depau/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-28 05:01:49 +00:00
Code Issues Releases Wiki Activity

[jsinterp] Various tweaks

Browse source 
* treat Infinity like NaN
* cache operator list
This commit is contained in:
dirkf 2024-06-20 02:57:52 +01:00
parent fd8242e3ef
commit e0094e63c3
1 changed files with 16 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

31
youtube_dl/jsinterp.py
View file

@ -54,15 +54,16 @@ def wraps_op(op):
# NB In principle NaN cannot be checked by membership. # NB In principle NaN cannot be checked by membership.
# Here all NaN values are actually this one, so _NaN is _NaN, # Here all NaN values are actually this one, so _NaN is _NaN,
# although _NaN != _NaN. # although _NaN != _NaN. Ditto Infinity.
_NaN = float('nan') _NaN = float('nan')
_Infinity = float('inf')
def _js_bit_op(op): def _js_bit_op(op):
def zeroise(x): def zeroise(x):
return 0 if x in (None, JS_Undefined, _NaN) else x return 0 if x in (None, JS_Undefined, _NaN, _Infinity) else x
@wraps_op(op) @wraps_op(op)
def wrapped(a, b): def wrapped(a, b):
@ -85,7 +86,7 @@ def _js_arith_op(op):
def _js_div(a, b): def _js_div(a, b):
if JS_Undefined in (a, b) or not (a or b): if JS_Undefined in (a, b) or not (a or b):
return _NaN return _NaN
return operator.truediv(a or 0, b) if b else float('inf') return operator.truediv(a or 0, b) if b else _Infinity
def _js_mod(a, b): def _js_mod(a, b):
@ -344,8 +345,7 @@ class JSInterpreter(object):
def __op_chars(cls): def __op_chars(cls):
op_chars = set(';,[') op_chars = set(';,[')
for op in cls._all_operators(): for op in cls._all_operators():
for c in op[0]: op_chars.update(op[0])
op_chars.add(c)
return op_chars return op_chars
def _named_object(self, namespace, obj): def _named_object(self, namespace, obj):
@ -363,9 +363,8 @@ class JSInterpreter(object):
# collections.Counter() is ~10% slower in both 2.7 and 3.9 # collections.Counter() is ~10% slower in both 2.7 and 3.9
counters = dict((k, 0) for k in _MATCHING_PARENS.values()) counters = dict((k, 0) for k in _MATCHING_PARENS.values())
start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1 start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1
in_quote, escaping, skipping = None, False, 0 in_quote, escaping, after_op, in_regex_char_group = None, False, True, False
after_op, in_regex_char_group = True, False skipping = 0
for idx, char in enumerate(expr): for idx, char in enumerate(expr):
paren_delta = 0 paren_delta = 0
if not in_quote: if not in_quote:
@ -419,10 +418,12 @@ class JSInterpreter(object):
return separated[0][1:].strip(), separated[1].strip() return separated[0][1:].strip(), separated[1].strip()
@staticmethod @staticmethod
def _all_operators(): def _all_operators(_cached=[]):
return itertools.chain( if not _cached:
# Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence _cached.extend(itertools.chain(
_SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS) # Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence
_SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS))
return _cached
def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion): def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion):
if op in ('||', '&&'): if op in ('||', '&&'):
@ -549,7 +550,6 @@ class JSInterpreter(object):
expr = self._dump(inner, local_vars) + outer expr = self._dump(inner, local_vars) + outer
if expr.startswith('('): if expr.startswith('('):
m = re.match(r'\((?P<d>[a-z])%(?P<e>[a-z])\.length\+(?P=e)\.length\)%(?P=e)\.length', expr) m = re.match(r'\((?P<d>[a-z])%(?P<e>[a-z])\.length\+(?P=e)\.length\)%(?P=e)\.length', expr)
if m: if m:
# short-cut eval of frequently used `(d%e.length+e.length)%e.length`, worth ~6% on `pytest -k test_nsig` # short-cut eval of frequently used `(d%e.length+e.length)%e.length`, worth ~6% on `pytest -k test_nsig`
@ -731,7 +731,7 @@ class JSInterpreter(object):
(?P<op>{_OPERATOR_RE})? (?P<op>{_OPERATOR_RE})?
=(?!=)(?P<expr>.*)$ =(?!=)(?P<expr>.*)$
)|(?P<return> )|(?P<return>
(?!if|return|true|false|null|undefined)(?P<name>{_NAME_RE})$ (?!if|return|true|false|null|undefined|NaN|Infinity)(?P<name>{_NAME_RE})$
)|(?P<indexing> )|(?P<indexing>
(?P<in>{_NAME_RE})\[(?P<idx>.+)\]$ (?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
)|(?P<attribute> )|(?P<attribute>
@ -765,11 +765,12 @@ class JSInterpreter(object):
raise JS_Break() raise JS_Break()
elif expr == 'continue': elif expr == 'continue':
raise JS_Continue() raise JS_Continue()
elif expr == 'undefined': elif expr == 'undefined':
return JS_Undefined, should_return return JS_Undefined, should_return
elif expr == 'NaN': elif expr == 'NaN':
return _NaN, should_return return _NaN, should_return
elif expr == 'Infinity':
return _Infinity, should_return
elif md.get('return'): elif md.get('return'):
return local_vars[m.group('name')], should_return return local_vars[m.group('name')], should_return