[jsinterp] Beef up and add tests

In preparation for #4822, extend jsinterp by a lot. (We may even have to/want to write a proper interpreter with actual parsing)
This commit is contained in:
Philipp Hagemeister 2015-02-01 22:38:26 +01:00
parent 4a3da4ebdb
commit 9e3f19919a
2 changed files with 221 additions and 46 deletions

95
test/test_jsinterp.py Normal file
View File

@ -0,0 +1,95 @@
#!/usr/bin/env python
from __future__ import unicode_literals
# Allow direct execution
import os
import sys
import unittest
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from youtube_dl.jsinterp import JSInterpreter
class TestJSInterpreter(unittest.TestCase):
def test_basic(self):
jsi = JSInterpreter('function x(){;}')
self.assertEqual(jsi.call_function('x'), None)
jsi = JSInterpreter('function x3(){return 42;}')
self.assertEqual(jsi.call_function('x3'), 42)
def test_calc(self):
jsi = JSInterpreter('function x4(a){return 2*a+1;}')
self.assertEqual(jsi.call_function('x4', 3), 7)
def test_empty_return(self):
jsi = JSInterpreter('function f(){return; y()}')
self.assertEqual(jsi.call_function('f'), None)
def test_morespace(self):
jsi = JSInterpreter('function x (a) { return 2 * a + 1 ; }')
self.assertEqual(jsi.call_function('x', 3), 7)
jsi = JSInterpreter('function f () { x = 2 ; return x; }')
self.assertEqual(jsi.call_function('f'), 2)
def test_strange_chars(self):
jsi = JSInterpreter('function $_xY1 ($_axY1) { var $_axY2 = $_axY1 + 1; return $_axY2; }')
self.assertEqual(jsi.call_function('$_xY1', 20), 21)
def test_operators(self):
jsi = JSInterpreter('function f(){return 1 << 5;}')
self.assertEqual(jsi.call_function('f'), 32)
jsi = JSInterpreter('function f(){return 19 & 21;}')
self.assertEqual(jsi.call_function('f'), 17)
jsi = JSInterpreter('function f(){return 11 >> 2;}')
self.assertEqual(jsi.call_function('f'), 2)
def test_array_access(self):
jsi = JSInterpreter('function f(){var x = [1,2,3]; x[0] = 4; x[0] = 5; x[2] = 7; return x;}')
self.assertEqual(jsi.call_function('f'), [5, 2, 7])
def test_parens(self):
jsi = JSInterpreter('function f(){return (1) + (2) * ((( (( (((((3)))))) )) ));}')
self.assertEqual(jsi.call_function('f'), 7)
jsi = JSInterpreter('function f(){return (1 + 2) * 3;}')
self.assertEqual(jsi.call_function('f'), 9)
def test_assignments(self):
jsi = JSInterpreter('function f(){var x = 20; x = 30 + 1; return x;}')
self.assertEqual(jsi.call_function('f'), 31)
jsi = JSInterpreter('function f(){var x = 20; x += 30 + 1; return x;}')
self.assertEqual(jsi.call_function('f'), 51)
jsi = JSInterpreter('function f(){var x = 20; x -= 30 + 1; return x;}')
self.assertEqual(jsi.call_function('f'), -11)
def test_comments(self):
jsi = JSInterpreter('''
function x() {
var x = /* 1 + */ 2;
var y = /* 30
* 40 */ 50;
return x + y;
}
''')
self.assertEqual(jsi.call_function('x'), 52)
def test_precedence(self):
jsi = JSInterpreter('''
function x() {
var a = [10, 20, 30, 40, 50];
var b = 6;
a[0]=a[b%a.length];
return a;
}''')
self.assertEqual(jsi.call_function('x'), [20, 20, 30, 40, 50])
if __name__ == '__main__':
unittest.main()

