summaryrefslogtreecommitdiff
path: root/youtube_dl/jsinterp.py
blob: 8e119d08a3b42e1b9b7c9d09a9a17072151b37c1 (plain)
    1 from __future__ import unicode_literals
    2 
    3 import itertools
    4 import json
    5 import math
    6 import operator
    7 import re
    8 
    9 from .utils import (
   10     ExtractorError,
   11     js_to_json,
   12     remove_quotes,
   13     unified_timestamp,
   14 )
   15 from .compat import (
   16     compat_collections_chain_map as ChainMap,
   17     compat_itertools_zip_longest as zip_longest,
   18     compat_str,
   19 )
   20 
   21 _NAME_RE = r'[a-zA-Z_$][\w$]*'
   22 
   23 _UNDEFINED = object()
   24 
   25 
   26 def _js_bit_op(op):
   27 
   28     def wrapped(a, b):
   29         def zeroise(x):
   30             return 0 if x in (None, _UNDEFINED) else x
   31         return op(zeroise(a), zeroise(b))
   32 
   33     return wrapped
   34 
   35 
   36 def _js_arith_op(op):
   37 
   38     def wrapped(a, b):
   39         if _UNDEFINED in (a, b):
   40             return float('nan')
   41         return op(a or 0, b or 0)
   42 
   43     return wrapped
   44 
   45 
   46 def _js_div(a, b):
   47     if _UNDEFINED in (a, b) or not (a and b):
   48         return float('nan')
   49     return float('inf') if not b else operator.truediv(a or 0, b)
   50 
   51 
   52 def _js_mod(a, b):
   53     if _UNDEFINED in (a, b) or not b:
   54         return float('nan')
   55     return (a or 0) % b
   56 
   57 
   58 def _js_exp(a, b):
   59     if not b:
   60         # even 0 ** 0 !!
   61         return 1
   62     if _UNDEFINED in (a, b):
   63         return float('nan')
   64     return (a or 0) ** b
   65 
   66 
   67 def _js_eq_op(op):
   68 
   69     def wrapped(a, b):
   70         if set((a, b)) <= set((None, _UNDEFINED)):
   71             return op(a, a)
   72         return op(a, b)
   73 
   74     return wrapped
   75 
   76 
   77 def _js_comp_op(op):
   78 
   79     def wrapped(a, b):
   80         if _UNDEFINED in (a, b):
   81             return False
   82         return op(a or 0, b or 0)
   83 
   84     return wrapped
   85 
   86 
   87 # (op, definition) in order of binding priority, tightest first
   88 # avoid dict to maintain order
   89 # definition None => Defined in JSInterpreter._operator
   90 _DOT_OPERATORS = (
   91     ('.', None),
   92     # TODO: ('?.', None),
   93 )
   94 
   95 _OPERATORS = (
   96     ('>>', _js_bit_op(operator.rshift)),
   97     ('<<', _js_bit_op(operator.lshift)),
   98     ('+', _js_arith_op(operator.add)),
   99     ('-', _js_arith_op(operator.sub)),
  100     ('*', _js_arith_op(operator.mul)),
  101     ('/', _js_div),
  102     ('%', _js_mod),
  103     ('**', _js_exp),
  104 )
  105 
  106 _COMP_OPERATORS = (
  107     ('===', operator.is_),
  108     ('==', _js_eq_op(operator.eq)),
  109     ('!==', operator.is_not),
  110     ('!=', _js_eq_op(operator.ne)),
  111     ('<=', _js_comp_op(operator.le)),
  112     ('>=', _js_comp_op(operator.ge)),
  113     ('<', _js_comp_op(operator.lt)),
  114     ('>', _js_comp_op(operator.gt)),
  115 )
  116 
  117 _LOG_OPERATORS = (
  118     ('|', _js_bit_op(operator.or_)),
  119     ('^', _js_bit_op(operator.xor)),
  120     ('&', _js_bit_op(operator.and_)),
  121 )
  122 
  123 _SC_OPERATORS = (
  124     ('?', None),
  125     ('??', None),
  126     ('||', None),
  127     ('&&', None),
  128 )
  129 
  130 _OPERATOR_RE = '|'.join(map(lambda x: re.escape(x[0]), _OPERATORS + _LOG_OPERATORS))
  131 
  132 _MATCHING_PARENS = dict(zip(*zip('()', '{}', '[]')))
  133 _QUOTES = '\'"'
  134 
  135 
  136 def _ternary(cndn, if_true=True, if_false=False):
  137     """Simulate JS's ternary operator (cndn?if_true:if_false)"""
  138     if cndn in (False, None, 0, '', _UNDEFINED):
  139         return if_false
  140     try:
  141         if math.isnan(cndn):  # NB: NaN cannot be checked by membership
  142             return if_false
  143     except TypeError:
  144         pass
  145     return if_true
  146 
  147 
  148 class JS_Break(ExtractorError):
  149     def __init__(self):
  150         ExtractorError.__init__(self, 'Invalid break')
  151 
  152 
  153 class JS_Continue(ExtractorError):
  154     def __init__(self):
  155         ExtractorError.__init__(self, 'Invalid continue')
  156 
  157 
  158 class LocalNameSpace(ChainMap):
  159     def __getitem__(self, key):
  160         try:
