Python网站源码下载器

声明:本程序基于Python3.11解释器制作

会有闪退的现象出现不清楚是为什么，有能力请自行修改

(不一定100%会有，但是有概率出现)

import os, re, requests, subprocess, sys, threading, webbrowser, time, random
from urllib.parse import urljoin, urlparse
from PyQt5 import QtWidgets, QtCore

def install(package):
    subprocess.check_call([sys.executable, "-m", "pip", "install", package])

def check_and_install_packages():
    for package in ['requests', 'PyQt5']:
        try:
            __import__(package)
        except ImportError:
            install(package)

def log_message(message):
    log_text.append(message)

def get_website_source(url):
    try:
        response = requests.get(url, timeout=10)
        response.raise_for_status()
        log_message(f"成功获取网站源代码: {url}")
        return response.text
    except requests.RequestException as e:
        log_message(f"无法访问网址: {url}. 错误: {e}")
        return None

def extract_resources(html_content, base_url):
    links = re.findall(r'<link[^>]+href="([^"]+)"|<script[^>]+src="([^"]+)"|<img[^>]+src="([^"]+)"|<a[^>]+href="([^"]+\.html?)"', html_content)
    return {urljoin(base_url, link) for group in links for link in group if link}

def download_resource(url, folder, delay):
    try:
        time.sleep(delay)
        response = requests.get(url, stream=True, timeout=10)
        response.raise_for_status()
        filename = os.path.join(folder, os.path.basename(urlparse(url).path) or 'index.html')
        with open(filename, 'wb') as file:
            for data in response.iter_content(8192):
                file.write(data)
        log_message(f"下载完成: {url}")
    except Exception as e:
        log_message(f"下载失败: {url}. 错误: {e}")

def process_resources(url, folder, delay):
    html_content = get_website_source(url)
    if html_content:
        resources = extract_resources(html_content, url)
        log_message(f"找到 {len(resources)} 个资源，开始下载...")
        for resource in resources:
            threading.Thread(target=download_resource, args=(resource, folder, delay)).start()

def start_download():
    url, folder = url_entry.text(), path_entry.text()
    if url and folder:
        delay = 1 if high_frequency_var.isChecked() and QtWidgets.QMessageBox.question(window, "警告", "您选择了高频访问，可能会被防火墙拦截。是否继续？") == QtWidgets.QMessageBox.Yes else random.randint(10, 20)
        log_message("开始下载...")
        process_resources(url, folder, delay)
        log_message("下载任务已提交，请等待完成。")

def browse_folder():
    folder_selected = QtWidgets.QFileDialog.getExistingDirectory(window)
    if folder_selected:
        path_entry.setText(folder_selected)

def exit_program():
    QtWidgets.qApp.quit()

def open_contact_link():
    webbrowser.open("mailto:[email protected]")

def show_warning_message():
    msg = QtWidgets.QMessageBox()
    msg.setIcon(QtWidgets.QMessageBox.Warning)
    msg.setText("本工具仅供学习，禁止用于违法用途，禁止将爬取的网站资源文件私自使用，一切版权归属于原网站")
    msg.setWindowTitle("警告")
    msg.setStyleSheet("QLabel { color : red; }")
    msg.setStandardButtons(QtWidgets.QMessageBox.Ok | QtWidgets.QMessageBox.Cancel)
    msg.setWindowModality(QtCore.Qt.ApplicationModal)
    msg.setWindowFlags(msg.windowFlags() | QtCore.Qt.WindowStaysOnTopHint)
    return msg.exec_()

check_and_install_packages()
app = QtWidgets.QApplication(sys.argv)
window = QtWidgets.QWidget()
window.setWindowTitle("网站源码下载工具 v1.0测试版")
window.setFixedSize(600, 600)
window.setWindowFlags(QtCore.Qt.WindowStaysOnTopHint)

if show_warning_message() == QtWidgets.QMessageBox.Cancel:
    exit_program()

layout = QtWidgets.QVBoxLayout()
layout.addWidget(QtWidgets.QLabel("请输入网址(请使用快捷方式粘贴):"))
url_entry = QtWidgets.QLineEdit()
url_entry.setContextMenuPolicy(QtCore.Qt.NoContextMenu) 
layout.addWidget(url_entry)

layout.addWidget(QtWidgets.QLabel("请选择保存路径:"))
path_entry = QtWidgets.QLineEdit()
path_entry.setContextMenuPolicy(QtCore.Qt.NoContextMenu)
layout.addWidget(path_entry)

button_layout = QtWidgets.QHBoxLayout()
button_layout.addWidget(QtWidgets.QPushButton("浏览", clicked=browse_folder))
button_layout.addWidget(QtWidgets.QPushButton("下载资源", clicked=start_download))
button_layout.addWidget(QtWidgets.QPushButton("退出", clicked=exit_program))
layout.addLayout(button_layout)

layout.addWidget(QtWidgets.QPushButton("联系我", clicked=open_contact_link))
high_frequency_var = QtWidgets.QCheckBox("使用高频访问(可能会被防火墙拦截)")
layout.addWidget(high_frequency_var)

log_text = QtWidgets.QTextEdit()
log_text.setReadOnly(True)
log_text.setContextMenuPolicy(QtCore.Qt.NoContextMenu)
layout.addWidget(log_text)

window.setLayout(layout)
window.show()
sys.exit(app.exec_())

需要安装运行库，否则直接报错,解释器随便都可以

有问题请联系我，在本平台上不要通过分享代码中的联系方式,我并不会回复