2014-03-30 07:02:58 +02:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
import itertools
|
2014-07-24 10:41:12 +02:00
|
|
|
import json
|
2022-08-14 19:45:45 +02:00
|
|
|
import math
|
2015-02-01 22:38:26 +01:00
|
|
|
import operator
|
2014-03-30 07:02:58 +02:00
|
|
|
import re
|
|
|
|
|
|
|
|
from .utils import (
|
2022-08-19 12:45:04 +02:00
|
|
|
error_to_compat_str,
|
2014-03-30 07:02:58 +02:00
|
|
|
ExtractorError,
|
2022-08-14 19:45:45 +02:00
|
|
|
js_to_json,
|
2017-05-18 17:18:33 +02:00
|
|
|
remove_quotes,
|
2022-08-14 19:45:45 +02:00
|
|
|
unified_timestamp,
|
2014-03-30 07:02:58 +02:00
|
|
|
)
|
2021-11-02 12:18:39 +01:00
|
|
|
from .compat import (
|
2022-08-26 09:17:54 +02:00
|
|
|
compat_basestring,
|
2022-08-14 19:45:45 +02:00
|
|
|
compat_collections_chain_map as ChainMap,
|
|
|
|
compat_itertools_zip_longest as zip_longest,
|
2021-11-27 04:18:29 +01:00
|
|
|
compat_str,
|
2021-11-02 12:18:39 +01:00
|
|
|
)
|
|
|
|
|
2022-08-17 15:22:02 +02:00
|
|
|
|
|
|
|
def _js_bit_op(op):
|
|
|
|
|
2022-09-01 00:22:48 +02:00
|
|
|
def zeroise(x):
|
|
|
|
return 0 if x in (None, JS_Undefined) else x
|
|
|
|
|
2022-08-17 15:22:02 +02:00
|
|
|
def wrapped(a, b):
|
2022-09-01 00:22:48 +02:00
|
|
|
return op(zeroise(a), zeroise(b)) & 0xffffffff
|
2022-08-17 15:22:02 +02:00
|
|
|
|
|
|
|
return wrapped
|
|
|
|
|
|
|
|
|
|
|
|
def _js_arith_op(op):
|
|
|
|
|
|
|
|
def wrapped(a, b):
|
2022-08-19 16:34:33 +02:00
|
|
|
if JS_Undefined in (a, b):
|
2022-08-17 15:22:02 +02:00
|
|
|
return float('nan')
|
|
|
|
return op(a or 0, b or 0)
|
|
|
|
|
|
|
|
return wrapped
|
|
|
|
|
|
|
|
|
|
|
|
def _js_div(a, b):
|
2022-08-19 16:34:33 +02:00
|
|
|
if JS_Undefined in (a, b) or not (a and b):
|
2022-08-17 15:22:02 +02:00
|
|
|
return float('nan')
|
2022-09-01 00:22:48 +02:00
|
|
|
return operator.truediv(a or 0, b) if b else float('inf')
|
2022-08-17 15:22:02 +02:00
|
|
|
|
|
|
|
|
|
|
|
def _js_mod(a, b):
|
2022-08-19 16:34:33 +02:00
|
|
|
if JS_Undefined in (a, b) or not b:
|
2022-08-17 15:22:02 +02:00
|
|
|
return float('nan')
|
|
|
|
return (a or 0) % b
|
|
|
|
|
|
|
|
|
|
|
|
def _js_exp(a, b):
|
|
|
|
if not b:
|
2022-08-19 16:34:33 +02:00
|
|
|
return 1 # even 0 ** 0 !!
|
|
|
|
elif JS_Undefined in (a, b):
|
2022-08-17 15:22:02 +02:00
|
|
|
return float('nan')
|
|
|
|
return (a or 0) ** b
|
|
|
|
|
|
|
|
|
|
|
|
def _js_eq_op(op):
|
|
|
|
|
|
|
|
def wrapped(a, b):
|
2022-08-19 16:34:33 +02:00
|
|
|
if set((a, b)) <= set((None, JS_Undefined)):
|
2022-08-17 15:22:02 +02:00
|
|
|
return op(a, a)
|
|
|
|
return op(a, b)
|
|
|
|
|
|
|
|
return wrapped
|
|
|
|
|
|
|
|
|
|
|
|
def _js_comp_op(op):
|
|
|
|
|
|
|
|
def wrapped(a, b):
|
2022-08-19 16:34:33 +02:00
|
|
|
if JS_Undefined in (a, b):
|
2022-08-17 15:22:02 +02:00
|
|
|
return False
|
2022-08-26 09:17:54 +02:00
|
|
|
if isinstance(a, compat_basestring):
|
|
|
|
b = compat_str(b or 0)
|
|
|
|
elif isinstance(b, compat_basestring):
|
|
|
|
a = compat_str(a or 0)
|
2022-08-17 15:22:02 +02:00
|
|
|
return op(a or 0, b or 0)
|
|
|
|
|
|
|
|
return wrapped
|
|
|
|
|
|
|
|
|
2022-08-19 16:34:33 +02:00
|
|
|
def _js_ternary(cndn, if_true=True, if_false=False):
|
|
|
|
"""Simulate JS's ternary operator (cndn?if_true:if_false)"""
|
|
|
|
if cndn in (False, None, 0, '', JS_Undefined):
|
|
|
|
return if_false
|
|
|
|
try:
|
|
|
|
if math.isnan(cndn): # NB: NaN cannot be checked by membership
|
|
|
|
return if_false
|
|
|
|
except TypeError:
|
|
|
|
pass
|
|
|
|
return if_true
|
|
|
|
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
# (op, definition) in order of binding priority, tightest first
|
|
|
|
# avoid dict to maintain order
|
|
|
|
# definition None => Defined in JSInterpreter._operator
|
|
|
|
_OPERATORS = (
|
2022-08-17 15:22:02 +02:00
|
|
|
('>>', _js_bit_op(operator.rshift)),
|
|
|
|
('<<', _js_bit_op(operator.lshift)),
|
|
|
|
('+', _js_arith_op(operator.add)),
|
|
|
|
('-', _js_arith_op(operator.sub)),
|
|
|
|
('*', _js_arith_op(operator.mul)),
|
|
|
|
('%', _js_mod),
|
2022-09-02 17:11:39 +02:00
|
|
|
('/', _js_div),
|
2022-08-17 15:22:02 +02:00
|
|
|
('**', _js_exp),
|
2022-08-14 19:45:45 +02:00
|
|
|
)
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
_COMP_OPERATORS = (
|
|
|
|
('===', operator.is_),
|
|
|
|
('!==', operator.is_not),
|
2022-08-25 13:16:10 +02:00
|
|
|
('==', _js_eq_op(operator.eq)),
|
2022-08-17 15:22:02 +02:00
|
|
|
('!=', _js_eq_op(operator.ne)),
|
|
|
|
('<=', _js_comp_op(operator.le)),
|
|
|
|
('>=', _js_comp_op(operator.ge)),
|
|
|
|
('<', _js_comp_op(operator.lt)),
|
|
|
|
('>', _js_comp_op(operator.gt)),
|
2022-08-14 19:45:45 +02:00
|
|
