[YoutubeDL] format spec: correctly handle dashes and other unused operators

'mp4-baseline-16x9' must be handled as a single string, but the '-' was treated as an operator.

diff --git a/test/test_YoutubeDL.py b/test/test_YoutubeDL.py
index 9a3c28f8c101ff99cdc18504974bf1f69c29c410..0388c0bf32a003f41974957ef2f90ab2ab5e0240 100644 (file)
--- a/test/test_YoutubeDL.py
+++ b/test/test_YoutubeDL.py
@@ -105,6 +105,7 @@ class TestFormatSelection(unittest.TestCase):
     def test_format_selection(self):
         formats = [
             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
+            {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
@@ -136,6 +137,11 @@ class TestFormatSelection(unittest.TestCase):
         downloaded = ydl.downloaded_info_dicts[0]
         self.assertEqual(downloaded['format_id'], '35')
 
+        ydl = YDL({'format': 'example-with-dashes'})
+        ydl.process_ie_result(info_dict.copy())
+        downloaded = ydl.downloaded_info_dicts[0]
+        self.assertEqual(downloaded['format_id'], 'example-with-dashes')
+
     def test_format_selection_audio(self):
         formats = [
             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},

diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py
index c608ff91a91636bc40b4fa1c99c013aadcc80820..1446b3254f0085cf933d368b7109723af3720fba 100755 (executable)
--- a/youtube_dl/YoutubeDL.py
+++ b/youtube_dl/YoutubeDL.py
@@ -933,6 +933,37 @@ class YoutubeDL(object):
                 else:
                     filter_parts.append(string)
 
+        def _remove_unused_ops(tokens):
+            # Remove operators that we don't use and join them with the sourrounding strings
+            # for example: 'mp4' '-' 'baseline' '-' '16x9' is converted to 'mp4-baseline-16x9'
+            ALLOWED_OPS = ('/', '+', ',', '(', ')')
+            last_string, last_start, last_end, last_line = None, None, None, None
+            for type, string, start, end, line in tokens:
+                if type == tokenize.OP and string == '[':
+                    if last_string:
+                        yield tokenize.NAME, last_string, last_start, last_end, last_line
+                        last_string = None
+                    yield type, string, start, end, line
+                    # everything inside brackets will be handled by _parse_filter
+                    for type, string, start, end, line in tokens:
+                        yield type, string, start, end, line
+                        if type == tokenize.OP and string == ']':
+                            break
+                elif type == tokenize.OP and string in ALLOWED_OPS:
+                    if last_string:
+                        yield tokenize.NAME, last_string, last_start, last_end, last_line
+                        last_string = None
+                    yield type, string, start, end, line
+                elif type in [tokenize.NAME, tokenize.NUMBER, tokenize.OP]:
+                    if not last_string:
+                        last_string = string
+                        last_start = start
+                        last_end = end
+                    else:
+                        last_string += string
+            if last_string:
+                yield tokenize.NAME, last_string, last_start, last_end, last_line
+
         def _parse_format_selection(tokens, inside_merge=False, inside_choice=False, inside_group=False):
             selectors = []
             current_selector = None
@@ -1111,7 +1142,7 @@ class YoutubeDL(object):
 
         stream = io.BytesIO(format_spec.encode('utf-8'))
         try:
-            tokens = list(compat_tokenize_tokenize(stream.readline))
+            tokens = list(_remove_unused_ops(compat_tokenize_tokenize(stream.readline)))
         except tokenize.TokenError:
             raise syntax_error('Missing closing/opening brackets or parenthesis', (0, len(format_spec)))