  161             return super(LocalNameSpace, self).__getitem__(key)
  162         except KeyError:
  163             return _UNDEFINED
  164 
  165     def __setitem__(self, key, value):
  166         for scope in self.maps:
  167             if key in scope:
  168                 scope[key] = value
  169                 return
  170         self.maps[0][key] = value
  171 
  172     def __delitem__(self, key):
  173         raise NotImplementedError('Deleting is not supported')
  174 
  175     def __contains__(self, key):
  176         try:
  177             super(LocalNameSpace, self).__getitem__(key)
  178             return True
  179         except KeyError:
  180             return False
  181 
  182     def __repr__(self):
  183         return 'LocalNameSpace%s' % (self.maps, )
  184 
  185 
  186 class JSInterpreter(object):
  187     __named_object_counter = 0
  188 
  189     undefined = _UNDEFINED
  190 
  191     def __init__(self, code, objects=None):
  192         self.code, self._functions = code, {}
  193         self._objects = {} if objects is None else objects
  194 
  195     class Exception(ExtractorError):
  196         def __init__(self, msg, *args, **kwargs):
  197             expr = kwargs.pop('expr', None)
  198             if expr is not None:
  199                 msg = '{0} in: {1!r}'.format(msg.rstrip(), expr[:100])
  200             super(JSInterpreter.Exception, self).__init__(msg, *args, **kwargs)
  201 
  202     def _named_object(self, namespace, obj):
  203         self.__named_object_counter += 1
  204         name = '__youtube_dl_jsinterp_obj%d' % (self.__named_object_counter, )
  205         namespace[name] = obj
  206         return name
  207 
  208     @staticmethod
  209     def _separate(expr, delim=',', max_split=None, skip_delims=None):
  210         if not expr:
  211             return
  212         counters = {k: 0 for k in _MATCHING_PARENS.values()}
  213         start, splits, pos, skipping, delim_len = 0, 0, 0, 0, len(delim) - 1
  214         in_quote, escaping = None, False
  215         for idx, char in enumerate(expr):
  216             if not in_quote:
  217                 if char in _MATCHING_PARENS:
  218                     counters[_MATCHING_PARENS[char]] += 1
  219                 elif char in counters:
  220                     counters[char] -= 1
  221             if not escaping:
  222                 if char in _QUOTES and in_quote in (char, None):
  223                     in_quote = None if in_quote else char
  224                 else:
  225                     escaping = in_quote and char == '\\'
  226             else:
  227                 escaping = False
  228 
  229             if char != delim[pos] or any(counters.values()) or in_quote:
  230                 pos = skipping = 0
  231                 continue
  232             elif skipping > 0:
  233                 skipping -= 1
  234                 continue
  235             elif pos == 0 and skip_delims:
  236                 here = expr[idx:]
  237                 for s in skip_delims if isinstance(skip_delims, (list, tuple)) else [skip_delims]:
  238                     if here.startswith(s) and s:
  239                         skipping = len(s) - 1
  240                         break
  241                 if skipping > 0:
  242                     continue
  243             if pos < delim_len:
  244                 pos += 1
  245                 continue
  246             yield expr[start: idx - delim_len]
  247             start, pos = idx + 1, 0
  248             splits += 1
  249             if max_split and splits >= max_split:
  250                 break
  251         yield expr[start:]
  252 
  253     @classmethod
  254     def _separate_at_paren(cls, expr, delim):
  255         separated = list(cls._separate(expr, delim, 1))
  256 
  257         if len(separated) < 2:
  258             raise cls.Exception('No terminating paren {delim} in {expr}'.format(**locals()))
  259         return separated[0][1:].strip(), separated[1].strip()
  260 
  261     @staticmethod
  262     def _all_operators():
  263         return itertools.chain(
  264             # Ref: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Operator_Precedence
  265             _SC_OPERATORS, _LOG_OPERATORS, _COMP_OPERATORS, _OPERATORS)
  266 
  267     def _operator(self, op, left_val, right_expr, expr, local_vars, allow_recursion):
  268         if op in ('||', '&&'):
