Spaces:
Sleeping
Sleeping
import gradio as gr | |
import asyncio | |
import os | |
import json | |
import urllib.request | |
from openai import AsyncOpenAI, OpenAI | |
# 第一个功能:检查YouTube视频是否具有Creative Commons许可证 | |
def get_youtube_id(youtube_url): | |
if 'youtube.com' in youtube_url: | |
video_id = youtube_url.split('v=')[-1] | |
video_id = video_id.split('&')[0] # 移除可能的额外参数 | |
elif 'youtu.be' in youtube_url: | |
video_id = youtube_url.split('/')[-1].split('?')[0] | |
else: | |
video_id = '' | |
return video_id | |
def check_cc_license(youtube_url): | |
# 从URL中提取视频ID | |
video_id = get_youtube_id(youtube_url) | |
if not video_id: | |
return "Invalid YouTube URL." | |
# YouTube Data API URL,用于获取视频详情 | |
api_url = f'https://www.googleapis.com/youtube/v3/videos?id={video_id}&part=status&key={API_KEY}' | |
try: | |
# 获取视频详情 | |
response = urllib.request.urlopen(api_url) | |
data = json.load(response) | |
# 检查许可证状态 | |
if 'items' in data and len(data['items']) > 0: | |
item = data['items'][0] | |
if item['status']['license'] == 'creativeCommon': | |
return "Yes." | |
else: | |
return "No." | |
else: | |
return "Video not found." | |
except Exception as e: | |
return f"An error occurred: {str(e)}" | |
# 第二个功能:为多项选择题生成干扰项 | |
# 从您的模块中导入必要的函数 | |
from utils.generate_distractors import prepare_q_inputs, construct_prompt_textonly, generate_distractors | |
from utils.api_utils import generate_from_openai_chat_completion | |
def generate_distractors_sync(model_name: str, | |
queries: list, | |
n: int=1, | |
max_tokens: int=4096): | |
assert model_name in ["gpt-4o-mini", "gpt-4-turbo", "gpt-4o", "gpt-4o-2024-08-06"], "Invalid model name" | |
client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"), base_url="https://yanlp.zeabur.app/v1") | |
messages = prepare_q_inputs(queries) | |
# 同步调用,不使用异步函数 | |
responses = generate_from_openai_chat_completion( | |
client, | |
messages=messages, | |
engine_name=model_name, | |
n=n, | |
max_tokens=max_tokens, | |
requests_per_minute=30, | |
json_format=True | |
) | |
for query, response in zip(queries, responses): | |
new_options = response | |
if new_options and "distractors" in new_options: | |
query["option_5"] = new_options["distractors"].get("E", "") | |
query["option_6"] = new_options["distractors"].get("F", "") | |
query["option_7"] = new_options["distractors"].get("G", "") | |
query["distractor_analysis"] = new_options["distractors"].get("analysis_of_distractors", "") | |
else: | |
query["option_5"] = "" | |
query["option_6"] = "" | |
query["option_7"] = "" | |
query["distractor_analysis"] = "" | |
return queries | |
# 处理生成干扰项的同步版本 | |
def generate_distractors_gradio(question, option1, option2, option3, option4, answer, answer_analysis): | |
is_valid, message = validate_inputs(question, option1, option2, option3, option4, answer, answer_analysis) | |
if not is_valid: | |
return {"error": message}, "" # Output error message | |
query = { | |
'question': question, | |
'option_1': option1, | |
'option_2': option2, | |
'option_3': option3, | |
'option_4': option4, | |
'answer': answer, | |
'answer_analysis': answer_analysis | |
} | |
queries = [query] | |
# 调用同步生成干扰项的函数 | |
results = generate_distractors_sync( | |
model_name="gpt-4o", | |
queries=queries, | |
n=1, | |
max_tokens=4096 | |
) | |
result = results[0] | |
new_options = { | |
'E': result.get('option_5', ''), | |
'F': result.get('option_6', ''), | |
'G': result.get('option_7', '') | |
} | |
new_option_str = f"E: {new_options['E']}\nF:{new_options['F']}\nG:{new_options['G']}" | |
distractor_analysis = result.get('distractor_analysis', '') | |
return new_option_str, distractor_analysis | |
def validate_inputs(question, option1, option2, option3, option4, answer, analysis): | |
if not question: | |
return False, "问题不能为空" | |
if not option1: | |
return False, "选项A不能为空" | |
if not option2: | |
return False, "选项B不能为空" | |
if not option3: | |
return False, "选项C不能为空" | |
if not option4: | |
return False, "选项D不能为空" | |
if not answer: | |
return False, "正确答案不能为空" | |
if not analysis: | |
return False, "答案解析不能为空" | |
return True, "" | |
with gr.Blocks() as demo: | |
gr.Markdown("# CC检查器和干扰项生成器") | |
with gr.Tabs(): | |
with gr.TabItem("YouTube Creative Commons检查器"): | |
gr.Markdown("## 检查YouTube视频是否具有Creative Commons许可证") | |
youtube_url_input = gr.Textbox(label="YouTube视频URL") | |
cc_license_output = gr.Textbox(label="是否为Creative Commons许可证?") | |
check_button = gr.Button("检查许可证") | |
check_button.click( | |
fn=check_cc_license, | |
inputs=youtube_url_input, | |
outputs=cc_license_output | |
) | |
with gr.TabItem("多项选择题干扰项生成器"): | |
gr.Markdown("## 为多项选择题生成干扰项") | |
with gr.Row(): | |
question_input = gr.Textbox(label="问题", lines=2) | |
with gr.Row(): | |
option1_input = gr.Textbox(label="选项A") | |
option2_input = gr.Textbox(label="选项B") | |
option3_input = gr.Textbox(label="选项C") | |
option4_input = gr.Textbox(label="选项D") | |
with gr.Row(): | |
answer_input = gr.Textbox(label="正确答案(A/B/C/D)") | |
with gr.Row(): | |
answer_analysis_input = gr.Textbox(label="答案解析", lines=3) | |
generate_button = gr.Button("生成干扰项") | |
output_options = gr.Textbox(label="生成的干扰选项") | |
output_analysis = gr.Textbox(label="干扰项解析", lines=5) | |
generate_button.click( | |
fn=generate_distractors_gradio, | |
inputs=[question_input, option1_input, option2_input, option3_input, option4_input, answer_input, answer_analysis_input], | |
outputs=[output_options, output_analysis] | |
) | |
# 运行Gradio应用 | |
demo.launch() | |