smooth-dl.py: factor out a download_file() function

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

diff --git a/smooth-dl.py b/smooth-dl.py
index 16ee966..9530392 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
 __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
 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):
 
 
 def get_chunk_data(data):


@@ -91,31 +93,40 @@ def write_wav_header(out_file, fmt, codec_private_data, data_len):
     out_file.write("data")
     out_file.write(struct.pack('<L', data_len))
 
     out_file.write("data")
     out_file.write(struct.pack('<L', data_len))
 

+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

 
 

-def get_manifest(base_url, dest_dir=tempfile.gettempdir(),
-        manifest_file='Manifest'):
+    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)
 
     """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'

 
 

-        response = urllib2.urlopen(manifest_url)
-        data = response.read()
-
-        manifest_path = os.path.join(dest_dir, manifest_file)
-        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:
     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":
 
     version = manifest.getroot().attrib['MajorVersion']
     if version != "2":


@@ -124,11 +135,13 @@ 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")
     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
 
     except:
         pass
 

+    manifest_pattern = re.compile("/manifest$", re.IGNORECASE)
+    base_url = manifest_pattern.sub("", manifest_url)
+

     return (manifest, base_url)
 
 
     return (manifest, base_url)
 
 


@@ -216,12 +229,7 @@ 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
 
         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()
         else:
             f = open(chunk_file, "rb")
             data = f.read()