修改了生成逻辑

This commit is contained in:
jinye_huang 2025-04-18 11:08:54 +08:00
parent f547c3bfdc
commit 6d09ec65f8
10 changed files with 112 additions and 70 deletions

View File

@ -1,12 +1,13 @@
import os
from openai import OpenAI
import time
import random
class AI_Agent():
"""AI代理类负责与AI模型交互生成文本内容"""
def __init__(self, base_url, model_name, api):
def __init__(self, base_url, model_name, api, timeout=10, max_retries=3):
self.url_list = {
"ali": "https://dashscope.aliyuncs.com/compatible-mode/v1",
"kimi": "https://api.moonshot.cn/v1",
@ -18,11 +19,13 @@ class AI_Agent():
self.base_url = self.url_list[base_url] if base_url in self.url_list else base_url
self.api = api
self.model_name = model_name
self.timeout = timeout # 设置超时时间(秒)
self.max_retries = max_retries # 最大重试次数
self.client = OpenAI(
api_key=self.api,
base_url=self.base_url,
# timeout=10
timeout=self.timeout # 设置OpenAI客户端超时
)
def generate_text(self, system_prompt, user_prompt, temperature, top_p, presence_penalty):
@ -35,29 +38,68 @@ class AI_Agent():
print(f"Base URL: {self.base_url}")
print(f"Model: {self.model_name}")
response = self.client.chat.completions.create(
model=self.model_name,
messages=[{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt}],
temperature=temperature,
top_p=top_p,
# presence_penalty=presence_penalty,
stream=True,
max_tokens=8192,
extra_body={
"repetition_penalty": 1.05,
},
)
retry_count = 0
max_retry_wait = 10 # 最大重试等待时间(秒)
# 收集完整的输出内容
full_response = ""
for chunk in response:
if chunk.choices and len(chunk.choices) > 0 and chunk.choices[0].delta.content is not None:
content = chunk.choices[0].delta.content
full_response += content
print(content, end="", flush=True)
if chunk.choices and len(chunk.choices) > 0 and chunk.choices[0].finish_reason == "stop":
break
while retry_count <= self.max_retries:
try:
response = self.client.chat.completions.create(
model=self.model_name,
messages=[{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt}],
temperature=temperature,
top_p=top_p,
# presence_penalty=presence_penalty,
stream=True,
max_tokens=8192,
timeout=self.timeout, # 设置请求超时
extra_body={
"repetition_penalty": 1.05,
},
)
# 收集完整的输出内容
full_response = ""
try:
for chunk in response:
if chunk.choices and len(chunk.choices) > 0 and chunk.choices[0].delta.content is not None:
content = chunk.choices[0].delta.content
full_response += content
print(content, end="", flush=True)
if chunk.choices and len(chunk.choices) > 0 and chunk.choices[0].finish_reason == "stop":
break
# 成功完成,跳出重试循环
break
except:
# 处理流式响应中的超时
print(f"\n接收响应时超时")
if len(full_response) > 0:
print(f"已接收部分响应({len(full_response)}字符)")
# 如果已接收足够内容,可以考虑使用已有内容
if len(full_response) > 100: # 假设至少需要100个字符才有意义
print("使用已接收的部分内容继续处理")
break
# 否则准备重试
retry_count += 1
if retry_count <= self.max_retries:
wait_time = min(2 ** retry_count + random.random(), max_retry_wait) # 指数退避
print(f"\n等待 {wait_time:.2f} 秒后重试({retry_count}/{self.max_retries})...")
time.sleep(wait_time)
continue
except Exception as e:
print(f"\n请求发生错误: {e}")
retry_count += 1
if retry_count <= self.max_retries:
wait_time = min(2 ** retry_count + random.random(), max_retry_wait) # 指数退避
print(f"\n等待 {wait_time:.2f} 秒后重试({retry_count}/{self.max_retries})...")
time.sleep(wait_time)
else:
print(f"已达到最大重试次数({self.max_retries}),放弃请求")
return "请求失败,无法生成内容。", 0
print("\n完成生成,正在处理结果...")

View File

@ -24,7 +24,7 @@
</title>
<content>
(正文内容,不含带#的TAG内容
(正文内容)
(TAG内容如#周末去哪)
(最后一段是TAG内容如#周末去哪)
</content>

View File

@ -24,7 +24,7 @@
</title>
<content>
(正文内容,不含带#的TAG内容
(正文内容)
(TAG内容如#周末去哪)
(最后一段是TAG内容如#周末去哪)
</content>

View File

@ -28,7 +28,7 @@
</title>
<content>
(正文内容,不含带#的TAG内容
(正文内容)
(TAG内容如#周末去哪)
(最后一段是TAG内容如#周末去哪)
</content>

View File

@ -32,5 +32,5 @@
<content>
(正文内容)
(最后一段是TAG内容如#周末去哪)
最后一段是TAG内容如#周末去哪)
</content>

