[ie] Add _search_nuxt_json helper (#13386)

* Adds InfoExtractor._search_nuxt_json for webpage extraction
* Adds InfoExtractor._resolve_nuxt_array for direct use with payload JSON
* Adds yt_dlp.utils.jslib module for Python solutions to common JavaScript libraries
* Adds devalue.parse and devalue.parse_iter to jslib utils

Ref:
* 9e503be0f2
* f3fd2aa93d/src/parse.js

Authored by: bashonly, Grub4K

Co-authored-by: Simon Sawicki <contact@grub4k.dev>

yt_dlp/extractor/common.py

@@ -101,6 +101,7 @@ from ..utils import (
     xpath_with_ns,
 )
 from ..utils._utils import _request_dump_filename
+from ..utils.jslib import devalue
 
 
 class InfoExtractor:
@@ -1795,6 +1796,63 @@ class InfoExtractor:
         ret = self._parse_json(js, video_id, transform_source=functools.partial(js_to_json, vars=args), fatal=fatal)
         return traverse_obj(ret, traverse) or {}
 
+    def _resolve_nuxt_array(self, array, video_id, *, fatal=True, default=NO_DEFAULT):
+        """Resolves Nuxt rich JSON payload arrays"""
+        # Ref: https://github.com/nuxt/nuxt/commit/9e503be0f2a24f4df72a3ccab2db4d3e63511f57
+        #      https://github.com/nuxt/nuxt/pull/19205
+        if default is not NO_DEFAULT:
+            fatal = False
+
+        if not isinstance(array, list) or not array:
+            error_msg = 'Unable to resolve Nuxt JSON data: invalid input'
+            if fatal:
+                raise ExtractorError(error_msg, video_id=video_id)
+            elif default is NO_DEFAULT:
+                self.report_warning(error_msg, video_id=video_id)
+            return {} if default is NO_DEFAULT else default
+
+        def indirect_reviver(data):
+            return data
+
+        def json_reviver(data):
+            return json.loads(data)
+
+        gen = devalue.parse_iter(array, revivers={
+            'NuxtError': indirect_reviver,
+            'EmptyShallowRef': json_reviver,
+            'EmptyRef': json_reviver,
+            'ShallowRef': indirect_reviver,
+            'ShallowReactive': indirect_reviver,
+            'Ref': indirect_reviver,
+            'Reactive': indirect_reviver,
+        })
+
+        while True:
+            try:
+                error_msg = f'Error resolving Nuxt JSON: {gen.send(None)}'
+                if fatal:
+                    raise ExtractorError(error_msg, video_id=video_id)
+                elif default is NO_DEFAULT:
+                    self.report_warning(error_msg, video_id=video_id, only_once=True)
+                else:
+                    self.write_debug(f'{video_id}: {error_msg}', only_once=True)
+            except StopIteration as error:
+                return error.value or ({} if default is NO_DEFAULT else default)
+
+    def _search_nuxt_json(self, webpage, video_id, *, fatal=True, default=NO_DEFAULT):
+        """Parses metadata from Nuxt rich JSON payloads embedded in HTML"""
+        passed_default = default is not NO_DEFAULT
+
+        array = self._search_json(
+            r'<script\b[^>]+\bid="__NUXT_DATA__"[^>]*>', webpage,
+            'Nuxt JSON data', video_id, contains_pattern=r'\[(?s:.+)\]',
+            fatal=fatal, default=NO_DEFAULT if not passed_default else None)
+
+        if not array:
+            return default if passed_default else {}
+
+        return self._resolve_nuxt_array(array, video_id, fatal=fatal, default=default)
+
     @staticmethod
     def _hidden_inputs(html):
         html = re.sub(r'<!--(?:(?!<!--).)*-->', '', html)