  269             if (op == '&&') ^ _ternary(left_val):
  270                 return left_val  # short circuiting
  271         elif op == '??':
  272             if left_val not in (None, self.undefined):
  273                 return left_val
  274         elif op == '?':
  275             right_expr = _ternary(left_val, *self._separate(right_expr, ':', 1))
  276 
  277         right_val = self.interpret_expression(right_expr, local_vars, allow_recursion)
  278         opfunc = op and next((v for k, v in self._all_operators() if k == op), None)
  279         if not opfunc:
  280             return right_val
  281 
  282         try:
  283             return opfunc(left_val, right_val)
  284         except Exception as e:
  285             raise self.Exception('Failed to evaluate {left_val!r} {op} {right_val!r}'.format(**locals()), expr, cause=e)
  286 
  287     def _index(self, obj, idx, allow_undefined=False):
  288         if idx == 'length':
  289             return len(obj)
  290         try:
  291             return obj[int(idx)] if isinstance(obj, list) else obj[idx]
  292         except Exception as e:
  293             if allow_undefined:
  294                 return self.undefined
  295             raise self.Exception('Cannot get index {idx}'.format(**locals()), expr=repr(obj), cause=e)
  296 
  297     def _dump(self, obj, namespace):
  298         try:
  299             return json.dumps(obj)
  300         except TypeError:
  301             return self._named_object(namespace, obj)
  302 
  303     def interpret_statement(self, stmt, local_vars, allow_recursion=100):
  304         if allow_recursion < 0:
  305             raise self.Exception('Recursion limit reached')
  306         allow_recursion -= 1
  307 
  308         should_return = False
  309         sub_statements = list(self._separate(stmt, ';')) or ['']
  310         expr = stmt = sub_statements.pop().strip()
  311         for sub_stmt in sub_statements:
  312             ret, should_return = self.interpret_statement(sub_stmt, local_vars, allow_recursion)
  313             if should_return:
  314                 return ret, should_return
  315 
  316         m = re.match(r'(?P<var>(?:var|const|let)\s)|return(?:\s+|$)', stmt)
  317         if m:
  318             expr = stmt[len(m.group(0)):].strip()
  319             should_return = not m.group('var')
  320         if not expr:
  321             return None, should_return
  322 
  323         if expr[0] in _QUOTES:
  324             inner, outer = self._separate(expr, expr[0], 1)
  325             inner = json.loads(js_to_json(inner + expr[0]))  # , strict=True))
  326             if not outer:
  327                 return inner, should_return
  328             expr = self._named_object(local_vars, inner) + outer
  329 
  330         if expr.startswith('new '):
  331             obj = expr[4:]
  332             if obj.startswith('Date('):
  333                 left, right = self._separate_at_paren(obj[4:], ')')
  334                 expr = unified_timestamp(
  335                     self.interpret_expression(left, local_vars, allow_recursion), False)
  336                 if not expr:
  337                     raise self.Exception('Failed to parse date {left!r}'.format(**locals()), expr=expr)
  338                 expr = self._dump(int(expr * 1000), local_vars) + right
  339             else:
  340                 raise self.Exception('Unsupported object {obj}'.format(**locals()), expr=expr)
  341 
  342         if expr.startswith('void '):
  343             left = self.interpret_expression(expr[5:], local_vars, allow_recursion)
  344             return None, should_return
  345 
  346         if expr.startswith('{'):
  347             inner, outer = self._separate_at_paren(expr, '}')
  348             # try for object expression
  349             sub_expressions = [list(self._separate(sub_expr.strip(), ':', 1)) for sub_expr in self._separate(inner)]
  350             if all(len(sub_expr) == 2 for sub_expr in sub_expressions):
  351                 return dict(
  352                     (key_expr if re.match(_NAME_RE, key_expr) else key_expr,
  353                      self.interpret_expression(val_expr, local_vars, allow_recursion))
  354                     for key_expr, val_expr in sub_expressions), should_return
  355             # or statement list
  356             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
