|
|
|
@ -4,68 +4,6 @@ import re
|
|
|
|
|
from CommonUtil import *
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def update_json_with_ai(client, markdown_content, json_obj):
|
|
|
|
|
"""
|
|
|
|
|
更新 JSON 对象,将 title 属性生成描述信息填充到 text 属性中
|
|
|
|
|
"""
|
|
|
|
|
# 如果 json_obj 是字符串,则将其解析为字典
|
|
|
|
|
if isinstance(json_obj, str):
|
|
|
|
|
json_obj = json.loads(json_obj)
|
|
|
|
|
|
|
|
|
|
# 检查是否存在嵌套的 data 对象
|
|
|
|
|
if "data" in json_obj and isinstance(json_obj["data"], dict):
|
|
|
|
|
data = json_obj["data"]
|
|
|
|
|
if "title" in data:
|
|
|
|
|
title = data["title"]
|
|
|
|
|
description = generate_description_with_ai(client, markdown_content, title)
|
|
|
|
|
data["text"] = description # 将生成的描述信息填充到 text 属性中
|
|
|
|
|
elif "title" in json_obj:
|
|
|
|
|
# 如果 title 直接存在于 json_obj 中
|
|
|
|
|
title = json_obj["title"]
|
|
|
|
|
description = generate_description_with_ai(client, markdown_content, title)
|
|
|
|
|
json_obj["text"] = description
|
|
|
|
|
|
|
|
|
|
# 递归处理 items 中的子对象
|
|
|
|
|
if "items" in json_obj:
|
|
|
|
|
for item in json_obj["items"]:
|
|
|
|
|
update_json_with_ai(client, markdown_content, item)
|
|
|
|
|
|
|
|
|
|
return json_obj
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def generate_description_with_ai(client, markdown_content, title):
|
|
|
|
|
"""
|
|
|
|
|
使用 AI 生成一句话的描述信息
|
|
|
|
|
"""
|
|
|
|
|
# 构造提示词
|
|
|
|
|
prompt = f"以下是 Markdown 内容:\n{markdown_content}\n\n请为以下标题生成一句话的描述信息,描述信息应简洁明了,且与标题内容相关,不要包含任何序号(如 1.、2. 等)或 Markdown 语法(如 #、- 等),也一定不要重复标题:\n{title}"
|
|
|
|
|
|
|
|
|
|
try:
|
|
|
|
|
# 调用 AI 模型
|
|
|
|
|
response = client.chat.completions.create(
|
|
|
|
|
model=MODEL_NAME, # 根据需要选择合适的模型
|
|
|
|
|
messages=[
|
|
|
|
|
{"role": "system", "content": "你是一个专业的助手,能够根据上下文生成简洁的描述信息。"},
|
|
|
|
|
{"role": "user", "content": prompt}
|
|
|
|
|
],
|
|
|
|
|
max_tokens=50 # 限制生成内容的长度
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
# 提取生成的描述信息
|
|
|
|
|
if response.choices and response.choices[0].message.content:
|
|
|
|
|
description = response.choices[0].message.content.strip()
|
|
|
|
|
# 去除序号和 Markdown 语法(如 1.、#、- 等)
|
|
|
|
|
description = re.sub(r'[\d.]+', '', description).strip() # 去除序号
|
|
|
|
|
description = re.sub(r'[#-]', '', description).strip() # 去除 Markdown 语法
|
|
|
|
|
return description
|
|
|
|
|
else:
|
|
|
|
|
print(f"AI 未返回有效描述信息,标题:{title}")
|
|
|
|
|
return title # 如果 AI 未返回有效描述,则返回标题本身
|
|
|
|
|
except Exception as e:
|
|
|
|
|
print(f"调用 AI 生成描述信息时出错:{e}")
|
|
|
|
|
return title # 如果调用失败,则返回标题本身
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# 流式生成数据的函数
|
|
|
|
|
async def generate_stream_markdown(course_name: str):
|
|
|
|
|
"""
|
|
|
|
@ -100,12 +38,6 @@ async def generate_stream_markdown(course_name: str):
|
|
|
|
|
print(full_markdown)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# 测试函数
|
|
|
|
|
async def test_generate_stream_markdown():
|
|
|
|
|
async for chunk in generate_stream_markdown("三角形面积"):
|
|
|
|
|
print(chunk.decode("utf-8"), end="", flush=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def extract_level1_title(markdown_content):