|
)
|
|
|
|
|
|
|
|
_LOG_OPERATORS = (
|
2022-08-17 15:22:02 +02:00
|
|
|
('|', _js_bit_op(operator.or_)),
|
|
|
|
('^', _js_bit_op(operator.xor)),
|
|
|
|
('&', _js_bit_op(operator.and_)),
|
2022-08-14 19:45:45 +02:00
|
|
|
)
|
|
|
|
|
|
|
|
_SC_OPERATORS = (
|
|
|
|
('?', None),
|
2022-08-17 15:22:02 +02:00
|
|
|
('??', None),
|
2022-08-14 19:45:45 +02:00
|
|
|
('||', None),
|
|
|
|
('&&', None),
|
|
|
|
)
|
|
|
|
|
|
|
|
_OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS))
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2022-08-19 16:34:33 +02:00
|
|
|
_NAME_RE = r'[a-zA-Z_$][\w$]*'
|
2021-12-10 20:14:54 +01:00
|
|
|
_MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]')))
|
2022-08-19 12:45:04 +02:00
|
|
|
_QUOTES = '\'"/'
|
2022-08-14 19:45:45 +02:00
|
|
|
|
|
|
|
|
2022-08-19 16:34:33 +02:00
|
|
|
class JS_Undefined(object):
|
|
|
|
pass
|
2021-12-10 20:14:54 +01:00
|
|
|
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
class JS_Break(ExtractorError):
|
|
|
|
def __init__(self):
|
|
|
|
ExtractorError.__init__(self, 'Invalid break')
|
|
|
|
|
|
|
|
|
|
|
|
class JS_Continue(ExtractorError):
|
|
|
|
def __init__(self):
|
|
|
|
ExtractorError.__init__(self, 'Invalid continue')
|
|
|
|
|
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
class JS_Throw(ExtractorError):
|
|
|
|
def __init__(self, e):
|
|
|
|
self.error = e
|
|
|
|
ExtractorError.__init__(self, 'Uncaught exception ' + error_to_compat_str(e))
|
|
|
|
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
class LocalNameSpace(ChainMap):
|
2022-08-17 15:22:02 +02:00
|
|
|
def __getitem__(self, key):
|
|
|
|
try:
|
|
|
|
return super(LocalNameSpace, self).__getitem__(key)
|
|
|
|
except KeyError:
|
2022-08-19 16:34:33 +02:00
|
|
|
return JS_Undefined
|
2022-08-17 15:22:02 +02:00
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
def __setitem__(self, key, value):
|
2022-08-14 19:45:45 +02:00
|
|
|
for scope in self.maps:
|
2021-11-02 12:18:39 +01:00
|
|
|
if key in scope:
|
|
|
|
scope[key] = value
|
2022-08-14 19:45:45 +02:00
|
|
|
return
|
|
|
|
self.maps[0][key] = value
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
def __delitem__(self, key):
|
|
|
|
raise NotImplementedError('Deleting is not supported')
|
|
|
|
|
|
|
|
def __repr__(self):
|
2022-08-14 19:45:45 +02:00
|
|
|
return 'LocalNameSpace%s' % (self.maps, )
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
|
2014-03-30 07:02:58 +02:00
|
|
|
class JSInterpreter(object):
|
2022-08-14 19:45:45 +02:00
|
|
|
__named_object_counter = 0
|
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
_OBJ_NAME = '__youtube_dl_jsinterp_obj'
|
|
|
|
|
|
|
|
OP_CHARS = None
|
|
|
|
|
2015-02-01 22:38:26 +01:00
|
|
|
def __init__(self, code, objects=None):
|
2022-08-14 19:45:45 +02:00
|
|
|
self.code, self._functions = code, {}
|
|
|
|
self._objects = {} if objects is None else objects
|
2022-08-19 12:45:04 +02:00
|
|
|
if type(self).OP_CHARS is None:
|
|
|
|
type(self).OP_CHARS = self.OP_CHARS = self.__op_chars()
|
2022-08-14 19:45:45 +02:00
|
|
|
|
|
|
|
class Exception(ExtractorError):
|
|
|
|
def __init__(self, msg, *args, **kwargs):
|
|
|
|
expr = kwargs.pop('expr', None)
|
|
|
|
if expr is not None:
|
2023-02-04 02:53:47 +01:00
|
|
|
msg = '{0} in: {1!r:.100}'.format(msg.rstrip(), expr)
|
2022-08-14 19:45:45 +02:00
|
|
|
super(JSInterpreter.Exception, self).__init__(msg, *args, **kwargs)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2023-02-02 15:28:32 +01:00
|
|
|
class JS_RegExp(object):
|
|
|
|
_RE_FLAGS = {
|
|
|
|
# special knowledge: Python's re flags are bitmask values, current max 128
|
|
|
|
# invent new bitmask values well above that for literal parsing
|
|
|
|
# TODO: new pattern class to execute matches with these flags
|
|
|
|
'd': 1024, # Generate indices for substring matches
|
|
|
|
'g': 2048, # Global search
|
|
|
|
'i': re.I, # Case-insensitive search
|
|
|
|
'm': re.M, # Multi-line search
|
|
|
|
's': re.S, # Allows . to match newline characters
|
|
|
|
'u': re.U, # Treat a pattern as a sequence of unicode code points
|
|
|
|
'y': 4096, # Perform a "sticky" search that matches starting at the current position in the target string
|
|
|
|
}
|
|
|
|
|
|
|
|
def __init__(self, pattern_txt, flags=''):
|
|
|
|
if isinstance(flags, compat_str):
|
|
|
|
flags, _ = self.regex_flags(flags)
|
|
|
|
# Thx: https://stackoverflow.com/questions/44773522/setattr-on-python2-sre-sre-pattern
|
|
|
|
# First, avoid https://github.com/python/cpython/issues/74534
|
|
|
|
self.__self = re.compile(pattern_txt.replace('[[', r'[\['), flags)
|
|
|
|
for name in dir(self.__self):