View File

@ -1,59 +1,122 @@
from __future__ import unicode_literals from __future__ import unicode_literals
import json import json
import operator
import re import re
from .utils import ( from .utils import (
ExtractorError, ExtractorError,
) )
_OPERATORS = [
('|', operator.or_),
('^', operator.xor),
('&', operator.and_),
('>>', operator.rshift),
('<<', operator.lshift),
('-', operator.sub),
('+', operator.add),
('%', operator.mod),
('/', operator.div),
('*', operator.mul),
]
_ASSIGN_OPERATORS = [(op + '=', opfunc) for op, opfunc in _OPERATORS]
_ASSIGN_OPERATORS.append(('=', lambda cur, right: right))
_NAME_RE = r'[a-zA-Z_$][a-zA-Z_$0-9]*'
class JSInterpreter(object): class JSInterpreter(object):
def __init__(self, code): def __init__(self, code, objects=None):
self.code = code if objects is None:
objects = {}
self.code = self._remove_comments(code)
self._functions = {} self._functions = {}
self._objects = {} self._objects = objects
def interpret_statement(self, stmt, local_vars, allow_recursion=20): def _remove_comments(self, code):
return re.sub(r'(?s)/\*.*?\*/', '', code)
def interpret_statement(self, stmt, local_vars, allow_recursion=100):
if allow_recursion < 0: if allow_recursion < 0:
raise ExtractorError('Recursion limit reached') raise ExtractorError('Recursion limit reached')
if stmt.startswith('var '): should_abort = False
stmt = stmt[len('var '):] stmt = stmt.lstrip()
ass_m = re.match(r'^(?P<out>[a-z]+)(?:\[(?P<index>[^\]]+)\])?' + stmt_m = re.match(r'var\s', stmt)
r'=(?P<expr>.*)$', stmt) if stmt_m:
if ass_m: expr = stmt[len(stmt_m.group(0)):]
if ass_m.groupdict().get('index'):
def assign(val):
lvar = local_vars[ass_m.group('out')]
idx = self.interpret_expression(
ass_m.group('index'), local_vars, allow_recursion)
assert isinstance(idx, int)
lvar[idx] = val
return val
expr = ass_m.group('expr')
else: else:
def assign(val): return_m = re.match(r'return(?:\s+|$)', stmt)
local_vars[ass_m.group('out')] = val if return_m:
return val expr = stmt[len(return_m.group(0)):]
expr = ass_m.group('expr') should_abort = True
elif stmt.startswith('return '):
assign = lambda v: v
expr = stmt[len('return '):]
else: else:
# Try interpreting it as an expression # Try interpreting it as an expression
expr = stmt expr = stmt
assign = lambda v: v
v = self.interpret_expression(expr, local_vars, allow_recursion) v = self.interpret_expression(expr, local_vars, allow_recursion)
return assign(v) return v, should_abort
def interpret_expression(self, expr, local_vars, allow_recursion): def interpret_expression(self, expr, local_vars, allow_recursion):
expr = expr.strip()
if expr == '': # Empty expression
return None
if expr.startswith('('):
parens_count = 0
for m in re.finditer(r'[()]', expr):
if m.group(0) == '(':
parens_count += 1
else:
parens_count -= 1
if parens_count == 0:
sub_expr = expr[1:m.start()]
sub_result = self.interpret_expression(
sub_expr, local_vars, allow_recursion)
remaining_expr = expr[m.end():].strip()
if not remaining_expr:
return sub_result
else:
expr = json.dumps(sub_result) + remaining_expr
break
else:
raise ExtractorError('Premature end of parens in %r' % expr)
for op, opfunc in _ASSIGN_OPERATORS:
m = re.match(r'''(?x)
(?P<out>%s)(?:\[(?P<index>[^\]]+?)\])?
\s*%s
(?P<expr>.*)$''' % (_NAME_RE, re.escape(op)), expr)
if not m:
continue
right_val = self.interpret_expression(
m.group('expr'), local_vars, allow_recursion - 1)
if m.groupdict().get('index'):
lvar = local_vars[m.group('out')]
idx = self.interpret_expression(
m.group('index'), local_vars, allow_recursion)
assert isinstance(idx, int)
cur = lvar[idx]
val = opfunc(cur, right_val)
lvar[idx] = val
return val
else:
cur = local_vars.get(m.group('out'))
val = opfunc(cur, right_val)
local_vars[m.group('out')] = val
return val
if expr.isdigit(): if expr.isdigit():
return int(expr) return int(expr)
if expr.isalpha(): var_m = re.match(
