Merge pull request #50 from GrumpyOldTroll/master

Added a strict mode (strict=True optional parameter to M3U8 constructor), plus a couple of parse fixes.

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
     '''
@@ -42,36 +51,36 @@ 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):
             _parse_byterange(line, state)
             state['expect_segment'] = True
 
-        elif state['expect_segment']:
-            _parse_ts_chunk(line, data, state)
-            state['expect_segment'] = False
-
-        elif state['expect_playlist']:
-            _parse_variant_playlist(line, data, state)
-            state['expect_playlist'] = False
-
         elif line.startswith(protocol.ext_x_targetduration):
             _parse_simple_parameter(line, data, float)
+
         elif line.startswith(protocol.ext_x_media_sequence):
             _parse_simple_parameter(line, data, int)
+
         elif line.startswith(protocol.ext_x_program_date_time):
             _, program_date_time = _parse_simple_parameter_raw_value(line, cast_date_time)
             if not data.get('program_date_time'):
                 data['program_date_time'] = program_date_time
             state['current_program_date_time'] = program_date_time
+
         elif line.startswith(protocol.ext_x_discontinuity):
             state['discontinuity'] = True
+
         elif line.startswith(protocol.ext_x_cue_out):
             state['cue_out'] = True
+
         elif line.startswith(protocol.ext_x_version):
             _parse_simple_parameter(line, data)
+
         elif line.startswith(protocol.ext_x_allow_cache):
             _parse_simple_parameter(line, data)
 
@@ -102,6 +111,25 @@ def parse(content):
         elif line.startswith(protocol.ext_x_endlist):
             data['is_endlist'] = True
 
+        elif line.startswith('#'):
+            # comment
+            pass
+
+        elif line.strip() == '':
+            # blank lines are legal
+            pass
+
+        elif state['expect_segment']:
+            _parse_ts_chunk(line, data, state)
+            state['expect_segment'] = False
+
+        elif state['expect_playlist']:
+            _parse_variant_playlist(line, data, state)
+            state['expect_playlist'] = False
+
+        elif strict:
+            raise ParseError(lineno, line)
+
     return data
 
 def _parse_key(line):
@@ -114,7 +142,10 @@ def _parse_key(line):
 
 def _parse_extinf(line, data, state):
     duration, title = line.replace(protocol.extinf + ':', '').split(',')
-    state['segment'] = {'duration': float(duration), 'title': remove_quotes(title)}
+    if 'segment' not in state:
+        state['segment'] = {}
+    state['segment']['duration'] = float(duration)
+    state['segment']['title'] = remove_quotes(title)
 
 def _parse_ts_chunk(line, data, state):
     segment = state.pop('segment')
@@ -173,6 +204,8 @@ def _parse_variant_playlist(line, data, state):
     data['playlists'].append(playlist)
 
 def _parse_byterange(line, state):
+    if 'segment' not in state:
+        state['segment'] = {}
     state['segment']['byterange'] = line.replace(protocol.ext_x_byterange + ':', '')
 
 def _parse_simple_parameter_raw_value(line, cast_to=str, normalize=False):

tests/playlists.py

@@ -122,6 +122,25 @@
 #EXT-X-ENDLIST
 '''
 
+# reversing byterange and extinf from IFRAME.
+IFRAME_PLAYLIST2 = '''
+#EXTM3U
+#EXT-X-VERSION:4
+#EXT-X-TARGETDURATION:10
+#EXT-X-PLAYLIST-TYPE:VOD
+#EXT-X-I-FRAMES-ONLY
+#EXT-X-BYTERANGE:9400@376
+#EXTINF:4.12,
+segment1.ts
+#EXT-X-BYTERANGE:7144@47000
+#EXTINF:3.56,
+segment1.ts
+#EXT-X-BYTERANGE:10340@1880
+#EXTINF:3.82,
+segment2.ts
+#EXT-X-ENDLIST
+'''
+
 PLAYLIST_USING_BYTERANGES = '''
 #EXTM3U
 #EXT-X-MEDIA-SEQUENCE:0

tests/test_model.py

@@ -375,6 +375,14 @@ def test_dump_should_work_for_iframe_playlists():
 
     assert expected == obj.dumps().strip()
 
+    obj = m3u8.M3U8(playlists.IFRAME_PLAYLIST2)
+
+    expected = playlists.IFRAME_PLAYLIST.strip()
+
+    # expected that dump will reverse EXTINF and EXT-X-BYTERANGE,
+    # hence IFRAME_PLAYLIST dump from IFRAME_PLAYLIST2 parse.
+    assert expected == obj.dumps().strip()
+
 def test_dump_should_include_program_date_time():
     obj = m3u8.M3U8(playlists.SIMPLE_PLAYLIST_WITH_PROGRAM_DATE_TIME)