mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-30 12:01:28 +00:00
[extractor] Add _search_json
All fetching of JSON objects should eventually be done with this function but only `youtube` is being refactored for now
This commit is contained in:
parent
00bbc5f177
commit
b7c47b7438
|
@ -442,9 +442,10 @@ class YoutubeWebArchiveIE(InfoExtractor):
|
||||||
'only_matching': True
|
'only_matching': True
|
||||||
},
|
},
|
||||||
]
|
]
|
||||||
_YT_INITIAL_DATA_RE = r'(?:(?:(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*=\s*({.+?})\s*;)|%s)' % YoutubeBaseInfoExtractor._YT_INITIAL_DATA_RE
|
_YT_INITIAL_DATA_RE = YoutubeBaseInfoExtractor._YT_INITIAL_DATA_RE
|
||||||
_YT_INITIAL_PLAYER_RESPONSE_RE = r'(?:(?:(?:window\s*\[\s*["\']ytInitialPlayerResponse["\']\s*\]|ytInitialPlayerResponse)\s*=[(\s]*({.+?})[)\s]*;)|%s)' % YoutubeBaseInfoExtractor._YT_INITIAL_PLAYER_RESPONSE_RE
|
_YT_INITIAL_PLAYER_RESPONSE_RE = fr'''(?x)
|
||||||
_YT_INITIAL_BOUNDARY_RE = r'(?:(?:var\s+meta|</script|\n)|%s)' % YoutubeBaseInfoExtractor._YT_INITIAL_BOUNDARY_RE
|
(?:window\s*\[\s*["\']ytInitialPlayerResponse["\']\s*\]|ytInitialPlayerResponse)\s*=[(\s]*|
|
||||||
|
{YoutubeBaseInfoExtractor._YT_INITIAL_PLAYER_RESPONSE_RE}'''
|
||||||
|
|
||||||
_YT_DEFAULT_THUMB_SERVERS = ['i.ytimg.com'] # thumbnails most likely archived on these servers
|
_YT_DEFAULT_THUMB_SERVERS = ['i.ytimg.com'] # thumbnails most likely archived on these servers
|
||||||
_YT_ALL_THUMB_SERVERS = orderedSet(
|
_YT_ALL_THUMB_SERVERS = orderedSet(
|
||||||
|
@ -474,11 +475,6 @@ def _call_cdx_api(self, item_id, url, filters: list = None, collapse: list = Non
|
||||||
elif not isinstance(res, list) or len(res) != 0:
|
elif not isinstance(res, list) or len(res) != 0:
|
||||||
self.report_warning('Error while parsing CDX API response' + bug_reports_message())
|
self.report_warning('Error while parsing CDX API response' + bug_reports_message())
|
||||||
|
|
||||||
def _extract_yt_initial_variable(self, webpage, regex, video_id, name):
|
|
||||||
return self._parse_json(self._search_regex(
|
|
||||||
(fr'{regex}\s*{self._YT_INITIAL_BOUNDARY_RE}',
|
|
||||||
regex), webpage, name, default='{}'), video_id, fatal=False)
|
|
||||||
|
|
||||||
def _extract_webpage_title(self, webpage):
|
def _extract_webpage_title(self, webpage):
|
||||||
page_title = self._html_extract_title(webpage, default='')
|
page_title = self._html_extract_title(webpage, default='')
