def GET(self):
rdio, current_user, user_id = get_rdio_and_current_user()
disco_user, prefs = get_discoversong_user(user_id)
programs_page = etree.parse("http://www.kpfa.org/programs/music", etree.HTMLParser())
divs = try_get(programs_page, '//div[@class="node"][.//a[starts-with(@href, "/playlists/show/")]]')
shows = []
for div in divs:
title = try_get(div, './/h2/a[@title]/@title')
playlists_link = try_get(div, './/a[starts-with(@href, "/playlists/show/")]/@href')
show_num = int(playlists_link[len("/playlists/show/"):])
shows.append(show(title=title, number=show_num))
return render.list_shows(params=Params(user=disco_user, preferences=prefs, message="", environment_message=get_environment_message()),
shows=shows)
def GET(self, show_number, episode_number):
episode_url = "http://www.kpfa.org/playlist/id/%i" % int(episode_number)
playlist_page = etree.parse(episode_url, etree.HTMLParser())
show_title = try_get(playlist_page, '//div[@class="field-item"]/a/text()')
episode_title = try_get(playlist_page, '//div[@class="orangeheader"]/text()', xform=get_episode_title)
songs = EpisodeImporter().scrape_songs(episode_number)
rdio, current_user, user_id = get_rdio_and_current_user()
disco_user, prefs = get_discoversong_user(user_id)
return render.import_episode(params=Params(user=disco_user, preferences=prefs, message="", environment_message=get_environment_message()),
search=search,
importing=[],
show_title=show_title,
show_number=show_number,
episode_title=episode_title,
episode_number=episode_number,
show_url="/import/" + show_number,
songs=songs)
def GET(self, show_number):
inputs = web.input()
p_url = inputs.get('p')
rdio, current_user, user_id = get_rdio_and_current_user()
disco_user, prefs = get_discoversong_user(user_id)
show_url = "http://www.kpfa.org/playlists/show/" + show_number
episodes_page = etree.parse(show_url, etree.HTMLParser())
show_title = try_get(episodes_page, '//div[@class="orangeheader"]/text()')
show_title = show_title[:show_title.index(": Playlists")]
episode_links = try_get(episodes_page, '//a[starts-with(@href, "/playlist/id/")]')
episodes = []
for episode_link in episode_links:
title = try_get(episode_link, 'text()', xform=get_episode_title)
url = try_get(episode_link, '@href')
number = int(url[len("/playlist/id/"):])
episodes.append(episode(title=title, number=number))
return render.list_episodes(params=Params(user=disco_user, preferences=prefs, message="", playlist_url=p_url, environment_message=get_environment_message()),
show=show(title=show_title, number=show_number),
episodes=episodes)
def scrape_songs(self, episode_number):
episode_url = "http://www.kpfa.org/playlist/id/%i" % int(episode_number)
playlist_page = etree.parse(episode_url, etree.HTMLParser())
table_rows = try_get(playlist_page, '//div[contains(@class, "field")]/table//tr')
header = table_rows[0]
table_rows = table_rows[1:]
artist_pos = try_get(header, 'count(td[contains(text(), "rtist")]/preceding-sibling::td)+1', xform=int)
song_pos = try_get(header, 'count(td[contains(text(), "ong")]/preceding-sibling::td)+1', xform=int)
album_pos = try_get(header, 'count(td[contains(text(), "lbum")]/preceding-sibling::td)+1', xform=int)
label_pos = try_get(header, 'count(td[contains(text(), "abel")]/preceding-sibling::td)+1', xform=int)
year_pos = try_get(header, 'count(td[contains(text(), "ear")]/preceding-sibling::td)+1', xform=int)
songs = []
for row in table_rows:
artist = try_get(row, 'td[%i]/text()' % artist_pos, xform=strip_blanks)
title = try_get(row, 'td[%i]/text()' % song_pos, xform=strip_blanks)
album = try_get(row, 'td[%i]/text()' % album_pos, xform=strip_blanks)
label = try_get(row, 'td[%i]/text()' % label_pos, xform=strip_blanks)
year = try_get(row, 'td[%i]/text()' % year_pos, xform=strip_blanks)
if artist and title and artist.lower() != "talk" and title.lower() != "talk":
artists = artist.split('/') if '/' in artist else [artist]
artists = map(strip_blanks, artists)
titles = title.split('/') if '/' in title else [title]
titles = map(strip_blanks, titles)
for artist in artists:
for title in titles:
if artist and title:
songs.append(ScrapedSong(artist=artist, album=album, song=title, label=label, year=year))
return songs
