def _parse_show(self, link):
LOG.debug("Fetching show: %s" % link)
event_doc = html_util.fetch_and_parse(link)
event_detail = event_doc.get_element_by_id("mainColumn")
show = Show()
for performer in html_util.get_elements(event_detail, 'h1'):
name = performer.text_content().strip(' \n\r\t')
if name:
show.performers.append(Performer(name))
date_txt = html_util.get_first_element(event_detail, '.date').text_content()
event_match = self.EVENT_URL.match(link)
show.merge_key = event_match.group('page_id')
show.venue = self.venue()
show.show_time = date_util.parse_date_time(date_txt).replace(hour = 21)
LOG.debug('Date: %s' % show.date)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail)
for img_tag in event_detail.iter(tag = 'img'):
if 'main' in img_tag.get('src'):
show.resources.image_url = img_tag.get('src')
break
return show
def _trans_show(self, show_data):
LOG.debug("Checking event: %s" % show_data['EventName'])
if "Music" not in show_data['MajorGenre']:
LOG.debug("Skipping non music show")
return None
elif show_data.get('Canceled'):
LOG.debug("Skipping cancelled show")
return None
elif 'VIP Packages' in show_data['EventName']:
LOG.debug("Skipping VIP package")
return None
show = Show()
performers = []
for i, name in enumerate(lang_util.parse_performers(show_data['EventName'])):
performers.append(Performer(name, headliner = i == 0))
show.merge_key = show_data['EventId']
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_date_time(show_data['EventDate'])
#if show.show_time:
# show.show_time = timezone(show_data['Timezone']).localize(show.show_time)
if show_data['AttractionImage']:
show.resources.image_url = self._image_url(show_data, show_data['AttractionImage'][0])
return show
def _parse_show(self, link):
event_doc = html_util.fetch_and_parse(link, parse_500 = True)
event_detail = html_util.get_first_element(event_doc, ".event-detail")
artist_info = html_util.get_first_element(event_doc, ".artist-boxes")
date_txt = html_util.get_first_element(event_detail, ".dates").text_content()
performers = []
for el in html_util.get_elements(event_doc, '.headliners'):
for name in lang_util.parse_performers(el.text_content()):
performers.append(Performer(name, headliner = True))
for el in html_util.get_elements(event_doc, '.supports'):
for name in lang_util.parse_performers(el.text_content()):
performers.append(Performer(name, headliner = False))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_date_and_time(date_txt, html_util.get_first_element(event_detail, ".times").text_content())
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail, artist_info)
img = html_util.get_first_element(event_detail, "img", optional = True)
if img is not None:
show.resources.image_url = img.get('src')
return show
def _parse_show(self, link):
logging.debug('Parsing show from: %s' % link)
event_doc = html_util.fetch_and_parse(link)
event = html_util.get_first_element(event_doc, '.biglisting')
img = html_util.get_first_element(event, '.tonightinfo img', optional = True)
date_el = html_util.get_first_element(event, '.date')
date_match = self.DATE_RE.search(date_el.text_content())
if date_match:
date_txt = date_match.group(0)
else:
raise Exception('Unable to determine show date from: %s' % date_el.text_content())
performers = []
first_time = None
for det in event.cssselect('.showpage-details'):
header = None
for child in det.getchildren():
if child.tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6'):
header = child
if header is None:
logger.error('Unable to determine performer')
else:
time_txt = html_util.get_first_element(det, '.time').text_content()
time_match = date_util.TIME_RE.search(time_txt)
if time_match:
first_time = time_txt = time_match.group('time')
else:
time_txt = None
performers.append(Performer(header.text_content(), start_time = time_txt, headliner = header.tag in ('h1')))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_date_and_time(date_txt, first_time)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event)