|
|
|
|
# Only these? Obviously __class__, __init__.
|
|
|
|
# PyPy creates a __weakref__ attribute with value None
|
|
|
|
# that can't be setattr'd but also can't need to be copied.
|
|
|
|
if name in ('__class__', '__init__', '__weakref__'):
|
|
|
|
continue
|
|
|
|
setattr(self, name, getattr(self.__self, name))
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def regex_flags(cls, expr):
|
|
|
|
flags = 0
|
|
|
|
if not expr:
|
|
|
|
return flags, expr
|
|
|
|
for idx, ch in enumerate(expr):
|
|
|
|
if ch not in cls._RE_FLAGS:
|
|
|
|
break
|
|
|
|
flags |= cls._RE_FLAGS[ch]
|
|
|
|
return flags, expr[idx + 1:]
|
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
@classmethod
|
|
|
|
def __op_chars(cls):
|
2023-02-02 15:28:32 +01:00
|
|
|
op_chars = set(';,[')
|
2022-08-19 12:45:04 +02:00
|
|
|
for op in cls._all_operators():
|
|
|
|
for c in op[0]:
|
|
|
|
op_chars.add(c)
|
|
|
|
return op_chars
|
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
def _named_object(self, namespace, obj):
|
|
|
|
self.__named_object_counter += 1
|
2022-08-19 12:45:04 +02:00
|
|
|
name = '%s%d' % (self._OBJ_NAME, self.__named_object_counter)
|
2021-11-02 12:18:39 +01:00
|
|
|
namespace[name] = obj
|
|
|
|
return name
|
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
@classmethod
|
|
|
|
def _separate(cls, expr, delim=',', max_split=None, skip_delims=None):
|
2021-11-02 12:18:39 +01:00
|
|
|
if not expr:
|
|
|
|
return
|
2022-08-26 09:17:54 +02:00
|
|
|
# collections.Counter() is ~10% slower in both 2.7 and 3.9
|
2023-02-19 14:47:49 +01:00
|
|
|
counters = dict((k, 0) for k in _MATCHING_PARENS.values())
|
2022-08-19 12:45:04 +02:00
|
|
|
start, splits, pos, delim_len = 0, 0, 0, len(delim) - 1
|
|
|
|
in_quote, escaping, skipping = None, False, 0
|
|
|
|
after_op, in_regex_char_group, skip_re = True, False, 0
|
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
for idx, char in enumerate(expr):
|
2022-08-19 12:45:04 +02:00
|
|
|
if skip_re > 0:
|
|
|
|
skip_re -= 1
|
|
|
|
continue
|
2022-08-14 19:45:45 +02:00
|
|
|
if not in_quote:
|
|
|
|
if char in _MATCHING_PARENS:
|
|
|
|
counters[_MATCHING_PARENS[char]] += 1
|
|
|
|
elif char in counters:
|
|
|
|
counters[char] -= 1
|
2022-09-01 00:22:48 +02:00
|
|
|
if not escaping:
|
|
|
|
if char in _QUOTES and in_quote in (char, None):
|
|
|
|
if in_quote or after_op or char != '/':
|
|
|
|
in_quote = None if in_quote and not in_regex_char_group else char
|
|
|
|
elif in_quote == '/' and char in '[]':
|
|
|
|
in_regex_char_group = char == '['
|
2022-08-19 12:45:04 +02:00
|
|
|
escaping = not escaping and in_quote and char == '\\'
|
2023-02-01 05:09:49 +01:00
|
|
|
after_op = not in_quote and (char in cls.OP_CHARS or (char.isspace() and after_op))
|
2022-08-14 19:45:45 +02:00
|
|
|
|
|
|
|
if char != delim[pos] or any(counters.values()) or in_quote:
|
|
|
|
pos = skipping = 0
|
2021-12-10 20:14:54 +01:00
|
|
|
continue
|
2022-08-14 19:45:45 +02:00
|
|
|
elif skipping > 0:
|
|
|
|
skipping -= 1
|
|
|
|
continue
|
|
|
|
elif pos == 0 and skip_delims:
|
|
|
|
here = expr[idx:]
|
|
|
|
for s in skip_delims if isinstance(skip_delims, (list, tuple)) else [skip_delims]:
|
|
|
|
if here.startswith(s) and s:
|
|
|
|
skipping = len(s) - 1
|
|
|
|
break
|
|
|
|
if skipping > 0:
|
|
|
|
continue
|
|
|
|
if pos < delim_len:
|
2021-12-10 20:14:54 +01:00
|
|
|
pos += 1
|
|
|
|
continue
|
|
|
|
yield expr[start: idx - delim_len]
|
|
|
|
start, pos = idx + 1, 0
|
|
|
|
splits += 1
|
|
|
|
if max_split and splits >= max_split:
|
|
|
|
break
|
2021-11-02 12:18:39 +01:00
|
|
|
yield expr[start:]
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
@classmethod
|
2022-08-26 09:17:54 +02:00
|
|
|
def _separate_at_paren(cls, expr, delim=None):
|
|
|
|
if delim is None:
|
|
|
|
delim = expr and _MATCHING_PARENS[expr[0]]
|
2022-08-14 19:45:45 +02:00
|
|
|
separated = list(cls._separate(expr, delim, 1))
|
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
if len(separated) < 2:
|
2023-02-01 05:09:49 +01:00
|
|
|
raise cls.Exception('No terminating paren {delim} in {expr}'.format(**locals()))
|
2021-11-02 12:18:39 +01:00
|
|
|
return separated[0][1:].strip(), separated[1].strip()
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
@staticmethod
|
|
|
|
def _all_operators():
|
|
|
|
return itertools.chain(
|
2022-08-17 15:22:02 +02:00
|
|
|
# Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence
|
2022-08-14 19:45:45 +02:00
|
|
|
_SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS)
|
|
|
|
|
|
|
|
def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion):
|
|
|
|
if op in ('||', '&&'):
|
2022-08-19 16:34:33 +02:00
|
|
|
if (op == '&&') ^ _js_ternary(left_val):
|
2022-08-14 19:45:45 +02:00
|
|
|
return left_val # short circuiting
|
2022-08-17 15:22:02 +02:00
|
|
|
elif op == '??':
|
2022-08-19 16:34:33 +02:00
|
|
|
if left_val not in (None, JS_Undefined):
|
2022-08-17 15:22:02 +02:00
|
|
|
return left_val
|
2022-08-14 19:45:45 +02:00
|
|
|
elif op == '?':
|
2022-08-19 16:34:33 +02:00
|
|
|
right_expr = _js_ternary(left_val, *self._separate(right_expr, ':', 1))
|
2022-08-14 19:45:45 +02:00
|
|
|
|
|
|
|
right_val = self.interpret_expression(right_expr, local_vars, allow_recursion)
|
|
|
|
opfunc = op and next((v for k, v in self._all_operators() if k == op), None)
|
|
|
|
if not opfunc:
|
|
|
|
return right_val
|
|
|
|
|
|
|
|
try:
|
|
|
|
return opfunc(left_val, right_val)
|
|
|
|
except Exception as e:
|
2023-02-02 15:28:32 +01:00
|
|
|
raise self.Exception('Failed to evaluate {left_val!r:.50} {op} {right_val!r:.50}'.format(**locals()), expr, cause=e)
|
2022-08-14 19:45:45 +02:00
|
|
|
|
2022-08-17 15:22:02 +02:00
|
|
|
def _index(self, obj, idx, allow_undefined=False):
|
2022-08-14 19:45:45 +02:00
|
|
|
if idx == 'length':
|
|
|
|
return len(obj)
|
|
|
|
try:
|
|
|
|
return obj[int(idx)] if isinstance(obj, list) else obj[idx]
|
|
|
|
except Exception as e:
|
2022-08-17 15:22:02 +02:00
|
|
|
if allow_undefined:
|
2022-08-19 16:34:33 +02:00
|
|
|
return JS_Undefined
|
2023-02-02 15:28:32 +01:00
|
|
|
raise self.Exception('Cannot get index {idx:.100}'.format(**locals()), expr=repr(obj), cause=e)
