#!/usr/bin/env python # -*- coding: utf-8 -*- # Zusammenfassung von m3u8_get.py und ts_download.py # Testscript lokal: m3u8_get3.py # # OK: # https://mcdn.daserste.de/daserste/de/master_3744/00803/master_3744_00925.ts # # Schema master.m3u8: # #EXTM3U # #EXT-X-VERSION:3 # #EXT-X-INDEPENDENT-SEGMENTS # #EXT-X-STREAM-INF:BANDWIDTH=4237552,AVERAGE-BANDWIDTH=4118400,CODECS="avc1.640020,mp4a.40.2",RESOLUTION=1280x720,FRAME-RATE=50.000 # master_3744.m3u8 # Schema S-Liste: # #EXTM3U # #EXT-X-VERSION:3 # #EXT-X-TARGETDURATION:4 # #EXT-X-MEDIA-SEQUENCE:2011743 # #EXT-X-DISCONTINUITY-SEQUENCE:33741 # #EXT-X-PROGRAM-DATE-TIME:2020-07-16T12:02:28.120Z <- Zeitmarke (akt.: plus 2 Std.), Abdeckung 2 Std. zurück # #EXTINF:4.00000, # master_3744/00804/master_3744_01743.ts # #EXTINF:4.00000, # master_3744/00804/master_3744_01744.ts # # Details zum Livestream DasErste https://mcdn.daserste.de/daserste/de/master.m3u8: # zwischen den Marken X-PROGRAM-DATE-TIME (10 Min.=600 sec) liegen ca.150 ts-Einträge, Größe einz. # ts-Dateien: ca. 1.8 MByte, Länge in VLC 3 sec statt real 4, Verkettung von 10 ts-Downloads # ergeben 35 sec Video in VLC # Länge laut ffprobe 4 sec (ffprobe master_3744_01948.ts) # Zeitmarke: # Die ts-Dateien decken den Zeitraum bis 2 Std. zurück ab. Die Zeitmarken haben den übl. Versatz minus 2 Std. # Die letzte Zeitmarke liegt 10 Min. zurück - bis zum Ende liegen ebenfalls 150 ts-Einträge # Berechnung: 600 sec / 150 ts-Einträge = 4 sec/pro ts-Eintrag (wie Ausgabe ffprobe) # siehe ts_liste1.txt und ts_liste2.txt (Abstand 8 Min) # Berechnung entfällt bei Marke TARGETDURATION (Kopf ts-Liste) # # curl-header: # < Content-Type: video/mp2t # < Content-Length: 1900492 # < Connection: keep-alive # < Cache-Control: max-age=14400 # # s. https://www.dacast.com/blog/hls-streaming-protocol/ # In terms of technical functionality, HLS will play video encoded with the H.264 or HEVC/H.265 codecs. # It then chops video into 10-second segments. Remember, latency for delivery tends to be in the 30-second range. # # Playlist-Varianten: # https://github.com/globocom/m3u8/blob/7d9e2bde4ba1cb69f0f76d95e7e098f54fd58af8/tests/playlists.py # # Alternativen: # Download via VLC (google: vlc console recording settings): # https://blog.sourcefabric.org/en/news/blog/2077/Schedule-stream-recordings-from-the-command-line-Part-2.htm # Download via ffmpeg - s. LiveRecord (Kodi-Addon-ARDundZDF) from urllib import quote, unquote, quote_plus, unquote_plus, urlencode, urlretrieve from urllib2 import Request, urlopen, URLError from urlparse import urljoin, urlparse, urlunparse , urlsplit, parse_qs import os import sys, ssl, re import time, datetime def get_url_list(m3u8_url, body): # Url-Liste für Einzelauflösungen print "get_url_list:" urlgroup = urlparse(m3u8_url) host = urlgroup.scheme + '://' + urlgroup.hostname lines = body.split('\n') ts_url_list = [] for line in lines: if not line.startswith('#') and line != '': if line.startswith('http'): ts_url_list.append(line) else: if '/' in line: # normaler Pfad ts_url_list.append('%s/%s' % (host, line)) else: # rel. Pfad: url anpassen url_m3u8_path = m3u8_url.split('/')[-1] # Bsp.: ..de/master.m3u8 -> ..de/master_320.m3u8 line = m3u8_url.replace(url_m3u8_path, line) ts_url_list.append(line) return ts_url_list def get_m3u8_body(m3u8_url): # Master m3u8 print 'hole Inhalt m3u8-Datei:' + m3u8_url req = Request(m3u8_url, headers={'user-agent': 'Mozilla/5.0'}) r = urlopen(req) new_url = r.geturl() # follow redirects print("new_url: " + new_url) # -> msg s.u. page = r.read() return page def download_ts_file(ts_url): # TS-Listen