feat: 去掉版权信息

py-src/del_copyright_img.py

@@ -0,0 +1,70 @@
+# 依赖目录pip install python-pptx Pillow imagehash
+from pptx import Presentation
+from pptx.util import Inches
+from PIL import Image
+import imagehash
+import os
+
+def is_similar_image(image_path, reference_image_path, similarity_threshold=0.7):
+    """
+    检查两张图片之间的相似度
+    """
+    img1 = Image.open(image_path)
+    img2 = Image.open(reference_image_path)
+
+    # 计算哈希值
+    hash1 = imagehash.average_hash(img1)
+    hash2 = imagehash.average_hash(img2)
+
+    # 计算相似度（1 - 汉明距离 / 哈希位数）
+    similarity = 1 - (hash1 - hash2) / len(hash1.hash) ** 2
+    return similarity >= similarity_threshold
+
+def remove_copyright_images(pptx_path, copyright_image_paths):
+    # 打开 PowerPoint 文件
+    prs = Presentation(pptx_path)
+
+    # 使用一个列表来记录要删除的幻灯片索引
+    slides_to_delete = []
+
+    # 遍历每一页幻灯片
+    for index, slide in enumerate(prs.slides):
+        # 遍历幻灯片中的每个形状
+        for shape in slide.shapes:
+            # 检查形状是否为图片
+            if shape.shape_type == 13:  # 13 表示图片
+                # 临时保存图片
+                temp_image_path = 'temp_image.png'
+                with open(temp_image_path, 'wb') as f:
+                    f.write(shape.image.blob)
+
+                # 检查相似度
+                for copyright_image in copyright_image_paths:
+                    if is_similar_image(temp_image_path, copyright_image):
+                        print(f"有相似的图片,在第{index + 1}页")
+                        # 删除该形状
+                        slide.shapes._spTree.remove(shape._element)
+
+                        slides_to_delete.append(index)  # 记录要删除的幻灯片
+                        break  # 找到相似图片后可以跳出循环
+
+                # 删除临时图片
+                os.remove(temp_image_path)
+
+    # 删除记录中的幻灯片
+    for index in slides_to_delete:
+        slides = list(prs.slides._sldIdLst)
+        current_slide_to_delete = slides[index]
+        prs.slides._sldIdLst.remove(current_slide_to_delete)
+
+    if(len(slides_to_delete) > 0):
+        # 保存修改后的 PowerPoint 文件
+        prs.save(pptx_path)
+    else:
+        print(f"{pptx_path}没有版权信息，不需要删除")
+
+# 使用示例
+pptx_file_path = '/Users/minya/Desktop/test/1.pptx'  # 输入 PowerPoint 文件路径
+copyright_image_paths = ['/Users/minya/Desktop/test/1.png']  # 版权图片路径列表
+
+remove_copyright_images(pptx_file_path, copyright_image_paths)

py-src/del_ppt.py

@@ -2,10 +2,6 @@ import os
 from pptx import Presentation
 from config import Config
 
-# 指定包含PPT文件的文件夹
-folder_path = "C:/Users/Administrator/Desktop/测试/ppt1010"
-
-
 def del_slide(presentation, slide_index):
     """
     删除某一张幻灯片并返回该幻灯片对象