|
2022-08-14 19:45:45 +02:00
|
|
|
|
|
|
|
def _dump(self, obj, namespace):
|
|
|
|
try:
|
|
|
|
return json.dumps(obj)
|
|
|
|
except TypeError:
|
|
|
|
return self._named_object(namespace, obj)
|
|
|
|
|
2015-02-01 22:38:26 +01:00
|
|
|
def interpret_statement(self, stmt, local_vars, allow_recursion=100):
|
2014-03-30 07:02:58 +02:00
|
|
|
if allow_recursion < 0:
|
2022-08-14 19:45:45 +02:00
|
|
|
raise self.Exception('Recursion limit reached')
|
|
|
|
allow_recursion -= 1
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
should_return = False
|
2023-02-02 15:28:32 +01:00
|
|
|
# fails on (eg) if (...) stmt1; else stmt2;
|
2022-08-14 19:45:45 +02:00
|
|
|
sub_statements = list(self._separate(stmt, ';')) or ['']
|
|
|
|
expr = stmt = sub_statements.pop().strip()
|
2021-11-02 12:18:39 +01:00
|
|
|
for sub_stmt in sub_statements:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion)
|
|
|
|
if should_return:
|
|
|
|
return ret, should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
m = re.match(r'(?P<var>(?:var|const|let)\s)|return(?:\s+|(?=["\'])|$)|(?P<throw>throw\s+)', stmt)
|
2022-08-14 19:45:45 +02:00
|
|
|
if m:
|
|
|
|
expr = stmt[len(m.group(0)):].strip()
|
2022-08-19 12:45:04 +02:00
|
|
|
if m.group('throw'):
|
|
|
|
raise JS_Throw(self.interpret_expression(expr, local_vars, allow_recursion))
|
2022-08-14 19:45:45 +02:00
|
|
|
should_return = not m.group('var')
|
|
|
|
if not expr:
|
|
|
|
return None, should_return
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
if expr[0] in _QUOTES:
|
|
|
|
inner, outer = self._separate(expr, expr[0], 1)
|
2022-08-19 12:45:04 +02:00
|
|
|
if expr[0] == '/':
|
2023-02-02 15:28:32 +01:00
|
|
|
flags, outer = self.JS_RegExp.regex_flags(outer)
|
|
|
|
inner = self.JS_RegExp(inner[1:], flags=flags)
|
2022-08-19 12:45:04 +02:00
|
|
|
else:
|
|
|
|
inner = json.loads(js_to_json(inner + expr[0])) # , strict=True))
|
2022-08-14 19:45:45 +02:00
|
|
|
if not outer:
|
|
|
|
return inner, should_return
|
|
|
|
expr = self._named_object(local_vars, inner) + outer
|
|
|
|
|
2023-02-02 15:28:32 +01:00
|
|
|
new_kw, _, obj = expr.partition('new ')
|
|
|
|
if not new_kw:
|
|
|
|
for klass, konstr in (('Date', lambda x: int(unified_timestamp(x, False) * 1000)),
|
|
|
|
('RegExp', self.JS_RegExp),
|
|
|
|
('Error', self.Exception)):
|
|
|
|
if not obj.startswith(klass + '('):
|
|
|
|
continue
|
|
|
|
left, right = self._separate_at_paren(obj[len(klass):])
|
|
|
|
argvals = self.interpret_iter(left, local_vars, allow_recursion)
|
|
|
|
expr = konstr(*argvals)
|
2022-08-14 19:45:45 +02:00
|
|
|
if not expr:
|
2023-02-02 15:28:32 +01:00
|
|
|
raise self.Exception('Failed to parse {klass} {left!r:.100}'.format(**locals()), expr=expr)
|
|
|
|
expr = self._dump(expr, local_vars) + right
|
|
|
|
break
|
2022-08-14 19:45:45 +02:00
|
|
|
else:
|
2023-02-02 15:28:32 +01:00
|
|
|
raise self.Exception('Unsupported object {obj:.100}'.format(**locals()), expr=expr)
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
if expr.startswith('void '):
|
|
|
|
left = self.interpret_expression(expr[5:], local_vars, allow_recursion)
|
|
|
|
return None, should_return
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
if expr.startswith('{'):
|
2022-08-26 09:17:54 +02:00
|
|
|
inner, outer = self._separate_at_paren(expr)
|
2022-08-19 16:34:33 +02:00
|
|
|
# try for object expression (Map)
|
2022-08-17 15:22:02 +02:00
|
|
|
sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)]
|
|
|
|
if all(len(sub_expr) == 2 for sub_expr in sub_expressions):
|
|
|
|
return dict(
|
|
|
|
(key_expr if re.match(_NAME_RE, key_expr) else key_expr,
|
|
|
|
self.interpret_expression(val_expr, local_vars, allow_recursion))
|
|
|
|
for key_expr, val_expr in sub_expressions), should_return
|
|
|
|
# or statement list
|
2022-08-14 19:45:45 +02:00
|
|
|
inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
if not outer or should_abort:
|
2022-08-14 19:45:45 +02:00
|
|
|
return inner, should_abort or should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
else:
|
2022-08-14 19:45:45 +02:00
|
|
|
expr = self._dump(inner, local_vars) + outer
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2015-02-01 22:38:26 +01:00
|
|
|
if expr.startswith('('):
|
2022-08-26 09:17:54 +02:00
|
|
|
inner, outer = self._separate_at_paren(expr)
|
2022-08-14 19:45:45 +02:00
|
|
|
inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
|
|
|
|
if not outer or should_abort:
|
|
|
|
return inner, should_abort or should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
else:
|
2022-08-14 19:45:45 +02:00
|
|
|
expr = self._dump(inner, local_vars) + outer
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
if expr.startswith('['):
|
2022-08-26 09:17:54 +02:00
|
|
|
inner, outer = self._separate_at_paren(expr)
|
2021-11-02 12:18:39 +01:00
|
|
|
name = self._named_object(local_vars, [
|
|
|
|
self.interpret_expression(item, local_vars, allow_recursion)
|
|
|
|
for item in self._separate(inner)])
|
|
|
|
expr = name + outer
|
|
|
|
|
2022-08-19 12:45:04 +02:00
|
|
|
m = re.match(r'''(?x)
|
2022-08-26 09:17:54 +02:00
|
|
|
(?P<try>try)\s*\{|
|
2023-02-01 05:09:49 +01:00
|
|
|
(?P<if>if)\s*\(|
|
2022-08-26 09:17:54 +02:00
|
|
|
(?P<switch>switch)\s*\(|
|
2023-02-02 15:28:32 +01:00
|
|
|
(?P<for>for)\s*\(|
|
|
|
|
(?P<while>while)\s*\(
|
2022-08-26 09:17:54 +02:00
|
|
|
''', expr)
|
2022-08-14 19:45:45 +02:00
|
|
|
md = m.groupdict() if m else {}
|
2023-02-01 05:09:49 +01:00
|
|
|
if md.get('if'):
|
|
|
|
cndn, expr = self._separate_at_paren(expr[m.end() - 1:])
|
2023-02-02 15:28:32 +01:00
|
|
|
if expr.startswith('{'):
|
|
|
|
if_expr, expr = self._separate_at_paren(expr)
|
|
|
|
else:
|
|
|
|
# may lose ... else ... because of ll.368-374
|
|
|
|
if_expr, expr = self._separate_at_paren(expr, delim=';')
|
2023-02-01 05:09:49 +01:00
|
|
|
else_expr = None
|
2023-02-02 15:28:32 +01:00
|
|
|
m = re.match(r'else\s*(?P<block>\{)?', expr)
|
2023-02-01 05:09:49 +01:00
|
|
|
if m:
|
2023-02-02 15:28:32 +01:00
|
|
|
if m.group('block'):
|
|
|
|
else_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
|
|
|
|
else:
|
|
|
|
# handle subset ... else if (...) {...} else ...
|
|
|
|
# TODO: make interpret_statement do this properly, if possible
|
|
|
|
exprs = list(self._separate(expr[m.end():], delim='}', max_split=2))
|
|
|
|
if len(exprs) > 1:
|
|
|
|
if re.match(r'\s*if\s*\(', exprs[0]) and re.match(r'\s*else\b', exprs[1]):
|
|
|
|
else_expr = exprs[0] + '}' + exprs[1]
|
|
|
|
expr = (exprs[2] + '}') if len(exprs) == 3 else None
|
|
|
|
else:
|
|
|
|
else_expr = exprs[0]
|
|
|
|
exprs.append('')
|
|
|
|
expr = '}'.join(exprs[1:])
|
|
|
|
else:
|
|
|
|
else_expr = exprs[0]
|
|
|
|
expr = None
|
|
|
|
else_expr = else_expr.lstrip() + '}'
|
2023-02-01 05:09:49 +01:00
|
|
|
cndn = _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion))
|
|
|
|
ret, should_abort = self.interpret_statement(
|
|
|
|
if_expr if cndn else else_expr, local_vars, allow_recursion)
|
|
|
|
if should_abort:
|
|
|
|
return ret, True
|
|
|
|
|
2023-02-02 15:28:32 +01:00
|
|
|
elif md.get('try'):
|
2022-08-26 09:17:54 +02:00
|
|
|
try_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
|
|
|
|
err = None
|
2022-08-19 12:45:04 +02:00
|
|
|
try:
|
|
|
|
ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion)
|
|
|
|
if should_abort:
|
|
|
|
return ret, True
|
|
|
|
except Exception as e:
|
|
|
|
# XXX: This works for now, but makes debugging future issues very hard
|
2022-08-26 09:17:54 +02:00
|
|
|
err = e
|
|
|
|
|
|
|
|
pending = (None, False)
|
|
|
|
m = re.match(r'catch\s*(?P<err>\(\s*{_NAME_RE}\s*\))?\{{'.format(**globals()), expr)
|
|
|
|
if m:
|
|
|
|
sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
|
|
|
|
if err:
|
|
|
|
catch_vars = {}
|
|
|
|
if m.group('err'):
|
|
|
|
catch_vars[m.group('err')] = err.error if isinstance(err, JS_Throw) else err
|
|
|
|
catch_vars = local_vars.new_child(m=catch_vars)
|
|
|
|
err = None
|
|
|
|
pending = self.interpret_statement(sub_expr, catch_vars, allow_recursion)
|
|
|
|
|
|
|
|
m = re.match(r'finally\s*\{', expr)
|
|
|
|
if m:
|
|
|
|
sub_expr, expr = self._separate_at_paren(expr[m.end() - 1:])
|
|
|
|
ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
|
2022-08-19 12:45:04 +02:00
|
|
|
if should_abort:
|
|
|
|
return ret, True
|
|
|
|
|
2022-08-26 09:17:54 +02:00
|
|
|
ret, should_abort = pending
|
|
|
|
if should_abort:
|
|
|
|
return ret, True
|
2022-08-25 13:16:10 +02:00
|
|
|
|
2022-08-26 09:17:54 +02:00
|
|
|
if err:
|
|
|
|
raise err
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2023-02-02 15:28:32 +01:00
|
|
|
elif md.get('for') or md.get('while'):
|
|
|
|
init_or_cond, remaining = self._separate_at_paren(expr[m.end() - 1:])
|
2021-11-02 12:18:39 +01:00
|
|
|
if remaining.startswith('{'):
|
2022-08-26 09:17:54 +02:00
|
|
|
body, expr = self._separate_at_paren(remaining)
|
2021-11-02 12:18:39 +01:00
|
|
|
else:
|
2022-08-14 19:45:45 +02:00
|
|
|
switch_m = re.match(r'switch\s*\(', remaining) # FIXME
|
|
|
|
if switch_m:
|
2022-08-26 09:17:54 +02:00
|
|
|
switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:])
|
2021-11-02 12:18:39 +01:00
|
|
|
body, expr = self._separate_at_paren(remaining, '}')
|
|
|
|
body = 'switch(%s){%s}' % (switch_val, body)
|
2015-02-01 22:38:26 +01:00
|
|
|
else:
|
2021-11-02 12:18:39 +01:00
|
|
|
body, expr = remaining, ''
|
2023-02-02 15:28:32 +01:00
|
|
|
if md.get('for'):
|
|
|
|
start, cndn, increment = self._separate(init_or_cond, ';')
|
|
|
|
self.interpret_expression(start, local_vars, allow_recursion)
|
|
|
|
else:
|
|
|
|
cndn, increment = init_or_cond, None
|
|
|
|
while _js_ternary(self.interpret_expression(cndn, local_vars, allow_recursion)):
|
2021-11-02 12:18:39 +01:00
|
|
|
try:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
if should_abort:
|
2022-08-14 19:45:45 +02:00
|
|
|
return ret, True
|
2021-11-02 12:18:39 +01:00
|
|
|
except JS_Break:
|
|
|
|
break
|
|
|
|
except JS_Continue:
|
|
|
|
pass
|
2023-02-02 15:28:32 +01:00
|
|
|
if increment:
|
|
|
|
self.interpret_expression(increment, local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
elif md.get('switch'):
|
2022-08-26 09:17:54 +02:00
|
|
|
switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:])
|
2021-11-02 12:18:39 +01:00
|
|
|
switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion)
|
|
|
|
body, expr = self._separate_at_paren(remaining, '}')
|
2021-11-04 13:48:06 +01:00
|
|
|
items = body.replace('default:', 'case default:').split('case ')[1:]
|
|
|
|
for default in (False, True):
|
|
|
|
matched = False
|
|
|
|
for item in items:
|
2022-08-14 19:45:45 +02:00
|
|
|
case, stmt = (i.strip() for i in self._separate(item, ':', 1))
|
2021-11-04 13:48:06 +01:00
|
|
|
if default:
|
|
|
|
matched = matched or case == 'default'
|
|
|
|
elif not matched:
|
|
|
|
matched = (case != 'default'
|
|
|
|
and switch_val == self.interpret_expression(case, local_vars, allow_recursion))
|
|
|
|
if not matched:
|
|
|
|
continue
|
2021-11-02 12:18:39 +01:00
|
|
|
try:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
if should_abort:
|
|
|
|
return ret
|
|
|
|
except JS_Break:
|
2015-02-01 22:38:26 +01:00
|
|
|
break
|
2021-11-04 13:48:06 +01:00
|
|
|
if matched:
|
|
|
|
break
|
2022-08-26 09:17:54 +02:00
|
|
|
|
|
|
|
if md:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
|
|
|
|
return ret, should_abort or should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
# Comma separated statements
|
|
|
|
sub_expressions = list(self._separate(expr))
|
2022-08-14 19:45:45 +02:00
|
|
|
if len(sub_expressions) > 1:
|
|
|
|
for sub_expr in sub_expressions:
|
|
|
|
ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
|
|
|
|
if should_abort:
|
|
|
|
return ret, True
|
|
|
|
return ret, False
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
for m in re.finditer(r'''(?x)
|
2022-08-14 19:45:45 +02:00
|
|
|
(?P<pre_sign>\+\+|--)(?P<var1>{_NAME_RE})|
|
|
|
|
(?P<var2>{_NAME_RE})(?P<post_sign>\+\+|--)'''.format(**globals()), expr):
|
2021-11-02 12:18:39 +01:00
|
|
|
var = m.group('var1') or m.group('var2')
|
|
|
|
start, end = m.span()
|
|
|
|
sign = m.group('pre_sign') or m.group('post_sign')
|
|
|
|
ret = local_vars[var]
|
|
|
|
local_vars[var] += 1 if sign[0] == '+' else -1
|
|
|
|
if m.group('pre_sign'):
|
|
|
|
ret = local_vars[var]
|
2022-08-14 19:45:45 +02:00
|
|
|
expr = expr[:start] + self._dump(ret, local_vars) + expr[end:]
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
if not expr:
|
|
|
|
return None, should_return
|
|
|
|
|
|
|
|
m = re.match(r'''(?x)
|
|
|
|
(?P<assign>
|
|
|
|
(?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?\s*
|
|
|
|
(?P<op>{_OPERATOR_RE})?