|
||||||
# YouTube video pages appear to always have either 'YouTube -' as prefix or '- YouTube' as suffix.
|
# YouTube video pages appear to always have either 'YouTube -' as prefix or '- YouTube' as suffix.
|
||||||
|
@ -488,10 +484,11 @@ def _extract_webpage_title(self, webpage):
|
||||||
|
|
||||||
def _extract_metadata(self, video_id, webpage):
|
def _extract_metadata(self, video_id, webpage):
|
||||||
search_meta = ((lambda x: self._html_search_meta(x, webpage, default=None)) if webpage else (lambda x: None))
|
search_meta = ((lambda x: self._html_search_meta(x, webpage, default=None)) if webpage else (lambda x: None))
|
||||||
player_response = self._extract_yt_initial_variable(
|
player_response = self._search_json(
|
||||||
webpage, self._YT_INITIAL_PLAYER_RESPONSE_RE, video_id, 'initial player response') or {}
|
self._YT_INITIAL_PLAYER_RESPONSE_RE, webpage, 'initial player response',
|
||||||
initial_data = self._extract_yt_initial_variable(
|
video_id, fatal=False)
|
||||||
webpage, self._YT_INITIAL_DATA_RE, video_id, 'initial player response') or {}
|
initial_data = self._search_json(
|
||||||
|
self._YT_INITIAL_DATA_RE, webpage, 'initial data', video_id, fatal=False)
|
||||||
|
|
||||||
initial_data_video = traverse_obj(
|
initial_data_video = traverse_obj(
|
||||||
initial_data, ('contents', 'twoColumnWatchNextResults', 'results', 'results', 'contents', ..., 'videoPrimaryInfoRenderer'),
|
initial_data, ('contents', 'twoColumnWatchNextResults', 'results', 'results', 'contents', ..., 'videoPrimaryInfoRenderer'),
|
||||||
|
|
|
@ -35,6 +35,7 @@
|
||||||
ExtractorError,
|
ExtractorError,
|
||||||
GeoRestrictedError,
|
GeoRestrictedError,
|
||||||
GeoUtils,
|
GeoUtils,
|
||||||
|
LenientJSONDecoder,
|
||||||
RegexNotFoundError,
|
RegexNotFoundError,
|
||||||
UnsupportedError,
|
UnsupportedError,
|
||||||
age_restricted,
|
age_restricted,
|
||||||
|
@ -930,19 +931,10 @@ def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True):
|
||||||
else:
|
else:
|
||||||
self.report_warning(errmsg + str(ve))
|
self.report_warning(errmsg + str(ve))
|
||||||
|
|
||||||
def _parse_json(self, json_string, video_id, transform_source=None, fatal=True, lenient=False):
|
def _parse_json(self, json_string, video_id, transform_source=None, fatal=True, **parser_kwargs):
|
||||||
if transform_source:
|
|
||||||
json_string = transform_source(json_string)
|
|
||||||
try:
|
try:
|
||||||
try:
|
return json.loads(
|
||||||
return json.loads(json_string, strict=False)
|
json_string, cls=LenientJSONDecoder, strict=False, transform_source=transform_source, **parser_kwargs)
|
||||||
except json.JSONDecodeError as e:
|
|
||||||
if not lenient:
|
|
||||||
raise
|
|
||||||
try:
|
|
||||||
return json.loads(json_string[:e.pos], strict=False)
|
|
||||||
except ValueError:
|
|
||||||
raise e
|
|
||||||
except ValueError as ve:
|
except ValueError as ve:
|
||||||
errmsg = f'{video_id}: Failed to parse JSON'
|
errmsg = f'{video_id}: Failed to parse JSON'
|
||||||
if fatal:
|
if fatal:
|
||||||
|
@ -1196,6 +1188,14 @@ def _search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, f
|
||||||
self.report_warning('unable to extract %s' % _name + bug_reports_message())
|
self.report_warning('unable to extract %s' % _name + bug_reports_message())
|
||||||
return None
|
return None
|
||||||
|
|
||||||
|
def _search_json(self, start_pattern, string, name, video_id, *, end_pattern='', fatal=True, **kwargs):
|
||||||
|
"""Searches string for the JSON object specified by start_pattern"""
|
||||||
|
# NB: end_pattern is only used to reduce the size of the initial match
|
||||||
|
return self._parse_json(
|
||||||
|
self._search_regex(rf'{start_pattern}\s*(?P<json>{{.+}})\s*{end_pattern}',
|
||||||
|
string, name, group='json', fatal=fatal) or '{}',
|
||||||
|
video_id, fatal=fatal, ignore_extra=True, **kwargs) or {}
|
||||||
|
|
||||||
def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
|
def _html_search_regex(self, pattern, string, name, default=NO_DEFAULT, fatal=True, flags=0, group=None):
|
||||||
"""
|
"""
|
||||||
Like _search_regex, but strips HTML tags and unescapes entities.
|
Like _search_regex, but strips HTML tags and unescapes entities.
