X-Git-Url: https://git.ao2.it/smooth-dl.git/blobdiff_plain/64956443214ffb5ba10b4079a71954a2f6b23995..00716d449533774cb4ab83bdae6d980848160484:/smooth-dl.py

diff --git a/smooth-dl.py b/smooth-dl.py
index 91b4ec2..aa57c6e 100755
--- a/smooth-dl.py
+++ b/smooth-dl.py
@@ -30,18 +30,19 @@
 # basically, write a proper implementation of manifest parsing and chunk
 # downloading
 
-
-__description = "Download videos served using Smooth Streaming technology"
-__version = "0.x"
-__author_info = "Written by Antonio Ospite http://ao2.it"
-
 import os
+import re
 import sys
 import xml.etree.ElementTree as etree
 import urllib2
 import struct
 import tempfile
 from optparse import OptionParser
+from urlparse import urlparse, urlunparse
+
+__description = "Download videos served using Smooth Streaming technology"
+__version = "0.x"
+__author_info = "Written by Antonio Ospite http://ao2.it"
 
 
 def get_chunk_data(data):
@@ -52,7 +53,7 @@ def get_chunk_data(data):
     data_start = moof_size + 4 + len('mdat')
     data_size = mdat_size - 4 - len('mdat')
 
-    #print len(data[data_start:]), \
+    # print len(data[data_start:]), \
     #        len(data[data_start:data_start + data_size]), data_size
 
     assert(len(data[data_start:]) == data_size)
@@ -92,29 +93,41 @@ def write_wav_header(out_file, fmt, codec_private_data, data_len):
     out_file.write(struct.pack('<L', data_len))
 
 
-def get_manifest(base_url, dest_dir=tempfile.gettempdir()):
+def download_file(src_url, dest_file, mode):
+    try:
+        response = urllib2.urlopen(src_url)
+        data = response.read()
+    except urllib2.HTTPError as e:
+        sys.stderr.write("Error while dowloading URL: %s" % src_url)
+        raise
+
+    if dest_file:
+        f = open(dest_file, mode)
+        f.write(data)
+        f.close()
+
+    return data
+
+
+def get_manifest(url, dest_dir=tempfile.gettempdir()):
     """Returns the manifest and the new URL if this is changed"""
 
     if os.path.exists(dest_dir) == False:
         os.mkdir(dest_dir, 0755)
 
-    if base_url.startswith('http://'):
-
-        manifest_url = base_url
-        if not manifest_url.lower().endswith(('/manifest', '.ismc', '.csm')):
-            manifest_url += '/Manifest'
+    # Remove the querystring if present
+    manifest_url = urlunparse(urlparse(url)._replace(query=''))
 
-        response = urllib2.urlopen(manifest_url)
-        data = response.read()
+    if not manifest_url.lower().endswith(('/manifest', '.ismc', '.csm')):
+        manifest_url += '/Manifest'
 
-        manifest_path = os.path.join(dest_dir, 'Manifest')
-        f = open(manifest_path, "w")
-        f.write(data)
-        f.close()
+    if manifest_url.startswith('http://'):
+        local_manifest_path = os.path.join(dest_dir, 'Manifest')
+        download_file(manifest_url, local_manifest_path, "w")
     else:
-        manifest_path = base_url
+        local_manifest_path = url
 
-    manifest = etree.parse(manifest_path)
+    manifest = etree.parse(local_manifest_path)
 
     version = manifest.getroot().attrib['MajorVersion']
     if version != "2":
@@ -123,11 +136,14 @@ def get_manifest(base_url, dest_dir=tempfile.gettempdir()):
     try:
         # if some intermediate client Manifest is used, like in Rai Replay
         clip = manifest.find("Clip")
-        actual_manifest_url = clip.attrib["Url"]
-        base_url = actual_manifest_url.lower().replace("/manifest", "")
-    except:
+        manifest_url = clip.attrib["Url"]
+        manifest = download_file(manifest_url, None, None)
+    except AttributeError:
         pass
 
+    manifest_pattern = re.compile("/manifest$", re.IGNORECASE)
+    base_url = manifest_pattern.sub("", manifest_url)
+
     return (manifest, base_url)
 
 
@@ -154,8 +170,9 @@ def print_manifest_info(manifest):
                 channels = q.attrib["Channels"]
                 sampling_rate = q.attrib["SamplingRate"]
                 bits_per_sample = q.attrib["BitsPerSample"]
-                print "\t%2s: %4s %sHz %sbits %sch @ %7s bps" % (i, fourcc,
-                        sampling_rate, bits_per_sample, channels, bitrate)
+                print "\t%2s: %4s %sHz %sbits %sch @ %7s bps" % \
+                    (i, fourcc, sampling_rate, bits_per_sample, channels,
+                     bitrate)
 
     print
 
@@ -177,7 +194,8 @@ def get_chunk_quality_string(stream, quality_level):
     url = stream.attrib["Url"]
 
     chunks_quality = url.split('/')[0].replace("{bitrate}", bitrate)
-    chunks_quality = chunks_quality.replace("{CustomAttributes}", custom_attributes)
+    chunks_quality = chunks_quality.replace("{CustomAttributes}",
+                                            custom_attributes)
 
