[utils] Support 'dur' field in TTML

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index de09b53b2d81150e355788f8f870dcaf112ae296..5439fcb35d3212fc6ea66f31b6eab9d86bda2301 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -327,13 +327,6 @@ def sanitize_path(s):
     return os.path.join(*sanitized_path)
 
 
-def sanitize_url_path_consecutive_slashes(url):
-    """Collapses consecutive slashes in URLs' path"""
-    parsed_url = list(compat_urlparse.urlparse(url))
-    parsed_url[2] = re.sub(r'/{2,}', '/', parsed_url[2])
-    return compat_urlparse.urlunparse(parsed_url)
-
-
 def orderedSet(iterable):
     """ Remove all duplicates from the input iterable """
     res = []
@@ -1873,10 +1866,14 @@ def dfxp2srt(dfxp_data):
     paras = dfxp.findall(_x('.//ttml:p'))
 
     for para, index in zip(paras, itertools.count(1)):
+        begin_time = parse_dfxp_time_expr(para.attrib['begin'])
+        end_time = parse_dfxp_time_expr(para.attrib.get('end'))
+        if not end_time:
+            end_time = begin_time + parse_dfxp_time_expr(para.attrib['dur'])
         out.append('%d\n%s --> %s\n%s\n\n' % (
             index,
-            format_srt_time(parse_dfxp_time_expr(para.attrib.get('begin'))),
-            format_srt_time(parse_dfxp_time_expr(para.attrib.get('end'))),
+            format_srt_time(begin_time),
+            format_srt_time(end_time),
             parse_node(para)))
 
     return ''.join(out)