[yt-dlp.git] / yt_dlp / extractor / nekohacker.py

import re

from .common import InfoExtractor
from ..utils import (
    ExtractorError,
    determine_ext,
    extract_attributes,
    get_element_by_class,
    get_element_text_and_html_by_tag,
    parse_duration,
    traverse_obj,
    try_call,
    url_or_none,
)


class NekoHackerIE(InfoExtractor):
    _VALID_URL = r'https?://(?:www\.)?nekohacker\.com/(?P<id>(?!free-dl)[\w-]+)'
    _TESTS = [{
        'url': 'https://nekohacker.com/nekoverse/',
        'info_dict': {
            'id': 'nekoverse',
            'title': 'Nekoverse',
        },
        'playlist': [
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2022/11/01-Spaceship.mp3',
                'md5': '44223701ebedba0467ebda4cc07fb3aa',
                'info_dict': {
                    'id': '1712',
                    'ext': 'mp3',
                    'title': 'Spaceship',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20221101',
                    'album': 'Nekoverse',
                    'artist': 'Neko Hacker',
                    'track': 'Spaceship',
                    'track_number': 1,
                    'duration': 195.0
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2022/11/02-City-Runner.mp3',
                'md5': '8f853c71719389d32bbbd3f1a87b3f08',
                'info_dict': {
                    'id': '1713',
                    'ext': 'mp3',
                    'title': 'City Runner',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20221101',
                    'album': 'Nekoverse',
                    'artist': 'Neko Hacker',
                    'track': 'City Runner',
                    'track_number': 2,
                    'duration': 148.0
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2022/11/03-Nature-Talk.mp3',
                'md5': '5a8a8ae852720cee4c0ac95c7d1a7450',
                'info_dict': {
                    'id': '1714',
                    'ext': 'mp3',
                    'title': 'Nature Talk',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20221101',
                    'album': 'Nekoverse',
                    'artist': 'Neko Hacker',
                    'track': 'Nature Talk',
                    'track_number': 3,
                    'duration': 174.0
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2022/11/04-Crystal-World.mp3',
                'md5': 'd8e59a48061764e50d92386a294abd50',
                'info_dict': {
                    'id': '1715',
                    'ext': 'mp3',
                    'title': 'Crystal World',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20221101',
                    'album': 'Nekoverse',
                    'artist': 'Neko Hacker',
                    'track': 'Crystal World',
                    'track_number': 4,
                    'duration': 199.0
                }
            }
        ]
    }, {
        'url': 'https://nekohacker.com/susume/',
        'info_dict': {
            'id': 'susume',
            'title': '進め！むじなカンパニー',
        },
        'playlist': [
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-feat.-六科なじむ-CV_-日高里菜-割戶真友-CV_-金元寿子-軽井沢ユキ-CV_-上坂すみれ-出稼ぎガルシア-CV_-金子彩花-.mp3',
                'md5': 'fb13f008aa81f26ba48f91fd2d6186ce',
                'info_dict': {
                    'id': '711',
                    'ext': 'mp3',
                    'title': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20210115',
                    'album': '進め！むじなカンパニー',
                    'artist': 'Neko Hacker',
                    'track': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
                    'track_number': 1,
                    'duration': None
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-feat.-六科なじむ-CV_-日高里菜-.mp3',
                'md5': '028803f70241df512b7764e73396fdd1',
                'info_dict': {
                    'id': '709',
                    'ext': 'mp3',
                    'title': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20210115',
                    'album': '進め！むじなカンパニー',
                    'artist': 'Neko Hacker',
                    'track': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
                    'track_number': 2,
                    'duration': None
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-instrumental.mp3',
                'md5': 'adde9e9a16e1da5e602b579c247d0fb9',
                'info_dict': {
                    'id': '710',
                    'ext': 'mp3',
                    'title': '進め！むじなカンパニー (instrumental)',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20210115',
                    'album': '進め！むじなカンパニー',
                    'artist': 'Neko Hacker',
                    'track': '進め！むじなカンパニー (instrumental)',
                    'track_number': 3,
                    'duration': None
                }
            },
            {
                'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-instrumental.mp3',
                'md5': 'ebb0443039cf5f9ff7fd557ed9b23599',
                'info_dict': {
                    'id': '712',
                    'ext': 'mp3',
                    'title': 'むじな de なじむ (instrumental)',
                    'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
                    'vcodec': 'none',
                    'acodec': 'mp3',
                    'release_date': '20210115',
                    'album': '進め！むじなカンパニー',
                    'artist': 'Neko Hacker',
                    'track': 'むじな de なじむ (instrumental)',
                    'track_number': 4,
                    'duration': None
                }
            }
        ]
    }]

    def _real_extract(self, url):
        playlist_id = self._match_id(url)

        webpage = self._download_webpage(url, playlist_id)
        playlist = get_element_by_class('playlist', webpage)

