stranger_renderer.py

   1 from bs4 import BeautifulSoup
   2 import datetime
   3 import file_writer
   4 import grab_bag
   5 import httplib
   6 import page_builder
   7 import profanity_filter
   8 import random
   9 import re
  10 import renderer
  11 import sets
  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.thestranger.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_none.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             "/events/?page=1&picks=true",
  90             "/events/?page=2&picks=true",
  91             "/events/?page=3&picks=true",
  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("/events/?start-date=%s&picks=true" % 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("/events/?start-date=%s&page=1&picks=true" % next_sat)
 104             feed_uris.append("/events/?start-date=%s&page=2&picks=true" % 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("/events/?start-date=%s&page=1&picks=true" % next_sun)
 110             feed_uris.append("/events/?start-date=%s&page=2&picks=true" % next_sun)
 111
 112         for uri in feed_uris:
 113             self.debug_print("fetching '%s'" % uri)
 114             self.conn = httplib.HTTPSConnection(self.feed_site)
 115             self.conn.request(
 116                 "GET",
 117                 uri,
 118                 None,
 119                 {"Accept-Charset": "utf-8"})
 120             response = self.conn.getresponse()
 121             if response.status != 200:
 122                 print("stranger: Failed, status %d" % (response.status))
 123                 continue
 124
 125             raw = response.read()
 126             soup = BeautifulSoup(raw, "html.parser")
 127             filter = profanity_filter.profanity_filter()
 128             for x in soup.find_all('div', class_='row event list-item mb-3 py-3'):
 129                 text = x.get_text();
 130                 if (filter.contains_bad_words(text)):
 131                     continue
 132
 133 #          <div class="row event list-item mb-3 py-3">
 134 #          <div class="col-12">
 135 #                <a class="category-tag" href="?category=on-demand">On Demand</a>
 136 #          </div> // col-12
 137 #          <div class="col-md-3 order-1 order-md-3">
 138 #              <a href="https://everout.thestranger.com/events/spliff-2020-on-demand/e24125/">
 139 #                  <img class="img-responsive" src="https://d2i729k8wyri5w.cloudfront.net/eyJidWNrZXQiOiAiZXZlcm91dC1pbWFnZXMtcHJvZHVjdGlvbiIsICJrZXkiOiAiaW1hZ2UtMTU5MTA2NTQxODU5NzA5My1vcmlnaW5hbC1sb2dvLmpwZWciLCAiZWRpdHMiOiB7InJlc2l6ZSI6IHsiZml0IjogImNvdmVyIiwgIndpZHRoIjogNDAwLCAiaGVpZ2h0IjogMzAwfX19">
 140 #              </a>
 141 #          </div> // col-md-3 order-1 order-md-3
 142 #          <div class="col-md-6 order-2 order-md-1 event-details">
 143 #             <h3 class="mb-0 event-title">
 144 #                 <a href="https://everout.thestranger.com/events/spliff-2020-on-demand/e24125/"><span class="staff-pick fas fa-star" aria-hidden="true"></span></a>
 145 #                 <a href="https://everout.thestranger.com/events/spliff-2020-on-demand/e24125/">
 146 #                 <span class="title-link">SPLIFF 2020 - On Demand</span>
 147 #                 </a>
 148 #             </h3>
 149 #             <div class="event-date">
 150 #               Every day
 151 #             </div> // event-date
 152 #             <div class="event-time">
 153 #             </div> // event-time
 154 #          </div> // col-md-6 order-2 order-md-1 event-details
 155 #          <div class="col-md-3 order-3 order-md-2 location-column">
 156 #            <div class="location-name">
 157 #              <i class="fad fa-map-marker-alt"></i> <a href="https://everout.thestranger.com/locations/the-stranger-online/l27660/">The Stranger (Online)</a>
 158 #            </div> // location-name
 159 #            <div class="location-region">
 160 #            </div> // location-region
 161 #            <ul class="event-tags">
 162 #              <li>$10 - $20</li>
 163 #            </ul>
 164 #          </div> // col-md-3 order-3 order-md-2 location-colum
 165 #        </div> // row event list-item mb-3 py-3
 166
 167                 raw = unicode(x)
 168                 raw = raw.replace('src="/',
 169                                   'align="left" src="https://www.thestranger.com/')
 170                 raw = raw.replace('href="/',
 171                                   'href="https://www.thestranger.com/')
 172                 raw = raw.replace('FREE', 'Free')
 173                 raw = raw.replace('Save Event', '')
 174                 raw = re.sub('^\s*$', '', raw, 0, re.MULTILINE)
 175                 raw = re.sub('\n+', '\n', raw)
 176                 raw = re.sub('<span[^<>]*class="calendar-post-ticket"[^<>]*>.*</#span>', '', raw, 0, re.DOTALL | re.IGNORECASE)
 177                 self.events.add(raw.encode('utf-8'))
 178             self.debug_print("fetched %d events so far." % self.events.size())
 179         return self.events.size() > 0
 180
 181 x = stranger_events_renderer({"Test", 123})
 182 x.periodic_render("Fetch Events")
 183 x.periodic_render("Shuffle Events")