[utils] Add {expected_type} and Iterable support to traverse_obj()

1 files changed, 124 insertions, 29 deletions

diff --git a/test/test_utils.py b/test/test_utils.py
index 5fab05f7c..1fc16ed05 100644
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -79,10 +79,12 @@ from youtube_dl.utils import (
     rot47,
     shell_quote,
     smuggle_url,
+    str_or_none,
     str_to_int,
     strip_jsonp,
     strip_or_none,
     subtitles_filename,
+    T,
     timeconvert,
     traverse_obj,
     try_call,
@@ -1566,6 +1568,7 @@ Line 1
         self.assertEqual(variadic('spam', allowed_types=[dict]), 'spam')
 
     def test_traverse_obj(self):
+        str = compat_str
         _TEST_DATA = {
             100: 100,
             1.2: 1.2,
@@ -1598,8 +1601,8 @@ Line 1
 
         # Test Ellipsis behavior
         self.assertCountEqual(traverse_obj(_TEST_DATA, Ellipsis),
-                              (item for item in _TEST_DATA.values() if item is not None),
-                              msg='`...` should give all values except `None`')
+                              (item for item in _TEST_DATA.values() if item not in (None, {})),
+                              msg='`...` should give all non discarded values')
         self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, Ellipsis)), _TEST_DATA['urls'][0].values(),
                               msg='`...` selection for dicts should select all values')
         self.assertEqual(traverse_obj(_TEST_DATA, (Ellipsis, Ellipsis, 'url')),
@@ -1607,13 +1610,51 @@ Line 1
                          msg='nested `...` queries should work')
         self.assertCountEqual(traverse_obj(_TEST_DATA, (Ellipsis, Ellipsis, 'index')), range(4),
                               msg='`...` query result should be flattened')
+        self.assertEqual(traverse_obj(iter(range(4)), Ellipsis), list(range(4)),
+                         msg='`...` should accept iterables')
 
         # Test function as key
         self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)),
                          [_TEST_DATA['urls']],
                          msg='function as query key should perform a filter based on (key, value)')
-        self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], compat_str)), ('str',),
-                              msg='exceptions in the query function should be caught')
+        self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'},
+                              msg='exceptions in the query function should be catched')
+        self.assertEqual(traverse_obj(iter(range(4)), lambda _, x: x % 2 == 0), [0, 2],
+                         msg='function key should accept iterables')
+        if __debug__:
+            with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
+                traverse_obj(_TEST_DATA, lambda a: Ellipsis)
+            with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
+                traverse_obj(_TEST_DATA, lambda a, b, c: Ellipsis)
+
+        # Test set as key (transformation/type, like `expected_type`)
+        self.assertEqual(traverse_obj(_TEST_DATA, (Ellipsis, T(str.upper), )), ['STR'],
+                         msg='Function in set should be a transformation')
+        self.assertEqual(traverse_obj(_TEST_DATA, (Ellipsis, T(str))), ['str'],
+                         msg='Type in set should be a type filter')
+        self.assertEqual(traverse_obj(_TEST_DATA, T(dict)), _TEST_DATA,
+                         msg='A single set should be wrapped into a path')
+        self.assertEqual(traverse_obj(_TEST_DATA, (Ellipsis, T(str.upper))), ['STR'],
+                         msg='Transformation function should not raise')
+        self.assertEqual(traverse_obj(_TEST_DATA, (Ellipsis, T(str_or_none))),
+                         [item for item in map(str_or_none, _TEST_DATA.values()) if item is not None],
+                         msg='Function in set should be a transformation')
+        if __debug__:
+            with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
+                traverse_obj(_TEST_DATA, set())
+            with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
+                traverse_obj(_TEST_DATA, {str.upper, str})
+
+        # Test `slice` as a key
+        _SLICE_DATA = [0, 1, 2, 3, 4]
+        self.assertEqual(traverse_obj(_TEST_DATA, ('dict', slice(1))), None,
+                         msg='slice on a dictionary should not throw')
+        self.assertEqual(traverse_obj(_SLICE_DATA, slice(1)), _SLICE_DATA[:1],
+                         msg='slice key should apply slice to sequence')
+        self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 2)), _SLICE_DATA[1:2],
+                         msg='slice key should apply slice to sequence')
+        self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 4, 2)), _SLICE_DATA[1:4:2],
+                         msg='slice key should apply slice to sequence')
 
         # Test alternative paths
         self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str',
@@ -1659,15 +1700,23 @@ Line 1
                          {0: ['https://www.example.com/1', 'https://www.example.com/0']},
                          msg='triple nesting in dict path should be treated as branches')
         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {},
-                         msg='remove `None` values when dict key')
+                         msg='remove `None` values when top level dict key fails')
         self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=Ellipsis), {0: Ellipsis},
-                         msg='do not remove `None` values if `default`')
-        self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {0: {}},
-                         msg='do not remove empty values when dict key')
-        self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=Ellipsis), {0: {}},
-                         msg='do not remove empty values when dict key and a default')
-        self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', Ellipsis)}), {0: []},
-                         msg='if branch in dict key not successful, return `[]`')
+                         msg='use `default` if key fails and `default`')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {},
+                         msg='remove empty values when dict key')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=Ellipsis), {0: Ellipsis},
+                         msg='use `default` when dict key and `default`')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}), {},
+                         msg='remove empty values when nested dict key fails')
+        self.assertEqual(traverse_obj(None, {0: 'fail'}), {},
+                         msg='default to dict if pruned')
+        self.assertEqual(traverse_obj(None, {0: 'fail'}, default=Ellipsis), {0: Ellipsis},
+                         msg='default to dict if pruned and default is given')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=Ellipsis), {0: {0: Ellipsis}},
+                         msg='use nested `default` when nested dict key fails and `default`')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', Ellipsis)}), {},
+                         msg='remove key if branch in dict key not successful')
 
