|
|
|
@ -2,12 +2,13 @@ import requests
|
|
|
|
from bs4 import BeautifulSoup
|
|
|
|
from bs4 import BeautifulSoup
|
|
|
|
from http.server import BaseHTTPRequestHandler, HTTPServer
|
|
|
|
from http.server import BaseHTTPRequestHandler, HTTPServer
|
|
|
|
|
|
|
|
|
|
|
|
overview_url = 'https://gruene-hohenlohe.de/kalender'
|
|
|
|
|
|
|
|
overview_html = requests.get(overview_url, timeout=60).text
|
|
|
|
|
|
|
|
overview_soup = BeautifulSoup(overview_html, 'html.parser')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def scrape():
|
|
|
|
def scrape():
|
|
|
|
|
|
|
|
output = ''
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
overview_url = 'https://gruene-hohenlohe.de/kalender'
|
|
|
|
|
|
|
|
overview_html = requests.get(overview_url, timeout=60).text
|
|
|
|
|
|
|
|
overview_soup = BeautifulSoup(overview_html, 'html.parser')
|
|
|
|
|
|
|
|
|
|
|
|
for href in overview_soup.select('.media-body h2 a'):
|
|
|
|
for href in overview_soup.select('.media-body h2 a'):
|
|
|
|
event_url = 'https://gruene-hohenlohe.de/' + href.attrs['href']
|
|
|
|
event_url = 'https://gruene-hohenlohe.de/' + href.attrs['href']
|
|
|
|
event_html = requests.get(event_url, timeout=60).text
|
|
|
|
event_html = requests.get(event_url, timeout=60).text
|
|
|
|
@ -15,8 +16,6 @@ def scrape():
|
|
|
|
|
|
|
|
|
|
|
|
data = event_soup.select('.calendarize dl dd')
|
|
|
|
data = event_soup.select('.calendarize dl dd')
|
|
|
|
|
|
|
|
|
|
|
|
output = ''
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# date
|
|
|
|
# date
|
|
|
|
output += data[0].text.strip()
|
|
|
|
output += data[0].text.strip()
|
|
|
|
output += ' '
|
|
|
|
output += ' '
|
|
|
|
@ -30,7 +29,7 @@ def scrape():
|
|
|
|
output += ', '
|
|
|
|
output += ', '
|
|
|
|
|
|
|
|
|
|
|
|
# place
|
|
|
|
# place
|
|
|
|
#output += data[2].text.strip()
|
|
|
|
output += data[2].text.strip()
|
|
|
|
output += ', '
|
|
|
|
output += ', '
|
|
|
|
|
|
|
|
|
|
|
|
# title
|
|
|
|
# title
|
|
|
|
@ -45,7 +44,9 @@ def scrape():
|
|
|
|
except IndexError:
|
|
|
|
except IndexError:
|
|
|
|
pass
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
|
|
return output
|
|
|
|
output += '<br>'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return output
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@ -54,7 +55,13 @@ class SimpleHTTPRequestHandler(BaseHTTPRequestHandler):
|
|
|
|
def do_GET(self):
|
|
|
|
def do_GET(self):
|
|
|
|
self.send_response(200)
|
|
|
|
self.send_response(200)
|
|
|
|
self.end_headers()
|
|
|
|
self.end_headers()
|
|
|
|
self.wfile.write(scrape().encode('utf-8'))
|
|
|
|
html = templateHtml.replace('--body--', scrape())
|
|
|
|
|
|
|
|
self.wfile.write(html.encode('utf-8'))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
with open('template.html', 'r') as templateFile:
|
|
|
|
|
|
|
|
templateHtml = templateFile.read()
|
|
|
|
|
|
|
|
|
|
|
|
httpd = HTTPServer(('', 8000), SimpleHTTPRequestHandler)
|
|
|
|
httpd = HTTPServer(('', 8000), SimpleHTTPRequestHandler)
|
|
|
|
httpd.serve_forever()
|
|
|
|
httpd.serve_forever()
|
|
|
|
|
