From 55bc96ac0fc8e233dbc3bd558d8802391c5f1de0 Mon Sep 17 00:00:00 2001 From: Enes Date: Tue, 24 Apr 2018 22:48:40 +0300 Subject: [PATCH 1/4] [izlesene] fixing #16233 --- youtube_dl/extractor/izlesene.py | 31 +++++++++---------------------- 1 file changed, 9 insertions(+), 22 deletions(-) diff --git a/youtube_dl/extractor/izlesene.py b/youtube_dl/extractor/izlesene.py index b1d72177d..0449d7c1f 100644 --- a/youtube_dl/extractor/izlesene.py +++ b/youtube_dl/extractor/izlesene.py @@ -72,7 +72,7 @@ class IzleseneIE(InfoExtractor): 'uploadDate', webpage, 'upload date')) duration = float_or_none(self._html_search_regex( - r'"videoduration"\s*:\s*"([^"]+)"', + r'videoduration\s*=\s*\'([^\']+)\'', webpage, 'duration', fatal=False), scale=1000) view_count = str_to_int(get_element_by_id('videoViewCount', webpage)) @@ -80,29 +80,16 @@ class IzleseneIE(InfoExtractor): r'comment_count\s*=\s*\'([^\']+)\';', webpage, 'comment_count', fatal=False) - content_url = self._html_search_meta( - 'contentURL', webpage, 'content URL', fatal=False) - ext = determine_ext(content_url, 'mp4') - - # Might be empty for some videos. - streams = self._html_search_regex( - r'"qualitylevel"\s*:\s*"([^"]+)"', webpage, 'streams', default='') - + streams = self._parse_json(self._html_search_regex( + r'_videoObj\s*=\s*(.*);', webpage, 'streams', default=''), video_id) formats = [] - if streams: - for stream in streams.split('|'): - quality, url = re.search(r'\[(\w+)\](.+)', stream).groups() - formats.append({ - 'format_id': '%sp' % quality if quality else 'sd', - 'url': compat_urllib_parse_unquote(url), - 'ext': ext, - }) - else: - stream_url = self._search_regex( - r'"streamurl"\s*:\s*"([^"]+)"', webpage, 'stream URL') + for stream in streams.get('media').get('level'): + url = stream.get('source') + ext = determine_ext(url) + quality = stream.get('value') formats.append({ - 'format_id': 'sd', - 'url': compat_urllib_parse_unquote(stream_url), + 'format_id': '%sp' % quality if quality else 'sd', + 'url': compat_urllib_parse_unquote(url), 'ext': ext, }) From 6f571ca5fddde15d465fcb5aad400e83a559ae97 Mon Sep 17 00:00:00 2001 From: Enes Date: Mon, 30 Apr 2018 16:18:02 +0300 Subject: [PATCH 2/4] [izlesene] improvement --- youtube_dl/extractor/izlesene.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/izlesene.py b/youtube_dl/extractor/izlesene.py index 0449d7c1f..b32e9d35e 100644 --- a/youtube_dl/extractor/izlesene.py +++ b/youtube_dl/extractor/izlesene.py @@ -81,14 +81,14 @@ class IzleseneIE(InfoExtractor): webpage, 'comment_count', fatal=False) streams = self._parse_json(self._html_search_regex( - r'_videoObj\s*=\s*(.*);', webpage, 'streams', default=''), video_id) + r'_videoObj\s*=\s*(.*);', webpage, 'streams'), video_id) formats = [] for stream in streams.get('media').get('level'): url = stream.get('source') ext = determine_ext(url) quality = stream.get('value') formats.append({ - 'format_id': '%sp' % quality if quality else 'sd', + 'format_id': '%sp' % quality, 'url': compat_urllib_parse_unquote(url), 'ext': ext, }) From df544e7ee9abeb99b393336e124d65af6d95c30b Mon Sep 17 00:00:00 2001 From: Enes Date: Tue, 1 May 2018 15:36:59 +0300 Subject: [PATCH 3/4] [izlesene] checked code with flake8 and fixed --- youtube_dl/extractor/izlesene.py | 2 -- 1 file changed, 2 deletions(-) diff --git a/youtube_dl/extractor/izlesene.py b/youtube_dl/extractor/izlesene.py index b32e9d35e..f42bcd094 100644 --- a/youtube_dl/extractor/izlesene.py +++ b/youtube_dl/extractor/izlesene.py @@ -1,8 +1,6 @@ # coding: utf-8 from __future__ import unicode_literals -import re - from .common import InfoExtractor from ..compat import compat_urllib_parse_unquote from ..utils import ( From 482e106e1cfe656181cb96ab16acc8f0deed7a67 Mon Sep 17 00:00:00 2001 From: Enes Date: Thu, 17 May 2018 13:12:29 +0300 Subject: [PATCH 4/4] [izlesene] replaced .* to .+ in streams regex and added default value of video extension --- youtube_dl/extractor/izlesene.py | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/youtube_dl/extractor/izlesene.py b/youtube_dl/extractor/izlesene.py index f42bcd094..5b2095490 100644 --- a/youtube_dl/extractor/izlesene.py +++ b/youtube_dl/extractor/izlesene.py @@ -78,12 +78,14 @@ class IzleseneIE(InfoExtractor): r'comment_count\s*=\s*\'([^\']+)\';', webpage, 'comment_count', fatal=False) - streams = self._parse_json(self._html_search_regex( - r'_videoObj\s*=\s*(.*);', webpage, 'streams'), video_id) + streams_json = self._html_search_regex( + r'_videoObj\s*=\s*(.+);', webpage, 'streams') + streams = self._parse_json(streams_json, video_id) + formats = [] for stream in streams.get('media').get('level'): url = stream.get('source') - ext = determine_ext(url) + ext = determine_ext(url, 'mp4') quality = stream.get('value') formats.append({ 'format_id': '%sp' % quality,