DaveM@0: #!/usr/bin/env python DaveM@0: DaveM@0: import requests DaveM@0: import re DaveM@0: from HTMLParser import HTMLParser DaveM@0: # from lxml import html DaveM@0: from bs4 import BeautifulSoup DaveM@0: DaveM@0: DaveM@0: url = "https://horoscopes.astro-seek.com/calculate-love-compatibility/?send_calculation=1&muz_narozeni_den=1&muz_narozeni_mesic=1&muz_narozeni_rok=1970&muz_narozeni_hodina=00&muz_narozeni_minuta=00&muz_narozeni_city=London%2C+United+Kingdom&muz_narozeni_mesto_hidden=London&muz_narozeni_stat_hidden=GB&muz_narozeni_podstat_kratky_hidden=England&muz_narozeni_podstat_hidden=England&muz_narozeni_podstat2_kratky_hidden=Greater+London&muz_narozeni_podstat3_kratky_hidden=undefined&muz_narozeni_input_hidden=&muz_narozeni_sirka_stupne=51&muz_narozeni_sirka_minuty=30&muz_narozeni_sirka_smer=0&muz_narozeni_delka_stupne=0&muz_narozeni_delka_minuty=8&muz_narozeni_delka_smer=1&muz_narozeni_timezone_form=auto&muz_narozeni_timezone_dst_form=auto&send_calculation=1&zena_narozeni_den=1&zena_narozeni_mesic=1&zena_narozeni_rok=1970&zena_narozeni_hodina=00&zena_narozeni_minuta=00&zena_narozeni_city=Berlin%2C+Germany&zena_narozeni_mesto_hidden=Berlin&zena_narozeni_stat_hidden=DE&zena_narozeni_podstat_kratky_hidden=Berlin&zena_narozeni_podstat_hidden=Berlin&zena_narozeni_podstat2_kratky_hidden=undefined&zena_narozeni_podstat3_kratky_hidden=undefined&zena_narozeni_input_hidden=&zena_narozeni_sirka_stupne=52&zena_narozeni_sirka_minuty=31&zena_narozeni_sirka_smer=0&zena_narozeni_delka_stupne=13&zena_narozeni_delka_minuty=24&zena_narozeni_delka_smer=0&zena_narozeni_timezone_form=auto&zena_narozeni_timezone_dst_form=auto&switch_interpretations=0&house_system=placidus&uhel_orbis=#tabs_redraw" DaveM@0: url = 'https://horoscopes.astro-seek.com/calculate-love-compatibility/' DaveM@0: payload = {'?send_calculation':'1','muz_narozeni_den':'25','muz_narozeni_mesic':'6','muz_narozeni_rok':'1988','muz_narozeni_hodina':'00','muz_narozeni_minuta':'00','muz_narozeni_city':'London%2C+United+Kingdom','zena_narozeni_den':'14','zena_narozeni_mesic':'3','zena_narozeni_rok':'1995','zena_narozeni_hodina':'00','zena_narozeni_minuta':'00'} DaveM@0: # {'muz_narozeni_den':'1','muz_narozeni_mesic':'1','muz_narozeni_rok':'1970'} DaveM@0: DaveM@0: # GET DaveM@0: r = requests.get(url) DaveM@0: # # GET with params in URL DaveM@0: r = requests.get(url, params=payload) DaveM@0: DaveM@0: # Response, status etc DaveM@0: # print r.text.encode('utf-8') DaveM@0: # tree = html.fromstring(r.content) DaveM@0: # tcTable = tree.xpath('//div[@class="detail-rozbor-clanek"]/strong/text()') DaveM@0: # print tcTable DaveM@0: DaveM@0: soup = BeautifulSoup(r.content, 'lxml') DaveM@0: tcCell = soup.find_all('div', attrs={'class':'tc'}) DaveM@0: # print tcCell DaveM@0: DaveM@0: DaveM@0: for cell in tcCell: DaveM@0: person = dict() DaveM@0: tableCell = cell.find_all('td'); DaveM@0: if len(tableCell) > 2: DaveM@0: C = tableCell[0].strong.contents[0].encode('utf-8') DaveM@0: D = tableCell[2].strong.contents[0].encode('utf-8') DaveM@0: print (C,D) DaveM@0: A = re.search("\/>(.*)
.*\(([0-9]*)\\xc2\\xb0([0-9]*)(.*)\)",str(tableCell[1])) DaveM@0: # A0 = A.group(1) DaveM@0: # A1 = A.group(2).split('\xc2\xb0')[0] DaveM@0: # A2 = A.group(2).split('\xc2\xb0')[1].split('\xe2')[0] DaveM@0: print (A.group(1),A.group(2),A.group(3)) DaveM@0: person[(C,D)] = (A.group(1),A.group(2),A.group(3)) DaveM@0: return person DaveM@0: # print re.search(">, (.*),
.*\(([0-9]*)\\xb([0-9]*)\\u",valueCell.contents.encode('utf-8')) DaveM@0: # except IndexError: DaveM@0: # print 'te' DaveM@0: # print "\n\n\n" DaveM@0: DaveM@0: DaveM@0: DaveM@0: DaveM@0: DaveM@0: DaveM@0: # # # # # DaveM@0: DaveM@0: payload = {'?send_calculation':'1', DaveM@0: 'muz_narozeni_den':'25', DaveM@0: 'muz_narozeni_mesic':'6', DaveM@0: 'muz_narozeni_rok':'1988', DaveM@0: 'muz_narozeni_hodina':'00', DaveM@0: 'muz_narozeni_minuta':'00', DaveM@0: 'muz_narozeni_city':'London%2C+United+Kingdom', DaveM@0: 'zena_narozeni_den':'14', DaveM@0: 'zena_narozeni_mesic':'3', DaveM@0: 'zena_narozeni_rok':'1995', DaveM@0: 'zena_narozeni_hodina':'00', DaveM@0: 'zena_narozeni_minuta':'00'}