if img is not None:
show.resources.image_url = img.get('src')
return show
def _parse_show(self, link):
event_doc = html_util.fetch_and_parse(link)
event_detail = event_doc.get_element_by_id("eventDetail")
title_txt = []
found_h_el = False
# Start parsing when we find the first h* element
# Stop parsing if we found an h* element, but then encounter anything else
for el in event_detail.getchildren():
if el.tag in ("h1", "h2"):
found_h_el = True
if el.text_content():
title_txt.append(el.text_content())
elif found_h_el:
break
"""
<span id="timeDetail">
Apr 24, 2010<br />
upstairs<br />
Doors @ 7 PM<br/>
$15.00 Adv. / $20 at the Door<br />
<a href="http://www.deadcellentertainment.tix.com/Schedule.asp?OrganizationNumber=2690" target="_blank">
<img src="/images/buyticket.png" alt="Purchase Tickets" />
</a>
</span>
"""
time_el = event_detail.get_element_by_id("timeDetail")
date_txt = time_el.text
time_txt = time_el.text_content()
performers = []
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = [Performer(p) for p in lang_util.parse_performers("/".join(title_txt))]
show.door_time = date_util.parse_show_time(date_txt, time_txt)
show.show_time = date_util.parse_door_time(date_txt, time_txt)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail)
img = html_util.get_first_element(event_detail, "img")
if img is not None:
show.resources.image_url = img.get("src")
return show
def _parse_show(self, link):
raw_url = self.raw_url(link)
match = self.EVENT_ID.search(link)
if not match:
raise Exception("Unable to locate event id in: %s" % link)
event_id = match.group(0)
logging.debug('Fetching show info: %s' % link)
event_doc = html_util.fetch_and_parse(link)
show_el = html_util.get_first_element(event_doc, '#content')
header_el = html_util.get_first_element(show_el, 'h1')
header_match = self.HEADER_PARSE.search(header_el.text_content())
if not header_match:
raise Exception("Unable to parse header: %s" % header_el.text_content())
date_txt = header_match.group('date').strip()
title = header_match.group('title').strip()
if date_txt.lower().startswith('tonight'):
date_txt = datetime.today().date().strftime('%F')
img = html_util.get_first_element(show_el, 'img', optional = True)
show = Show()
show.performers = [Performer(p) for p in lang_util.parse_performers(title)]
show.show_time = date_util.parse_date_and_time(date_txt, None)
show.merge_key = event_id
show.venue = self.venue()
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(show_el)
if img is not None:
show.resources.image_url = img.get('src')
return show
def _parse_show(self, api, event):
LOG.debug("Parsing event: %s" % event["id"])
show = Show()
show.merge_key = event["id"]
show.venue = self.venue()
show.performers = [Performer(p) for p in lang_util.parse_performers(event["name"])]
show.show_time = date_util.parse_date_time(event["start_time"])
html_doc = u"<html><body>%s</body></html>" % cgi.escape(event.get("description", ""))
doc = lxml.html.document_fromstring(html_doc)
show.resources.show_url = self.EVENT_URL % event["id"]
show.resources.image_url = self.PICTURE_URL % event["id"]
show.resources.resource_uris = self.resource_extractor.extract_resources(doc)
return show
def _parse_show(self, url, section):
doc = html_util.fetch_and_parse(url)
show_el = html_util.get_first_element(doc, '#detailPage')
date_txt = html_util.get_first_element(show_el, 'time.dtstart').get('datetime')
title = html_util.get_first_element(section, 'h4').text_content()
show = Show()
show.merge_key = url
show.venue = self.venue()
show.performers = [Performer(p) for p in lang_util.parse_performers(title)]
show.show_time = date_util.parse_date_time(date_txt)
show.resources.show_url = url
show.resources.resource_uris = self.resource_extractor.extract_resources(section, show_el)
return show
def _parse_show(self, show_url, event_detail, today):
show = Show()
