哪种大模型是程序员的福音？ChatGPT并非最佳Copilot 成功“弑神”

作者：AI研习所发布时间：2023-07-05

我们都知道ChatGPT4的综合能力肯定是大于ChatGPT3.5。

1：那么官方提供的 GPT3.5 API 能力和官方的 GPT3.5 能力相当吗？

（友情提示：ChatGPT3.5和ChatGPT3.5API是不一样的）

2：GPT4.0和GPT4.0配合联网搜索哪个效果好？

3：另外， Github 的 copilot chat 能力又怎样？

带着这些问题，我们用一个简单的编程问题，逐个进行测试，看看他们的表现如何？

编程问题：

帮我写一段python完整的代码，我希望用最快的效率爬取10000个网页，并存储在本地的MySQL数据库。

ChatGPT3.5 API

import requests

from bs4 import BeautifulSoup

import mysql.connector
# 设置MySQL数据库连接

db = mysql.connector.connect(

host="localhost",

user="your_username",

password="your_password",

database="your_database")

cursor = db.cursor()
# 爬取网页并存储到数据库

def crawl_and_store(url):

try:

response = requests.get(url)

soup = BeautifulSoup(response.text, 'html.parser')
# 解析网页内容并提取需要存储的数据

# 这里只是一个示例，你需要根据具体的网页结构进行解析 title = soup.title.text.strip()

content = soup.find('div', class_='content').text.strip()
# 将数据插入到数据库中

query = "INSERT INTO webpages (url, title, content) VALUES (%s, %s, %s)"

values = (url, title, content)

cursor.execute(query, values)

db.commit()
print(f"成功爬取并存储网页: {url}")

except Exception as e:

print(f"爬取网页失败: {url}")

print(f"错误信息: {str(e)}")
# 待爬取的网页列表

urls = ["https://example.com/page1","

https://example.com/page2",# 添加更多的URL...]
# 遍历网页列表进行爬取和存储for url in urls: crawl_and_store(url)
# 关闭数据库连接cursor.close()db.close()

总结：

GPT3.5 API的代码写的简单明了。没什么大问题，但是却忽略了我的要求，我要求需要以最快的效率，但是它忽略了我的要求，依然用了最简单方法写了代码。

ChatGPT3.5

import requests

import mysql.connector

from concurrent.futures import ThreadPoolExecutor

# 创建数据库连接

db = mysql.connector.connect(

host="localhost",

user="your_username",

password="your_password",

database="your_database")

cursor = db.cursor()
# 创建表格

cursor.execute("CREATE TABLE IF NOT EXISTS web_pages (id INT AUTO_INCREMENT PRIMARY KEY, url VARCHAR(255), content TEXT)")
def fetch_and_store(url):

try:

response = requests.get(url)

content = response.text
# 存储到数据库

sql = "INSERT INTO web_pages (url, content) VALUES (%s, %s)"

val = (url, content)

cursor.execute(sql, val)

db.commit()
print(f"成功爬取并存储网页: {url}")

except Exception as e:

print(f"爬取网页 {url} 失败: {str(e)}")

# 要爬取的网页列表urls = ["https://example.com/page1", "https://example.com/page2", ...] # 添加更多的URL
# 创建线程池

executor = ThreadPoolExecutor(max_workers=10) # 设置适当的线程数
# 使用线程池并发地爬取和存储网页for url in urls: executor.submit(fetch_and_store, url)
# 等待所有任务完成executor.shutdown()
# 关闭数据库连接cursor.close()db.close()

总结：

ChatGPT3.5的网页版本是比API的效果要好的。有两点，第一，它分析出来，我在插入MySQL的之前是要创建表的，并且它注意到了我强调了要高效率的实现，代码也引入了多线程。