|
2022-08-17 15:22:02 +02:00
|
|
|
=(?!=)(?P<expr>.*)$
|
2022-08-14 19:45:45 +02:00
|
|
|
)|(?P<return>
|
|
|
|
(?!if|return|true|false|null|undefined)(?P<name>{_NAME_RE})$
|
|
|
|
)|(?P<indexing>
|
|
|
|
(?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
|
|
|
|
)|(?P<attribute>
|
2022-08-17 15:22:02 +02:00
|
|
|
(?P<var>{_NAME_RE})(?:(?P<nullish>\?)?\.(?P<member>[^(]+)|\[(?P<member2>[^\]]+)\])\s*
|
2022-08-14 19:45:45 +02:00
|
|
|
)|(?P<function>
|
|
|
|
(?P<fname>{_NAME_RE})\((?P<args>.*)\)$
|
|
|
|
)'''.format(**globals()), expr)
|
|
|
|
md = m.groupdict() if m else {}
|
|
|
|
if md.get('assign'):
|
|
|
|
left_val = local_vars.get(m.group('out'))
|
|
|
|
|
|
|
|
if not m.group('index'):
|
|
|
|
local_vars[m.group('out')] = self._operator(
|
|
|
|
m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion)
|
|
|
|
return local_vars[m.group('out')], should_return
|
2022-08-19 16:34:33 +02:00
|
|
|
elif left_val in (None, JS_Undefined):
|
2022-08-14 19:45:45 +02:00
|
|
|
raise self.Exception('Cannot index undefined variable ' + m.group('out'), expr=expr)
|
|
|
|
|
|
|
|
idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion)
|
|
|
|
if not isinstance(idx, (int, float)):
|
|
|
|
raise self.Exception('List index %s must be integer' % (idx, ), expr=expr)
|
|
|
|
idx = int(idx)
|
|
|
|
left_val[idx] = self._operator(
|
2022-08-19 12:45:04 +02:00
|
|
|
m.group('op'), self._index(left_val, idx), m.group('expr'), expr, local_vars, allow_recursion)
|
2022-08-14 19:45:45 +02:00
|
|
|
return left_val[idx], should_return
|
|
|
|
|
|
|
|
elif expr.isdigit():
|
|
|
|
return int(expr), should_return
|
|
|
|
|
|
|
|
elif expr == 'break':
|
2021-11-02 12:18:39 +01:00
|
|
|
raise JS_Break()
|
|
|
|
elif expr == 'continue':
|
|
|
|
raise JS_Continue()
|
|
|
|
|
2022-08-17 15:22:02 +02:00
|
|
|
elif expr == 'undefined':
|
2022-08-19 16:34:33 +02:00
|
|
|
return JS_Undefined, should_return
|
2022-09-01 00:22:48 +02:00
|
|
|
elif expr == 'NaN':
|
|
|
|
return float('NaN'), should_return
|
2022-08-17 15:22:02 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
elif md.get('return'):
|
|
|
|
return local_vars[m.group('name')], should_return
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2014-07-24 10:41:12 +02:00
|
|
|
try:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret = json.loads(js_to_json(expr)) # strict=True)
|
|
|
|
if not md.get('attribute'):
|
|
|
|
return ret, should_return
|
2014-07-24 10:41:12 +02:00
|
|
|
except ValueError:
|
|
|
|
pass
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
if md.get('indexing'):
|
2017-05-18 17:18:33 +02:00
|
|
|
val = local_vars[m.group('in')]
|
2021-11-02 12:18:39 +01:00
|
|
|
idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion)
|
2022-08-14 19:45:45 +02:00
|
|
|
return self._index(val, idx), should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
for op, _ in self._all_operators():
|
|
|
|
# hackety: </> have higher priority than <</>>, but don't confuse them
|
2022-08-17 15:22:02 +02:00
|
|
|
skip_delim = (op + op) if op in '<>*?' else None
|
|
|
|
if op == '?':
|
|
|
|
skip_delim = (skip_delim, '?.')
