hypem.py 2.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869
  1. from __future__ import unicode_literals
  2. import json
  3. import re
  4. import time
  5. from .common import InfoExtractor
  6. from ..utils import (
  7. compat_str,
  8. compat_urllib_parse,
  9. compat_urllib_request,
  10. ExtractorError,
  11. )
  12. class HypemIE(InfoExtractor):
  13. _VALID_URL = r'http://(?:www\.)?hypem\.com/track/([^/]+)/([^/]+)'
  14. _TEST = {
  15. 'url': 'http://hypem.com/track/1v6ga/BODYWORK+-+TAME',
  16. 'md5': 'b9cc91b5af8995e9f0c1cee04c575828',
  17. 'info_dict': {
  18. 'id': '1v6ga',
  19. 'ext': 'mp3',
  20. 'title': 'Tame',
  21. 'uploader': 'BODYWORK',
  22. }
  23. }
  24. def _real_extract(self, url):
  25. mobj = re.match(self._VALID_URL, url)
  26. track_id = mobj.group(1)
  27. data = {'ax': 1, 'ts': time.time()}
  28. data_encoded = compat_urllib_parse.urlencode(data)
  29. complete_url = url + "?" + data_encoded
  30. request = compat_urllib_request.Request(complete_url)
  31. response, urlh = self._download_webpage_handle(
  32. request, track_id, 'Downloading webpage with the url')
  33. cookie = urlh.headers.get('Set-Cookie', '')
  34. html_tracks = self._html_search_regex(
  35. r'(?ms)<script type="application/json" id="displayList-data">\s*(.*?)\s*</script>',
  36. response, 'tracks')
  37. try:
  38. track_list = json.loads(html_tracks)
  39. track = track_list['tracks'][0]
  40. except ValueError:
  41. raise ExtractorError('Hypemachine contained invalid JSON.')
  42. key = track['key']
  43. track_id = track['id']
  44. artist = track['artist']
  45. title = track['song']
  46. serve_url = "http://hypem.com/serve/source/%s/%s" % (track_id, key)
  47. request = compat_urllib_request.Request(
  48. serve_url, '', {'Content-Type': 'application/json'})
  49. request.add_header('cookie', cookie)
  50. song_data = self._download_json(request, track_id, 'Downloading metadata')
  51. final_url = song_data["url"]
  52. return {
  53. 'id': track_id,
  54. 'url': final_url,
  55. 'ext': 'mp3',
  56. 'title': title,
  57. 'uploader': artist,
  58. }