Add basic support for custom tags

Main parser function now receives a function that can be
used to parse the line content and save it into the data store object.

m3u8/__init__.py

@@ -24,44 +24,44 @@
            'Segment', 'loads', 'load', 'parse', 'ParseError')
 
 
-def loads(content, uri=None):
+def loads(content, uri=None, custom_tags_parser=None):
     '''
     Given a string with a m3u8 content, returns a M3U8 object.
     Optionally parses a uri to set a correct base_uri on the M3U8 object.
     Raises ValueError if invalid content
     '''
 
     if uri is None:
-        return M3U8(content)
+        return M3U8(content, custom_tags_parser)
     else:
         base_uri = _parsed_url(uri)
-        return M3U8(content, base_uri=base_uri)
+        return M3U8(content, base_uri=base_uri, custom_tags_parser=custom_tags_parser)
 
 
-def load(uri, timeout=None, headers={}):
+def load(uri, timeout=None, headers={}, custom_tags_parser=None):
     '''
     Retrieves the content from a given URI and returns a M3U8 object.
     Raises ValueError if invalid content or IOError if request fails.
     Raises socket.timeout(python 2.7+) or urllib2.URLError(python 2.6) if
     timeout happens when loading from uri
     '''
     if is_url(uri):
-        return _load_from_uri(uri, timeout, headers)
+        return _load_from_uri(uri, timeout, headers, custom_tags_parser)
     else:
-        return _load_from_file(uri)
+        return _load_from_file(uri, custom_tags_parser)
 
 # Support for python3 inspired by https://github.com/szemtiv/m3u8/
 
 
-def _load_from_uri(uri, timeout=None, headers={}):
+def _load_from_uri(uri, timeout=None, headers={}, custom_tags_parser=None):
     request = Request(uri, headers=headers)
     resource = urlopen(request, timeout=timeout)
     base_uri = _parsed_url(resource.geturl())
     if PYTHON_MAJOR_VERSION < (3,):
         content = _read_python2x(resource)
     else:
         content = _read_python3x(resource)
-    return M3U8(content, base_uri=base_uri)
+    return M3U8(content, base_uri=base_uri, custom_tags_parser=custom_tags_parser)
 
 
 def _parsed_url(url):
@@ -81,8 +81,8 @@ def _read_python3x(resource):
     )
 
 
-def _load_from_file(uri):
+def _load_from_file(uri, custom_tags_parser=None):
     with open(uri) as fileobj:
         raw_content = fileobj.read().strip()
     base_uri = os.path.dirname(uri)
-    return M3U8(raw_content, base_uri=base_uri)
+    return M3U8(raw_content, base_uri=base_uri, custom_tags_parser=custom_tags_parser)

m3u8/model.py

@@ -136,9 +136,9 @@ class M3U8(object):
         ('discontinuity_sequence', 'discontinuity_sequence')
     )
 
-    def __init__(self, content=None, base_path=None, base_uri=None, strict=False):
+    def __init__(self, content=None, base_path=None, base_uri=None, strict=False, custom_tags_parser=None):
         if content is not None:
-            self.data = parse(content, strict)
+            self.data = parse(content, strict, custom_tags_parser)
         else:
             self.data = {}
         self._base_uri = base_uri

m3u8/parser.py

@@ -35,8 +35,7 @@ def __str__(self):
         return 'Syntax error in manifest on line %d: %s' % (self.lineno, self.line)
 
 
-
-def parse(content, strict=False):
+def parse(content, strict=False, custom_tags_parser=None):
     '''
     Given a M3U8 playlist content returns a dictionary with all data found
     '''
@@ -153,8 +152,8 @@ def parse(content, strict=False):
 
         # Comments and whitespace
         elif line.startswith('#'):
-            # comment
-            pass
+            if callable(custom_tags_parser):
+                custom_tags_parser(line, data, lineno)
 
         elif line.strip() == '':
             # blank lines are legal

tests/playlists.py

@@ -673,6 +673,14 @@
 #EXT-X-ENDLIST
 '''
 
+SIMPLE_PLAYLIST_WITH_CUSTOM_TAGS = '''#EXTM3U
+#EXT-X-MOVIE: million dollar baby
+#EXT-X-TARGETDURATION:5220
+#EXTINF:5220,
+http://media.example.com/entire.ts
+#EXT-X-ENDLIST
+'''
+
 RELATIVE_PLAYLIST_FILENAME = abspath(join(dirname(__file__), 'playlists/relative-playlist.m3u8'))
 
 RELATIVE_PLAYLIST_URI = TEST_HOST + '/path/to/relative-playlist.m3u8'

tests/test_parser.py

@@ -313,3 +313,12 @@ def test_should_parse_start_with_precise():
 def test_simple_playlist_with_discontinuity_sequence():
     data = m3u8.parse(playlists.SIMPLE_PLAYLIST_WITH_DISCONTINUITY_SEQUENCE)
     assert data['discontinuity_sequence'] == 123
+
+def test_simple_playlist_with_custom_tags():
+    def get_movie(line, data, lineno):
+        custom_tag = line.split(':')
+        if len(custom_tag) == 2:
+            data['movie'] = custom_tag[1].strip()
+
+    data = m3u8.parse(playlists.SIMPLE_PLAYLIST_WITH_CUSTOM_TAGS, strict=False, custom_tags_parser=get_movie)
+    assert data['movie'] == 'million dollar baby'