from selenium import webdriver 
browser = webdriver.Edge()
data = browser.get('http://www.baidu.com') #请求页面，会打开一个窗口
  text = browser.page_source #获取当前呈现页面的源码
  browser.quit() #关闭浏览器
  print(text) #输出

from selenium import webdriver
  from selenium.webdriver.common.by import By
  import time
  """
  由于懂车帝页面是JavaScript动态加载的，因此直接爬取页面html无法得到动态加载的数据
  因此需要模拟浏览器运行的库来应对，这里使用Selenium库
"""
  class Parsing:
    def __init__(self, url):
        # 设置WebDriver路径（以Edge为例）
        self.browser = webdriver.Edge()
        self.url = url
  
    def parsing(self, **kwargs):
        self.browser.get(self.url)
        for i in range(10):  # 页面滑动次数
            self.browser.execute_script("window.scrollTo(0, document.body.scrollHeight);")
            time.sleep(2)  # 滑动等待时间
        for key, value in kwargs.items():
            elements = self.browser.find_elements(By.CSS_SELECTOR, value) #find elements by CCS selector
            kwargs[key] = [element.text for element in elements] #must translate to text
        return kwargs
        self.browser.quit()

if __name__ == '__main__':
    parsing1 =  Parsing('https://www.dongchedi.com/sales/sale-x-1000-x-x-x-x')
    my_dict = {'name': 'a.tw-font-semibold',
               'brand_type': 'span.tw-text-color-gray-700.tw-ml-6',
               'sold_count': 'p.tw-text-18.tw-font-semibold.tw-leading-28',
               'prize': 'p.tw-leading-22.tw-text-color-red-500.tw-font-semibold'}
    res = parsing1.parsing(**my_dict)
    print(res)

def pre_process(car_dict, energy_list):
    #split the 'brand/type' into 'brand' and 'type'
    car_dict['brand'] = []
    car_dict['type'] = []
    for element in car_dict['brand_type']:
        b_t = element.split('/')
        if len(b_t)>=2:
            car_dict['brand'].append(b_t[0])
            car_dict['type'].append(b_t[1])
        else:
            car_dict['brand'].append(b_t[0])
            car_dict['type'].append('未知')
    del car_dict['brand_type']
  
    #del the ',' in sold_count
    car_dict['sold_count'] = [int(element.replace(',','')) for element in car_dict['sold_count']]
  
    #translate 'min_prize-max_prize万' intp 'min_prize' and 'max_prize'
    car_dict['min_prize'] = []
    car_dict['max_prize'] = []
    for element in car_dict['prize']:
        prz = element.replace('万', '')
        prz = prz.split('-')
        car_dict['min_prize'].append(float(prz[0]))
        car_dict['max_prize'].append(float(prz[1]))
    del car_dict['prize']
  
    #add the ranking
    rows = len(car_dict['name'])
    car_dict['rank'] = [i+1 for i in range(rows)]
  
    #new energy or not
    car_dict['new_energy'] = []
    for element in car_dict['name']:
        if element in energy_list:
            car_dict['new_energy'].append(1)
        else:
            car_dict['new_energy'].append(0)

import pymysql.cursors
  import pymysql
  
  class Connect_DB:
    def __init__(self, **kwargs):
        self.host = kwargs.get('host', 'localhost')
        self.port = kwargs.get('port', 3306)
        self.user = kwargs.get('user', 'root')
        self.passwd = kwargs.get('passwd', '123456')
        self.db = kwargs.get('db', 'demodb')
        self.charset = kwargs.get('charset', 'utf8')
        self.connection = None
        self.cursor = None
  
    def connect(self):
            self.connection = pymysql.Connect(
                host=self.host,
                port=self.port,
                user=self.user,
                passwd=self.passwd,
                db=self.db,
                charset=self.charset
            )
            self.cursor = self.connection.cursor()
            print('数据库连接成功')
    def disconnect(self):
        if self.cursor:
            self.cursor.close()
        if self.connection:
            self.connection.close()
            print('数据库关闭成功')
  
    def insert(self, data, table):
        # 确保数据库连接已经建立
        if not self.connection or not self.cursor:
            self.connect()
  
        # 接下来的操作
        try:
            self.cursor.execute(f"""create table if not exists {table} (
                                    `rank` int primary key,
                                    `name` varchar(255),
                                    `brand` varchar(255),
                                    `type` varchar(255),
                                    sold_count int,
                                    min_prize float,
                                    max_prize float,
                                    new_energy int)""")
            self.connection.commit()
  
            for i in range(len(data['name'])):
                self.cursor.execute(f"""insert into {table} values 
                                    ({data['rank'][i]}, "{data['name'][i]}", "{data['brand'][i]}", "{data['type'][i]}",
                                     {data['sold_count'][i]}, {data['min_prize'][i]}, {data['max_prize'][i]},{data['new_energy'][i]})""")
            self.connection.commit()
            print('插入数据库成功')
        except Exception as e:
            print(f"发生错误: {e}")
        finally:
            self.disconnect()
  
