[sandia] Add new extractor (#4974)

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 03566d22357a086bb1019bb1187db9b5425b83f0..238b6556beb514f2cc4755f4ad767c44509d68f6 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -62,6 +62,11 @@ std_headers = {
 }
 
 
+ENGLISH_MONTH_NAMES = [
+    'January', 'February', 'March', 'April', 'May', 'June',
+    'July', 'August', 'September', 'October', 'November', 'December']
+
+
 def preferredencoding():
     """Get preferred encoding.
 
@@ -666,26 +671,27 @@ class YoutubeDLHTTPSHandler(compat_urllib_request.HTTPSHandler):
             req, **kwargs)
 
 
-def parse_iso8601(date_str, delimiter='T'):
+def parse_iso8601(date_str, delimiter='T', timezone=None):
     """ Return a UNIX timestamp from the given date """
 
     if date_str is None:
         return None
 
-    m = re.search(
-        r'(\.[0-9]+)?(?:Z$| ?(?P<sign>\+|-)(?P<hours>[0-9]{2}):?(?P<minutes>[0-9]{2})$)',
-        date_str)
-    if not m:
-        timezone = datetime.timedelta()
-    else:
-        date_str = date_str[:-len(m.group(0))]
-        if not m.group('sign'):
+    if timezone is None:
+        m = re.search(
+            r'(\.[0-9]+)?(?:Z$| ?(?P<sign>\+|-)(?P<hours>[0-9]{2}):?(?P<minutes>[0-9]{2})$)',
+            date_str)
+        if not m:
             timezone = datetime.timedelta()
         else:
-            sign = 1 if m.group('sign') == '+' else -1
-            timezone = datetime.timedelta(
-                hours=sign * int(m.group('hours')),
-                minutes=sign * int(m.group('minutes')))
+            date_str = date_str[:-len(m.group(0))]
+            if not m.group('sign'):
+                timezone = datetime.timedelta()
+            else:
+                sign = 1 if m.group('sign') == '+' else -1
+                timezone = datetime.timedelta(
+                    hours=sign * int(m.group('hours')),
+                    minutes=sign * int(m.group('minutes')))
     date_format = '%Y-%m-%d{0}%H:%M:%S'.format(delimiter)
     dt = datetime.datetime.strptime(date_str, date_format) - timezone
     return calendar.timegm(dt.timetuple())
@@ -1184,11 +1190,18 @@ def get_term_width():
 def month_by_name(name):
     """ Return the number of a month by (locale-independently) English name """
 
-    ENGLISH_NAMES = [
-        'January', 'February', 'March', 'April', 'May', 'June',
-        'July', 'August', 'September', 'October', 'November', 'December']
     try:
-        return ENGLISH_NAMES.index(name) + 1
+        return ENGLISH_MONTH_NAMES.index(name) + 1
+    except ValueError:
+        return None
+
+
+def month_by_abbreviation(abbrev):
+    """ Return the number of a month by (locale-independently) English
+        abbreviations """
+
+    try:
+        return [s[:3] for s in ENGLISH_MONTH_NAMES].index(abbrev) + 1
     except ValueError:
         return None
 
@@ -1547,8 +1560,8 @@ def js_to_json(code):
         return '"%s"' % v
 
     res = re.sub(r'''(?x)
-        "(?:[^"\\]*(?:\\\\|\\")?)*"|
-        '(?:[^'\\]*(?:\\\\|\\')?)*'|
+        "(?:[^"\\]*(?:\\\\|\\['"nu]))*[^"\\]*"|
+        '(?:[^'\\]*(?:\\\\|\\['"nu]))*[^'\\]*'|
         [a-zA-Z_][.a-zA-Z_0-9]*
         ''', fix_kv, code)
     res = re.sub(r',(\s*\])', lambda m: m.group(1), res)
@@ -1603,6 +1616,15 @@ def args_to_str(args):
     return ' '.join(shlex_quote(a) for a in args)
 
 
+def mimetype2ext(mt):
+    _, _, res = mt.rpartition('/')
+
+    return {
+        'x-ms-wmv': 'wmv',
+        'x-mp4-fragmented': 'mp4',
+    }.get(res, res)
+
+
 def urlhandle_detect_ext(url_handle):
     try:
         url_handle.headers
@@ -1618,7 +1640,7 @@ def urlhandle_detect_ext(url_handle):
             if e:
                 return e
 
-    return getheader('Content-Type').split("/")[1]
+    return mimetype2ext(getheader('Content-Type'))
 
 
 def age_restricted(content_limit, age_limit):