[utils] traverse_obj: Move is_user_input into output template (#8673)

Authored by: Grub4K

test/test_utils.py

@@ -2317,23 +2317,6 @@ Line 1
         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))}
-        self.assertEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3'),
-                                      is_user_input=True), 3,
-                         msg='allow for string indexing if `is_user_input`')
-        self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3:'),
-                                           is_user_input=True), tuple(range(8))[3:],
-                              msg='allow for string slice if `is_user_input`')
-        self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':4:2'),
-                                           is_user_input=True), tuple(range(8))[:4:2],
-                              msg='allow step in string slice if `is_user_input`')
-        self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':'),
-                                           is_user_input=True), range(8),
-                              msg='`:` should be treated as `...` if `is_user_input`')
-        with self.assertRaises(TypeError, msg='too many params should result in error'):
-            traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':::'), is_user_input=True)
-
         # Test re.Match as input obj
         mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123')
         self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None],

yt_dlp/YoutubeDL.py

@@ -1201,6 +1201,15 @@ class YoutubeDL:
                 (?:\|(?P<default>.*?))?
             )$''')
 
+        def _from_user_input(field):
+            if field == ':':
+                return ...
+            elif ':' in field:
+                return slice(*map(int_or_none, field.split(':')))
+            elif int_or_none(field) is not None:
+                return int(field)
+            return field
+
         def _traverse_infodict(fields):
             fields = [f for x in re.split(r'\.({.+?})\.?', fields)
                       for f in ([x] if x.startswith('{') else x.split('.'))]
@@ -1210,11 +1219,12 @@ class YoutubeDL:
 
             for i, f in enumerate(fields):
                 if not f.startswith('{'):
+                    fields[i] = _from_user_input(f)
                     continue
                 assert f.endswith('}'), f'No closing brace for {f} in {fields}'
-                fields[i] = {k: k.split('.') for k in f[1:-1].split(',')}
+                fields[i] = {k: list(map(_from_user_input, k.split('.'))) for k in f[1:-1].split(',')}
 
-            return traverse_obj(info_dict, fields, is_user_input=True, traverse_string=True)
+            return traverse_obj(info_dict, fields, traverse_string=True)
 
         def get_value(mdict):
             # Object traversal

yt_dlp/extractor/pr0gramm.py

@@ -4,7 +4,14 @@ from urllib.parse import unquote
 
 from .common import InfoExtractor
 from ..compat import functools
-from ..utils import ExtractorError, make_archive_id, urljoin
+from ..utils import (
+    ExtractorError,
+    float_or_none,
+    int_or_none,
+    make_archive_id,
+    mimetype2ext,
+    urljoin,
+)
 from ..utils.traversal import traverse_obj
 
 
@@ -26,6 +33,7 @@ class Pr0grammIE(InfoExtractor):
             'dislike_count': int,
             'age_limit': 0,
             'thumbnail': r're:^https://thumb\.pr0gramm\.com/.*\.jpg',
+            '_old_archive_ids': ['pr0grammstatic 5466437'],
         },
     }, {
         # Tags require account
@@ -43,6 +51,7 @@ class Pr0grammIE(InfoExtractor):
             'dislike_count': int,
             'age_limit': 0,
             'thumbnail': r're:^https://thumb\.pr0gramm\.com/.*\.jpg',
+            '_old_archive_ids': ['pr0grammstatic 3052805'],
         },
     }, {
         # Requires verified account
@@ -60,6 +69,7 @@ class Pr0grammIE(InfoExtractor):
             'dislike_count': int,
             'age_limit': 18,
             'thumbnail': r're:^https://thumb\.pr0gramm\.com/.*\.jpg',
+            '_old_archive_ids': ['pr0grammstatic 5848332'],
         },
     }, {
         'url': 'https://pr0gramm.com/static/5466437',
@@ -110,37 +120,61 @@ class Pr0grammIE(InfoExtractor):
 
         return data
 
+    @staticmethod
+    def _create_source_url(path):
+        return urljoin('https://img.pr0gramm.com', path)
+
     def _real_extract(self, url):
         video_id = self._match_id(url)
         video_info = traverse_obj(
             self._call_api('get', video_id, {'id': video_id, 'flags': self._maximum_flags}),
             ('items', 0, {dict}))
 
