def get_rel_url(js):
js = js[31:-9]
# js=js.replace('location','\"http://bbs.21ic.com/icview-2545904-1-1.html\"')
js = 'location= function(){return "http://bbs.21ic.com/icview-2545904-1-1.html\";};' + js
js = 'window = function(){return "bbs.21ic.com";};' + js
for st in [
'window', '\'location\'', '\'assign\'', '\'href\'', '\'replace\''
]:
equal = re.findall('[_A-Za-z0-9= ]+{};'.format(st), js) # 找到变量赋值等式
if equal == []:
continue
else:
var = equal[0].split('=')
left = var[0].strip()
ri = var[1].strip()
js = js.replace(left + ';', ri)
# js = js.replace(equal, st)
# print(js)
js = js.replace("['{}']".format(st).strip("'"),
'.{}'.format(st).strip("'"))
if re.findall('window.href=\"http://www.php1.cn/\">', js):
js = js.replace('window.href=\"http://www.php1.cn/\">', js)
js = js.replace('location.href=\"http://www.php1.cn/\">', js)
ctxt = JSContext()
ctxt.__enter__()
x = (ctxt.eval(js))
vars = ctxt.locals
url = 'http://bbs.21ic.com' + str(vars.location['href'])
return url
class CommonJS():
_js_path = [
os.path.dirname(__file__),
os.path.join(os.path.dirname(__file__), 'core')
]
_js_logger = logger().instance()
def __init__(self):
self._js_threadlock = False
self._js_ctx = JSContext(self)
self._js_modules = {}
self._loaded_modules = {}
for jsroot in CommonJS._js_path:
for (root, dirs, files) in os.walk(jsroot):
#print files
for _file in files:
m = re.compile('(.*)\.js$').match(_file)
relpath = os.path.relpath(root, jsroot)
namespace = re.sub(r'^\.', '', relpath)
namespace = re.sub(r'^\\', '/', namespace)
if (namespace):
namespace = namespace + '/'
if (m):
self._js_modules.update({
namespace + m.group(1):
os.path.join(root, _file)
})
self.execute("var exports;")
@staticmethod
def append(path):
if (path not in CommonJS._js_path):
CommonJS._js_path.append(path)
def require(self, module):
if (not self._js_modules.has_key(module)):
raise Exception, "unknown module `" + module + "`"
path = self._js_modules[module]
if (not self._loaded_modules.has_key(path)):
self._js_logger.info("loading module <%s>...", module)
code = file(path).read()
try:
code = code.decode('utf-8')
if (platform.system() == 'Windows'):
code = code.encode('utf-8')
self._js_ctx.eval(code)
except JSError, ex:
self._js_logger.error(ex)
self._js_logger.debug(ex.stackTrace)
raise Exception, ex
self._loaded_modules[path] = self._js_ctx.locals.exports
return self._loaded_modules[path]
else:
class CommonJS:
_js_path = [os.path.dirname(__file__), os.path.join(os.path.dirname(__file__), "core")]
_js_logger = logger().instance()
def __init__(self):
self._js_threadlock = False
self._js_ctx = JSContext(self)
self._js_modules = {}
self._loaded_modules = {}
for jsroot in CommonJS._js_path:
for (root, dirs, files) in os.walk(jsroot):
# print files
for _file in files:
m = re.compile("(.*)\.js$").match(_file)
relpath = os.path.relpath(root, jsroot)
namespace = re.sub(r"^\.", "", relpath)
namespace = re.sub(r"^\\", "/", namespace)
if namespace:
namespace = namespace + "/"
if m:
self._js_modules.update({namespace + m.group(1): os.path.join(root, _file)})
self.execute("var exports;")
@staticmethod
def append(path):
if path not in CommonJS._js_path:
CommonJS._js_path.append(path)
def require(self, module):
if not self._js_modules.has_key(module):
raise Exception, "unknown module `" + module + "`"
path = self._js_modules[module]
if not self._loaded_modules.has_key(path):
self._js_logger.info("loading module <%s>...", module)
code = file(path).read()
try:
code = code.decode("utf-8")
if platform.system() == "Windows":
code = code.encode("utf-8")
self._js_ctx.eval(code)
except JSError, ex:
self._js_logger.error(ex)
self._js_logger.debug(ex.stackTrace)
raise Exception, ex
self._loaded_modules[path] = self._js_ctx.locals.exports
return self._loaded_modules[path]
else:
def evalExpr(self, expr):
with JSContext(self.g) as c:
try:
ret = c.eval(expr)
except Exception, e:
raise ExprEvalError(e, [])