ChatGPT4

import concurrent.futures

import requests

import pymysql.cursors
# TODO: 替换为实际的网址列表

url_list = ["http://www.example.com/1", "http://www.example.com/2", "..."]
# 爬取单个网页的函数

def fetch_url(url):

response = requests.get(url)

return url, response.text
# 连接到 MySQL

connection = pymysql.connect(host='localhost', user='your_username', password='your_password',

database='your_database',

charset='utf8mb4',

cursorclass=pymysql.cursors.DictCursor)
try:# 创建一个线程池

with concurrent.futures.ThreadPoolExecutor(max_workers=10) as executor:

# 使用线程池异步地爬取所有网页

future_to_url = {executor.submit(fetch_url, url): url for url in url_list}

for future in concurrent.futures.as_completed(future_to_url):

url = future_to_url[future]try: data = future.result()except Exception as exc: print('%r generated an exception: %s' % (url, exc))else: print('%r page is %d bytes' % (url, len(data[1])))
# 存储到数据库

try:with connection.cursor() as cursor:

sql = "INSERT INTO `your_table` (`url`, `content`) VALUES (%s, %s)" cursor.execute(sql, (data[0], data[1]))

connection.commit()

except Exception as error: print('Failed to update database: {}'.format(error))

finally: connection.close()

总结：

ChatGPT4比ChatGPT3.5又升级了一版，理解了我要高效率的意思，不仅用了多线程还在注释里面强调：使用线程池异步地爬取所有网页。异步使得程序的抓取效率又上升了一个档次，而且还考虑到插入表的时候会报错。还打印出来了报错信息。可谓是非常规范了。

科大星火

import requests
from bs4 import BeautifulSoup
import pymysql

# 连接MySQL数据库
def connect_db():
db = pymysql.connect(host='localhost', user='root', password='your_password', database='your_database', charset='utf8')
return db

# 爬取网页内容
def get_html(url):
try:
response = requests.get(url)
response.raise_for_status()
response.encoding = response.apparent_encoding
return response.text
except:
print("获取网页失败")
return None

# 解析网页内容
def parse_html(html):
soup = BeautifulSoup(html, 'html.parser')
# 根据网页结构提取所需信息，这里需要根据实际情况修改
titles = soup.find_all('h1')
for title in titles:
yield title.text

# 将数据存储到MySQL数据库
def save_to_db(data):
db = connect_db()
cursor = db.cursor()
sql = "INSERT INTO your_table (title) VALUES (%s)"
try:
cursor.executemany(sql, data)
db.commit()
except:
db.rollback()
print("保存数据失败")
finally:
cursor.close()
db.close()

if __name__ == '__main__':
urls = ['https://www.example.com/page{}'.format(i) for i in range(1, 10001)]
data = []
for url in urls:
html = get_html(url)
if html:
parsed_data = parse_html(html)
data.extend(parsed_data)
if len(data) >= 10000:
save_to_db(data)
data = []
else:
print("跳过无效网页")
continue
if data:
save_to_db(data)

总结：

基本效果等于ChatGPT3.5API

Copilot

import asyncioimport aiohttpimport pymysqlimport aiomysql
# MySQL数据库配置信息DB_SETTINGS = {"host": "localhost","database": "your_database","user": "your_username","password": "your_password",}
# 你需要爬取的URL列表URLS = ["http://www.example.com"] * 10000
async def fetch(session, url):async with session.get(url) as response:return await response.text()
async def save_to_db(pool, url, content):async with pool.acquire() as conn:async with conn.cursor() as cur:# 根据你的数据表结构，写入具体的SQL语句 insert_sql = "INSERT INTO table_name(url, content) VALUES(%s, %s)"await cur.execute(insert_sql, (url, content))await conn.commit()
async def fetch_and_save(pool, session, url): content = await fetch(session, url)await save_to_db(pool, url, content)
async def main():async with aiohttp.ClientSession() as session: pool = await aiomysql.create_pool(**DB_SETTINGS) tasks = []for url in URLS: task = asyncio.create_task(fetch_and_save(pool, session, url)) tasks.append(task)await asyncio.gather(*tasks)
if __name__ == "__main__": asyncio.run(main())

总结：