oe1archive.py

   1 #!/usr/bin/env python3
   2
   3 """A simple tool to query the Oe1 7 Tage archive."""
   4
   5 __version__ = "2.1"
   6 __author__ = "Stefan Huber"
   7
   8
   9 import urllib.request
  10 import simplejson
  11 import dateutil.parser
  12 import sys
  13 import getopt
  14 import re
  15 import os
  16
  17
  18 class Archive:
  19
  20     def __init__(self):
  21         self.json = read_json("http://audioapi.orf.at/oe1/json/2.0/broadcasts/")
  22
  23     def get_days(self):
  24         return map(_json_to_day, self.json)
  25
  26     def get_broadcasts(self, day):
  27         bjson = self.json[day]['broadcasts']
  28         return map(_json_to_broadcast, bjson)
  29
  30     def get_broadcast(self, day, broadcast):
  31         return _json_to_broadcast(self.json[day]['broadcasts'][broadcast])
  32
  33     def get_player_url(self, day, broadcast):
  34         date = self.json[day]['day']
  35         pk = self.json[day]['broadcasts'][broadcast]['programKey']
  36         url = "http://oe1.orf.at/player/%d/%s"
  37         return url % (date, pk)
  38
  39     def get_broadcast_subtitle(self, day, broadcast):
  40         return self.json[day]['broadcasts'][broadcast]['subtitle']
  41
  42     def get_broadcast_pk(self, day, broadcast):
  43         return self.json[day]['broadcasts'][broadcast]['programKey']
  44
  45     def get_broadcast_url(self, day, broadcast):
  46         date = self.json[day]['day']
  47         pk = self.json[day]['broadcasts'][broadcast]['programKey']
  48
  49         burl = 'https://audioapi.orf.at/oe1/api/json/current/broadcast/%s/%d'
  50         bjson = read_json(burl % (pk, date))
  51
  52         sjson = bjson['streams']
  53         if len(sjson) == 0:
  54             return None
  55
  56         sid = sjson[0]['loopStreamId']
  57         surl = 'https://loopstream01.apa.at/?channel=oe1&shoutcast=0&id=%s'
  58         return surl % sid
  59
  60     def get_broadcast_description(self, day, broadcast):
  61         date = self.json[day]['day']
  62         pk = self.json[day]['broadcasts'][broadcast]['programKey']
  63
  64         burl = 'https://audioapi.orf.at/oe1/api/json/current/broadcast/%s/%d'
  65         bjson = read_json(burl % (pk, date))
  66
  67         description = bjson['description']
  68         akm = bjson['akm']
  69         if description is None:
  70             description = ""
  71         if akm is None:
  72             akm = ""
  73         return description + "<br>" + akm;
  74
  75     def get_broadcasts_by_regex(self, key):
  76         rex = re.compile(key, re.IGNORECASE)
  77
  78         res = []
  79         for d, djson in enumerate(self.json):
  80             for b, bjson in enumerate(djson['broadcasts']):
  81                 if rex.search(bjson['title']) is not None:
  82                     res.append((d, b))
  83                 elif bjson['subtitle'] is not None and rex.search(bjson['subtitle']) is not None:
  84                     res.append((d, b))
  85         return res
  86
  87 def _json_to_day(djson):
  88     return dateutil.parser.parse(djson['dateISO'])
  89
  90 def _json_to_broadcast(bjson):
  91     dt = dateutil.parser.parse(bjson['startISO'])
  92     return (dt, bjson['title'])
  93
  94
  95 def read_json(url):
  96     with urllib.request.urlopen(url) as f:
  97         dec = simplejson.JSONDecoder()
  98         return dec.decode(f.read())
  99
 100 def input_index(prompt, li):
 101     while True:
 102         try:
 103             idx = int(input(prompt))
 104             if idx < 0 or idx >= len(li):
 105                 print("Out out range!")
 106             else:
 107                 return idx
 108
 109         except ValueError:
 110             print("Unknown input.")