# TODO: this causes variabels typeof == 'undefined' to become null instead.
for key in c.locals.keys():
setattr(self.g, key, c.locals[key])
return ret
def __init__(self):
self._js_threadlock = False
self._js_ctx = JSContext(self)
self._js_modules = {}
self._loaded_modules = {}
for jsroot in CommonJS._js_path:
for (root, dirs, files) in os.walk(jsroot):
#print files
for _file in files:
m = re.compile('(.*)\.js$').match(_file)
relpath = os.path.relpath(root, jsroot)
namespace = re.sub(r'^\.', '', relpath)
namespace = re.sub(r'^\\', '/', namespace)
if (namespace):
namespace = namespace + '/'
if (m):
self._js_modules.update({
namespace + m.group(1):
os.path.join(root, _file)
})
self.execute("var exports;")
def __init__(self):
self._js_threadlock = False
self._js_ctx = JSContext(self)
self._js_modules = {}
self._loaded_modules = {}
for jsroot in CommonJS._js_path:
for (root, dirs, files) in os.walk(jsroot):
# print files
for _file in files:
m = re.compile("(.*)\.js$").match(_file)
relpath = os.path.relpath(root, jsroot)
namespace = re.sub(r"^\.", "", relpath)
namespace = re.sub(r"^\\", "/", namespace)
if namespace:
namespace = namespace + "/"
if m:
self._js_modules.update({namespace + m.group(1): os.path.join(root, _file)})
self.execute("var exports;")
def __init__(self, template):
super(BemhtmlRenderer, self).__init__()
with JSLocker():
self.ctx = JSContext()
with self.ctx as ctx:
ctx.eval(template)
output = self.evalExpr("(%s)" % contentNode.get("expr"))
elif len(contentNode) == 0:
output = contentNode.xpath("normalize-space(./text())")
elif len(contentNode) > 0:
output = contentNode.xpath("./*")
else:
self.logger.error("Line %s: error when parsing content node." % contentNode.sourceline)
return
return output
def evalExpr(self, expr):
return self[expr]
def execExpr(self, expr):
# TODO: should fail on exmode=strict
self.logger.warn("The script element is ignored by the xpath datamodel.")
# raise DataModelError("multiline expressions can't be executed on the xpath datamodel.")
if __name__ == "__main__":
# import PyV8 #@UnresolvedImport
d = ECMAScriptDataModel()
d.evalExpr("var a = [1,2,3]")
c = JSContext(d.g)
c.enter()
print d["a"]
# d = DataModel()
elif node_name == "script":
try:
src = node.text or self.script_src.get(node) or ""
self.execExpr(src)
except ExprEvalError, e:
raise ExecutableError(e, node)
elif node_name == "if":
self.parseIf(node)
elif node_name == "foreach":
startIndex = 0 if self.datamodel != "xpath" else 1
try:
array = self.getExprValue(node.get("array"))
if self.datamodel == "ecmascript":
from PyV8 import JSContext
c = JSContext(self.dm.g)
c.enter()
# if self.datamodel == "xpath":
# assert all(map(lambda x: x, array))
except ExprEvalError, e:
raise AttributeEvalError(e, node, "array")
except TypeError, e:
err = DataModelError(e)
raise AttributeEvalError(err, node, "array")
for index, item in enumerate(array, startIndex):
try:
# if it's not a correct QName: crash.
etree.QName(node.get("item"))
if self.datamodel != "xpath":
self.dm[node.get("item")] = item
elif node_name == "script":
try:
src = node.text or self.script_src.get(node) or ""
self.execExpr(src)
except ExprEvalError, e:
raise ExecutableError(e, node)
elif node_name == "if":
self.parseIf(node)
elif node_name == "foreach":
startIndex = 0 if self.datamodel != "xpath" else 1
try:
array = self.getExprValue(node.get("array"))
if self.datamodel == "ecmascript":
from PyV8 import JSContext
c = JSContext(self.dm.g)
c.enter()
itr = enumerate(array, startIndex)
# if self.datamodel == "xpath":
# assert all(map(lambda x: x, array))
except ExprEvalError, e:
raise AttributeEvalError(e, node, "array")
except TypeError, e:
err = DataModelError(e)
raise AttributeEvalError(err, node, "array")
for index, item in itr:
try:
if self.datamodel != "xpath":
self.dm[node.get("item")] = item
