|
|
|
@ -65,6 +65,65 @@ def extract_level2_titles(markdown_content):
|
|
|
|
|
return unique_matches # 返回去重后的二级目录标题数组
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def extract_level2_and_level3(markdown_content):
|
|
|
|
|
"""
|
|
|
|
|
遍历输出所有二级目录及其三级目录,并生成指定格式的 JSON 对象
|
|
|
|
|
"""
|
|
|
|
|
# 使用正则表达式匹配所有二级目录和三级目录及其内容
|
|
|
|
|
matches = re.findall(r'^(##|###)\s+(.+)$|^-\s+(.+)$', markdown_content, re.MULTILINE)
|
|
|
|
|
|
|
|
|
|
# 遍历匹配结果,组织层级关系
|
|
|
|
|
current_level2 = None
|
|
|
|
|
current_level3 = None
|
|
|
|
|
level3_items = [] # 用于存储三级目录下的内容
|
|
|
|
|
|
|
|
|
|
for level, title, item in matches:
|
|
|
|
|
if level == '##': # 二级目录
|
|
|
|
|
if current_level2:
|
|
|
|
|
print() # 输出空行分隔不同的二级目录
|
|
|
|
|
current_level2 = title
|
|
|
|
|
# 生成二级目录的 JSON 对象
|
|
|
|
|
level2_json = {
|
|
|
|
|
"type": "transition",
|
|
|
|
|
"data": {
|
|
|
|
|
"title": current_level2,
|
|
|
|
|
"text": current_level2
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
print(json.dumps(level2_json, ensure_ascii=False))
|
|
|
|
|
elif level == '###': # 三级目录
|
|
|
|
|
if current_level3 and level3_items:
|
|
|
|
|
# 生成三级目录的 JSON 对象
|
|
|
|
|
level3_json = {
|
|
|
|
|
"type": "content",
|
|
|
|
|
"data": {
|
|
|
|
|
"title": current_level3,
|
|
|
|
|
"text": current_level3,
|
|
|
|
|
"items": level3_items
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
print(json.dumps(level3_json, ensure_ascii=False))
|
|
|
|
|
level3_items = [] # 清空三级目录下的内容
|
|
|
|
|
current_level3 = title
|
|
|
|
|
elif item: # 三级目录下的内容
|
|
|
|
|
# 去掉内容中的 "-" 符号
|
|
|
|
|
cleaned_item = item.lstrip('- ').strip()
|
|
|
|
|
# 生成三级目录下内容的 JSON 对象
|
|
|
|
|
item_json = {"title": cleaned_item, "text": cleaned_item}
|
|
|
|
|
level3_items.append(item_json)
|
|
|
|
|
|
|
|
|
|
# 处理最后一个三级目录
|
|
|
|
|
if current_level3 and level3_items:
|
|
|
|
|
level3_json = {
|
|
|
|
|
"type": "content",
|
|
|
|
|
"data": {
|
|
|
|
|
"title": current_level3,
|
|
|
|
|
"text": current_level3,
|
|
|
|
|
"items": level3_items
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
print(json.dumps(level3_json, ensure_ascii=False))
|
|
|
|
|
|
|
|
|
|
# 运行应用
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
|
# 使用 asyncio.run 运行异步函数
|
|
|
|
@ -83,3 +142,6 @@ if __name__ == "__main__":
|
|
|
|
|
contents = extract_level2_titles(markdown_content)
|
|
|
|
|
ret = {"type": "contents", "data": {"items": contents}}
|
|
|
|
|
print(json.dumps(ret, ensure_ascii=False))
|
|
|
|
|
|
|
|
|
|
# 二级目录和三级目录
|
|
|
|
|
extract_level2_and_level3(markdown_content)
|