From 2af3edf3cc110d10febd3b5ea45664e9996bbcde Mon Sep 17 00:00:00 2001 From: Breno Lipi Date: Thu, 11 Mar 2021 00:16:21 -0300 Subject: [PATCH 1/5] Adding to extractors --- youtube_dl/extractor/extractors.py | 1 + 1 file changed, 1 insertion(+) diff --git a/youtube_dl/extractor/extractors.py b/youtube_dl/extractor/extractors.py index 07a8af055..871767e81 100644 --- a/youtube_dl/extractor/extractors.py +++ b/youtube_dl/extractor/extractors.py @@ -170,6 +170,7 @@ from .carambatv import ( CarambaTVIE, CarambaTVPageIE, ) +from .captiongenerator import CaptionGeneratorIE from .cartoonnetwork import CartoonNetworkIE from .cbc import ( CBCIE, From 625a551f84f7a59a1abb4de1d1f50ea51b4e75f3 Mon Sep 17 00:00:00 2001 From: Breno Lipi Date: Thu, 11 Mar 2021 00:16:56 -0300 Subject: [PATCH 2/5] captiongenerator.com extractor --- youtube_dl/extractor/captiongenerator.py | 40 ++++++++++++++++++++++++ 1 file changed, 40 insertions(+) create mode 100644 youtube_dl/extractor/captiongenerator.py diff --git a/youtube_dl/extractor/captiongenerator.py b/youtube_dl/extractor/captiongenerator.py new file mode 100644 index 000000000..0d6ba0600 --- /dev/null +++ b/youtube_dl/extractor/captiongenerator.py @@ -0,0 +1,40 @@ +# coding: utf-8 +from __future__ import unicode_literals + +from .common import InfoExtractor + + +class CaptionGeneratorIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?captiongenerator\.com/(?P[0-9]+)' + _TEST = { + 'url': 'https://d34ov3vwfhhb30.cloudfront.net/Hitler+Reacts+-+No+Subtitles.mp4', + 'info_dict': { + 'id': '128', + 'ext': 'mp4', + 'title': 'Team building...', + } + } + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + video_url = self._html_search_regex(r'(https?://[a-z0-9]+\.cloudfront\.net/.+?\.mp4)', webpage, 'videoUrl') + print(video_url) + + vtt_page = self._download_webpage('https://www.captiongenerator.com/videos/%s.vtt' % video_id, video_id) + + vtt_captions = open("128.vtt", "w") + vtt_captions.write(vtt_page) + vtt_captions.close() + + title = self._html_search_regex(r'(.*)', webpage, 'videoTitle') + + return { + 'id': video_id, + 'title': title, + 'url': video_url, + 'description': self._og_search_description(webpage), + 'http_headers': {"Referer": "https://www.captiongenerator.com/"} + } From 15480a5083ef7024e55abb043139e025af56ce56 Mon Sep 17 00:00:00 2001 From: Breno Lipi Date: Thu, 11 Mar 2021 01:07:04 -0300 Subject: [PATCH 3/5] Non capture group, removed description, test value --- youtube_dl/extractor/captiongenerator.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/captiongenerator.py b/youtube_dl/extractor/captiongenerator.py index 0d6ba0600..1c7e1b133 100644 --- a/youtube_dl/extractor/captiongenerator.py +++ b/youtube_dl/extractor/captiongenerator.py @@ -5,13 +5,14 @@ from .common import InfoExtractor class CaptionGeneratorIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?captiongenerator\.com/(?P[0-9]+)' + _VALID_URL = r'(?:https?://(?:www\.)?captiongenerator\.com/(?P[0-9]+))' _TEST = { 'url': 'https://d34ov3vwfhhb30.cloudfront.net/Hitler+Reacts+-+No+Subtitles.mp4', 'info_dict': { 'id': '128', 'ext': 'mp4', 'title': 'Team building...', + 'http_headers': {"Referer": "https://www.captiongenerator.com/"} } } @@ -35,6 +36,5 @@ class CaptionGeneratorIE(InfoExtractor): 'id': video_id, 'title': title, 'url': video_url, - 'description': self._og_search_description(webpage), 'http_headers': {"Referer": "https://www.captiongenerator.com/"} } From 245ba7602330121cbf8ea280a1ae53db39808d3a Mon Sep 17 00:00:00 2001 From: Breno Lipi Date: Thu, 11 Mar 2021 01:12:53 -0300 Subject: [PATCH 4/5] [CaptionGenerator] Add new extractor --- youtube_dl/extractor/captiongenerator.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/youtube_dl/extractor/captiongenerator.py b/youtube_dl/extractor/captiongenerator.py index 1c7e1b133..ee9de335f 100644 --- a/youtube_dl/extractor/captiongenerator.py +++ b/youtube_dl/extractor/captiongenerator.py @@ -12,7 +12,7 @@ class CaptionGeneratorIE(InfoExtractor): 'id': '128', 'ext': 'mp4', 'title': 'Team building...', - 'http_headers': {"Referer": "https://www.captiongenerator.com/"} + 'http_headers': 'Referer: https://www.captiongenerator.com/' } } From edb1ec29858255f29da9aad12e143e59dc68193c Mon Sep 17 00:00:00 2001 From: Breno Lipi Date: Thu, 11 Mar 2021 01:53:55 -0300 Subject: [PATCH 5/5] Added more tests --- youtube_dl/extractor/captiongenerator.py | 34 ++++++++++++++++++++---- 1 file changed, 29 insertions(+), 5 deletions(-) diff --git a/youtube_dl/extractor/captiongenerator.py b/youtube_dl/extractor/captiongenerator.py index ee9de335f..5e5f3f107 100644 --- a/youtube_dl/extractor/captiongenerator.py +++ b/youtube_dl/extractor/captiongenerator.py @@ -6,15 +6,39 @@ from .common import InfoExtractor class CaptionGeneratorIE(InfoExtractor): _VALID_URL = r'(?:https?://(?:www\.)?captiongenerator\.com/(?P[0-9]+))' - _TEST = { - 'url': 'https://d34ov3vwfhhb30.cloudfront.net/Hitler+Reacts+-+No+Subtitles.mp4', + _TESTS = [{ + 'url': 'https://www.captiongenerator.com/128/Team-building', 'info_dict': { 'id': '128', 'ext': 'mp4', 'title': 'Team building...', - 'http_headers': 'Referer: https://www.captiongenerator.com/' - } - } + 'http_headers': {'Referer': 'https://www.captiongenerator.com/'} + }, + }, { + 'url': 'https://www.captiongenerator.com/2153287/Man-finds-phone', + 'info_dict': { + 'id': '2153287', + 'ext': 'mp4', + 'title': 'Man finds phone', + 'http_headers': {'Referer': 'https://www.captiongenerator.com/'} + }, + }, { + 'url': 'https://www.captiongenerator.com/2140517/Haavisto-Pietarissa', + 'info_dict': { + 'id': '2140517', + 'ext': 'mp4', + 'title': 'Haavisto Pietarissa', + 'http_headers': {'Referer': 'https://www.captiongenerator.com/'} + }, + }, { + 'url': 'https://www.captiongenerator.com/2153147/81-Pay', + 'info_dict': { + 'id': '2153147', + 'ext': 'mp4', + 'title': '81 Pay', + 'http_headers': {'Referer': 'https://www.captiongenerator.com/'} + }, + }] def _real_extract(self, url): video_id = self._match_id(url)