郵便物検索結果の状態を取ってきて吐き出すスクリプト

config.example.yaml

config: 
    color: true
goods:
    - URL: http://tracking.post.japanpost.jp/service/detail_search.do?searchKind=M004&reqCode=HOGEHOGE123
      kind: 健康かつ快活になれる海外サプリメント

trace.py

# -*- coding: utf-8 -*-

import yaml
import urllib
import sys
from termcolor import colored
from bs4 import BeautifulSoup


class InterStatus(object):

    def __init__(self, conf):
        print(u'Connecting....')
        html = urllib.urlopen(conf['URL']).read()
        print(u'Get URL soruce')
        print(conf['URL'])
        soup = BeautifulSoup(html)
        try:
            self.table_data = soup.find_all('table')[7].find_all('tr')
        except IndexError:
            print('!!!!! Invalid Input Error !!!!!')
            print('Often, URL Error or Maintenance')
            sys.exit(1)
        self.kind = conf['kind']
        self.is_pass_line = False

    def colorize(self, string, color):
        if self.is_color:
            string = colored(string, color)
        return string

    def render(self, config):
        self.is_color = (
            'color' in config and config['color'])
        p_str_line = self.colorize(
            u'=' * 40, 'yellow')
        p_str_name = self.colorize(
            u'【' + self.kind + u'】\n', 'cyan')
        print (
            p_str_line + "\n" +
            p_str_name +
            p_str_line)

        for tr in self.table_data:
            line = self._generate_line(tr)
            if line:
                print line
        print

    def _generate_line(self, tr):
        line = ""
        first_line = True
        for td in tr.find_all('td'):
            if 'rowspan' in td.attrs and td['rowspan'] == "2":
                self.is_pass_line = True
            elif self.is_pass_line and not 'rowspan' in td:
                self.is_pass_line = False
                continue
            td_text = td.text.rstrip().strip()
            if td_text:
                if first_line:
                    line = '%-15s|' % td_text
                    first_line = False
                else:
                    line += u'{0: <20}'.format(td_text)
        return line


def factory_InterStatus(configs):
    return_interstatus = []
    for conf in configs:
        return_interstatus.append(
            InterStatus(conf))
    return return_interstatus


def parse_config(filepath):
    conffile = open(filepath).read().decode('utf8')
    conf = yaml.load(conffile)
    return conf


def main():
    yaml = parse_config('config.yaml')
    config = yaml['config']
    goods = yaml['goods']
    status = factory_InterStatus(goods)

    for state in status:
        state.render(config)

if __name__ == "__main__":
    main()