Update utils/docx_builder.py

utils/docx_builder.py

@@ -3,14 +3,14 @@ import io
 import mistune
 from bs4 import BeautifulSoup
 from docx import Document
-from docx.shared import Pt
+from docx.shared import Inches, Pt
 
 class DocxRenderer(mistune.HTMLRenderer):
     def __init__(self):
         super().__init__()
         self.doc = Document()
         self.h_counters = [0] * 6
-        self.is_first_h1 = True
+        self.is_first_h1 = True # 新增：用于捕获首个H1（论文题目），避免为其编号
         try:
             style = self.doc.styles['Normal']
             style.font.name = 'Times New Roman'
@@ -33,8 +33,10 @@ class DocxRenderer(mistune.HTMLRenderer):
     def heading(self, text, level):
         clean_text = BeautifulSoup(text, 'html.parser').get_text().strip()
         lower_text = clean_text.lower()
+        # 黑名单：严格跳过这些关键字的自动编号
         skip_keywords = ["abstract", "keywords", "references", "摘要", "关键词", "关键字", "参考文献", "reference list", "аннотация", "ключевые слова", "түйіндеме", "кілт сөздер"]
         
+        # 拦截逻辑：若是黑名单、或者是文档首个H1（论文题目），直接输出不加编号
         if any(kw in lower_text for kw in skip_keywords) or (level == 1 and self.is_first_h1):
             if level == 1: self.is_first_h1 = False
             self.doc.add_heading(clean_text, level=1)
@@ -42,11 +44,15 @@ class DocxRenderer(mistune.HTMLRenderer):
 
         if level == 1: self.is_first_h1 = False
 
+        # === 核心自动编号逻辑 ===
         idx = level - 1
         if 0 <= idx < 6:
             self.h_counters[idx] += 1
+            # 将当前层级以下的子级计数器全部清零
             for i in range(idx + 1, 6): 
                 self.h_counters[i] = 0
+            
+            # 拼接数组生成编号，如 [2, 1, 0] -> "2.1"
             nums = self.h_counters[0:idx+1]
             num_str = ".".join(map(str, nums))
             self.doc.add_heading(f"{num_str} {clean_text}", level=level)