  357             if not outer or should_abort:
  358                 return inner, should_abort or should_return
  359             else:
  360                 expr = self._dump(inner, local_vars) + outer
  361 
  362         if expr.startswith('('):
  363             inner, outer = self._separate_at_paren(expr, ')')
  364             inner, should_abort = self.interpret_statement(inner, local_vars, allow_recursion)
  365             if not outer or should_abort:
  366                 return inner, should_abort or should_return
  367             else:
  368                 expr = self._dump(inner, local_vars) + outer
  369 
  370         if expr.startswith('['):
  371             inner, outer = self._separate_at_paren(expr, ']')
  372             name = self._named_object(local_vars, [
  373                 self.interpret_expression(item, local_vars, allow_recursion)
  374                 for item in self._separate(inner)])
  375             expr = name + outer
  376 
  377         m = re.match(r'(?P<try>try|finally)\s*|(?:(?P<catch>catch)|(?P<for>for)|(?P<switch>switch))\s*\(', expr)
  378         md = m.groupdict() if m else {}
  379         if md.get('try'):
  380             if expr[m.end()] == '{':
  381                 try_expr, expr = self._separate_at_paren(expr[m.end():], '}')
  382             else:
  383                 try_expr, expr = expr[m.end() - 1:], ''
  384             ret, should_abort = self.interpret_statement(try_expr, local_vars, allow_recursion)
  385             if should_abort:
  386                 return ret, True
  387             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
  388             return ret, should_abort or should_return
  389 
  390         elif md.get('catch'):
  391             # We ignore the catch block
  392             _, expr = self._separate_at_paren(expr, '}')
  393             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
  394             return ret, should_abort or should_return
  395 
  396         elif md.get('for'):
  397             constructor, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
  398             if remaining.startswith('{'):
  399                 body, expr = self._separate_at_paren(remaining, '}')
  400             else:
  401                 switch_m = re.match(r'switch\s*\(', remaining)  # FIXME
  402                 if switch_m:
  403                     switch_val, remaining = self._separate_at_paren(remaining[switch_m.end() - 1:], ')')
  404                     body, expr = self._separate_at_paren(remaining, '}')
  405                     body = 'switch(%s){%s}' % (switch_val, body)
  406                 else:
  407                     body, expr = remaining, ''
  408             start, cndn, increment = self._separate(constructor, ';')
  409             self.interpret_expression(start, local_vars, allow_recursion)
  410             while True:
  411                 if not _ternary(self.interpret_expression(cndn, local_vars, allow_recursion)):
  412                     break
  413                 try:
  414                     ret, should_abort = self.interpret_statement(body, local_vars, allow_recursion)
  415                     if should_abort:
  416                         return ret, True
  417                 except JS_Break:
  418                     break
  419                 except JS_Continue:
  420                     pass
  421                 self.interpret_expression(increment, local_vars, allow_recursion)
  422             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
  423             return ret, should_abort or should_return
  424 
  425         elif md.get('switch'):
  426             switch_val, remaining = self._separate_at_paren(expr[m.end() - 1:], ')')
  427             switch_val = self.interpret_expression(switch_val, local_vars, allow_recursion)
  428             body, expr = self._separate_at_paren(remaining, '}')
  429             items = body.replace('default:', 'case default:').split('case ')[1:]
  430             for default in (False, True):
  431                 matched = False
  432                 for item in items:
  433                     case, stmt = (i.strip() for i in self._separate(item, ':', 1))
  434                     if default:
  435                         matched = matched or case == 'default'
  436                     elif not matched:
  437                         matched = (case != 'default'
