Fix for #49: Added optional scrict=True parameter to avoid silently a…

…ccepting things like html by raising ParseError. (Defaults to false for backward compatibility.)

m3u8/__init__.py

@@ -20,10 +20,10 @@
 
 
 from m3u8.model import M3U8, Playlist, IFramePlaylist, Media, Segment
-from m3u8.parser import parse, is_url
+from m3u8.parser import parse, is_url, ParseError
 
 __all__ = ('M3U8', 'Playlist', 'IFramePlaylist', 'Media',
-           'Segment', 'loads', 'load', 'parse')
+           'Segment', 'loads', 'load', 'parse', 'ParseError')
 
 def loads(content):
     '''

m3u8/model.py

@@ -120,9 +120,9 @@ class M3U8(object):
         ('playlist_type',    'playlist_type')
         )
 
-    def __init__(self, content=None, base_path=None, base_uri=None):
+    def __init__(self, content=None, base_path=None, base_uri=None, strict=False):
         if content is not None:
-            self.data = parser.parse(content)
+            self.data = parser.parse(content, strict)
         else:
             self.data = {}
         self._base_uri = base_uri

m3u8/parser.py

@@ -8,6 +8,7 @@
 import itertools
 import re
 from m3u8 import protocol
+import exceptions
 
 '''
 http://tools.ietf.org/html/draft-pantos-http-live-streaming-08#section-3.2
@@ -21,7 +22,15 @@ def cast_date_time(value):
 def format_date_time(value):
     return value.isoformat()
 
-def parse(content):
+class ParseError(exceptions.Exception):
+    def __init__(self, lineno, line):
+        self.lineno = lineno
+        self.line = line
+
+    def __str__(self):
+        return 'Syntax error in manifest on line %d: %s' % (self.lineno, self.line)
+
+def parse(content, strict=False):
     '''
     Given a M3U8 playlist content returns a dictionary with all data found
     '''
@@ -42,7 +51,9 @@ def parse(content):
         'expect_playlist': False,
         }
 
+    lineno = 0
     for line in string_to_lines(content):
+        lineno += 1
         line = line.strip()
 
         if line.startswith(protocol.ext_x_byterange):
@@ -100,6 +111,13 @@ def parse(content):
         elif line.startswith(protocol.ext_x_endlist):
             data['is_endlist'] = True
 
+        elif line.startswith('#'):
+            # comment
+            pass
+
+        elif strict:
+            raise ParseError(lineno, line)
+
     return data
 
 def _parse_key(line):