|
|
|
|
|
"""
|
|
|
|
|
从 Markdown 字符串中提取一级目录的文本内容
|
|
|
|
@ -183,22 +115,42 @@ def extract_level2_and_level3(markdown_content):
|
|
|
|
|
|
|
|
|
|
return result
|
|
|
|
|
|
|
|
|
|
def add_descriptions_to_structure(client, markdown_content, structure):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def convert_structure_to_json(structure):
|
|
|
|
|
"""
|
|
|
|
|
遍历结构化数据,为每个二级和三级文本生成描述信息,并赋值给 text 属性
|
|
|
|
|
将结构化数据转换为指定的 JSON 格式
|
|
|
|
|
"""
|
|
|
|
|
result = []
|
|
|
|
|
|
|
|
|
|
for level2 in structure:
|
|
|
|
|
# 为二级目录生成描述信息
|
|
|
|
|
# 处理二级目录
|
|
|
|
|
level2_title = level2["title"]
|
|
|
|
|
level2["text"] = generate_description_with_ai(client, markdown_content, level2_title)
|
|
|
|
|
|
|
|
|
|
# 遍历三级目录
|
|
|
|
|
level2_json = {
|
|
|
|
|
"type": "transition",
|
|
|
|
|
"data": {
|
|
|
|
|
"title": level2_title,
|
|
|
|
|
"text": level2_title
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
result.append(level2_json)
|
|
|
|
|
|
|
|
|
|
# 处理三级目录
|
|
|
|
|
for level3 in level2["children"]:
|
|
|
|
|
# 为三级目录生成描述信息
|
|
|
|
|
level3_title = level3["title"]
|
|
|
|
|
level3["text"] = generate_description_with_ai(client, markdown_content, level3_title)
|
|
|
|
|
items = [{"title": item, "text": item} for item in level3["items"]]
|
|
|
|
|
level3_json = {
|
|
|
|
|
"type": "content",
|
|
|
|
|
"data": {
|
|
|
|
|
"title": level3_title,
|
|
|
|
|
"text": level3_title,
|
|
|
|
|
"items": items
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
result.append(level3_json)
|
|
|
|
|
|
|
|
|
|
return result
|
|
|
|
|
|
|
|
|
|
return structure
|
|
|
|
|
|
|
|
|
|
# 运行应用
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
@ -209,22 +161,23 @@ if __name__ == "__main__":
|
|
|
|
|
with open("Sample.md", "r", encoding="utf-8") as file:
|
|
|
|
|
markdown_content = file.read()
|
|
|
|
|
|
|
|
|
|
listAll = []
|
|
|
|
|
# 一级名称
|
|
|
|
|
level1_title = extract_level1_title(markdown_content)
|
|
|
|
|
json_obj = {"type": "cover", "data": {"title": level1_title, "text": ""}}
|
|
|
|
|
# print(json.dumps(json_obj, ensure_ascii=False))
|
|
|
|
|
# 更新 JSON 对象
|
|
|
|
|
updated_json = update_json_with_ai(client, markdown_content, json_obj)
|
|
|
|
|
print(json.dumps(updated_json, ensure_ascii=False, indent=2))
|
|
|
|
|
listAll.append(json_obj)
|
|
|
|
|
|
|
|
|
|
# 二级名称列表
|
|
|
|
|
contents = extract_level2_titles(markdown_content)
|
|
|
|
|
json_obj = {"type": "contents", "data": {"items": contents}}
|
|
|
|
|
print(json.dumps(json_obj, ensure_ascii=False))
|
|
|
|
|
listAll.append(json_obj)
|
|
|
|
|
|
|
|
|
|
# 二级目录和三级目录
|
|
|
|
|
result = extract_level2_and_level3(markdown_content)
|
|
|
|
|
print(json.dumps(result, ensure_ascii=False, indent=2))
|
|
|
|
|
json_obj = convert_structure_to_json(result)
|
|
|
|
|
# 遍历结果并一个对象一行打印
|
|
|
|
|
for item in json_obj:
|
|
|
|
|
listAll.append(item)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
result=add_descriptions_to_structure(client, markdown_content, result)
|
|
|
|
|
print(json.dumps(result, ensure_ascii=False, indent=2))
|
|
|
|
|
print(listAll)
|
|
|
|
|