  438                                    and switch_val == self.interpret_expression(case, local_vars, allow_recursion))
  439                     if not matched:
  440                         continue
  441                     try:
  442                         ret, should_abort = self.interpret_statement(stmt, local_vars, allow_recursion)
  443                         if should_abort:
  444                             return ret
  445                     except JS_Break:
  446                         break
  447                 if matched:
  448                     break
  449             ret, should_abort = self.interpret_statement(expr, local_vars, allow_recursion)
  450             return ret, should_abort or should_return
  451 
  452         # Comma separated statements
  453         sub_expressions = list(self._separate(expr))
  454         if len(sub_expressions) > 1:
  455             for sub_expr in sub_expressions:
  456                 ret, should_abort = self.interpret_statement(sub_expr, local_vars, allow_recursion)
  457                 if should_abort:
  458                     return ret, True
  459             return ret, False
  460 
  461         for m in re.finditer(r'''(?x)
  462                 (?P<pre_sign>\+\+|--)(?P<var1>{_NAME_RE})|
  463                 (?P<var2>{_NAME_RE})(?P<post_sign>\+\+|--)'''.format(**globals()), expr):
  464             var = m.group('var1') or m.group('var2')
  465             start, end = m.span()
  466             sign = m.group('pre_sign') or m.group('post_sign')
  467             ret = local_vars[var]
  468             local_vars[var] += 1 if sign[0] == '+' else -1
  469             if m.group('pre_sign'):
  470                 ret = local_vars[var]
  471             expr = expr[:start] + self._dump(ret, local_vars) + expr[end:]
  472 
  473         if not expr:
  474             return None, should_return
  475 
  476         m = re.match(r'''(?x)
  477             (?P<assign>
  478                 (?P<out>{_NAME_RE})(?:\[(?P<index>[^\]]+?)\])?\s*
  479                 (?P<op>{_OPERATOR_RE})?
  480                 =(?!=)(?P<expr>.*)$
  481             )|(?P<return>
  482                 (?!if|return|true|false|null|undefined)(?P<name>{_NAME_RE})$
  483             )|(?P<indexing>
  484                 (?P<in>{_NAME_RE})\[(?P<idx>.+)\]$
  485             )|(?P<attribute>
  486                 (?P<var>{_NAME_RE})(?:(?P<nullish>\?)?\.(?P<member>[^(]+)|\[(?P<member2>[^\]]+)\])\s*
  487             )|(?P<function>
  488                 (?P<fname>{_NAME_RE})\((?P<args>.*)\)$
  489             )'''.format(**globals()), expr)
  490         md = m.groupdict() if m else {}
  491         if md.get('assign'):
  492             left_val = local_vars.get(m.group('out'))
  493 
  494             if not m.group('index'):
  495                 local_vars[m.group('out')] = self._operator(
  496                     m.group('op'), left_val, m.group('expr'), expr, local_vars, allow_recursion)
  497                 return local_vars[m.group('out')], should_return
  498             elif left_val in (None, self.undefined):
  499                 raise self.Exception('Cannot index undefined variable ' + m.group('out'), expr=expr)
  500 
  501             idx = self.interpret_expression(m.group('index'), local_vars, allow_recursion)
  502             if not isinstance(idx, (int, float)):
  503                 raise self.Exception('List index %s must be integer' % (idx, ), expr=expr)
  504             idx = int(idx)
  505             left_val[idx] = self._operator(
  506                 m.group('op'), left_val[idx], m.group('expr'), expr, local_vars, allow_recursion)
  507             return left_val[idx], should_return
  508 
  509         elif expr.isdigit():
  510             return int(expr), should_return
  511 
  512         elif expr == 'break':
  513             raise JS_Break()
  514         elif expr == 'continue':
  515             raise JS_Continue()
  516 
  517         elif expr == 'undefined':
  518             return self.undefined, should_return
  519 
  520         elif md.get('return'):
  521             return local_vars[m.group('name')], should_return
  522 
  523         try:
  524             ret = json.loads(js_to_json(expr))  # strict=True)
  525             if not md.get('attribute'):
  526                 return ret, should_return
  527         except ValueError:
  528             pass
  529 