# Union hall will have duplicate instances of #unionhall_performer
# some may or may not have links, but those that do have links are tagged
# with the same id again ie: <div id="unionhall_performer"><a href="#" id="#unionhall_performer"> ...
performers = [Performer(p.text_content()) for p in event_detail.cssselect("#unionhall_performer") if p.tag != 'a']
performers[0].headliner = True
ticket_link = html_util.get_first_element(event_detail, '#ticket_link a', optional = True)
show.venue = self.venue()
show.performers = performers
if ticket_link is not None:
show.merge_key = ticket_link.get('href')
# Format: THU 3/25: 6pm / $15
date_tag = event_detail.get_element_by_id("unionhall_date")
date_match = self.DATE_RE.match(date_tag.text_content())
time_match = self.TIME_RE.search(date_tag.text_content())
if date_match and time_match:
month, day = (int(d) for d in (date_match.group('month'), date_match.group('day')))
show_date = datetime.now().replace(month = month, day = day)
show.show_time = date_util.parse_date_and_time(show_date.strftime('%F'), time_match.group('time'))
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail)
for img_tag in event_detail.iter(tag = 'img'):
show.resources.image_url = img_tag.get('src')
break
date_util.adjust_fuzzy_years(show, self._parse_started)
return show
def _parse_show(self, link):
event_doc = html_util.fetch_and_parse(link)
event = html_util.get_first_element(event_doc, '#tfly-center-column-wide')
event_detail = html_util.get_first_element(event, '#details')
"""
<div class="info">
Sat, May 22, 2010<br />
Doors: 6:00 PM / Show: 7:00 PM <br />
$5.00<br />
</div>
"""
event_info = html_util.get_first_element(event_detail, ".info")
date_txt = event_info.text
time_txt = event_info.getchildren()[0].tail
img = html_util.get_first_element(event_detail, "img")
performers = []
for tag in ('h1', 'h2', 'h3', 'h4'):
for h in event_detail.iter(tag = tag):
performers.extend(self._parse_performers(h))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_show_time(date_txt, time_txt)
show.door_time = date_util.parse_door_time(date_txt, time_txt)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event)
if img is not None:
show.resources.image_url = img.get('src')
return show
def _parse_show(self, link):
event_doc = html_util.fetch_and_parse(link)
event_detail = event_doc.get_element_by_id("detail")
show = Show()
strong_iter = event_detail.iter(tag = 'strong')
date_tag, title_tag, blank_tag, desc_tag = strong_iter.next(), strong_iter.next(), strong_iter.next(), strong_iter.next()
date_txt = date_tag.text_content()
if desc_tag.getnext().tail:
time_match = self.TIME_RE.search(desc_tag.getnext().tail)
else:
time_match = None
if time_match:
time_txt = time_match.group('time')
else:
time_txt = None
show.merge_key = link
show.venue = self.venue()
show.title = title_tag.text_content()
show.show_time = date_util.parse_date_and_time(date_txt, time_txt)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail)
for img_tag in event_detail.iter(tag = 'img'):
if 'main' in img_tag.get('src'):
show.resources.image_url = img_tag.get('src')
break
return show
def _trans_show(self, event):
LOG.debug("Transforming show: %s" % event.get_title())
show = Show()
performers = []
artists = event.get_artists()
for i, artist in enumerate(artists):
performers.append(Performer(artist.get_name(), headliner = i == 0))
if artist.get_cover_image(size = pylast.COVER_MEGA):
show.resources.image_url = artist.get_cover_image(size = pylast.COVER_MEGA)
show.merge_key = event.get_id()
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_date_time(event.get_start_date())
show.resources.show_url = event.get_url()
return show
def _parse_show(self, link, show_section):
show_doc = html_util.fetch_and_parse(link)
show_detail = html_util.get_first_element(show_doc, "#content .event-detail")
date_txt = html_util.get_first_element(show_detail, ".dates").text_content()
time_txt = html_util.get_first_element(show_detail, ".times").text_content()
sold_out = html_util.get_first_element(show_detail, '.sold-out', optional = True)