return local_vars[expr] r'(?!if|return|true|false)(?P<name>%s)$' % _NAME_RE,
expr)
if var_m:
return local_vars[var_m.group('name')]
try: try:
return json.loads(expr) return json.loads(expr)
@ -61,7 +124,7 @@ def interpret_expression(self, expr, local_vars, allow_recursion):
pass pass
m = re.match( m = re.match(
r'^(?P<var>[$a-zA-Z0-9_]+)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$', r'(?P<var>%s)\.(?P<member>[^(]+)(?:\(+(?P<args>[^()]*)\))?$' % _NAME_RE,
expr) expr)
if m: if m:
variable = m.group('var') variable = m.group('var')
@ -114,23 +177,31 @@ def interpret_expression(self, expr, local_vars, allow_recursion):
return obj[member](argvals) return obj[member](argvals)
m = re.match( m = re.match(
r'^(?P<in>[a-z]+)\[(?P<idx>.+)\]$', expr) r'(?P<in>%s)\[(?P<idx>.+)\]$' % _NAME_RE, expr)
if m: if m:
val = local_vars[m.group('in')] val = local_vars[m.group('in')]
idx = self.interpret_expression( idx = self.interpret_expression(
m.group('idx'), local_vars, allow_recursion - 1) m.group('idx'), local_vars, allow_recursion - 1)
return val[idx] return val[idx]
m = re.match(r'^(?P<a>.+?)(?P<op>[%])(?P<b>.+?)$', expr) for op, opfunc in _OPERATORS:
if m: m = re.match(r'(?P<x>.+?)%s(?P<y>.+)' % re.escape(op), expr)
a = self.interpret_expression( if not m:
m.group('a'), local_vars, allow_recursion) continue
b = self.interpret_expression( x, abort = self.interpret_statement(
m.group('b'), local_vars, allow_recursion) m.group('x'), local_vars, allow_recursion - 1)
return a % b if abort:
raise ExtractorError(
'Premature left-side return of %s in %r' % (op, expr))
y, abort = self.interpret_statement(
m.group('y'), local_vars, allow_recursion - 1)
if abort:
raise ExtractorError(
'Premature right-side return of %s in %r' % (op, expr))
return opfunc(x, y)
m = re.match( m = re.match(
r'^(?P<func>[a-zA-Z$]+)\((?P<args>[a-z0-9,]+)\)$', expr) r'^(?P<func>%s)\((?P<args>[a-zA-Z0-9_$,]+)\)$' % _NAME_RE, expr)
if m: if m:
fname = m.group('func') fname = m.group('func')
argvals = tuple([ argvals = tuple([
@ -139,6 +210,7 @@ def interpret_expression(self, expr, local_vars, allow_recursion):
if fname not in self._functions: if fname not in self._functions:
self._functions[fname] = self.extract_function(fname) self._functions[fname] = self.extract_function(fname)
return self._functions[fname](argvals) return self._functions[fname](argvals)
raise ExtractorError('Unsupported JS expression %r' % expr) raise ExtractorError('Unsupported JS expression %r' % expr)
def extract_object(self, objname): def extract_object(self, objname):
@ -162,9 +234,11 @@ def extract_object(self, objname):
def extract_function(self, funcname): def extract_function(self, funcname):
func_m = re.search( func_m = re.search(
(r'(?:function %s|[{;]%s\s*=\s*function)' % ( r'''(?x)
re.escape(funcname), re.escape(funcname))) + (?:function\s+%s|[{;]%s\s*=\s*function)\s*
r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}', \((?P<args>[^)]*)\)\s*
\{(?P<code>[^}]+)\}''' % (
re.escape(funcname), re.escape(funcname)),
self.code) self.code)
if func_m is None: if func_m is None:
raise ExtractorError('Could not find JS function %r' % funcname) raise ExtractorError('Could not find JS function %r' % funcname)
@ -172,10 +246,16 @@ def extract_function(self, funcname):
return self.build_function(argnames, func_m.group('code')) return self.build_function(argnames, func_m.group('code'))
def call_function(self, funcname, *args):
f = self.extract_function(funcname)
return f(args)
def build_function(self, argnames, code): def build_function(self, argnames, code):
def resf(args): def resf(args):
local_vars = dict(zip(argnames, args)) local_vars = dict(zip(argnames, args))
for stmt in code.split(';'): for stmt in code.split(';'):
res = self.interpret_statement(stmt, local_vars) res, abort = self.interpret_statement(stmt, local_vars)
if abort:
break
return res return res
return resf return resf