parse wanfangh live ER status board

wanfangh.py

#!/usr/bin/env python
#coding:UTF-8
import requests, re, json, os
from datetime import datetime
os.environ['TZ'] = 'ROC'
html = requests.get('http://www.wanfang.gov.tw/p11_information.aspx')

update_time = re.findall(u"日期：(.+?)</div",html.text)[0]
pending = re.findall(u"\'>(.+?)</div",html.text)[:-1]
# prase like ["<div class='p11_con'><div class='p11_icon05'>否", '1', '0', '16', '1']
pending[0] = pending[0][-1] #take "否" or "是"
values = [int(ele) for ele in pending[1:]]
keys = ['pending_doctor','pending_bed', 'pending_ward', 'pending_icu']
report = { key:value for value, key in zip(values, keys) }
report["hospital_sn"] = '1301200010'
report['full_reported'] = False if pending[0]==u'否' else True
report["update_time"] = datetime.strptime(update_time, '%Y/%m/%d %H:%M').strftime('%s')
#using .timestamp() if py3 else .strftime('%s')

print ( json.dumps(report, ensure_ascii=False) )
#[{"update_time": "1407225600", "pending_ICU": 2, "pending_bed": 0, "full_reported": false, "Hosptial_SN": "1301200010", "pending_doctor": 2, "pending_ward": 19}]

!/usr/bin/env python

coding:UTF-8

import requests, re, json
html = requests.get('http://www.wanfang.gov.tw/W402008web_new/epd_query.asp')
html.encoding='big5'

update_time = re.findall(u'日期:(.) ，系統',html.text)
pending = re.findall(u'：(.)',html.text)
keys = ['full_reported','pending_doctor','pending_bed', 'pending_ward', 'pending_ICU']

report = { key:value.replace(u' 人','').strip() for value, key in zip(pending, keys) }
update_time = update_time[0].replace(u'：',':').replace('\u3000',' ')
report["Hosptial_SN"] = 1301200010
report["update_time"] = update_time
report = [report]

print ( json.dumps(report, ensure_ascii=False) )

fix int value, strftime()