mirror of
https://github.com/JefferyHcool/BiliNote.git
synced 2026-05-22 00:30:28 +08:00
ab8cdc416a
## 主要更新 ### 新增功能 - 新增 SiliconFlow_provider.py 专用提供商 - 添加硅基流动 API 集成文档 - 实现 Cherry Studio 风格的连接测试 ### 错误处理优化 - 修复前端 Form.tsx 错误显示问题 - 改进 universal_gpt.py 异常处理逻辑 - 统一 URL 格式处理，避免路径重复 ### 兼容性改进 - 优化 OpenAI 兼容提供商 URL 处理 - 增强模型列表获取的容错性 - 添加详细的调试日志 ### 安全性提升 - 更新 .gitignore 保护敏感信息 - 移除示例配置文件 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>
252 lines
9.9 KiB
Python
252 lines
9.9 KiB
Python
from app.gpt.base import GPT
|
||
from app.gpt.prompt_builder import generate_base_prompt
|
||
from app.models.gpt_model import GPTSource
|
||
from app.gpt.prompt import BASE_PROMPT, AI_SUM, SCREENSHOT, LINK
|
||
from app.gpt.utils import fix_markdown
|
||
from app.models.transcriber_model import TranscriptSegment
|
||
from datetime import timedelta
|
||
from typing import List
|
||
import json
|
||
import logging
|
||
import math
|
||
|
||
logger = logging.getLogger(__name__)
|
||
# 设置分段处理的参数
|
||
MAX_SEGMENTS_PER_CHUNK = 250 # 每块最多包含的段落数
|
||
MAX_CONTENT_LENGTH = 30000 # 字符数
|
||
|
||
class UniversalGPT(GPT):
|
||
def __init__(self, client, model: str, temperature: float = 0.7):
|
||
self.client = client
|
||
self.model = model
|
||
self.temperature = temperature
|
||
self.screenshot = False
|
||
self.link = False
|
||
|
||
def _format_time(self, seconds: float) -> str:
|
||
return str(timedelta(seconds=int(seconds)))[2:]
|
||
|
||
def _build_segment_text(self, segments: List[TranscriptSegment]) -> str:
|
||
return "\n".join(
|
||
f"{self._format_time(seg.start)} - {seg.text.strip()}"
|
||
for seg in segments
|
||
)
|
||
|
||
def ensure_segments_type(self, segments) -> List[TranscriptSegment]:
|
||
return [TranscriptSegment(**seg) if isinstance(seg, dict) else seg for seg in segments]
|
||
|
||
def create_messages(self, segments: List[TranscriptSegment], **kwargs):
|
||
content_text = generate_base_prompt(
|
||
title=kwargs.get('title'),
|
||
segment_text=self._build_segment_text(segments),
|
||
tags=kwargs.get('tags'),
|
||
_format=kwargs.get('_format'),
|
||
style=kwargs.get('style'),
|
||
extras=kwargs.get('extras'),
|
||
)
|
||
|
||
# 检查文本长度
|
||
if len(content_text) > MAX_CONTENT_LENGTH:
|
||
# 保留前部分和后部分内容
|
||
first_part = int(MAX_CONTENT_LENGTH * 0.3)
|
||
second_part = MAX_CONTENT_LENGTH - first_part - 100 # 预留100字符给提示文本
|
||
|
||
truncated_text = (
|
||
content_text[:first_part] +
|
||
"\n\n[内容过长，中间部分已省略]\n\n" +
|
||
content_text[-second_part:]
|
||
)
|
||
content_text = truncated_text
|
||
print(f"内容已截断，原长度: {len(content_text)}，截断后: {len(truncated_text)}")
|
||
|
||
# 组装 content 数组，支持 text + image_url 混合
|
||
content = [{"type": "text", "text": content_text}]
|
||
video_img_urls = kwargs.get('video_img_urls', [])
|
||
|
||
# 限制图片数量
|
||
if len(video_img_urls) > 5:
|
||
video_img_urls = video_img_urls[:5]
|
||
print("图片数量过多，已限制为5张")
|
||
|
||
for url in video_img_urls:
|
||
content.append({
|
||
"type": "image_url",
|
||
"image_url": {
|
||
"url": url,
|
||
"detail": "auto"
|
||
}
|
||
})
|
||
|
||
# 正确格式：整体包在一个 message 里，role + content array
|
||
messages = [{
|
||
"role": "user",
|
||
"content": content
|
||
}]
|
||
|
||
return messages
|
||
|
||
def list_models(self):
|
||
return self.client.models.list()
|
||
|
||
def summarize(self, source: GPTSource) -> str:
|
||
self.screenshot = source.screenshot
|
||
self.link = source.link
|
||
source.segment = self.ensure_segments_type(source.segment)
|
||
|
||
# 如果段落数量超过阈值，使用分段处理方法
|
||
if len(source.segment) > MAX_SEGMENTS_PER_CHUNK:
|
||
print(f"段落过多({len(source.segment)}个)，将使用分段处理方法")
|
||
return self._process_long_content_by_chunks(source)
|
||
|
||
# 正常处理较短的内容
|
||
try:
|
||
messages = self.create_messages(
|
||
source.segment,
|
||
title=source.title,
|
||
tags=source.tags,
|
||
video_img_urls=source.video_img_urls,
|
||
_format=source._format,
|
||
style=source.style,
|
||
extras=source.extras
|
||
)
|
||
|
||
# 检查消息大小
|
||
messages_json = json.dumps(messages)
|
||
if len(messages_json) > 100000: # API限制
