Python采集药典2020版中药饮片数据

pip3 install pandas
pip3 install Pypinyin
pip3 install beautifulsoup4
pip3 install requests

import requests,time
from bs4 import BeautifulSoup
from pypinyin import lazy_pinyin, Style
import pandas as pd

#添加拼音缩写
def add_initials_to_chinese_name(name):
    if not name:
        return name
    initials = lazy_pinyin(name, style=Style.FIRST_LETTER)
    initials_str = ''.join(initials).lower()
    modified_name = f"{initials_str}-{name}"
    return modified_name

response=requests.get(url="https://ydz.chp.org.cn/front-api/children?bookId=1&directoryId=9")
data=response.json()
yc=None

ycdata=[]
if data['code']==200:
    yc=data['data']['children']
    excel_path = 'output.xlsx'

    for item in yc:
        oneyc=[]
        herb_name=item['title']
        modified_herb_name = add_initials_to_chinese_name(herb_name)
        print(modified_herb_name)
        oneyc.append(modified_herb_name)
        response=requests.get(url="https://ydz.chp.org.cn/front-api/entry/"+str(item['id']))
        if response.status_code==200:
            content=response.json()
            if content['code']==200:
                html_doc=content['data']['htmlContent']
                soup = BeautifulSoup(html_doc, 'html.parser')
                # 查找【性味与归经】的内容
                try:
                    # 尝试寻找【性味与归经】
                    xing_wei_gui_jing = soup.find('p', string='【性味与归经】').find_next_sibling('p').text
                except AttributeError:
                    try:
                        # 如果上面的寻找失败，则尝试寻找性味
                        xing_wei_gui_jing = soup.find('p', string='【性味】').find_next_sibling('p').text
                    except AttributeError:
                        # 如果两次尝试都失败了，则设置一个默认值
                        xing_wei_gui_jing = '自行替换'
                # 查找【功能与主治】的内容
                try:
                    gong_neng_zhu_zhi = soup.find('p', string='【功能与主治】').find_next_sibling('p').text
                except AttributeError:
                    gong_neng_zhu_zhi="没有功能与主治"
                try:
                    yong_fa_yong_liang = soup.find('p', string='【用法与用量】').find_next_sibling('p').text
                except AttributeError:
                    yong_fa_yong_liang="没有用法与用量"
                try:
                    zhu_yi = soup.find('p', string='【注意】').find_next_sibling('p').text
                except AttributeError:
                    zhu_yi="没有注意事项"
                print(xing_wei_gui_jing)
                print(gong_neng_zhu_zhi)
                print(yong_fa_yong_liang)
                print(zhu_yi)
                oneyc.append(xing_wei_gui_jing)
                oneyc.append(gong_neng_zhu_zhi)
                oneyc.append(yong_fa_yong_liang)
                oneyc.append(zhu_yi)
                ycdata.append(oneyc)
                time.sleep(1)
    result = pd.DataFrame(ycdata, columns=['名称', '性味', '功能','法量','注意'])
    result.to_csv("C:\\Users\daimadog\Desktop\ycdata.csv", index=False)
       
else:
    print("error")