Upload app.py

app.py

@@ -37,83 +37,52 @@ def format_docx(file, chapter_keywords):
         # 合併所有模式
         combined_pattern = '|'.join(patterns)
         
-        # 先標記空段落，準備刪除多餘的換行
-        paragraphs_list = list(doc.paragraphs)
-        to_delete = []
+        # 處理文件：先收集所有段落內容和類型
+        content_list = []
         
-        for i in range(len(paragraphs_list)):
-            current_para = paragraphs_list[i]
-            current_empty = len(current_para.text.strip()) == 0
+        for paragraph in doc.paragraphs:
+            text = paragraph.text.strip()
             
-            # 如果是空段落，檢查前一個段落
-            if current_empty and i > 0:
-                prev_para = paragraphs_list[i-1]
-                prev_empty = len(prev_para.text.strip()) == 0
-                
-                # 如果前一個段落也是空的，標記當前段落刪除
-                if prev_empty:
-                    to_delete.append(current_para)
+            # 檢查是否為章節標題
+            if text and re.search(combined_pattern, text):
+                content_list.append(('heading', text))
+            elif text:  # 有內容的一般段落
+                content_list.append(('paragraph', text))
+            else:  # 空段落
+                # 只有在前一個不是空段落時才添加空段落
+                if not content_list or content_list[-1][0] != 'empty':
+                    content_list.append(('empty', ''))
         
-        # 刪除多餘的空段落（從後往前刪除）
-        for para in reversed(to_delete):
-            p = para._element
+        # 清空整個文件
+        for paragraph in doc.paragraphs:
+            p = paragraph._element
             p.getparent().remove(p)
         
-        # 處理每個段落
-        for paragraph in doc.paragraphs:
-            # 跳過空段落
-            if len(paragraph.text.strip()) == 0:
-                continue
-                
-            # 1. 將樣式改為無間距
-            paragraph.style.paragraph_format.space_before = Cm(0)
-            paragraph.style.paragraph_format.space_after = Cm(0)
-            paragraph.style.paragraph_format.line_spacing = 1.0
-            
-            # 2. 首行縮排兩個字元位置（約0.7公分）
-            paragraph.paragraph_format.left_indent = Cm(0)  # 整體不縮排
-            paragraph.paragraph_format.first_line_indent = Cm(0.7)  # 只有首行縮排
-            
-            # 3. 檢查是否為章節標題
-            if re.search(combined_pattern, paragraph.text):
-                # 保存原始文字
-                original_text = paragraph.text
-                
-                # 嘗試設定標題 1 樣式（處理不同語言版本的樣式名稱）
-                try:
-                    # 英文版本
-                    paragraph.style = doc.styles['Heading 1']
-                except KeyError:
-                    try:
-                        # 中文版本
-                        paragraph.style = doc.styles['標題 1']
-                    except KeyError:
-                        try:
-                            # 其他可能的名稱
-                            paragraph.style = doc.styles['Heading1']
-                        except KeyError:
-                            # 如果都找不到，創建自定義標題樣式
-                            from docx.enum.style import WD_STYLE_TYPE
-                            try:
-                                heading_style = doc.styles.add_style('CustomHeading1', WD_STYLE_TYPE.PARAGRAPH)
-                                heading_style.font.bold = True
-                                heading_style.font.size = Pt(16)
-                                paragraph.style = heading_style
-                            except:
-                                # 最後手段：直接設定格式
-                                paragraph.style = doc.styles['Normal']
-                
-                # 確保粗體設定和字體大小
-                for run in paragraph.runs:
-                    run.font.bold = True
-                    run.font.size = Pt(16)
+        # 重新建立文件內容
+        for content_type, text in content_list:
+            if content_type == 'heading':
+                # 使用 add_heading 方法建立標題
+                heading = doc.add_heading(text, level=1)
+                # 在標題前分頁
+                heading.paragraph_format.page_break_before = True
+                # 重置標題的縮排
+                heading.paragraph_format.left_indent = Cm(0)
+                heading.paragraph_format.first_line_indent = Cm(0)
                 
-                # 4. 在章節前分頁
-                paragraph.paragraph_format.page_break_before = True
+            elif content_type == 'paragraph':
+                # 添加一般段落
+                para = doc.add_paragraph(text)
+                # 設定樣式
+                para.style.paragraph_format.space_before = Cm(0)
+                para.style.paragraph_format.space_after = Cm(0)
+                para.style.paragraph_format.line_spacing = 1.0
+                # 首行縮排
+                para.paragraph_format.left_indent = Cm(0)
+                para.paragraph_format.first_line_indent = Cm(0.7)
                 
-                # 重置章節標題的縮排（標題不需要首行縮排）
-                paragraph.paragraph_format.left_indent = Cm(0)
-                paragraph.paragraph_format.first_line_indent = Cm(0)
+            elif content_type == 'empty':
+                # 添加空段落
+                doc.add_paragraph('')
         
         # 儲存處理後的文件
         output_path = tempfile.mktemp(suffix='.docx')