|
2022-08-14 19:45:45 +02:00
|
|
|
separated = list(self._separate(expr, op, skip_delims=skip_delim))
|
2021-11-02 12:18:39 +01:00
|
|
|
if len(separated) < 2:
|
|
|
|
continue
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
right_expr = separated.pop()
|
|
|
|
while op == '-' and len(separated) > 1 and not separated[-1].strip():
|
|
|
|
right_expr = '-' + right_expr
|
|
|
|
separated.pop()
|
|
|
|
left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion)
|
2022-08-17 15:22:02 +02:00
|
|
|
return self._operator(op, left_val, right_expr, expr, local_vars, allow_recursion), should_return
|
2014-07-15 22:46:39 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
if md.get('attribute'):
|
2022-08-17 15:22:02 +02:00
|
|
|
variable, member, nullish = m.group('var', 'member', 'nullish')
|
2022-08-14 19:45:45 +02:00
|
|
|
if not member:
|
|
|
|
member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
arg_str = expr[m.end():]
|
|
|
|
if arg_str.startswith('('):
|
2022-08-26 09:17:54 +02:00
|
|
|
arg_str, remaining = self._separate_at_paren(arg_str)
|
2014-07-24 10:41:12 +02:00
|
|
|
else:
|
2021-11-02 12:18:39 +01:00
|
|
|
arg_str, remaining = None, arg_str
|
|
|
|
|
|
|
|
def assertion(cndn, msg):
|
|
|
|
""" assert, but without risk of getting optimized out """
|
|
|
|
if not cndn:
|
2022-09-01 00:22:48 +02:00
|
|
|
memb = member
|
2023-02-04 02:53:47 +01:00
|
|
|
raise self.Exception('{memb} {msg}'.format(**locals()), expr=expr)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
def eval_method():
|
2022-08-14 19:45:45 +02:00
|
|
|
if (variable, member) == ('console', 'debug'):
|
|
|
|
return
|
|
|
|
types = {
|
|
|
|
'String': compat_str,
|
|
|
|
'Math': float,
|
|
|
|
}
|
2022-08-17 15:22:02 +02:00
|
|
|
obj = local_vars.get(variable)
|
2022-08-19 16:34:33 +02:00
|
|
|
if obj in (JS_Undefined, None):
|
|
|
|
obj = types.get(variable, JS_Undefined)
|
|
|
|
if obj is JS_Undefined:
|
2022-08-17 15:22:02 +02:00
|
|
|
try:
|
|
|
|
if variable not in self._objects:
|
|
|
|
self._objects[variable] = self.extract_object(variable)
|
|
|
|
obj = self._objects[variable]
|
|
|
|
except self.Exception:
|
|
|
|
if not nullish:
|
|
|
|
raise
|
|
|
|
|
2022-08-19 16:34:33 +02:00
|
|
|
if nullish and obj is JS_Undefined:
|
|
|
|
return JS_Undefined
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
# Member access
|
2021-11-02 12:18:39 +01:00
|
|
|
if arg_str is None:
|
2022-08-17 15:22:02 +02:00
|
|
|
return self._index(obj, member, nullish)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
# Function call
|
|
|
|
argvals = [
|
2014-07-24 10:41:12 +02:00
|
|
|
self.interpret_expression(v, local_vars, allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
for v in self._separate(arg_str)]
|
|
|
|
|
2021-11-27 04:18:29 +01:00
|
|
|
if obj == compat_str:
|
2021-11-02 12:18:39 +01:00
|
|
|
if member == 'fromCharCode':
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
|
|
|
return ''.join(map(chr, argvals))
|
2022-08-14 19:45:45 +02:00
|
|
|
raise self.Exception('Unsupported string method ' + member, expr=expr)
|
|
|
|
elif obj == float:
|
|
|
|
if member == 'pow':
|
|
|
|
assertion(len(argvals) == 2, 'takes two arguments')
|
|
|
|
return argvals[0] ** argvals[1]
|
|
|
|
raise self.Exception('Unsupported Math method ' + member, expr=expr)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
if member == 'split':
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
2022-08-14 19:45:45 +02:00
|
|
|
assertion(len(argvals) == 1, 'with limit argument is not implemented')
|
|
|
|
return obj.split(argvals[0]) if argvals[0] else list(obj)
|
2021-11-02 12:18:39 +01:00
|
|
|
elif member == 'join':
|
|
|
|
assertion(isinstance(obj, list), 'must be applied on a list')
|
|
|
|
assertion(len(argvals) == 1, 'takes exactly one argument')
|
|
|
|
return argvals[0].join(obj)
|
|
|
|
elif member == 'reverse':
|
|
|
|
assertion(not argvals, 'does not take any arguments')
|
|
|
|
obj.reverse()
|
|
|
|
return obj
|
|
|
|
elif member == 'slice':
|
|
|
|
assertion(isinstance(obj, list), 'must be applied on a list')
|
|
|
|
assertion(len(argvals) == 1, 'takes exactly one argument')
|
|
|
|
return obj[argvals[0]:]
|
|
|
|
elif member == 'splice':
|
|
|
|
assertion(isinstance(obj, list), 'must be applied on a list')
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
2021-11-27 03:06:13 +01:00
|
|
|
index, howMany = map(int, (argvals + [len(obj)])[:2])
|
2021-11-02 12:18:39 +01:00
|
|
|
if index < 0:
|
|
|
|
index += len(obj)
|
|
|
|
add_items = argvals[2:]
|
|
|
|
res = []
|
|
|
|
for i in range(index, min(index + howMany, len(obj))):
|
|
|
|
res.append(obj.pop(index))
|
|
|
|
for i, item in enumerate(add_items):
|
|
|
|
obj.insert(index + i, item)
|
|
|
|
return res
|
|
|
|
elif member == 'unshift':
|
|
|
|
assertion(isinstance(obj, list), 'must be applied on a list')
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
|
|
|
for item in reversed(argvals):
|
|
|
|
obj.insert(0, item)
|
|
|
|
return obj
|
|
|
|
elif member == 'pop':
|
|
|
|
assertion(isinstance(obj, list), 'must be applied on a list')
|
|
|
|
assertion(not argvals, 'does not take any arguments')
|
|
|
|
if not obj:
|
|
|
|
return
|
|
|
|
return obj.pop()
|
|
|
|
elif member == 'push':
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
|
|
|
obj.extend(argvals)
|
|
|
|
return obj
|
|
|
|
elif member == 'forEach':
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
|
|
|
assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
|
|
|
|
f, this = (argvals + [''])[:2]
|
2022-08-14 19:45:45 +02:00
|
|
|
return [f((item, idx, obj), {'this': this}, allow_recursion) for idx, item in enumerate(obj)]
|
2021-11-02 12:18:39 +01:00
|
|
|
elif member == 'indexOf':
|
|
|
|
assertion(argvals, 'takes one or more arguments')
|
|
|
|
assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
|
|
|
|
idx, start = (argvals + [0])[:2]
|
|
|
|
try:
|
|
|
|
return obj.index(idx, start)
|
|
|
|
except ValueError:
|
|
|
|
return -1
|
2022-09-01 00:22:48 +02:00
|
|
|
elif member == 'charCodeAt':
|
|
|
|
assertion(isinstance(obj, compat_str), 'must be applied on a string')
|
|
|
|
# assertion(len(argvals) == 1, 'takes exactly one argument') # but not enforced
|
|
|
|
idx = argvals[0] if isinstance(argvals[0], int) else 0
|
|
|
|
if idx >= len(obj):
|
|
|
|
return None
|
|
|
|
return ord(obj[idx])
|
2023-02-02 15:28:32 +01:00
|
|
|
elif member == 'replace':
|
|
|
|
assertion(isinstance(obj, compat_str), 'must be applied on a string')
|
|
|
|
assertion(len(argvals) == 2, 'takes exactly two arguments')