  530         if md.get('indexing'):
  531             val = local_vars[m.group('in')]
  532             idx = self.interpret_expression(m.group('idx'), local_vars, allow_recursion)
  533             return self._index(val, idx), should_return
  534 
  535         for op, _ in self._all_operators():
  536             # hackety: </> have higher priority than <</>>, but don't confuse them
  537             skip_delim = (op + op) if op in '<>*?' else None
  538             if op == '?':
  539                 skip_delim = (skip_delim, '?.')
  540             separated = list(self._separate(expr, op, skip_delims=skip_delim))
  541             if len(separated) < 2:
  542                 continue
  543 
  544             right_expr = separated.pop()
  545             while op == '-' and len(separated) > 1 and not separated[-1].strip():
  546                 right_expr = '-' + right_expr
  547                 separated.pop()
  548             left_val = self.interpret_expression(op.join(separated), local_vars, allow_recursion)
  549             return self._operator(op, left_val, right_expr, expr, local_vars, allow_recursion), should_return
  550 
  551         if md.get('attribute'):
  552             variable, member, nullish = m.group('var', 'member', 'nullish')
  553             if not member:
  554                 member = self.interpret_expression(m.group('member2'), local_vars, allow_recursion)
  555             arg_str = expr[m.end():]
  556             if arg_str.startswith('('):
  557                 arg_str, remaining = self._separate_at_paren(arg_str, ')')
  558             else:
  559                 arg_str, remaining = None, arg_str
  560 
  561             def assertion(cndn, msg):
  562                 """ assert, but without risk of getting optimized out """
  563                 if not cndn:
  564                     raise ExtractorError('{member} {msg}'.format(**locals()), expr=expr)
  565 
  566             def eval_method():
  567                 if (variable, member) == ('console', 'debug'):
  568                     return
  569                 types = {
  570                     'String': compat_str,
  571                     'Math': float,
  572                 }
  573                 obj = local_vars.get(variable)
  574                 if obj in (self.undefined, None):
  575                     obj = types.get(variable, self.undefined)
  576                 if obj is self.undefined:
  577                     try:
  578                         if variable not in self._objects:
  579                             self._objects[variable] = self.extract_object(variable)
  580                         obj = self._objects[variable]
  581                     except self.Exception:
  582                         if not nullish:
  583                             raise
  584 
  585                 if nullish and obj is self.undefined:
  586                     return self.undefined
  587 
  588                 # Member access
  589                 if arg_str is None:
  590                     return self._index(obj, member, nullish)
  591 
  592                 # Function call
  593                 argvals = [
  594                     self.interpret_expression(v, local_vars, allow_recursion)
  595                     for v in self._separate(arg_str)]
  596 
  597                 if obj == compat_str:
  598                     if member == 'fromCharCode':
  599                         assertion(argvals, 'takes one or more arguments')
  600                         return ''.join(map(chr, argvals))
  601                     raise self.Exception('Unsupported string method ' + member, expr=expr)
  602                 elif obj == float:
  603                     if member == 'pow':
  604                         assertion(len(argvals) == 2, 'takes two arguments')
  605                         return argvals[0] ** argvals[1]
  606                     raise self.Exception('Unsupported Math method ' + member, expr=expr)
  607 
  608                 if member == 'split':
  609                     assertion(argvals, 'takes one or more arguments')
  610                     assertion(len(argvals) == 1, 'with limit argument is not implemented')
  611                     return obj.split(argvals[0]) if argvals[0] else list(obj)
  612                 elif member == 'join':
  613                     assertion(isinstance(obj, list), 'must be applied on a list')