|
||
print(f"消息体积过大: {len(messages_json)} 字节，将使用分段处理方法")
|
||
return self._process_long_content_by_chunks(source)
|
||
|
||
response = self.client.chat.completions.create(
|
||
model=self.model,
|
||
messages=messages,
|
||
temperature=0.7
|
||
)
|
||
return response.choices[0].message.content.strip()
|
||
except Exception as e:
|
||
error_msg = f"总结失败: {str(e)}"
|
||
print(error_msg)
|
||
# 记录详细错误信息到日志
|
||
import logging
|
||
logging.error(f"GPT单块处理失败 - {error_msg}")
|
||
# 如果处理失败，尝试使用分段处理方法
|
||
print("尝试使用分段处理方法")
|
||
try:
|
||
return self._process_long_content_by_chunks(source)
|
||
except Exception as fallback_error:
|
||
# 如果分段处理也失败，抛出异常
|
||
logging.error(f"GPT分段处理也失败 - {str(fallback_error)}")
|
||
raise Exception(f"视频处理完全失败：主处理失败({str(e)})，分段处理也失败({str(fallback_error)})")
|
||
|
||
def _process_long_content_by_chunks(self, source: GPTSource) -> str:
|
||
"""
|
||
将长内容分成多个块，分别处理后再整合
|
||
"""
|
||
segments = source.segment
|
||
total_segments = len(segments)
|
||
|
||
# 计算需要多少块
|
||
num_chunks = math.ceil(total_segments / MAX_SEGMENTS_PER_CHUNK)
|
||
chunk_size = math.ceil(total_segments / num_chunks)
|
||
|
||
print(f"将内容分为{num_chunks}块进行处理，每块约{chunk_size}个段落")
|
||
|
||
chunk_summaries = []
|
||
|
||
# 处理每个块
|
||
for i in range(num_chunks):
|
||
start_idx = i * chunk_size
|
||
end_idx = min(start_idx + chunk_size, total_segments)
|
||
|
||
print(f"处理第{i+1}/{num_chunks}块 (段落 {start_idx} 到 {end_idx-1})")
|
||
|
||
# 创建此块的子源
|
||
chunk_source = GPTSource(
|
||
title=f"{source.title} - 第{i+1}/{num_chunks}部分",
|
||
segment=segments[start_idx:end_idx],
|
||
tags=source.tags,
|
||
screenshot=False, # 中间块不需要截图
|
||
video_img_urls=[], # 中间块不需要图片
|
||
link=False, # 中间块不需要链接
|
||
_format=[],
|
||
style=source.style,
|
||
extras=f"这是内容的第{i+1}部分，共{num_chunks}部分。请仅总结这部分内容的要点，无需引言和结论。"
|
||
)
|
||
|
||
try:
|
||
# 处理这个块
|
||
chunk_messages = self.create_messages(
|
||
chunk_source.segment,
|
||
title=chunk_source.title,
|
||
tags=chunk_source.tags,
|
||
video_img_urls=chunk_source.video_img_urls,
|
||
_format=chunk_source._format,
|
||
style=chunk_source.style,
|
||
extras=chunk_source.extras
|
||
)
|
||
|
||
chunk_response = self.client.chat.completions.create(
|
||
model=self.model,
|
||
messages=chunk_messages,
|
||
temperature=0.7
|
||
)
|
||
|
||
chunk_summary = chunk_response.choices[0].message.content.strip()
|
||
chunk_summaries.append(f"### 第{i+1}部分内容总结\n\n{chunk_summary}")
|
||
|
||
print(f"第{i+1}块处理完成")
|
||
|
||
except Exception as e:
|
||
error_msg = f"处理第{i+1}块时出错: {str(e)}"
|
||
print(error_msg)
|
||
# 记录详细错误信息到日志
|
||
import logging
|
||
logging.error(f"GPT处理失败 - {error_msg}")
|
||
# 如果某块处理失败，抛出异常停止整个处理流程
|
||
raise Exception(f"视频处理失败：第{i+1}块GPT调用失败 - {str(e)}")
|
||
|
||
# 合并所有块的总结
|
||
all_summaries = "\n\n".join(chunk_summaries)
|
||
|
||
# 创建最终总结请求
|
||
final_segment = TranscriptSegment(start=0, end=0, text=all_summaries)
|
||
final_source = GPTSource(
|
||
title=source.title,
|
||
segment=[final_segment],
|
||
tags=source.tags,
|
||
screenshot=source.screenshot,
|
||
video_img_urls=source.video_img_urls,
|
||
link=source.link,
|
||
_format=source._format,
|
||
style=source.style,
|
||
extras="以下是视频各部分的总结，请将它们整合为一篇完整、连贯的笔记。"
|
||
)
|
||
|
||
try:
|
||
# 最终合并处理
|
||
final_messages = self.create_messages(
|
||
final_source.segment,
|
||
title=final_source.title,
|
||
tags=final_source.tags,
|
||
video_img_urls=final_source.video_img_urls,
|
||
_format=final_source._format,
|
||
style=final_source.style,
|
||
extras=final_source.extras
|
||
)
|
||
|
||
final_response = self.client.chat.completions.create(
|
||
model=self.model,
|
||
messages=final_messages,
|
||
temperature=0.7
|
||
)
|
||
|
||
return final_response.choices[0].message.content.strip()
|
||
|
||
except Exception as e:
|
||
error_msg = f"最终合并处理时出错: {str(e)}"
|
||
print(error_msg)
|
||
# 记录详细错误信息到日志
|
||
import logging
|
||
logging.error(f"GPT最终合并失败 - {error_msg}")
|
||
# 如果最终合并失败，抛出异常
|
||
raise Exception(f"视频处理失败：GPT最终合并失败 - {str(e)}")
|