[youtube & jsinterp] Fix signature extraction (fixes #3255)

Some functions are defined now inside an object, the jsinterp will search its definition if the variable is not defined in the local namespace.
pull/8/head
Jaime Marquínez Ferrándiz 10 years ago
parent bd1f325b42
commit ad25aee245

@ -33,6 +33,12 @@ _TESTS = [
90, 90,
u']\\[@?>=<;:/.-,+*)(\'&%$#"hZYXWVUTSRQPONMLKJIHGFEDCBAzyxwvutsrqponmlkjiagfedcb39876', u']\\[@?>=<;:/.-,+*)(\'&%$#"hZYXWVUTSRQPONMLKJIHGFEDCBAzyxwvutsrqponmlkjiagfedcb39876',
), ),
(
u'https://s.ytimg.com/yts/jsbin/html5player-en_US-vfl0Cbn9e.js',
u'js',
84,
u'O1I3456789abcde0ghijklmnopqrstuvwxyzABCDEFGHfJKLMN2PQRSTUVW@YZ!"#$%&\'()*+,-./:;<=',
),
( (
u'https://s.ytimg.com/yts/jsbin/html5player-en_US-vflXGBaUN.js', u'https://s.ytimg.com/yts/jsbin/html5player-en_US-vflXGBaUN.js',
u'js', u'js',

@ -11,6 +11,7 @@ class JSInterpreter(object):
def __init__(self, code): def __init__(self, code):
self.code = code self.code = code
self._functions = {} self._functions = {}
self._objects = {}
def interpret_statement(self, stmt, local_vars, allow_recursion=20): def interpret_statement(self, stmt, local_vars, allow_recursion=20):
if allow_recursion < 0: if allow_recursion < 0:
@ -55,7 +56,19 @@ class JSInterpreter(object):
m = re.match(r'^(?P<in>[a-z]+)\.(?P<member>.*)$', expr) m = re.match(r'^(?P<in>[a-z]+)\.(?P<member>.*)$', expr)
if m: if m:
member = m.group('member') member = m.group('member')
val = local_vars[m.group('in')] variable = m.group('in')
if variable not in local_vars:
if variable not in self._objects:
self._objects[variable] = self.extract_object(variable)
obj = self._objects[variable]
key, args = member.split('(', 1)
args = args.strip(')')
argvals = [int(v) if v.isdigit() else local_vars[v]
for v in args.split(',')]
return obj[key](argvals)
val = local_vars[variable]
if member == 'split("")': if member == 'split("")':
return list(val) return list(val)
if member == 'join("")': if member == 'join("")':
@ -97,6 +110,25 @@ class JSInterpreter(object):
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):
obj = {}
obj_m = re.search(
(r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) +
r'\s*(?P<fields>([a-zA-Z$]+\s*:\s*function\(.*?\)\s*\{.*?\})*)' +
r'\}\s*;',
self.code)
fields = obj_m.group('fields')
# Currently, it only supports function definitions
fields_m = re.finditer(
r'(?P<key>[a-zA-Z$]+)\s*:\s*function'
r'\((?P<args>[a-z,]+)\){(?P<code>[^}]+)}',
fields)
for f in fields_m:
argnames = f.group('args').split(',')
obj[f.group('key')] = self.build_function(argnames, f.group('code'))
return obj
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'(?:function %s|[{;]%s\s*=\s*function)' % (
@ -107,10 +139,12 @@ class JSInterpreter(object):
raise ExtractorError('Could not find JS function %r' % funcname) raise ExtractorError('Could not find JS function %r' % funcname)
argnames = func_m.group('args').split(',') argnames = func_m.group('args').split(',')
return self.build_function(argnames, func_m.group('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 func_m.group('code').split(';'): for stmt in code.split(';'):
res = self.interpret_statement(stmt, local_vars) res = self.interpret_statement(stmt, local_vars)
return res return res
return resf return resf

Loading…
Cancel
Save