|
|
|
|
return re.sub(argvals[0], argvals[1], obj)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
idx = int(member) if isinstance(obj, list) else member
|
|
|
|
return obj[idx](argvals, allow_recursion=allow_recursion)
|
2021-11-02 12:18:39 +01:00
|
|
|
|
|
|
|
if remaining:
|
2022-08-14 19:45:45 +02:00
|
|
|
ret, should_abort = self.interpret_statement(
|
2021-11-02 12:18:39 +01:00
|
|
|
self._named_object(local_vars, eval_method()) + remaining,
|
|
|
|
local_vars, allow_recursion)
|
2022-08-14 19:45:45 +02:00
|
|
|
return ret, should_return or should_abort
|
2021-11-02 12:18:39 +01:00
|
|
|
else:
|
2022-08-14 19:45:45 +02:00
|
|
|
return eval_method(), should_return
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
elif md.get('function'):
|
|
|
|
fname = m.group('fname')
|
|
|
|
argvals = [self.interpret_expression(v, local_vars, allow_recursion)
|
|
|
|
for v in self._separate(m.group('args'))]
|
2021-11-02 12:18:39 +01:00
|
|
|
if fname in local_vars:
|
2022-08-14 19:45:45 +02:00
|
|
|
return local_vars[fname](argvals, allow_recursion=allow_recursion), should_return
|
2021-11-02 12:18:39 +01:00
|
|
|
elif fname not in self._functions:
|
2016-06-20 13:29:13 +02:00
|
|
|
self._functions[fname] = self.extract_function(fname)
|
2022-08-14 19:45:45 +02:00
|
|
|
return self._functions[fname](argvals, allow_recursion=allow_recursion), should_return
|
2015-02-01 22:38:26 +01:00
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
raise self.Exception(
|
|
|
|
'Unsupported JS expression ' + (expr[:40] if expr != stmt else ''), expr=stmt)
|
|
|
|
|
|
|
|
def interpret_expression(self, expr, local_vars, allow_recursion):
|
|
|
|
ret, should_return = self.interpret_statement(expr, local_vars, allow_recursion)
|
|
|
|
if should_return:
|
|
|
|
raise self.Exception('Cannot return from an expression', expr)
|
|
|
|
return ret
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2023-02-02 15:28:32 +01:00
|
|
|
def interpret_iter(self, list_txt, local_vars, allow_recursion):
|
|
|
|
for v in self._separate(list_txt):
|
|
|
|
yield self.interpret_expression(v, local_vars, allow_recursion)
|
|
|
|
|
2014-07-15 22:46:39 +02:00
|
|
|
def extract_object(self, objname):
|
2017-05-18 17:18:33 +02:00
|
|
|
_FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
|
2014-07-15 22:46:39 +02:00
|
|
|
obj = {}
|
|
|
|
obj_m = re.search(
|
2017-05-18 17:57:38 +02:00
|
|
|
r'''(?x)
|
|
|
|
(?<!this\.)%s\s*=\s*{\s*
|
|
|
|
(?P<fields>(%s\s*:\s*function\s*\(.*?\)\s*{.*?}(?:,\s*)?)*)
|
|
|
|
}\s*;
|
|
|
|
''' % (re.escape(objname), _FUNC_NAME_RE),
|
2014-07-15 22:46:39 +02:00
|
|
|
self.code)
|
2022-08-14 19:45:45 +02:00
|
|
|
if not obj_m:
|
|
|
|
raise self.Exception('Could not find object ' + objname)
|
2014-07-15 22:46:39 +02:00
|
|
|
fields = obj_m.group('fields')
|
|
|
|
# Currently, it only supports function definitions
|
|
|
|
fields_m = re.finditer(
|
2017-05-18 17:57:38 +02:00
|
|
|
r'''(?x)
|
2022-08-14 19:45:45 +02:00
|
|
|
(?P<key>%s)\s*:\s*function\s*\((?P<args>(?:%s|,)*)\){(?P<code>[^}]+)}
|
|
|
|
''' % (_FUNC_NAME_RE, _NAME_RE),
|
2014-07-15 22:46:39 +02:00
|
|
|
fields)
|
|
|
|
for f in fields_m:
|
2022-08-14 19:45:45 +02:00
|
|
|
argnames = self.build_arglist(f.group('args'))
|
2017-05-18 17:18:33 +02:00
|
|
|
obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code'))
|
2014-07-15 22:46:39 +02:00
|
|
|
|
|
|
|
return obj
|
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
def extract_function_code(self, funcname):
|
|
|
|
""" @returns argnames, code """
|
2014-03-30 07:02:58 +02:00
|
|
|
func_m = re.search(
|
2022-08-14 19:45:45 +02:00
|
|
|
r'''(?xs)
|
|
|
|
(?:
|
|
|
|
function\s+%(name)s|
|
|
|
|
[{;,]\s*%(name)s\s*=\s*function|
|
|
|
|
(?:var|const|let)\s+%(name)s\s*=\s*function
|
|
|
|
)\s*
|
2015-02-01 22:38:26 +01:00
|
|
|
\((?P<args>[^)]*)\)\s*
|
2022-08-14 19:45:45 +02:00
|
|
|
(?P<code>{.+})''' % {'name': re.escape(funcname)},
|
2014-03-30 07:02:58 +02:00
|
|
|
self.code)
|
2022-08-26 09:17:54 +02:00
|
|
|
code, _ = self._separate_at_paren(func_m.group('code')) # refine the match
|
2014-03-30 07:15:14 +02:00
|
|
|
if func_m is None:
|
2022-08-14 19:45:45 +02:00
|
|
|
raise self.Exception('Could not find JS function "{funcname}"'.format(**locals()))
|
|
|
|
return self.build_arglist(func_m.group('args')), code
|
2014-03-30 07:02:58 +02:00
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
def extract_function(self, funcname):
|
|
|
|
return self.extract_function_from_code(*self.extract_function_code(funcname))
|
|
|
|
|
|
|
|
def extract_function_from_code(self, argnames, code, *global_stack):
|
|
|
|
local_vars = {}
|
|
|
|
while True:
|
|
|
|
mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
|
|
|
|
if mobj is None:
|
|
|
|
break
|
|
|
|
start, body_start = mobj.span()
|
2022-09-01 00:22:48 +02:00
|
|
|
body, remaining = self._separate_at_paren(code[body_start - 1:], '}')
|
|
|
|
name = self._named_object(local_vars, self.extract_function_from_code(
|
|
|
|
[x.strip() for x in mobj.group('args').split(',')],
|
|
|
|
body, local_vars, *global_stack))
|
2021-11-02 12:18:39 +01:00
|
|
|
code = code[:start] + name + remaining
|
|
|
|
return self.build_function(argnames, code, local_vars, *global_stack)
|
2014-07-15 22:46:39 +02:00
|
|
|
|
2015-02-01 22:38:26 +01:00
|
|
|
def call_function(self, funcname, *args):
|
2021-11-02 12:18:39 +01:00
|
|
|
return self.extract_function(funcname)(args)
|
|
|
|
|
2022-08-14 19:45:45 +02:00
|
|
|
@classmethod
|
|
|
|
def build_arglist(cls, arg_text):
|
|
|
|
if not arg_text:
|
|
|
|
return []
|
2022-08-17 15:22:02 +02:00
|
|
|
|
|
|
|
def valid_arg(y):
|
|
|
|
y = y.strip()
|
|
|
|
if not y:
|
|
|
|
raise cls.Exception('Missing arg in "%s"' % (arg_text, ))
|
|
|
|
return y
|
|
|
|
|
|
|
|
return [valid_arg(x) for x in cls._separate(arg_text)]
|
2022-08-14 19:45:45 +02:00
|
|
|
|
2021-11-02 12:18:39 +01:00
|
|
|
def build_function(self, argnames, code, *global_stack):
|
|
|
|
global_stack = list(global_stack) or [{}]
|
2022-08-14 19:45:45 +02:00
|
|
|
argnames = tuple(argnames)
|
|
|
|
|
|
|
|
def resf(args, kwargs={}, allow_recursion=100):
|
|
|
|
global_stack[0].update(
|
|
|
|
zip_longest(argnames, args, fillvalue=None))
|
|
|
|
global_stack[0].update(kwargs)
|
|
|
|
var_stack = LocalNameSpace(*global_stack)
|
2022-09-01 00:22:48 +02:00
|
|
|
ret, should_abort = self.interpret_statement(code.replace('\n', ' '), var_stack, allow_recursion - 1)
|
2022-08-14 19:45:45 +02:00
|
|
|
if should_abort:
|
|
|
|
return ret
|
2014-03-30 07:02:58 +02:00
|
|
|
return resf
|