Akkariin/youtube-dl
1
0
Fork 0
mirror of https://github.com/l1ving/youtube-dl synced 2025-03-11 05:27:14 +08:00
Code Issues Releases Wiki Activity

[ciscolive] Exclude results without valid video URL

Browse Source
This commit is contained in:
Austin de Coup-Crank 2018-10-28 14:03:36 -07:00
parent 70f767da43
commit fe9791254e
1 changed files with 33 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
youtube_dl/extractor/ciscolive.py
View File

@ -96,17 +96,31 @@ class CiscoLiveIE(InfoExtractor):
"creator": rf_result.get("sess_pres_name"), "creator": rf_result.get("sess_pres_name"),
"description": rf_result.get("sess_desc"), "description": rf_result.get("sess_desc"),
"series": rf_result.get("event_name"), "series": rf_result.get("event_name"),
"duration": rf_result["sess_duration"], "duration": rf_result.get("sess_duration"),
"location": rf_result["sess_location"] "location": rf_result.get("sess_location")
} }
self.to_screen("Session: %s [%s]" % (rf_result["sess_title"], rf_result["sess_abbr"])) self.to_screen("Session: %s [%s]" % (rf_result["sess_title"], rf_result["sess_abbr"]))
self.to_screen("Presenter: %s, %s" % (rf_result["sess_pres_name"], rf_result["sess_pres_title"])) self.to_screen("Presenter: %s, %s" % (rf_result["sess_pres_name"], rf_result["sess_pres_title"]))
self.to_screen("Presentation PDF: %s" % rf_result["sess_pdf_url"]) self.to_screen("Presentation PDF: %s" % rf_result["sess_pdf_url"])
return self.url_result( return self.url_result(smuggle_url(rf_result["sess_bc_url"], metadata),
smuggle_url(rf_result["sess_bc_url"], metadata),
'BrightcoveNew', rf_result["sess_bc_id"], 'BrightcoveNew', rf_result["sess_bc_id"],
rf_result["sess_title"]) rf_result["sess_title"])
def _check_bc_url_exists(self, rf_item):
""" Checks for the existence of a Brightcove URL
"""
msg = "Skipping session that does not include a valid video URL: %s" % rf_item.get("title", "Unknown title")
try:
bc_id = rf_item["videos"][0]["url"]
mobj = re.match(r'\d+', bc_id)
if mobj:
return rf_item
else:
self.report_warning(msg)
except IndexError:
self.report_warning(msg)
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
rf_api_headers = { rf_api_headers = {
@ -128,19 +142,20 @@ class CiscoLiveIE(InfoExtractor):
rf_api_args['data'] = compat_urllib_parse_urlencode({'id': rf_id}) rf_api_args['data'] = compat_urllib_parse_urlencode({'id': rf_id})
self.to_screen('Video for session ID %s' % rf_id) self.to_screen('Video for session ID %s' % rf_id)
rf_api_result = self._download_json(**rf_api_args) rf_api_result = self._download_json(**rf_api_args)
rf_item = rf_api_result['items'][0] rf_item = self._check_bc_url_exists(rf_api_result['items'][0])
return self._parse_rf_item(rf_item) return self._parse_rf_item(rf_item)
else: else:
# Filter query URL (multiple videos) # Filter query URL (multiple videos)
if mobj.group('query'): if mobj.group('query'):
rf_query = mobj.group('query') rf_query = str(rf_query + '&type=session&size=1000')
rf_query = str(rf_query + "&type=session&size=1000")
data = rf_query
rf_api_args['url_or_request'] = self.RAINFOCUS_API_URL % 'search' rf_api_args['url_or_request'] = self.RAINFOCUS_API_URL % 'search'
rf_api_args['data'] = data rf_api_args['data'] = rf_query
# Query JSON results offer no obvious way to ID the search # Query JSON results offer no obvious way to ID the search
rf_api_args['video_id'] = "Filter query" rf_api_args['video_id'] = 'Filter query'
self.to_screen('Video collection for query %s' % rf_query) self.to_screen('Video collection for filter query "%s"' % rf_query)
rf_api_result = self._download_json(**rf_api_args) rf_api_results = self._download_json(**rf_api_args)
entries = [self._parse_rf_item(rf_item) for rf_item in rf_api_result['sectionList'][0]['items']] # Not all sessions have videos; filter them out before moving on
rf_video_results = [rf_item for rf_item in rf_api_results["sectionList"][0]["items"]
if self._check_bc_url_exists(rf_item)]
entries = [self._parse_rf_item(rf_item) for rf_item in rf_video_results]
return self.playlist_result(entries) return self.playlist_result(entries)