|
||||||
|
|
|
@ -397,9 +397,8 @@ def _check_login_required(self):
|
||||||
if self._LOGIN_REQUIRED and not self._cookies_passed:
|
if self._LOGIN_REQUIRED and not self._cookies_passed:
|
||||||
self.raise_login_required('Login details are needed to download this content', method='cookies')
|
self.raise_login_required('Login details are needed to download this content', method='cookies')
|
||||||
|
|
||||||
_YT_INITIAL_DATA_RE = r'(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*=\s*({.+})\s*;'
|
_YT_INITIAL_DATA_RE = r'(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*='
|
||||||
_YT_INITIAL_PLAYER_RESPONSE_RE = r'ytInitialPlayerResponse\s*=\s*({.+})\s*;'
|
_YT_INITIAL_PLAYER_RESPONSE_RE = r'ytInitialPlayerResponse\s*='
|
||||||
_YT_INITIAL_BOUNDARY_RE = r'(?:var\s+meta|</script|\n)'
|
|
||||||
|
|
||||||
def _get_default_ytcfg(self, client='web'):
|
def _get_default_ytcfg(self, client='web'):
|
||||||
return copy.deepcopy(INNERTUBE_CLIENTS[client])
|
return copy.deepcopy(INNERTUBE_CLIENTS[client])
|
||||||
|
@ -476,12 +475,8 @@ def _call_api(self, ep, query, video_id, fatal=True, headers=None,
|
||||||
data=json.dumps(data).encode('utf8'), headers=real_headers,
|
data=json.dumps(data).encode('utf8'), headers=real_headers,
|
||||||
query={'key': api_key or self._extract_api_key(), 'prettyPrint': 'false'})
|
query={'key': api_key or self._extract_api_key(), 'prettyPrint': 'false'})
|
||||||
|
|
||||||
def extract_yt_initial_data(self, item_id, webpage, fatal=True):
|
def extract_yt_initial_data(self, item_id, webpage):
|
||||||
data = self._search_regex(
|
return self._search_json(self._YT_INITIAL_DATA_RE, webpage, 'yt initial data', item_id, fatal=True)
|
||||||
(fr'{self._YT_INITIAL_DATA_RE}\s*{self._YT_INITIAL_BOUNDARY_RE}',
|
|
||||||
self._YT_INITIAL_DATA_RE), webpage, 'yt initial data', fatal=fatal)
|
|
||||||
if data:
|
|
||||||
return self._parse_json(data, item_id, fatal=fatal)
|
|
||||||
|
|
||||||
def _extract_yt_initial_variable(self, webpage, regex, video_id, name):
|
def _extract_yt_initial_variable(self, webpage, regex, video_id, name):
|
||||||
return self._parse_json(self._search_regex(
|
return self._parse_json(self._search_regex(
|
||||||
|
@ -3052,9 +3047,8 @@ def _get_requested_clients(self, url, smuggled_data):
|
||||||
def _extract_player_responses(self, clients, video_id, webpage, master_ytcfg):
|
def _extract_player_responses(self, clients, video_id, webpage, master_ytcfg):
|
||||||
initial_pr = None
|
initial_pr = None
|
||||||
if webpage:
|
if webpage:
|
||||||
initial_pr = self._extract_yt_initial_variable(
|
initial_pr = self._search_json(
|
||||||
webpage, self._YT_INITIAL_PLAYER_RESPONSE_RE,
|
self._YT_INITIAL_PLAYER_RESPONSE_RE, webpage, 'initial player response', video_id, fatal=False)
|
||||||
video_id, 'initial player response')
|
|
||||||
|
|
||||||
all_clients = set(clients)
|
all_clients = set(clients)
|
||||||
clients = clients[::-1]
|
clients = clients[::-1]
|
||||||
|
@ -3678,9 +3672,8 @@ def process_language(container, base_url, lang_code, sub_name, query):
|
||||||
|
|
||||||
initial_data = None
|
initial_data = None
|
||||||
if webpage:
|
if webpage:
|
||||||
initial_data = self._extract_yt_initial_variable(
|
initial_data = self._search_json(
|
||||||
webpage, self._YT_INITIAL_DATA_RE, video_id,
|
self._YT_INITIAL_DATA_RE, webpage, 'yt initial data', video_id, fatal=False)
|
||||||
'yt initial data')
|
|
||||||
if not initial_data:
|
if not initial_data:
|
||||||
query = {'videoId': video_id}
|
query = {'videoId': video_id}
|
||||||
query.update(self._get_checkok_params())
|
query.update(self._get_checkok_params())
|
||||||
|
|
|
@ -594,6 +594,19 @@ def clean_html(html):
|
||||||
return html.strip()
|
return html.strip()
|
||||||
|
|
||||||
|
|
||||||
|
class LenientJSONDecoder(json.JSONDecoder):
|
||||||
|
def __init__(self, *args, transform_source=None, ignore_extra=False, **kwargs):
|
||||||
|
self.transform_source, self.ignore_extra = transform_source, ignore_extra
|
||||||
|
super().__init__(*args, **kwargs)
|
||||||
|
|
||||||
|
def decode(self, s):
|
||||||
|
if self.transform_source:
|
||||||
|
s = self.transform_source(s)
|
||||||
|
if self.ignore_extra:
|
||||||
|
return self.raw_decode(s.lstrip())[0]
|
||||||
|
return super().decode(s)
|
||||||
|
|
||||||
|
|
||||||
def sanitize_open(filename, open_mode):
|
def sanitize_open(filename, open_mode):
|
||||||
"""Try to open the given filename, and slightly tweak it if this fails.
|
"""Try to open the given filename, and slightly tweak it if this fails.
|
||||||
|
|
||||||
|
|
Loading…
Reference in a new issue