  614                     assertion(len(argvals) == 1, 'takes exactly one argument')
  615                     return argvals[0].join(obj)
  616                 elif member == 'reverse':
  617                     assertion(not argvals, 'does not take any arguments')
  618                     obj.reverse()
  619                     return obj
  620                 elif member == 'slice':
  621                     assertion(isinstance(obj, list), 'must be applied on a list')
  622                     assertion(len(argvals) == 1, 'takes exactly one argument')
  623                     return obj[argvals[0]:]
  624                 elif member == 'splice':
  625                     assertion(isinstance(obj, list), 'must be applied on a list')
  626                     assertion(argvals, 'takes one or more arguments')
  627                     index, howMany = map(int, (argvals + [len(obj)])[:2])
  628                     if index < 0:
  629                         index += len(obj)
  630                     add_items = argvals[2:]
  631                     res = []
  632                     for i in range(index, min(index + howMany, len(obj))):
  633                         res.append(obj.pop(index))
  634                     for i, item in enumerate(add_items):
  635                         obj.insert(index + i, item)
  636                     return res
  637                 elif member == 'unshift':
  638                     assertion(isinstance(obj, list), 'must be applied on a list')
  639                     assertion(argvals, 'takes one or more arguments')
  640                     for item in reversed(argvals):
  641                         obj.insert(0, item)
  642                     return obj
  643                 elif member == 'pop':
  644                     assertion(isinstance(obj, list), 'must be applied on a list')
  645                     assertion(not argvals, 'does not take any arguments')
  646                     if not obj:
  647                         return
  648                     return obj.pop()
  649                 elif member == 'push':
  650                     assertion(argvals, 'takes one or more arguments')
  651                     obj.extend(argvals)
  652                     return obj
  653                 elif member == 'forEach':
  654                     assertion(argvals, 'takes one or more arguments')
  655                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
  656                     f, this = (argvals + [''])[:2]
  657                     return [f((item, idx, obj), {'this': this}, allow_recursion) for idx, item in enumerate(obj)]
  658                 elif member == 'indexOf':
  659                     assertion(argvals, 'takes one or more arguments')
  660                     assertion(len(argvals) <= 2, 'takes at-most 2 arguments')
  661                     idx, start = (argvals + [0])[:2]
  662                     try:
  663                         return obj.index(idx, start)
  664                     except ValueError:
  665                         return -1
  666 
  667                 idx = int(member) if isinstance(obj, list) else member
  668                 return obj[idx](argvals, allow_recursion=allow_recursion)
  669 
  670             if remaining:
  671                 ret, should_abort = self.interpret_statement(
  672                     self._named_object(local_vars, eval_method()) + remaining,
  673                     local_vars, allow_recursion)
  674                 return ret, should_return or should_abort
  675             else:
  676                 return eval_method(), should_return
  677 
  678         elif md.get('function'):
  679             fname = m.group('fname')
  680             argvals = [self.interpret_expression(v, local_vars, allow_recursion)
  681                        for v in self._separate(m.group('args'))]
  682             if fname in local_vars:
  683                 return local_vars[fname](argvals, allow_recursion=allow_recursion), should_return
  684             elif fname not in self._functions:
  685                 self._functions[fname] = self.extract_function(fname)
  686             return self._functions[fname](argvals, allow_recursion=allow_recursion), should_return
  687 
  688         raise self.Exception(
  689             'Unsupported JS expression ' + (expr[:40] if expr != stmt else ''), expr=stmt)
  690 
  691     def interpret_expression(self, expr, local_vars, allow_recursion):