     return chunks_quality
 
@@ -215,12 +233,7 @@ def download_chunks(base_url, manifest, stream_index, quality_level, dest_dir):
 
         if os.path.exists(chunk_file) == False:
             chunk_url = base_url + '/' + chunks_quality + '/' + chunk_name
-            response = urllib2.urlopen(chunk_url)
-            data = response.read()
-
-            f = open(chunk_file, "wb")
-            f.write(data)
-            f.close()
+            data = download_file(chunk_url, chunk_file, "wb")
         else:
             f = open(chunk_file, "rb")
             data = f.read()
@@ -233,9 +246,9 @@ def download_chunks(base_url, manifest, stream_index, quality_level, dest_dir):
 
 
 def rebuild_stream(manifest, stream_index, quality_level, src_dir,
-        dest_file_name, final_dest_file=None):
+                   dest_file_name, final_dest_file=None):
 
-    if final_dest_file == None:
+    if final_dest_file is None:
         final_dest_file = dest_file_name
 
     stream = manifest.findall('.//StreamIndex')[stream_index]
@@ -307,7 +320,7 @@ def calc_tracks_delay(manifest, stream1_index, stream2_index):
 
     # calc difference in seconds
     delay = s2_start_time / s2_timescale - \
-            s1_start_time / s1_timescale
+        s1_start_time / s1_timescale
 
     return delay
 
@@ -320,39 +333,39 @@ def get_clip_duration(manifest):
 
 
 def smooth_download(url, manifest, dest_dir=tempfile.gettempdir(),
-        video_stream_index=0, audio_stream_index=1,
-        video_quality_level=0, audio_quality_level=0,
-        chunks_dir=None, download=True,
-        out_video_file='_video.vc1', out_audio_file='_audio.raw'):
+                    video_stream_index=0, audio_stream_index=1,
+                    video_quality_level=0, audio_quality_level=0,
+                    chunks_dir=None, download=True,
+                    out_video_file='_video.vc1', out_audio_file='_audio.raw'):
 
-        if chunks_dir == None:
+        if chunks_dir is None:
             chunks_dir = dest_dir
 
         if download:
             download_chunks(url, manifest, video_stream_index,
-                    video_quality_level, chunks_dir)
+                            video_quality_level, chunks_dir)
             download_chunks(url, manifest, audio_stream_index,
-                    audio_quality_level, chunks_dir)
+                            audio_quality_level, chunks_dir)
 
         dest_video = os.path.join(dest_dir, out_video_file)
         dest_audio = os.path.join(dest_dir, out_audio_file)
 
         rebuild_stream(manifest, video_stream_index, video_quality_level,
-                chunks_dir, dest_video)
+                       chunks_dir, dest_video)
         rebuild_stream(manifest, audio_stream_index, audio_quality_level,
-                chunks_dir, dest_audio, dest_audio + '.wav')
+                       chunks_dir, dest_audio, dest_audio + '.wav')
 
-        #duration = get_clip_duration(manifest)
+        # duration = get_clip_duration(manifest)
 
         delay = calc_tracks_delay(manifest, video_stream_index,
-                audio_stream_index)
+                                  audio_stream_index)
 
         # optionally encode audio to vorbis:
         # ffmpeg -i _audio.raw.wav -acodec libvorbis -aq 60 audio.ogg
         mux_command = ("ffmpeg -i %s \\\n" +
-                      "  -itsoffset %f -async 1 -i %s \\\n" +
-                      "  -vcodec copy -acodec copy ffout.mkv") % \
-                      (dest_video, delay, dest_audio + '.wav')
+                       "  -itsoffset %f -async 1 -i %s \\\n" +
+                       "  -vcodec copy -acodec copy ffout.mkv") % \
+            (dest_video, delay, dest_audio + '.wav')
 
         print mux_command
 
@@ -361,7 +374,7 @@ def options_parser():
     version = "%%prog %s" % __version
     usage = "usage: %prog [options] <manifest URL or file>"
     parser = OptionParser(usage=usage, version=version,
-            description=__description, epilog=__author_info)
+                          description=__description, epilog=__author_info)
     parser.add_option("-i", "--info",
                       action="store_true", dest="info_only",
                       default=False, help="print Manifest info and exit")
@@ -413,8 +426,8 @@ if __name__ == "__main__":
 
     if options.sync_delay:
         print calc_tracks_delay(manifest,
-                options.video_stream_index,
-                options.audio_stream_index)
+                                options.video_stream_index,
+                                options.audio_stream_index)
         parser.exit(0)
 
     if options.info_only:
@@ -424,6 +437,6 @@ if __name__ == "__main__":
     print_manifest_info(manifest)
 
     smooth_download(url, manifest, options.dest_dir,
-            options.video_stream_index, options.audio_stream_index,
-            options.video_quality_level, options.audio_quality_level,
-            options.chunks_dir, options.download)
+                    options.video_stream_index, options.audio_stream_index,
+                    options.video_quality_level, options.audio_quality_level,
+                    options.chunks_dir, options.download)