stranger_renderer.py

   1 #!/usr/bin/env python3
   2
   3 import datetime
   4 import http.client
   5 import logging
   6 import re
   7 from typing import Dict
   8
   9 from bs4 import BeautifulSoup  # type: ignore
  10 from pyutilz import profanity_filter
  11
  12 import file_writer
  13 import grab_bag
  14 import page_builder
  15 import renderer
  16
  17
  18 logger = logging.getLogger(__file__)
  19
  20
  21 class stranger_events_renderer(renderer.abstaining_renderer):
  22     def __init__(self, name_to_timeout_dict: Dict[str, int]):
  23         super().__init__(name_to_timeout_dict)
  24         self.feed_site = "everout.com"
  25         self.events = grab_bag.grab_bag()
  26         self.pfilter = profanity_filter.ProfanityFilter()
  27
  28     def debug_prefix(self) -> str:
  29         return "stranger"
  30
  31     def periodic_render(self, key: str) -> bool:
  32         logger.debug("called for action %s" % key)
  33         if key == "Fetch Events":
  34             return self.fetch_events()
  35         elif key == "Shuffle Events":
  36             return self.shuffle_events()
  37         else:
  38             raise Exception("Unknown operaiton")
  39
  40     def get_style(self):
  41         return """
  42 <STYLE>
  43 .calendar-post {
  44   line-height: 96%;
  45 }
  46 .calendar-post-title a {
  47   text-decoration: none;
  48   color:black;
  49   font-size:125%
  50   line-height:104%;
  51 }
  52 .calendar-post-date {
  53 }
  54 .calendar-post-category {
  55 }
  56 .calendar-post-location {
  57 }
  58 .calendar-post-price {
  59 }
  60 .calendar-touch-link {
  61 }
  62 .calendar-category {
  63   background-color:lightyellow;
  64   color:black;
  65   border:none;
  66   font-size:50%;
  67   padding:1px;
  68 }
  69 .calendar-post-price-mobile {
  70   visibility: hidden;
  71 }
  72 .img-responsive {
  73   float: left;
  74   margin: 10px 10px 10px 10px;
  75 }
  76 </STYLE>"""
  77
  78     def shuffle_events(self) -> bool:
  79         layout = page_builder.page_builder()
  80         layout.set_layout(page_builder.page_builder.LAYOUT_FOUR_ITEMS)
  81         layout.set_title("Stranger Events")
  82         layout.set_style(self.get_style())
  83         subset = self.events.subset(4)
  84         if subset is None:
  85             logger.debug("Not enough events to build page.")
  86             return False
  87
  88         for msg in subset:
  89             layout.add_item(msg)
  90         with file_writer.file_writer("stranger-events_2_36000.html") as f:
  91             layout.render_html(f)
  92         return True
  93
  94     def fetch_events(self) -> bool:
  95         self.events.clear()
  96         feed_uris = []
  97         now = datetime.datetime.now()
  98         ts = now + datetime.timedelta(1)
  99         tomorrow = datetime.datetime.strftime(ts, "%Y-%m-%d")
 100         feed_uris.append(f"/seattle/events/?start-date={tomorrow}")
 101         delta = 5 - now.weekday()
 102         if delta <= 0:
 103             delta += 7
 104         if delta > 1:
 105             ts = now + datetime.timedelta(delta)
 106             next_sat = datetime.datetime.strftime(ts, "%Y-%m-%d")
 107             feed_uris.append(f"/seattle/events/?start-date={next_sat}&page=1")
 108             feed_uris.append(f"/seattle/events/?start-date={next_sat}&page=2")
 109         delta += 1
 110         if delta > 1:
 111             ts = now + datetime.timedelta(delta)
 112             next_sun = datetime.datetime.strftime(ts, "%Y-%m-%d")
 113             feed_uris.append(f"/seattle/events/?start-date={next_sun}&page=1")
 114             feed_uris.append(f"/seattle/events/?start-date={next_sun}&page=2")
 115
 116         for uri in feed_uris:
 117             try:
 118                 logger.debug("fetching 'https://%s%s'" % (self.feed_site, uri))
 119                 self.conn = http.client.HTTPSConnection(self.feed_site)
 120                 self.conn.request("GET", uri, None, {"Accept-Charset": "utf-8"})
 121                 response = self.conn.getresponse()
 122                 if response.status != 200:
 123                     logger.debug("Connection failed, status %d" % (response.status))
 124                     logger.debug(str(response.getheaders()))
 125                     continue
 126                 raw = response.read()
 127             except Exception:
 128                 logger.debug("Exception talking to the stranger, ignoring.")
 129                 continue
 130
 131             soup = BeautifulSoup(raw, "html.parser")
 132             for x in soup.find_all("div", class_="row event list-item mb-3 py-3"):
 133                 text = x.get_text()
 134                 if self.pfilter.contains_bad_word(text):
 135                     continue
 136                 raw_str = str(x)
 137                 raw_str = raw_str.replace(
 138                     'src="/', 'align="left" src="https://www.thestranger.com/'
 139                 )
 140                 raw_str = raw_str.replace('href="/', 'href="https://www.thestranger.com/')
 141                 raw_str = raw_str.replace("FREE", "Free")
 142                 raw_str = raw_str.replace("Save Event", "")
 143                 raw_str = re.sub("^\s*$", "", raw_str, 0, re.MULTILINE)
 144                 raw_str = re.sub(
 145                     '<span[^<>]*class="calendar-post-ticket"[^<>]*>.*</#span>',
 146                     "",
 147                     raw_str,
 148                     0,
 149                     re.DOTALL | re.IGNORECASE,
 150                 )
 151                 self.events.add(raw_str)
 152             logger.debug(f"fetched {self.events.size()} events so far.")
 153         return self.events.size() > 0
 154
 155
 156 # Test
 157 x = stranger_events_renderer({"Test", 123})
 158 x.periodic_render("Fetch Events")
 159 x.periodic_render("Shuffle Events")