-        source = urljoin('https://img.pr0gramm.com', video_info.get('image'))
+        source = video_info.get('image')
         if not source or not source.endswith('mp4'):
             self.raise_no_formats('Could not extract a video', expected=bool(source), video_id=video_id)
 
         tags = None
         if self._is_logged_in:
-            metadata = self._call_api('info', video_id, {'itemId': video_id})
+            metadata = self._call_api('info', video_id, {'itemId': video_id}, note='Downloading tags')
             tags = traverse_obj(metadata, ('tags', ..., 'tag', {str}))
             # Sorted by "confidence", higher confidence = earlier in list
             confidences = traverse_obj(metadata, ('tags', ..., 'confidence', ({int}, {float})))
             if confidences:
                 tags = [tag for _, tag in sorted(zip(confidences, tags), reverse=True)]
 
+        formats = traverse_obj(video_info, ('variants', ..., {
+            'format_id': ('name', {str}),
+            'url': ('path', {self._create_source_url}),
+            'ext': ('mimeType', {mimetype2ext}),
+            'vcodec': ('codec', {str}),
+            'width': ('width', {int_or_none}),
+            'height': ('height', {int_or_none}),
+            'bitrate': ('bitRate', {float_or_none}),
+            'filesize': ('fileSize', {int_or_none}),
+        })) if video_info.get('variants') else [{
+            'ext': 'mp4',
+            'format_id': 'source',
+            **traverse_obj(video_info, {
+                'url': ('image', {self._create_source_url}),
+                'width': ('width', {int_or_none}),
+                'height': ('height', {int_or_none}),
+            }),
+        }]
+
+        subtitles = {}
+        for subtitle in traverse_obj(video_info, ('subtitles', lambda _, v: v['language'])):
+            subtitles.setdefault(subtitle['language'], []).append(traverse_obj(subtitle, {
+                'url': ('path', {self._create_source_url}),
+                'note': ('label', {str}),
+            }))
+
         return {
             'id': video_id,
             'title': f'pr0gramm-{video_id} by {video_info.get("user")}',
-            'formats': [{
-                'url': source,
-                'ext': 'mp4',
-                **traverse_obj(video_info, {
-                    'width': ('width', {int}),
-                    'height': ('height', {int}),
-                }),
-            }],
             'tags': tags,
+            'formats': formats,
+            'subtitles': subtitles,
             'age_limit': 18 if traverse_obj(video_info, ('flags', {0b110.__and__})) else 0,
             '_old_archive_ids': [make_archive_id('Pr0grammStatic', video_id)],
             **traverse_obj(video_info, {

yt_dlp/utils/traversal.py

@@ -8,7 +8,7 @@ from ._utils import (
     IDENTITY,
     NO_DEFAULT,
     LazyList,
-    int_or_none,
+    deprecation_warning,
     is_iterable_like,
     try_call,
     variadic,
@@ -17,7 +17,7 @@ from ._utils import (
 
 def traverse_obj(
         obj, *paths, default=NO_DEFAULT, expected_type=None, get_all=True,
-        casesense=True, is_user_input=False, traverse_string=False):
+        casesense=True, is_user_input=NO_DEFAULT, traverse_string=False):
     """
     Safely traverse nested `dict`s and `Iterable`s
 
@@ -63,10 +63,8 @@ def traverse_obj(
     @param get_all          If `False`, return the first matching result, otherwise all matching ones.
     @param casesense        If `False`, consider string dictionary keys as case insensitive.
 
-    The following are only meant to be used by YoutubeDL.prepare_outtmpl and are not part of the API
+    `traverse_string` is only meant to be used by YoutubeDL.prepare_outtmpl and is not part of the API
 
-    @param is_user_input    Whether the keys are generated from user input.
-                            If `True` strings get converted to `int`/`slice` if needed.
     @param traverse_string  Whether to traverse into objects as strings.
                             If `True`, any non-compatible object will first be
                             converted into a string and then traversed into.
@@ -80,6 +78,9 @@ def traverse_obj(
                             If no `default` is given and the last path branches, a `list` of results
                             is always returned. If a path ends on a `dict` that result will always be a `dict`.
     """
+    if is_user_input is not NO_DEFAULT:
+        deprecation_warning('The is_user_input parameter is deprecated and no longer works')
+
     casefold = lambda k: k.casefold() if isinstance(k, str) else k
 
     if isinstance(expected_type, type):
@@ -195,14 +196,6 @@ def traverse_obj(
 
         key = None
         for last, key in lazy_last(variadic(path, (str, bytes, dict, set))):
-            if is_user_input and isinstance(key, str):
-                if key == ':':
-                    key = ...
-                elif ':' in key:
-                    key = slice(*map(int_or_none, key.split(':')))
-                elif int_or_none(key) is not None:
-                    key = int(key)
-
             if not casesense and isinstance(key, str):
                 key = key.casefold()