70
main.py
View File

@ -11,7 +11,7 @@ import core.posterGen as posterGen
import core.simple_collage as simple_collage
from utils.resource_loader import ResourceLoader
from utils.tweet_generator import prepare_topic_generation, generate_topics, generate_single_content
import random
def main():
config_file = {
"date": "4月17日",
@ -77,60 +77,56 @@ def main():
# 直接使用同一个AI Agent实例
for i in range(len(tweet_topic_record.topics_list)):
tweet_content_list = []
for j in range(config_file["variants"]):
tweet_content, gen_result = generate_single_content(
ai_agent, content_system_prompt, tweet_topic_record.topics_list[i],
config_file["prompts_dir"], config_file["resource_dir"],
output_dir, run_id, i+1, j+1, config_file["content_temperature"]
)
tweet_content_list.append(tweet_content.get_json_file())
if not tweet_content:
print(f"生成第{i+1}篇文章的第{j+1}个变体失败,跳过")
continue
object_name = tweet_topic_record.topics_list[i]["object"]
try:
object_name = object_name.split(".")[0]
except:
pass
try:
object_name = object_name.split("景点信息-")[1]
except:
pass
object_name = tweet_topic_record.topics_list[i]["object"]
try:
object_name = object_name.split(".")[0]
except:
pass
try:
object_name = object_name.split("景点信息-")[1]
except:
pass
# 处理对象名称中可能包含的"+"等特殊字符
# if "+" in object_name:
# # 取第一个景点名称作为主要对象
# object_name = object_name.split("+")[0].strip()
# print(f"对象名称包含多个景点,使用第一个景点:{object_name}")
# 检查图片路径是否存在
img_dir_path = f"/root/autodl-tmp/sanming_img/modify/{object_name}"
if not os.path.exists(img_dir_path):
print(f"图片目录不存在:{img_dir_path},跳过该对象")
continue
img_dir = os.path.join(output_dir, f"{i+1}_{j+1}")
info_directory = [
# 检查图片路径是否存在
img_dir_path = f"/root/autodl-tmp/sanming_img/modify/{object_name}"
if not os.path.exists(img_dir_path):
print(f"图片目录不存在:{img_dir_path},跳过该对象")
continue
img_dir = os.path.join(output_dir, f"{i+1}_{j_index+1}")
info_directory = [
f"/root/autodl-tmp/sanming_img/相机/{object_name}/description.txt"
]
# 检查描述文件是否存在
if not os.path.exists(info_directory[0]):
print(f"描述文件不存在:{info_directory[0]},使用生成的内容替代")
info_directory = []
# 检查描述文件是否存在
if not os.path.exists(info_directory[0]):
print(f"描述文件不存在:{info_directory[0]},使用生成的内容替代")
info_directory = []
poster_num = 1
tweet_content_json = tweet_content.get_json_file()
tweet_content_str = f"""
{tweet_content_json}
"""
input_dir = img_dir_path # 使用前面检查过的目录路径
# img_dir = output_dir
target_size = (900, 1200)
result_path = []
poster_num = config_file["variants"]
content_gen = contentGen.ContentGenerator()
response = content_gen.run(info_directory, poster_num, tweet_content_str)
print(response)
input_dir = img_dir_path # 使用前面检查过的目录路径
target_size = (900, 1200)
result_path = []
content_gen = contentGen.ContentGenerator()
response = content_gen.run(info_directory, poster_num, tweet_content_list)
print(response)
for j_index in range(config_file["variants"]):
try:
# 创建输出目录
collage_output_dir = os.path.join(img_dir, "collage_img")
@ -170,7 +166,7 @@ def main():
}
img_path = img_list[index]['path']
print(f"使用图片路径: {img_path}")
output_path = os.path.join(poster_output_dir, f"img_{i+1}_{j+1}_{index}.jpg")
output_path = os.path.join(poster_output_dir, f"img_{i+1}_{j_index+1}_{index}.jpg")
result_path.append(poster_gen.create_poster(img_path, text_data, output_path))
except Exception as e:
print(f"海报生成过程中出错: {e}")

View File

@ -1,5 +1,5 @@
import os
import random
class ResourceLoader:
"""资源加载器,用于加载提示词和参考资料"""
@ -20,7 +20,7 @@ class ResourceLoader:
return ""
@staticmethod
def load_all_refer_files(refer_dir):
def load_all_refer_files(refer_dir, refer_content_length=50):
"""加载Refer目录下的所有文件内容"""
refer_content = ""
try:
@ -30,6 +30,10 @@ class ResourceLoader:
file_path = os.path.join(refer_dir, file)
if os.path.isfile(file_path):
content = ResourceLoader.load_file_content(file_path)
# 用\n分割content取前length条
content_lines = content.split("\n")
content_lines = random.sample(content_lines, refer_content_length)
content = "\n".join(content_lines)
refer_content += f"## {file}\n{content}\n\n"
return refer_content
except Exception as e: