From ad25aee2458527c38bdd3a8f1dec318ea807052c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Tue, 15 Jul 2014 22:46:39 +0200 Subject: [PATCH] [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. --- test/test_youtube_signature.py | 6 +++++ youtube_dl/jsinterp.py | 40 +++++++++++++++++++++++++++++++--- 2 files changed, 43 insertions(+), 3 deletions(-) diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py index 8d46fe108..d95533959 100644 --- a/test/test_youtube_signature.py +++ b/test/test_youtube_signature.py @@ -33,6 +33,12 @@ 90, 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'js', diff --git a/youtube_dl/jsinterp.py b/youtube_dl/jsinterp.py index 3bbb07704..ae5bca2e6 100644 --- a/youtube_dl/jsinterp.py +++ b/youtube_dl/jsinterp.py @@ -11,6 +11,7 @@ class JSInterpreter(object): def __init__(self, code): self.code = code self._functions = {} + self._objects = {} def interpret_statement(self, stmt, local_vars, allow_recursion=20): if allow_recursion < 0: @@ -55,7 +56,19 @@ def interpret_expression(self, expr, local_vars, allow_recursion): m = re.match(r'^(?P[a-z]+)\.(?P.*)$', expr) if m: member = m.group('member') - val = local_vars[m.group('in')] + variable = m.group('in') + + if variable not in local_vars: + if variable not in self._objects: + self._objects[variable] = self.extract_object(variable) + obj = self._objects[variable] + key, args = member.split('(', 1) + args = args.strip(')') + argvals = [int(v) if v.isdigit() else local_vars[v] + for v in args.split(',')] + return obj[key](argvals) + + val = local_vars[variable] if member == 'split("")': return list(val) if member == 'join("")': @@ -97,6 +110,25 @@ def interpret_expression(self, expr, local_vars, allow_recursion): return self._functions[fname](argvals) raise ExtractorError('Unsupported JS expression %r' % expr) + def extract_object(self, objname): + obj = {} + obj_m = re.search( + (r'(?:var\s+)?%s\s*=\s*\{' % re.escape(objname)) + + r'\s*(?P([a-zA-Z$]+\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[a-zA-Z$]+)\s*:\s*function' + r'\((?P[a-z,]+)\){(?P[^}]+)}', + 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): func_m = re.search( (r'(?:function %s|[{;]%s\s*=\s*function)' % ( @@ -107,10 +139,12 @@ def extract_function(self, funcname): raise ExtractorError('Could not find JS function %r' % funcname) argnames = func_m.group('args').split(',') + return self.build_function(argnames, func_m.group('code')) + + def build_function(self, argnames, code): def resf(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) return res return resf -