Differentiate better between url, manifest_url and base_url

[smooth-dl.git] / smooth-dl.py

diff --git a/smooth-dl.py b/smooth-dl.py
index 91b4ec2..3297102 100755 (executable)
--- a/smooth-dl.py
+++ b/smooth-dl.py
@@ -36,12 +36,14 @@ __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
 
 
 def get_chunk_data(data):
@@ -92,29 +94,32 @@ 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 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://'):
+    # Remove the querystring if present
+    manifest_url = urlunparse(urlparse(url)._replace(query=''))
 
-        manifest_url = base_url
-        if not manifest_url.lower().endswith(('/manifest', '.ismc', '.csm')):
-            manifest_url += '/Manifest'
+    if not manifest_url.lower().endswith(('/manifest', '.ismc', '.csm')):
+        manifest_url += '/Manifest'
+
+    if manifest_url.startswith('http://'):
 
         response = urllib2.urlopen(manifest_url)
         data = response.read()
 
-        manifest_path = os.path.join(dest_dir, 'Manifest')
-        f = open(manifest_path, "w")
+        local_manifest_path = os.path.join(dest_dir, 'Manifest')
+
+        f = open(local_manifest_path, "w")
         f.write(data)
         f.close()
     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 +128,13 @@ 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", "")
+        manifest_url = clip.attrib["Url"]
     except:
         pass
 
+    manifest_pattern = re.compile("/manifest$", re.IGNORECASE)
+    base_url = manifest_pattern.sub("", manifest_url)
+
     return (manifest, base_url)
 
 
@@ -215,12 +222,16 @@ 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()
+            try:
+                response = urllib2.urlopen(chunk_url)
+                data = response.read()
+
+                f = open(chunk_file, "wb")
+                f.write(data)
+                f.close()
+            except Exception as e:
+                print e
 
-            f = open(chunk_file, "wb")
-            f.write(data)
-            f.close()
         else:
             f = open(chunk_file, "rb")
             data = f.read()