Merge branch 'master' into yt-live-from-start-range

yt_dlp/utils/_legacy.py

@@ -1,4 +1,6 @@
 """No longer used and new code should not use. Exists only for API compat."""
+import asyncio
+import atexit
 import platform
 import struct
 import sys
@@ -32,6 +34,77 @@ has_certifi = bool(certifi)
 has_websockets = bool(websockets)
 
 
+class WebSocketsWrapper:
+    """Wraps websockets module to use in non-async scopes"""
+    pool = None
+
+    def __init__(self, url, headers=None, connect=True, **ws_kwargs):
+        self.loop = asyncio.new_event_loop()
+        # XXX: "loop" is deprecated
+        self.conn = websockets.connect(
+            url, extra_headers=headers, ping_interval=None,
+            close_timeout=float('inf'), loop=self.loop, ping_timeout=float('inf'), **ws_kwargs)
+        if connect:
+            self.__enter__()
+        atexit.register(self.__exit__, None, None, None)
+
+    def __enter__(self):
+        if not self.pool:
+            self.pool = self.run_with_loop(self.conn.__aenter__(), self.loop)
+        return self
+
+    def send(self, *args):
+        self.run_with_loop(self.pool.send(*args), self.loop)
+
+    def recv(self, *args):
+        return self.run_with_loop(self.pool.recv(*args), self.loop)
+
+    def __exit__(self, type, value, traceback):
+        try:
+            return self.run_with_loop(self.conn.__aexit__(type, value, traceback), self.loop)
+        finally:
+            self.loop.close()
+            self._cancel_all_tasks(self.loop)
+
+    # taken from https://github.com/python/cpython/blob/3.9/Lib/asyncio/runners.py with modifications
+    # for contributors: If there's any new library using asyncio needs to be run in non-async, move these function out of this class
+    @staticmethod
+    def run_with_loop(main, loop):
+        if not asyncio.iscoroutine(main):
+            raise ValueError(f'a coroutine was expected, got {main!r}')
+
+        try:
+            return loop.run_until_complete(main)
+        finally:
+            loop.run_until_complete(loop.shutdown_asyncgens())
+            if hasattr(loop, 'shutdown_default_executor'):
+                loop.run_until_complete(loop.shutdown_default_executor())
+
+    @staticmethod
+    def _cancel_all_tasks(loop):
+        to_cancel = asyncio.all_tasks(loop)
+
+        if not to_cancel:
+            return
+
+        for task in to_cancel:
+            task.cancel()
+
+        # XXX: "loop" is removed in python 3.10+
+        loop.run_until_complete(
+            asyncio.gather(*to_cancel, loop=loop, return_exceptions=True))
+
+        for task in to_cancel:
+            if task.cancelled():
+                continue
+            if task.exception() is not None:
+                loop.call_exception_handler({
+                    'message': 'unhandled exception during asyncio.run() shutdown',
+                    'exception': task.exception(),
+                    'task': task,
+                })
+
+
 def load_plugins(name, suffix, namespace):
     from ..plugins import load_plugins
     ret = load_plugins(name, suffix)

yt_dlp/utils/_utils.py

@@ -1,5 +1,3 @@
-import asyncio
-import atexit
 import base64
 import binascii
 import calendar
@@ -54,7 +52,7 @@ from ..compat import (
     compat_os_name,
     compat_shlex_quote,
 )
-from ..dependencies import websockets, xattr
+from ..dependencies import xattr
 
 __name__ = __name__.rsplit('.', 1)[0]  # Pretend to be the parent module
 
@@ -560,7 +558,7 @@ class LenientJSONDecoder(json.JSONDecoder):
                     s = self._close_object(e)
                     if s is not None:
                         continue
-                raise type(e)(f'{e.msg} in {s[e.pos-10:e.pos+10]!r}', s, e.pos)
+                raise type(e)(f'{e.msg} in {s[e.pos - 10:e.pos + 10]!r}', s, e.pos)
         assert False, 'Too many attempts to decode JSON'
 
 
@@ -638,7 +636,7 @@ def sanitize_filename(s, restricted=False, is_id=NO_DEFAULT):
         elif char in '\\/|*<>':
             return '\0_'
         if restricted and (char in '!&\'()[]{}$;`^,#' or char.isspace() or ord(char) > 127):
-            return '\0_'
+            return '' if unicodedata.category(char)[0] in 'CM' else '\0_'
         return char
 