         # Testing default parameter behavior
         _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []}
@@ -1691,20 +1740,55 @@ Line 1
                          msg='if branched but not successful return `[]`, not `default`')
         self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', Ellipsis)), [],
                          msg='if branched but object is empty return `[]`, not `default`')
+        self.assertEqual(traverse_obj(None, Ellipsis), [],
+                         msg='if branched but object is `None` return `[]`, not `default`')
+        self.assertEqual(traverse_obj({0: None}, (0, Ellipsis)), [],
+                         msg='if branched but state is `None` return `[]`, not `default`')
+
+        branching_paths = [
+            ('fail', Ellipsis),
+            (Ellipsis, 'fail'),
+            100 * ('fail',) + (Ellipsis,),
+            (Ellipsis,) + 100 * ('fail',),
+        ]
+        for branching_path in branching_paths:
+            self.assertEqual(traverse_obj({}, branching_path), [],
+                             msg='if branched but state is `None`, return `[]` (not `default`)')
+            self.assertEqual(traverse_obj({}, 'fail', branching_path), [],
+                             msg='if branching in last alternative and previous did not match, return `[]` (not `default`)')
+            self.assertEqual(traverse_obj({0: 'x'}, 0, branching_path), 'x',
+                             msg='if branching in last alternative and previous did match, return single value')
+            self.assertEqual(traverse_obj({0: 'x'}, branching_path, 0), 'x',
+                             msg='if branching in first alternative and non-branching path does match, return single value')
+            self.assertEqual(traverse_obj({}, branching_path, 'fail'), None,
+                             msg='if branching in first alternative and non-branching path does not match, return `default`')
 
         # Testing expected_type behavior
         _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0}
-        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=compat_str), 'str',
-                         msg='accept matching `expected_type` type')
-        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int), None,
-                         msg='reject non matching `expected_type` type')
-        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: compat_str(x)), '0',
-                         msg='transform type using type function')
-        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str',
-                                      expected_type=lambda _: 1 / 0), None,
-                         msg='wrap expected_type function in try_call')
-        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, Ellipsis, expected_type=compat_str), ['str'],
-                         msg='eliminate items that expected_type fails on')
+        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str),
+                         'str', msg='accept matching `expected_type` type')
+        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int),
+                         None, msg='reject non matching `expected_type` type')
+        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)),
+                         '0', msg='transform type using type function')
+        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0),
+                         None, msg='wrap expected_type function in try_call')
+        self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, Ellipsis, expected_type=str),
+                         ['str'], msg='eliminate items that expected_type fails on')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int),
+                         {0: 100}, msg='type as expected_type should filter dict values')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none),
+                         {0: '100', 1: '1.2'}, msg='function as expected_type should transform dict values')
+        self.assertEqual(traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int),
+                         1, msg='expected_type should not filter non final dict values')
+        self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int),
+                         {0: {0: 100}}, msg='expected_type should transform deep dict values')
+        self.assertEqual(traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(Ellipsis)),
+                         [{0: Ellipsis}, {0: Ellipsis}], msg='expected_type should transform branched dict values')
+        self.assertEqual(traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int),
+                         [4], msg='expected_type regression for type matching in tuple branching')
+        self.assertEqual(traverse_obj(_TEST_DATA, ['data', Ellipsis], expected_type=int),
+                         [], msg='expected_type regression for type matching in dict result')
 
         # Test get_all behavior
         _GET_ALL_DATA = {'key': [0, 1, 2]}
@@ -1749,14 +1833,23 @@ Line 1
                                       _traverse_string=True), '.',
                          msg='traverse into converted data if `traverse_string`')
         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', Ellipsis),
-                                      _traverse_string=True), list('str'),
-                         msg='`...` branching into string should result in list')
+                                      _traverse_string=True), 'str',
+                         msg='`...` should result in string (same value) if `traverse_string`')
+        self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)),
+                                      _traverse_string=True), 'sr',
+                         msg='`slice` should result in string if `traverse_string`')
+        self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"),
+                                      _traverse_string=True), 'str',
+                         msg='function should result in string if `traverse_string`')
         self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)),
                                       _traverse_string=True), ['s', 'r'],
-                         msg='branching into string should result in list')
-        self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda _, x: x),
-                                      _traverse_string=True), list('str'),
-                         msg='function branching into string should result in list')
+                         msg='branching should result in list if `traverse_string`')
+        self.assertEqual(traverse_obj({}, (0, Ellipsis), _traverse_string=True), [],
+                         msg='branching should result in list if `traverse_string`')
+        self.assertEqual(traverse_obj({}, (0, lambda x, y: True), _traverse_string=True), [],
+                         msg='branching should result in list if `traverse_string`')
+        self.assertEqual(traverse_obj({}, (0, slice(1)), _traverse_string=True), [],
+                         msg='branching should result in list if `traverse_string`')
 
         # Test is_user_input behavior
         _IS_USER_INPUT_DATA = {'range8': list(range(8))}
@@ -1793,6 +1886,8 @@ Line 1
                          msg='failing str key on a `re.Match` should return `default`')
         self.assertEqual(traverse_obj(mobj, 8), None,
                          msg='failing int key on a `re.Match` should return `default`')
+        self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 'group')), ['0123', '3'],
+                         msg='function on a `re.Match` should give group name as well')
 
     def test_get_first(self):
         self.assertEqual(get_first([{'a': None}, {'a': 'spam'}], 'a'), 'spam')