 111         except EOFError:
 112             sys.exit(1)
 113
 114 def screen_help():
 115     print("""Usage:
 116     {0} -h, --help
 117     {0} -c, --choose
 118     {0} -s, --search TITLE""".format(sys.argv[0]))
 119
 120 def screen_choose():
 121     a = Archive()
 122
 123     print("Choose a date:")
 124     days = list(a.get_days())
 125     for i, date in enumerate(days):
 126         print("  [%d]  %s" % (i, date.strftime("%a %d. %b %Y")))
 127     day = input_index("Date: ", days)
 128     chosen_datetime = days[day]
 129     print()
 130
 131     print("Choose a broadcast:")
 132     broadcasts = list(a.get_broadcasts(day))
 133     for i, b in enumerate(broadcasts):
 134         date, title = b
 135         print("  [%2d]  %s  %s" % (i, date.strftime("%H:%M:%S"), title))
 136     broadcast = input_index("Broadcast: ", broadcasts)
 137     print()
 138
 139     print_broadcast_info(a, day, broadcast)
 140     print()
 141
 142     url = a.get_broadcast_url(day, broadcast)
 143     if url is not None:
 144         answer = input("Do you want to download the chosen broadcast? (y/N) ")
 145         if answer in ["y", "Y", "j", "J"]:
 146             name = input("Download directory (prefix): ")
 147
 148             try:
 149                 dirname = get_directory_name(name, chosen_datetime)
 150                 print("Downloading to %s..." % dirname)
 151
 152                 make_directory(name, chosen_datetime)
 153
 154                 description = a.get_broadcast_description(day, broadcast)
 155                 write_html_file(name, chosen_datetime, description)
 156
 157                 write_mp3_file(name, chosen_datetime, url)
 158
 159             except OSError as e:
 160                 print("Error creating directory.")
 161                 print(e)
 162
 163             except requests.exceptions.RequestException as e:
 164                 print("Request getting mp3 failed.")
 165
 166             except Exception as e:
 167                 print("Error downloading mp3.")
 168                 print(e)
 169
 170 def get_directory_name(name, datetime):
 171     prefix = ""
 172     if len(name) > 0:
 173         prefix = name + "_"
 174
 175     return prefix + datetime.strftime("%d-%m-%Y")
 176
 177 def make_directory(name, datetime):
 178     """Creates the download subdirectory for the given name and datetime."""
 179     dirname = get_directory_name(name, datetime)
 180     if not os.path.exists(dirname):
 181         os.makedirs(dirname)
 182
 183 def write_html_file(name, datetime, description):
 184     """Stores broadcast description into a html file."""
 185
 186     longname = get_directory_name(name, datetime)
 187     filepath = os.path.join(longname, longname + ".html")
 188     file = open(filepath, 'w+')
 189     file.write("<!DOCTYPE html>\n")
 190     file.write("<html>\n")
 191     file.write("<head>\n")
 192     file.write("<title>\n")
 193     file.write("%s %s\n" % (name, datetime.strftime("%d.%m.%Y")))
 194     file.write("</title>\n")
 195     file.write("<meta charset = \"utf-8\">\n")
 196     file.write("</head>\n")
 197     file.write("<body>\n")
 198     file.write("%s %s" % (name, datetime.strftime("%d.%m.%Y")))
 199     file.write(description)
 200     file.write("</body>\n")
 201     file.write("</html>")
 202     file.close()
 203
 204 def write_mp3_file(name, datetime, url):
 205     import requests
 206
 207     longname = get_directory_name(name, datetime)
 208     filepath = os.path.join(longname, longname + ".mp3")
 209
 210     print("Fetching mp3...")
 211     r = requests.get(url, stream=True)
 212     if r.status_code == 200:
 213         with open(filepath, 'wb') as f:
 214             f.write(r.content)
 215     else:
 216         print("Error downloading mp3. Status code: %d" % r.status_code)
 217
 218 def screen_search(key):
 219     a = Archive()
 220     for d, b in a.get_broadcasts_by_regex(key):
 221         print_broadcast_info(a, d, b)
 222         print()
 223
 224 def print_broadcast_info(archive, day, broadcast):
 225     a, d, b = archive, day, broadcast
 226     date, title = a.get_broadcast(d, b)
 227
 228     print("%s   %s" % (date.strftime("%a %d.%m.%Y  %H:%M:%S"), title))
 229     print("  %s" % a.get_broadcast_subtitle(d, b))
 230     print("  Broadcast: %s" % a.get_broadcast_url(d, b))
 231     print("  Player: %s" % a.get_player_url(d, b))
 232     print("  Program key: %s" % a.get_broadcast_pk(d, b))
 233
 234 if __name__ == "__main__":
 235
 236     try:
 237         opts, args = getopt.getopt(sys.argv[1:], "hcs:",
 238                 ["help", "choose", "search="])
 239     except getopt.GetoptError as err:
 240         print(err)
 241         screen_help()
 242         sys.exit(2)
 243
 244     for o, a in opts:
 245         if o in ["-h", "--help"]:
 246             screen_help()
 247         if o in ["-c", "--choose"]:
 248             screen_choose()
 249         if o in ["-s", "--search"]:
 250             screen_search(a)