  692         ret, should_return = self.interpret_statement(expr, local_vars, allow_recursion)
  693         if should_return:
  694             raise self.Exception('Cannot return from an expression', expr)
  695         return ret
  696 
  697     def extract_object(self, objname):
  698         _FUNC_NAME_RE = r'''(?:[a-zA-Z$0-9]+|"[a-zA-Z$0-9]+"|'[a-zA-Z$0-9]+')'''
  699         obj = {}
  700         obj_m = re.search(
  701             r'''(?x)
  702                 (?<!this\.)%s\s*=\s*{\s*
  703                     (?P<fields>(%s\s*:\s*function\s*\(.*?\)\s*{.*?}(?:,\s*)?)*)
  704                 }\s*;
  705             ''' % (re.escape(objname), _FUNC_NAME_RE),
  706             self.code)
  707         if not obj_m:
  708             raise self.Exception('Could not find object ' + objname)
  709         fields = obj_m.group('fields')
  710         # Currently, it only supports function definitions
  711         fields_m = re.finditer(
  712             r'''(?x)
  713                 (?P<key>%s)\s*:\s*function\s*\((?P<args>(?:%s|,)*)\){(?P<code>[^}]+)}
  714             ''' % (_FUNC_NAME_RE, _NAME_RE),
  715             fields)
  716         for f in fields_m:
  717             argnames = self.build_arglist(f.group('args'))
  718             obj[remove_quotes(f.group('key'))] = self.build_function(argnames, f.group('code'))
  719 
  720         return obj
  721 
  722     def extract_function_code(self, funcname):
  723         """ @returns argnames, code """
  724         func_m = re.search(
  725             r'''(?xs)
  726                 (?:
  727                     function\s+%(name)s|
  728                     [{;,]\s*%(name)s\s*=\s*function|
  729                     (?:var|const|let)\s+%(name)s\s*=\s*function
  730                 )\s*
  731                 \((?P<args>[^)]*)\)\s*
  732                 (?P<code>{.+})''' % {'name': re.escape(funcname)},
  733             self.code)
  734         code, _ = self._separate_at_paren(func_m.group('code'), '}')  # refine the match
  735         if func_m is None:
  736             raise self.Exception('Could not find JS function "{funcname}"'.format(**locals()))
  737         return self.build_arglist(func_m.group('args')), code
  738 
  739     def extract_function(self, funcname):
  740         return self.extract_function_from_code(*self.extract_function_code(funcname))
  741 
  742     def extract_function_from_code(self, argnames, code, *global_stack):
  743         local_vars = {}
  744         while True:
  745             mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
  746             if mobj is None:
  747                 break
  748             start, body_start = mobj.span()
  749             body, remaining = self._separate_at_paren(code[body_start - 1:], '}')
  750             name = self._named_object(
  751                 local_vars,
  752                 self.extract_function_from_code(
  753                     self.build_arglist(mobj.group('args')),
  754                     body, local_vars, *global_stack))
  755             code = code[:start] + name + remaining
  756         return self.build_function(argnames, code, local_vars, *global_stack)
  757 
  758     def call_function(self, funcname, *args):
  759         return self.extract_function(funcname)(args)
  760 
  761     @classmethod
  762     def build_arglist(cls, arg_text):
  763         if not arg_text:
  764             return []
  765 
  766         def valid_arg(y):
  767             y = y.strip()
  768             if not y:
  769                 raise cls.Exception('Missing arg in "%s"' % (arg_text, ))
  770             return y
  771 
  772         return [valid_arg(x) for x in cls._separate(arg_text)]
  773 
  774     def build_function(self, argnames, code, *global_stack):
  775         global_stack = list(global_stack) or [{}]
  776         argnames = tuple(argnames)
  777 
  778         def resf(args, kwargs={}, allow_recursion=100):
  779             global_stack[0].update(
  780                 zip_longest(argnames, args, fillvalue=None))
  781             global_stack[0].update(kwargs)
  782             var_stack = LocalNameSpace(*global_stack)
  783             ret, should_abort = self.interpret_statement(code.replace('\n', ''), var_stack, allow_recursion - 1)
  784             if should_abort:
  785                 return ret
  786         return resf

Generated by cgit