für Einzelauflösungen print 'hole Inhalt ts-Datei: ' + ts_url req = Request(ts_url, headers={'user-agent': 'Mozilla/5.0'}) r = urlopen(req) new_url = r.geturl() # follow redirects print("new_url: " + new_url) # -> msg s.u. page = r.read() print "len(page): " + str(len(page)) return page def download_ts(host, ts_page, dest_video, duration): print 'hole ts-Dateien:' + dest_video lines = ts_page.splitlines() if 'PROGRAM-DATE-TIME:' in ts_page: # kann fehlen dateLines=[] for line in lines: # Abstand zw. 2 Zeitmarken ermitteln if 'PROGRAM-DATE-TIME:' in line: # kann ev. entfallen d = line.split('DATE-TIME:')[1] dateLines.append(d) print "dateLines: " + str(dateLines) if len(dateLines) > 0: date_format = "%Y-%m-%dT%H:%M:%S" unix_time1 = dateLines[0].split('.')[0] unix_time1 = time.mktime(time.strptime(unix_time1, date_format)) unix_time2 = dateLines[1].split('.')[0] unix_time2 = time.mktime(time.strptime(unix_time2, date_format)) diff = unix_time2 - unix_time1 print "diff time2-time1: " + str(diff) # Differenz in Sek. try: ts_dur = re.search('TARGETDURATION:(\d+)', ts_page).group(1) except Exception as exception: print (str(exception)) ts_dur=4 # Default ARD & Co print "ts_dur: " + ts_dur #return #Debug f = open(dest_video, 'wb') dt = datetime.datetime.now() # s. EPG get_unixtime now = time.mktime(dt.timetuple()) start = str(now).split('.')[0] lines = ts_page.splitlines() cnt=1; for line in lines: #print line if 'PROGRAM-DATE-TIME:' in line or line.startswith('#') == False: if 'PROGRAM-DATE-TIME:' in line: # print line pass else: # hier Check ob der Download in Realtime erfolgt. Falls schneller, dann # Verzicht auf now, stattdessen einfach 10 sec addieren dt = datetime.datetime.now() # s. EPG get_unixtime now = time.mktime(dt.timetuple()) now = str(now).split('.')[0] diff = int(now) - int(start) print "diff: " + str(diff) if diff > duration: # Dauer erreicht break if line.startswith('http'): ts_url = line else: ts_url = "%s/%s" % (host, line) print "ts_url: " + ts_url req = Request(ts_url, headers={'user-agent': 'Mozilla/5.0'}) gcontext = ssl.create_default_context() gcontext.check_hostname = False r = urlopen(req, context=gcontext, timeout=3) new_url = r.geturl() # follow redirects meta = r.info() print "%d. Download %s, ts: %s" % (cnt, dest_video, ts_url) # os.system('cls') # clear Shell file_size_dl = 0 block_sz = 8192 cnt=cnt+1 while True: buffer = r.read(block_sz) if not buffer: break file_size_dl += len(buffer) f.write(buffer) time.sleep(int(ts_dur)) # DasErste je ts 4 sec laut ffprobe f.close() def main(m3u8_url, dest_video, duration): body = get_m3u8_body(m3u8_url) # gesamte m3u8-Seite print (body) ts_url_list = get_url_list(m3u8_url, body) # alle TS-Listen ts_url = ts_url_list[0] # 1. Qual. #ts_url = ts_url_list[-1] # Debug letzte Qual. print 'Anzahl ts-Quellen:', len(ts_url_list) print "ts_url_list: " + ts_url ts_page = download_ts_file(ts_url) # nur 1. Liste (höchste Qual.) with open("./ts_liste.txt",'w') as f: # Debug f.write(ts_page) pos=m3u8_url.rfind('/') host=m3u8_url[:pos] print "host: " + host download_ts(host, ts_page, dest_video, duration) # ts-Dateien laden + verketten #----------------------------------------------------------------------- duration = 60 # Dauer in sec. #m3u8_url = "https://mcdn.daserste.de/daserste/de/master.m3u8" # DasErste m3u8_url = "https://ndrfs-lh.akamaihd.net/i/ndrfs_mv@430232/master.m3u8" # NDR #m3u8_url = "https://rbmn-live.akamaized.net/hls/live/2002830/geoSTVDEweb/master.m3u8" # ServusTV dest_video = "./m3u8_Test/Video.mp4" main(m3u8_url, dest_video, duration)