Bus parser

bus.py

# -*- encoding: utf-8 -*-
import requests
import bs4


def ebus(rid):
    # use rid to form the url
    url = 'http://e-bus.tpc.gov.tw/pda/online.php?rid=%s' % rid

    # there are two arrays which are go and back
    go_route = []
    back_route = []

    response = requests.get(url)
    if response.content:
        soup = bs4.BeautifulSoup(response.content, 'lxml')
        for tr in soup.find_all('tr'):
            # there are four td in each tr
            # <go place> <go time> <back place> <back time>
            tds = [td.get_text() for td in list(tr.find_all('td'))]
            if tds and len(tds) == 4:
                if tds[0] and tds[1]:
                    go_route.append({'place': tds[0], 'time': tds[1]})
                if tds[2] and tds[3]:
                    back_route.append({'place': tds[2], 'time': tds[3]})
    return go_route, back_route


def loveBus(name):
    # use name to form the url
    url = 'http://pda.5284.com.tw/MQS/businfo2.jsp?routename=%s' % name

    # there are two arrays which are go and back
    go_route = []
    back_route = []

    response = requests.get(url)
    if response.content:
        soup = bs4.BeautifulSoup(response.content, 'lxml')

        # use counter to distinguish go(0) and back(1)
        counter = 0

        for td in soup.find_all('td', valign='top'):
            for tr in td.find_all('tr'):
                # place and time in tr.stripped_strings
                route = list(tr.stripped_strings)
                if route and len(route) == 2:
                    item = {'place': route[0], 'time': route[1]}
                    # go route
                    if counter == 0:
                        go_route.append(item)
                    # back route
                    elif counter == 1:
                        back_route.append(item)
            counter += 1
    return go_route, back_route


if __name__ == '__main__':
    # ebus('5127')
    loveBus('27')