stranger_renderer.py

   1 from bs4 import BeautifulSoup
   2 import datetime
   3 import file_writer
   4 import grab_bag
   5 import http.client
   6 import page_builder
   7 import profanity_filter
   8 import random
   9 import re
  10 import renderer
  11 import renderer_catalog
  12
  13 class stranger_events_renderer(renderer.debuggable_abstaining_renderer):
  14     def __init__(self, name_to_timeout_dict):
  15         super(stranger_events_renderer, self).__init__(name_to_timeout_dict, True)
  16         self.feed_site = "everout.com"
  17         self.events = grab_bag.grab_bag()
  18
  19     def debug_prefix(self):
  20         return "stranger"
  21
  22     def periodic_render(self, key):
  23         self.debug_print("called for action %s" % key)
  24         if key == "Fetch Events":
  25             return self.fetch_events()
  26         elif key == "Shuffle Events":
  27             return self.shuffle_events()
  28         else:
  29             raise error("Unknown operaiton")
  30
  31     def get_style(self):
  32         return """
  33 <STYLE>
  34 .calendar-post {
  35   line-height: 96%;
  36 }
  37 .calendar-post-title a {
  38   text-decoration: none;
  39   color:black;
  40   font-size:125%
  41   line-height:104%;
  42 }
  43 .calendar-post-date {
  44 }
  45 .calendar-post-category {
  46 }
  47 .calendar-post-location {
  48 }
  49 .calendar-post-price {
  50 }
  51 .calendar-touch-link {
  52 }
  53 .calendar-category {
  54   background-color:lightyellow;
  55   color:black;
  56   border:none;
  57   font-size:50%;
  58   padding:1px;
  59 }
  60 .calendar-post-price-mobile {
  61   visibility: hidden;
  62 }
  63 .img-responsive {
  64   float: left;
  65   margin: 10px 10px 10px 10px;
  66 }
  67 </STYLE>"""
  68
  69     def shuffle_events(self):
  70         layout = page_builder.page_builder()
  71         layout.set_layout(page_builder.page_builder.LAYOUT_FOUR_ITEMS)
  72         layout.set_title("Stranger Events")
  73         layout.set_style(self.get_style())
  74         subset = self.events.subset(4)
  75         if subset is None:
  76             self.debug_print("Not enough events to build page.")
  77             return False
  78
  79         for msg in subset:
  80             layout.add_item(msg)
  81         f = file_writer.file_writer('stranger-events_2_36000.html')
  82         layout.render_html(f)
  83         f.close()
  84         return True
  85
  86     def fetch_events(self):
  87         self.events.clear()
  88         feed_uris = [
  89             "/stranger-seattle/events/?page=1",
  90             "/stranger-seattle/events/?page=2",
  91             "/stranger-seattle/events/?page=3",
  92         ]
  93         now = datetime.datetime.now()
  94         ts = now + datetime.timedelta(1)
  95         tomorrow = datetime.datetime.strftime(ts, "%Y-%m-%d")
  96         feed_uris.append("/stranger-seattle/events/?start-date=%s" % tomorrow)
  97         delta = 5 - now.weekday()
  98         if delta <= 0:
  99             delta += 7
 100         if delta > 1:
 101             ts = now + datetime.timedelta(delta)
 102             next_sat = datetime.datetime.strftime(ts, "%Y-%m-%d")
 103             feed_uris.append("/stranger-seattle/events/?start-date=%s&page=1" % next_sat)
 104             feed_uris.append("/stranger-seattle/events/?start-date=%s&page=2" % next_sat)
 105         delta += 1
 106         if delta > 1:
 107             ts = now + datetime.timedelta(delta)
 108             next_sun = datetime.datetime.strftime(ts, "%Y-%m-%d")
 109             feed_uris.append("/stranger-seattle/events/?start-date=%s&page=1" % next_sun)
 110             feed_uris.append("/stranger-seattle/events/?start-date=%s&page=2" % next_sun)
 111
 112         for uri in feed_uris:
 113             try:
 114                 self.debug_print("fetching 'https://%s%s'" % (self.feed_site, uri))
 115                 self.conn = http.client.HTTPSConnection(self.feed_site)
 116                 self.conn.request(
 117                     "GET",
 118                     uri,
 119                     None,
 120                     {"Accept-Charset": "utf-8"})
 121                 response = self.conn.getresponse()
 122                 if response.status != 200:
 123                     self.debug_print("Connection failed, status %d" % (
 124                         response.status))
 125                     self.debug_print(response.getheaders())
 126                     continue
 127                 raw = response.read()
 128             except:
 129                 self.debug_print("Exception talking to the stranger, ignoring.")
 130                 continue
 131
 132             soup = BeautifulSoup(raw, "html.parser")
 133             filter = profanity_filter.profanity_filter()
 134             for x in soup.find_all('div', class_='row event list-item mb-3 py-3'):
 135                 text = x.get_text();
 136                 if (filter.contains_bad_words(text)):
 137                     continue
 138                 raw = str(x)
 139                 raw = raw.replace('src="/',
 140                                   'align="left" src="https://www.thestranger.com/')
 141                 raw = raw.replace('href="/',
 142                                   'href="https://www.thestranger.com/')
 143                 raw = raw.replace('FREE', 'Free')
 144                 raw = raw.replace('Save Event', '')
 145                 raw = re.sub('^\s*$', '', raw, 0, re.MULTILINE)
 146                 #raw = re.sub('\n+', '\n', raw)
 147                 raw = re.sub('<span[^<>]*class="calendar-post-ticket"[^<>]*>.*</#span>', '', raw, 0, re.DOTALL | re.IGNORECASE)
 148                 self.events.add(raw)
 149             self.debug_print("fetched %d events so far." % self.events.size())
 150         return self.events.size() > 0
 151
 152 # Test
 153 #x = stranger_events_renderer({"Test", 123})
 154 #x.periodic_render("Fetch Events")
 155 #x.periodic_render("Shuffle Events")