Python 爬取12306火车票

获取火车站

stations.py

#import certifi

#import urllib3

import re

import requests

from pprint import pprint

url = 'https://kyfw.12306.cn/otn/resources/js/framework/station_name.js?station_version=1.9001'

response = requests.get(url, verify=False)

stations = re.findall('([\u4e00-\u9fa5]+)\|([A-Z]+)',response.text)

print(dict(stations))

#pprint(dict(stations),indent=4)

根据出发地,目的地,时间查询

#!/usr/bin/env python

#coding:utf-8

"""命令行的火车票查看器

Usage:

    12306 [-gdtkz] <from> <to> <date>

Options:

    -h,--help  显示帮助

    -g           高铁

    -d         动车

    -t         特快

    -k         快速

    -z         直达

Example:

    12306 上海 北京 2016-12-16

    12306 -dg 上海北京 2016-12-16

pip install requests prettytable docopt colorama

#prettytable格式化工具

#docopt是python3命令行的位置参数解析工具

#colorama是添加显示颜色

"""

import requests

from docopt import docopt

import sys

#sys.path.append(r"/py/stations")

from stations import stations

import json

from prettytable import PrettyTable

class TrainsCollections:

    header = '车次 车站 时间 历时 一等 二等 软卧 硬卧 硬座 无座'.split()

    def __init__(self,available_tranins,options):

        self.available_tranins = available_tranins

        self.options = options

    def _get_duration(self,raw_train):

        durations = raw_train.get('lishi').replace(':','小时') + '分'

        return durations

    #装饰器,把此函数定义为属性,以被别的函数调用

    @property

    def trains(self):

        for raw_train in self.available_tranins:

            train_no = raw_train['station_train_code']

            initial = train_no[0].lower()

            if not self.options or initial  in self.options:

                train = [

                train_no,

                '\n'.join([raw_train['from_station_name'],raw_train['to_station_name']]),

                '\n'.join([raw_train['start_time'],raw_train['arrive_time']]),

                self._get_duration(raw_train),

                raw_train['zy_num'],

                raw_train['ze_num'],

                raw_train['rw_num'],

                raw_train['yw_num'],

                raw_train['yz_num'],

                raw_train['wz_num'],

                ]

                #print(train)

                #生成器,一次返回一项

                yield train

    def pretty_print(self):

        pt = PrettyTable()

        pt._set_field_names(self.header)

        for train in self.trains:

            pt.add_row(train)

            print(pt)

def cli():

    '''

    {

    '-d':True,

    '-g':True,

    '-k':False,

    '-t':False,

    '-z':False,

    '<date>':'2017-03-23',

    '<from>':'海口',

    '<to>':'北京'}

    '''

    arguments = docopt(__doc__)

    print(arguments)

    from_station = stations.get(arguments['<from>'])

    to_station = stations.get(arguments['<to>'])

    date = arguments['<date>']

    #获取参数,列表解析

    options = ''.join([k for k,v in arguments.items() if v is True])

    url = 'https://kyfw.12306.cn/otn/leftTicket/query?leftTicketDTO.train_date={}&leftTicketDTO.from_station={}&leftTicketDTO.to_station={}&purpose_codes=ADULT'.format(date,from_station,to_station)

    #url = 'https://kyfw.12306.cn/otn/leftTicket/query?leftTicketDTO.train_date={}&leftTicketDTO.from_station={}&leftTicketDTO.to_station={}&purpose_codes=ADULT'

    r = requests.get(url,verify=False)

    print(url)

    #jjj = json.loads(str(r))

    available_tranins1 = r.json()['data']

    available_tranins = []

    for available_tranins2 in available_tranins1:

       available_tranins.append(available_tranins2['queryLeftNewDTO'])

        #available_tranins = list(available_tranins.keys())

       # print(available_tranins)

        #TrainsCollections(available_tranins,options).pretty_print()

    #available_tranins = available_tranins2['queryLeftNewDTO']

    #print(available_tranins)

    #print(type(available_tranins[0]))

    TrainsCollections(available_tranins,options).pretty_print()

if __name__ == '__main__':

    #TrainsCollections.cli()

    cli()