     # Replace look-alike Unicode glyphs
@@ -1887,6 +1885,7 @@ def setproctitle(title):
     buf = ctypes.create_string_buffer(len(title_bytes))
     buf.value = title_bytes
     try:
+        # PR_SET_NAME = 15      Ref: /usr/include/linux/prctl.h
         libc.prctl(15, buf, 0, 0, 0)
     except AttributeError:
         return  # Strange libc, just skip this
@@ -2267,6 +2266,9 @@ class PagedList:
             raise self.IndexError()
         return entries[0]
 
+    def __bool__(self):
+        return bool(self.getslice(0, 1))
+
 
 class OnDemandPagedList(PagedList):
     """Download pages until a page with less than maximum results"""
@@ -4446,10 +4448,12 @@ def write_xattr(path, key, value):
             raise XAttrMetadataError(e.errno, e.strerror)
         return
 
-    # UNIX Method 1. Use xattrs/pyxattrs modules
+    # UNIX Method 1. Use os.setxattr/xattrs/pyxattrs modules
 
     setxattr = None
-    if getattr(xattr, '_yt_dlp__identifier', None) == 'pyxattr':
+    if callable(getattr(os, 'setxattr', None)):
+        setxattr = os.setxattr
+    elif getattr(xattr, '_yt_dlp__identifier', None) == 'pyxattr':
         # Unicode arguments are not supported in pyxattr until version 0.5.0
         # See https://github.com/ytdl-org/youtube-dl/issues/5498
         if version_tuple(xattr.__version__) >= (0, 5, 0):
@@ -4794,8 +4798,9 @@ def parse_http_range(range):
 
 
 def read_stdin(what):
-    eof = 'Ctrl+Z' if compat_os_name == 'nt' else 'Ctrl+D'
-    write_string(f'Reading {what} from STDIN - EOF ({eof}) to end:\n')
+    if what:
+        eof = 'Ctrl+Z' if compat_os_name == 'nt' else 'Ctrl+D'
+        write_string(f'Reading {what} from STDIN - EOF ({eof}) to end:\n')
     return sys.stdin
 
 
@@ -4926,77 +4931,6 @@ class Config:
         return self.parser.parse_args(self.all_args)
 
 
-class WebSocketsWrapper:
-    """Wraps websockets module to use in non-async scopes"""
-    pool = None
-
-    def __init__(self, url, headers=None, connect=True):
-        self.loop = asyncio.new_event_loop()
-        # XXX: "loop" is deprecated
-        self.conn = websockets.connect(
-            url, extra_headers=headers, ping_interval=None,
-            close_timeout=float('inf'), loop=self.loop, ping_timeout=float('inf'))
-        if connect:
-            self.__enter__()
-        atexit.register(self.__exit__, None, None, None)
-
-    def __enter__(self):
-        if not self.pool:
-            self.pool = self.run_with_loop(self.conn.__aenter__(), self.loop)
-        return self
-
-    def send(self, *args):
-        self.run_with_loop(self.pool.send(*args), self.loop)
-
-    def recv(self, *args):
-        return self.run_with_loop(self.pool.recv(*args), self.loop)
-
-    def __exit__(self, type, value, traceback):
-        try:
-            return self.run_with_loop(self.conn.__aexit__(type, value, traceback), self.loop)
-        finally:
-            self.loop.close()
-            self._cancel_all_tasks(self.loop)
-
-    # taken from https://github.com/python/cpython/blob/3.9/Lib/asyncio/runners.py with modifications
-    # for contributors: If there's any new library using asyncio needs to be run in non-async, move these function out of this class
-    @staticmethod
-    def run_with_loop(main, loop):
-        if not asyncio.iscoroutine(main):
-            raise ValueError(f'a coroutine was expected, got {main!r}')
-
-        try:
-            return loop.run_until_complete(main)
-        finally:
-            loop.run_until_complete(loop.shutdown_asyncgens())
-            if hasattr(loop, 'shutdown_default_executor'):
-                loop.run_until_complete(loop.shutdown_default_executor())
-
-    @staticmethod
-    def _cancel_all_tasks(loop):
-        to_cancel = asyncio.all_tasks(loop)
-
-        if not to_cancel:
-            return
-
-        for task in to_cancel:
-            task.cancel()
-
-        # XXX: "loop" is removed in python 3.10+
-        loop.run_until_complete(
-            asyncio.gather(*to_cancel, loop=loop, return_exceptions=True))
-
-        for task in to_cancel:
-            if task.cancelled():
-                continue
-            if task.exception() is not None:
-                loop.call_exception_handler({
-                    'message': 'unhandled exception during asyncio.run() shutdown',
-                    'exception': task.exception(),
-                    'task': task,
-                })
-
-
 def merge_headers(*dicts):
     """Merge dicts of http headers case insensitively, prioritizing the latter ones"""
     return {k.title(): v for k, v in itertools.chain.from_iterable(map(dict.items, dicts))}
@@ -5145,7 +5079,7 @@ def truncate_string(s, left, right=0):
     assert left > 3 and right >= 0
     if s is None or len(s) <= left + right:
         return s
-    return f'{s[:left-3]}...{s[-right:] if right else ""}'
+    return f'{s[:left - 3]}...{s[-right:] if right else ""}'
 