        if not playlist:
            iframe = try_call(lambda: get_element_text_and_html_by_tag('iframe', webpage)[1]) or ''
            iframe_src = url_or_none(extract_attributes(iframe).get('src'))
            if not iframe_src:
                raise ExtractorError('No playlist or embed found in webpage')
            elif re.match(r'https?://(?:\w+\.)?spotify\.com/', iframe_src):
                raise ExtractorError('Spotify embeds are not supported', expected=True)
            return self.url_result(url, 'Generic')

        entries = []
        for track_number, track in enumerate(re.findall(r'(<li[^>]+data-audiopath[^>]+>)', playlist), 1):
            entry = traverse_obj(extract_attributes(track), {
                'url': ('data-audiopath', {url_or_none}),
                'ext': ('data-audiopath', {determine_ext}),
                'id': 'data-trackid',
                'title': 'data-tracktitle',
                'track': 'data-tracktitle',
                'album': 'data-albumtitle',
                'duration': ('data-tracktime', {parse_duration}),
                'release_date': ('data-releasedate', {lambda x: re.match(r'\d{8}', x.replace('.', ''))}, 0),
                'thumbnail': ('data-albumart', {url_or_none}),
            })
            entries.append({
                **entry,
                'track_number': track_number,
                'artist': 'Neko Hacker',
                'vcodec': 'none',
                'acodec': 'mp3' if entry['ext'] == 'mp3' else None,
            })