  if __name__ == "__main__":
    pass

def save_csv(car_dict, filename):
    #如果文件夹不存在，则创建
    if not os.path.exists('./parsingFile'):
        os.mkdir('./parsingFile')
    rows = len(car_dict['name'])
    with open(f'./parsingFile/{filename}.csv', 'w', newline='', encoding='GBK') as csvfile:
        writer = csv.DictWriter(csvfile, fieldnames = ['rank','name','brand','type','sold_count','min_prize','max_prize','new_energy'])
        writer.writeheader()
  
        for i in range(rows-1):
            row ={key: car_dict[key][i] for key in car_dict}
            writer.writerow(row)
        print('文件已保存')

import pandas as pd
  import matplotlib.pyplot as plt
  
  # 设置字体为黑体，正确显示负号
  plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
  
class DrawPic:
    def __init__(self):
        pass
  
    def bar_chart(self, info, labels, values):
        # 绘制条形图
        plt.figure(figsize=(10, 11))
        plt.bar(labels, values)
  
        # 设置标题和标签
        plt.title(f'一年内不同{info}汽车销售总量')
        plt.xlabel(info)
        plt.xticks(rotation='vertical')
        plt.ylabel('一年内销量')
        plt.show()
  
    import pandas as pd
    import matplotlib.pyplot as plt
  
    # 设置字体为黑体，正确显示负号
    plt.rcParams['font.sans-serif'] = ['SimHei']
    plt.rcParams['axes.unicode_minus'] = False
  
    def bar_and_line_chart(self, car_data):
        # 获取车型名称、销售数量、最高价格和最低价格
        labels = car_data['name']
        sales = car_data['sold_count']
        max_price = car_data['max_prize']
        min_price = car_data['min_prize']
        new_energy_status = car_data['new_energy']
  
        # 设置颜色（新能源为浅绿色，非新能源为浅红色），通过调整 alpha 值来控制颜色的浅深
        colors = ['lightgreen' if new_energy else 'lightcoral' for new_energy in new_energy_status]
  
        # 创建图形和轴
        fig, ax1 = plt.subplots(figsize=(17, 10))
  
        # 绘制条形图，调整透明度
        ax1.bar(labels, sales, color=colors, alpha=0.6)
        ax1.set_xlabel('车型')
        ax1.set_ylabel('销量', color='b')
        ax1.tick_params(axis='y', labelcolor='b')
        ax1.set_xticklabels(labels, rotation='vertical')
  
        # 创建第二个轴来绘制折线图
        ax2 = ax1.twinx()
        ax2.plot(labels, max_price, 'o-', color='red', label='最高价格')
        ax2.plot(labels, min_price, 'o-', color='black', label='最低价格')
        ax2.set_ylabel('价格', color='g')
        ax2.tick_params(axis='y', labelcolor='g')
  
        # 添加图例
        ax2.legend(loc='upper left')
  
        # 设置标题
        plt.title('不同车型销售总量与价格范围（绿色是新能源车）')
  
        # 显示图形
        plt.show()
  
  if __name__ == '__main__':
    # 读取 CSV 文件
    car_data = pd.read_csv('./parsingFile/file2.csv', encoding='GBK')
    # 计算每个品牌的总销量，并按销量从高到低排序
    brand_totals = car_data.groupby('brand')['sold_count'].sum().sort_values(ascending=False)
    # 获取标签和值
    labels = brand_totals.index
    values = brand_totals.values
    # 创建 DrawPic 对象并绘图
    drawer = DrawPic()
    drawer.bar_chart('品牌', labels, values)
  
  
    # 计算每个类型的总销量，并按销量从高到低排序
    brand_totals = car_data.groupby('type')['sold_count'].sum().sort_values(ascending=False)
    # 获取标签和值
    labels = brand_totals.index
    values = brand_totals.values
    # 创建 DrawPic 对象并绘图
    drawer = DrawPic()
    drawer.bar_chart('车型',labels, values)
  
    drawer.bar_and_line_chart(car_data)

for i in range(10):  # 页面滑动次数
    self.browser.execute_script("window.scrollTo(0, document.body.scrollHeight);")
    time.sleep(2)  # 滑动等待时间

brand_totals = car_data.groupby('brand')['sold_count'].sum().sort_values(ascending=False)
  # 获取标签和值
  labels = brand_totals.index
values = brand_totals.values
  # 创建 DrawPic 对象并绘图
  drawer = DrawPic()
drawer.bar_chart('品牌', labels, values)

with open(f'./parsingFile/{filename}.csv', 'w', newline='', encoding='GBK') as csvfile:
    writer = csv.DictWriter(csvfile, fieldnames = ['rank','name','brand','type','sold_count','min_prize','max_prize','new_energy'])
    writer.writeheader()

# 创建第二个轴来绘制折线图
  ax2 = ax1.twinx()
ax2.plot(labels, max_price, 'o-', color='red', label='最高价格')
ax2.plot(labels, min_price, 'o-', color='black', label='最低价格')
ax2.set_ylabel('价格', color='g')
ax2.tick_params(axis='y', labelcolor='g')
  
  # 添加图例
  ax2.legend(loc='upper left')