 
 def orderedSet_from_options(options, alias_dict, *, use_regex=False, start=None):

yt_dlp/utils/networking.py

@@ -67,7 +67,7 @@ class HTTPHeaderDict(collections.UserDict, dict):
     def __setitem__(self, key, value):
         if isinstance(value, bytes):
             value = value.decode('latin-1')
-        super().__setitem__(key.title(), str(value))
+        super().__setitem__(key.title(), str(value).strip())
 
     def __getitem__(self, key):
         return super().__getitem__(key.title())
@@ -123,6 +123,7 @@ def clean_headers(headers: HTTPHeaderDict):
     if 'Youtubedl-No-Compression' in headers:  # compat
         del headers['Youtubedl-No-Compression']
         headers['Accept-Encoding'] = 'identity'
+    headers.pop('Ytdl-socks-proxy', None)
 
 
 def remove_dot_segments(path):

yt_dlp/utils/traversal.py

@@ -3,12 +3,13 @@ import contextlib
 import inspect
 import itertools
 import re
+import xml.etree.ElementTree
 
 from ._utils import (
     IDENTITY,
     NO_DEFAULT,
     LazyList,
-    int_or_none,
+    deprecation_warning,
     is_iterable_like,
     try_call,
     variadic,
@@ -17,13 +18,13 @@ 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
 
     >>> obj = [{}, {"key": "value"}]
     >>> traverse_obj(obj, (1, "key"))
-    "value"
+    'value'
 
     Each of the provided `paths` is tested and the first producing a valid result will be returned.
     The next path will also be tested if the path branched but no results could be found.
@@ -63,10 +64,8 @@ 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 +79,9 @@ 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):
@@ -117,7 +119,7 @@ def traverse_obj(
             branching = True
             if isinstance(obj, collections.abc.Mapping):
                 result = obj.values()
-            elif is_iterable_like(obj):
+            elif is_iterable_like(obj) or isinstance(obj, xml.etree.ElementTree.Element):
                 result = obj
             elif isinstance(obj, re.Match):
                 result = obj.groups()
@@ -131,7 +133,7 @@ def traverse_obj(
             branching = True
             if isinstance(obj, collections.abc.Mapping):
                 iter_obj = obj.items()
-            elif is_iterable_like(obj):
+            elif is_iterable_like(obj) or isinstance(obj, xml.etree.ElementTree.Element):
                 iter_obj = enumerate(obj)
             elif isinstance(obj, re.Match):
                 iter_obj = itertools.chain(
@@ -167,7 +169,7 @@ def traverse_obj(
                 result = next((v for k, v in obj.groupdict().items() if casefold(k) == key), None)
 
         elif isinstance(key, (int, slice)):
-            if is_iterable_like(obj, collections.abc.Sequence):
+            if is_iterable_like(obj, (collections.abc.Sequence, xml.etree.ElementTree.Element)):
                 branching = isinstance(key, slice)
                 with contextlib.suppress(IndexError):
                     result = obj[key]
@@ -175,6 +177,34 @@ def traverse_obj(
                 with contextlib.suppress(IndexError):
                     result = str(obj)[key]
 
+        elif isinstance(obj, xml.etree.ElementTree.Element) and isinstance(key, str):
+            xpath, _, special = key.rpartition('/')
+            if not special.startswith('@') and special != 'text()':
+                xpath = key
+                special = None
+
+            # Allow abbreviations of relative paths, absolute paths error
+            if xpath.startswith('/'):
+                xpath = f'.{xpath}'
+            elif xpath and not xpath.startswith('./'):
+                xpath = f'./{xpath}'
+
+            def apply_specials(element):
+                if special is None:
+                    return element
+                if special == '@':
+                    return element.attrib
+                if special.startswith('@'):
+                    return try_call(element.attrib.get, args=(special[1:],))
+                if special == 'text()':
+                    return element.text
+                assert False, f'apply_specials is missing case for {special!r}'
+
+            if xpath:
+                result = list(map(apply_specials, obj.iterfind(xpath)))
+            else:
+                result = apply_specials(obj)
+
         return branching, result if branching else (result,)
 
     def lazy_last(iterable):
@@ -195,14 +225,6 @@ 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()