        return self.playlist_result(entries, playlist_id, traverse_obj(entries, (0, 'album')))
Commit	Line	Data
489f5127	1	import re
	2
	3	from .common import InfoExtractor
	4	from ..utils import (
	5	ExtractorError,
	6	determine_ext,
	7	extract_attributes,
	8	get_element_by_class,
	9	get_element_text_and_html_by_tag,
	10	parse_duration,
	11	traverse_obj,
	12	try_call,
	13	url_or_none,
	14	)
	15
	16
	17	class NekoHackerIE(InfoExtractor):
	18	_VALID_URL = r'https?://(?:www\.)?nekohacker\.com/(?P<id>(?!free-dl)[\w-]+)'
	19	_TESTS = [{
	20	'url': 'https://nekohacker.com/nekoverse/',
	21	'info_dict': {
	22	'id': 'nekoverse',
	23	'title': 'Nekoverse',
	24	},
	25	'playlist': [
	26	{
	27	'url': 'https://nekohacker.com/wp-content/uploads/2022/11/01-Spaceship.mp3',
	28	'md5': '44223701ebedba0467ebda4cc07fb3aa',
	29	'info_dict': {
	30	'id': '1712',
	31	'ext': 'mp3',
	32	'title': 'Spaceship',
	33	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
	34	'vcodec': 'none',
	35	'acodec': 'mp3',
	36	'release_date': '20221101',
	37	'album': 'Nekoverse',
	38	'artist': 'Neko Hacker',
	39	'track': 'Spaceship',
	40	'track_number': 1,
	41	'duration': 195.0
	42	}
	43	},
	44	{
	45	'url': 'https://nekohacker.com/wp-content/uploads/2022/11/02-City-Runner.mp3',
	46	'md5': '8f853c71719389d32bbbd3f1a87b3f08',
	47	'info_dict': {
	48	'id': '1713',
	49	'ext': 'mp3',
	50	'title': 'City Runner',
	51	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
	52	'vcodec': 'none',
	53	'acodec': 'mp3',
	54	'release_date': '20221101',
	55	'album': 'Nekoverse',
	56	'artist': 'Neko Hacker',
	57	'track': 'City Runner',
	58	'track_number': 2,
	59	'duration': 148.0
	60	}
	61	},
	62	{
	63	'url': 'https://nekohacker.com/wp-content/uploads/2022/11/03-Nature-Talk.mp3',
	64	'md5': '5a8a8ae852720cee4c0ac95c7d1a7450',
65	'info_dict': {
66	'id': '1714',
67	'ext': 'mp3',
68	'title': 'Nature Talk',
69	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
70	'vcodec': 'none',
71	'acodec': 'mp3',
72	'release_date': '20221101',
73	'album': 'Nekoverse',
74	'artist': 'Neko Hacker',
75	'track': 'Nature Talk',
76	'track_number': 3,
77	'duration': 174.0
78	}
79	},
80	{
81	'url': 'https://nekohacker.com/wp-content/uploads/2022/11/04-Crystal-World.mp3',
82	'md5': 'd8e59a48061764e50d92386a294abd50',
83	'info_dict': {
84	'id': '1715',
85	'ext': 'mp3',
86	'title': 'Crystal World',
87	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2022/11/Nekoverse_Artwork-1024x1024.jpg',
88	'vcodec': 'none',
89	'acodec': 'mp3',
90	'release_date': '20221101',
91	'album': 'Nekoverse',
92	'artist': 'Neko Hacker',
93	'track': 'Crystal World',
94	'track_number': 4,
95	'duration': 199.0
96	}
97	}
98	]
99	}, {
100	'url': 'https://nekohacker.com/susume/',
101	'info_dict': {
102	'id': 'susume',
103	'title': '進め！むじなカンパニー',
104	},
105	'playlist': [
106	{
107	'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-feat.-六科なじむ-CV_-日高里菜-割戶真友-CV_-金元寿子-軽井沢ユキ-CV_-上坂すみれ-出稼ぎガルシア-CV_-金子彩花-.mp3',
108	'md5': 'fb13f008aa81f26ba48f91fd2d6186ce',
109	'info_dict': {
110	'id': '711',
111	'ext': 'mp3',
112	'title': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
113	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
114	'vcodec': 'none',
115	'acodec': 'mp3',
116	'release_date': '20210115',
117	'album': '進め！むじなカンパニー',
118	'artist': 'Neko Hacker',
119	'track': 'md5:1a5fcbc96ca3c3265b1c6f9f79f30fd0',
120	'track_number': 1,
121	'duration': None
122	}
123	},
124	{
125	'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-feat.-六科なじむ-CV_-日高里菜-.mp3',
126	'md5': '028803f70241df512b7764e73396fdd1',
127	'info_dict': {
128	'id': '709',
129	'ext': 'mp3',
130	'title': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
131	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
132	'vcodec': 'none',
133	'acodec': 'mp3',
134	'release_date': '20210115',
135	'album': '進め！むじなカンパニー',
136	'artist': 'Neko Hacker',
137	'track': 'むじな de なじむ feat. 六科なじむ (CV: 日高里菜 )',
138	'track_number': 2,
139	'duration': None
140	}
141	},
142	{
143	'url': 'https://nekohacker.com/wp-content/uploads/2021/01/進め！むじなカンパニー-instrumental.mp3',
144	'md5': 'adde9e9a16e1da5e602b579c247d0fb9',
145	'info_dict': {
146	'id': '710',
147	'ext': 'mp3',
148	'title': '進め！むじなカンパニー (instrumental)',
149	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
150	'vcodec': 'none',
151	'acodec': 'mp3',
152	'release_date': '20210115',
153	'album': '進め！むじなカンパニー',
154	'artist': 'Neko Hacker',
155	'track': '進め！むじなカンパニー (instrumental)',
156	'track_number': 3,
157	'duration': None
158	}
159	},
160	{
161	'url': 'https://nekohacker.com/wp-content/uploads/2021/01/むじな-de-なじむ-instrumental.mp3',
162	'md5': 'ebb0443039cf5f9ff7fd557ed9b23599',
163	'info_dict': {
164	'id': '712',
165	'ext': 'mp3',
166	'title': 'むじな de なじむ (instrumental)',
167	'thumbnail': 'https://nekohacker.com/wp-content/uploads/2021/01/OP表-1024x1024.png',
168	'vcodec': 'none',
169	'acodec': 'mp3',
170	'release_date': '20210115',
171	'album': '進め！むじなカンパニー',
172	'artist': 'Neko Hacker',
173	'track': 'むじな de なじむ (instrumental)',
174	'track_number': 4,
175	'duration': None
176	}
177	}
178	]
179	}]
180
181	def _real_extract(self, url):
182	playlist_id = self._match_id(url)
183
184	webpage = self._download_webpage(url, playlist_id)
185	playlist = get_element_by_class('playlist', webpage)
186
187	if not playlist:
188	iframe = try_call(lambda: get_element_text_and_html_by_tag('iframe', webpage)[1]) or ''
189	iframe_src = url_or_none(extract_attributes(iframe).get('src'))
190	if not iframe_src:
191	raise ExtractorError('No playlist or embed found in webpage')
192	elif re.match(r'https?://(?:\w+\.)?spotify\.com/', iframe_src):
193	raise ExtractorError('Spotify embeds are not supported', expected=True)
194	return self.url_result(url, 'Generic')
195
196	entries = []
197	for track_number, track in enumerate(re.findall(r'(<li[^>]+data-audiopath[^>]+>)', playlist), 1):
198	entry = traverse_obj(extract_attributes(track), {
199	'url': ('data-audiopath', {url_or_none}),
200	'ext': ('data-audiopath', {determine_ext}),
201	'id': 'data-trackid',
202	'title': 'data-tracktitle',
203	'track': 'data-tracktitle',
204	'album': 'data-albumtitle',
205	'duration': ('data-tracktime', {parse_duration}),
206	'release_date': ('data-releasedate', {lambda x: re.match(r'\d{8}', x.replace('.', ''))}, 0),
207	'thumbnail': ('data-albumart', {url_or_none}),
208	})
209	entries.append({
210	**entry,
211	'track_number': track_number,
212	'artist': 'Neko Hacker',
213	'vcodec': 'none',
214	'acodec': 'mp3' if entry['ext'] == 'mp3' else None,
215	})
216
217	return self.playlist_result(entries, playlist_id, traverse_obj(entries, (0, 'album')))