freq_index2.items():
+        freq_key = frozenset(freq.items())  # 将字典转换为可哈希的frozenset
+        freq_to_files2[freq_key].append(filename)
+    
+    # 检查folder1中的每个文件是否在folder2中有匹配
+    unmatched_files = []
+    for filename1, freq1 in freq_index1.items():
+        freq_key = frozenset(freq1.items())
+        if freq_key not in freq_to_files2:
+            unmatched_files.append(filename1)
+    
+    # 输出结果
+    if unmatched_files:
+        print("以下文件在目标文件夹中没有找到匹配项:")
+        for filename in unmatched_files:
+            print(f" - {filename}")
+    else:
+        print("所有文件都在目标文件夹中找到了匹配项")
+
+def compare_two_ldr_files(file1, file2):
+    """
+    详细比较两个.ldr文件的零件使用差异
+    
+    参数:
+        file1: 第一个.ldr文件路径
+        file2: 第二个.ldr文件路径
+    
+    返回:
+        dict: 差异报告字典
+    """
+    with open(file1, 'r') as f1, open(file2, 'r') as f2:
+        freq1 = count_part_frequencies(f1.read())
+        freq2 = count_part_frequencies(f2.read())
+    
+    # 获取所有零件类别的并集
+    all_parts = set(freq1.keys()).union(set(freq2.keys()))
+    
+    differences = {
+        'only_in_file1': {},
+        'only_in_file2': {},
+        'count_differences': {}
+    }
+    
+    for part in sorted(all_parts):
+        count1 = freq1.get(part, 0)
+        count2 = freq2.get(part, 0)
+        
+        if count1 == 0 and count2 > 0:
+            differences['only_in_file2'][part] = count2
+        elif count1 > 0 and count2 == 0:
+            differences['only_in_file1'][part] = count1
+        elif count1 != count2:
+            differences['count_differences'][part] = (count1, count2)
+    
+    return differences
+
+def print_differences(differences, file1_name, file2_name):
+    """
+    打印两个文件的比较结果
+    
+    参数:
+        differences: compare_two_ldr_files()的返回结果
+        file1_name: 第一个文件名（用于显示）
+        file2_name: 第二个文件名（用于显示）
+    """
+    print(f"\n详细比较结果 ({file1_name} vs {file2_name}):")
+    
+    if differences['only_in_file1']:
+        print("\n只在第一个文件中出现的零件:")
+        for part, count in differences['only_in_file1'].items():
+            print(f" - {part}: {count}次")
+    
+    if differences['only_in_file2']:
+        print("\n只在第二个文件中出现的零件:")
+        for part, count in differences['only_in_file2'].items():
+            print(f" - {part}: {count}次")
+    
+    if differences['count_differences']:
+        print("\n出现次数不同的零件:")
+        for part, (count1, count2) in differences['count_differences'].items():
+            print(f" - {part}: {file1_name}={count1}次, {file2_name}={count2}次")
+    
+    if not any(differences.values()):
+        print("两个文件使用的零件完全相同")
+
+# 使用示例
+folder1 = '/public/home/wangshuo/gap/assembly/cubedit/outputs/sample/'
+folder2 = '/public/home/wangshuo/gap/assembly/data/car_1k/subset_self/ldr_l30_rotrans_expand_wom'
+
+# 比较两个文件夹
+#compare_folders(folder1, folder2)
+
+# 比较两个具体文件
+file1 = '/public/home/wangshuo/gap/assembly/cubedit/outputs/sample/test_0_drp_101_r512_uncond.ldr'
+file2 = '/public/home/wangshuo/gap/assembly/data/car_1k/subset_self/ldr_l30_rotrans_expand_wom/modified_car_121_rot.ldr'
+diff_result = compare_two_ldr_files(file1, file2)
+print_differences(diff_result, os.path.basename(file1), os.path.basename(file2))