image = html_util.get_first_element(show_detail, 'img', optional = True)
# The image we want is generally the first one, but if the layout changes this may break
if image is not None:
image_url = image.get('src')
else:
image_url = None
performers = []
for tag in ('h1', 'h2', 'h3'):
for p in show_detail.iter(tag = tag):
if p.text_content():
performers.extend(self._parse_performers(p))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.door_time = date_util.parse_door_time(date_txt, time_txt)
show.show_time = date_util.parse_show_time(date_txt, time_txt)
show.soldout = sold_out is not None
show.resources.show_url = link
show.resources.image_url = image_url
show.resources.resource_uris = self.resource_extractor.extract_resources(show_detail)
return show
def _parse_show(self, link, show_section):
show_doc = html_util.fetch_and_parse(link)
show_detail = show_doc.get_element_by_id("content")
title = html_util.get_first_element(show_detail, '.title').text
date_txt = html_util.get_first_element(show_section, '.date').text
image_url = html_util.get_first_element(show_detail, '.left-view-header img').get('src')
performers = []
performer_detail = html_util.get_first_element(show_detail, '.performers')
performer_urls = []
for anchor in performer_detail.iter(tag = 'a'):
performers.extend(self._parse_performers(anchor))
if self.IS_ARTIST_URL_RE.match(anchor.get('href')):
performer_urls.append(anchor.get('href'))
resource_sections = [show_section, show_detail]
for url in performer_urls:
resource_sections.extend(self.fetch_performer_content(url))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_date_time(date_txt)
show.resources.show_url = link
show.resources.image_url = image_url
show.resources.resource_uris = self.resource_extractor.extract_resources(*resource_sections)
return show
def _trans_record(self, record):
show = Show()
show.venue = Venue(record.get('venue-name'), record.get('venue-url'))
show.title = record.get('title')
show.merge_key = record.get('merge-key')
performers = []
if record.get('performers'):
for performer in record['performers'].split(','):
performers.append(Performer(performer.strip()))
if record.get('tags'):
show.tags = [t.strip() for t in record['tags'].split(',')]
date_txt = record.get('show-date')
if not date_txt:
raise Exception('Show Date is required')
else:
show.date = date_util.parse_date_time(date_txt)
if performers:
show.performers = performers
if record.get('show-time'):
show.show_time = date_util.parse_date_and_time(date_txt, record.get('show-time'))
if record.get('door-time'):
show.show_time = date_util.parse_date_and_time(date_txt, record.get('door-time'))
show.resources.show_url = record.get('show-url')
show.resources.image_url = record.get('image-url')
show.resources.resource_uris = self.resource_extractor.extract_resources(self._create_resource_doc(record))
return show
def _parse_show(self, link):
event_doc = html_util.fetch_and_parse(link)
match = self.IS_EVENT.match(link)
event_id = int(match.group("event_id"))
event_detail = html_util.get_first_element(event_doc, ".tfly-event-id-%d" % event_id)
date_txt = html_util.get_first_element(event_doc, ".dates").text_content()
time_txt = html_util.get_first_element(event_doc, ".times").text_content()
img = html_util.get_first_element(event_detail, "img")
performers = []
for p in html_util.get_elements(event_detail, ".headliners"):
performers.append(Performer(p.text_content(), headliner=True))
for p in html_util.get_elements(event_detail, ".supports"):
for pi in lang_util.parse_performers(p.text_content()):
performers.append(Performer(pi, headliner=False))
show = Show()
show.merge_key = link
show.venue = self.venue()
show.performers = performers
show.show_time = date_util.parse_show_time(date_txt, time_txt)
show.door_time = date_util.parse_door_time(date_txt, time_txt)
show.resources.show_url = link
show.resources.resource_uris = self.resource_extractor.extract_resources(event_detail)
if img is not None:
show.resources.image_url = img.get("src")
return show
