def test_item_inheritance(self):
class BaseItem(Item):
class Meta:
find_query = '//player'
name = StringField('firstname')
age = IntegerField('age')
class ChildItem(BaseItem):
name = StringField('lastname')
items = list(BaseItem.find(xml_fromstring(XML)))
self.assertEqual(items[0].name, 'Ardeshir')
self.assertEqual(items[0].age, 19)
self.assertEqual(set(['name', 'age']), set(items[0]._fields.keys()))
items = list(ChildItem.find(xml_fromstring(XML)))
self.assertEqual(items[0].age, 19)
self.assertEqual(set(['name', 'age']), set(items[0]._fields.keys()))
def test_item_inheritance(self):
class BaseItem(Item):
class Meta:
find_query = '//player'
name = StringField('firstname')
age = IntegerField('age')
class ChildItem(BaseItem):
name = StringField('lastname')
items = list(BaseItem.find(xml_fromstring(XML)))
self.assertEqual(items[0].name, 'Ardeshir')
self.assertEqual(items[0].age, 19)
self.assertEqual(set(['name', 'age']), set(items[0]._fields.keys()))
items = list(ChildItem.find(xml_fromstring(XML)))
self.assertEqual(items[0].age, 19)
self.assertEqual(set(['name', 'age']), set(items[0]._fields.keys()))
def get_item(self):
return Player(xml_fromstring(XML))
def test_stringfield_multiple(self):
class GameItem(Item):
names = StringField('//game/@name', multiple=True)
game = GameItem(xml_fromstring(XML))
self.assertEqual(['quake1', 'quake2'], game.names)
def test_find(self):
games = list(GameItem.find(xml_fromstring(XML)))
self.assertEqual(['quake1', 'quake2'],
[x.name for x in games])
def get_item(self):
return Player(xml_fromstring(XML))
def test_stringfield_multiple(self):
class GameItem(Item):
names = StringField('//game/@name', multiple=True)
game = GameItem(xml_fromstring(XML))
self.assertEqual(['quake1', 'quake2'], game.names)
def test_find(self):
games = list(GameItem.find(xml_fromstring(XML)))
self.assertEqual(['quake1', 'quake2'], [x.name for x in games])
def get_data(self, url, force_type = False):
response_type = "unknown"
req = urllib2.Request(url = url, headers = self.headers)
data = False
log(self.__class__.__name__ + ": Trying to open %s." % url)
try:
response = urllib2.urlopen(req)
except (urllib2.URLError, urllib2.HTTPError) as e:
if DEBUG:
log(self.__class__.__name__ + ": %s" % str(e))
log(self.__class__.__name__ + ": Error while opening %s, Fatal." % url )
return(False, response_type)
if response.getcode() == 200:
response_info = response.info()
if "Content-Type" in response_info:
if (response_info["Content-Type"].lower().find('json')) > -1:
response_type="json"
elif (response_info["Content-Type"].lower().find('xml')) > -1:
response_type = "xml"
else:
print(response_info["Content-Type"])
if force_type:
if not force_type == response_type:
response_type = force_type
log(self.__class__.__name__ + ": Response-type data forced to '%s' from %s." % (response_type, url))
else:
log(self.__class__.__name__ + ": Response-type data '%s' from %s." % (response_type, url))
else:
log(self.__class__.__name__ + ": Response-type data '%s' from %s." % (response_type, url))
if "Content-Length" in response_info:
if int(response_info["Content-Length"]) < 5: # No json data available, #doesnotexist
log(self.__class__.__name__ + ": Getting %s bytes from %s" % (response_info["Content-Length"], url))
else:
log(self.__class__.__name__ + ": Getting %s bytes from %s" % (response_info["Content-Length"], url))
else:
log(self.__class__.__name__+ ": Getting ? bytes from %s." % (url))
try:
data = response.read()
log(self.__class__.__name__+ ": Got %i bytes from %s." % (len(data), url))
except:
log(self.__class__.__name__ + ": Error while reading data.")
return(False, response_type)
if not len(data) > 0:
log(self.__class__.__name__+ ": Data size to small (%i bytes) from %s." % (len(data), url))
return(False, response_type)
else:
log(self.__class__.__name__ + ": Did not get a 200 ok response, got %i." % (response.getcode()))
return(False, response_type)
if response_type == "xml":
try:
data = xml_fromstring(data)
log(self.__class__.__name__ + ": Converted raw data to xmletree object.")
walker=data.iter()
for item in walker:
print(item, item.tag, item.text)
except:
log(self.__class__.__name__ + ": Error while converting raw data to xml.")
return(False, response_type)
if response_type == "feed":
try:
data = feedparser.parse(data)
try:
data["feed"].pop("updated_parsed")
except:
pass
try:
for item in data["entries"]:
item.pop("updated_parsed")
except:
pass
log(self.__class__.__name__ + ": Converted raw data to feed object.")
except:
log(self.__class__.__name__ + ": Error while converting raw data to feed.")
return(False, response_type)
if response_type == "json":
try:
data = loads(data)
log(self.__class__.__name__ + ": Converted raw data to json.")
except:
log(self.__class__.__name__ + ": Error while converting raw data to json.")
return(False, response_type)
return(data, response_type)
def get_data(self, url, force_type = False):
response_type = "unknown"
req = urllib2.Request(url = url, headers = self.headers)
data = False
log(self.__class__.__name__ + ": Trying to open %s." % url)
try:
response = urllib2.urlopen(req)
except (urllib2.URLError, urllib2.HTTPError) as e:
if DEBUG:
log(self.__class__.__name__ + ": %s" % str(e))
log(self.__class__.__name__ + ": Error while opening %s, Fatal." % url )
return(False, response_type)
if response.getcode() == 200:
response_info = response.info()
if "Content-Type" in response_info:
if (response_info["Content-Type"].lower().find('json')) > -1:
response_type="json"
elif (response_info["Content-Type"].lower().find('xml')) > -1:
response_type = "xml"
else:
print(response_info["Content-Type"])
if force_type:
if not force_type == response_type:
response_type = force_type
log(self.__class__.__name__ + ": Response-type data forced to '%s' from %s." % (response_type, url))
else:
log(self.__class__.__name__ + ": Response-type data '%s' from %s." % (response_type, url))
else:
log(self.__class__.__name__ + ": Response-type data '%s' from %s." % (response_type, url))
if "Content-Length" in response_info:
if int(response_info["Content-Length"]) < 5: # No json data available, #doesnotexist
log(self.__class__.__name__ + ": Getting %s bytes from %s" % (response_info["Content-Length"], url))
else:
log(self.__class__.__name__ + ": Getting %s bytes from %s" % (response_info["Content-Length"], url))
else:
log(self.__class__.__name__+ ": Getting ? bytes from %s." % (url))
try:
data = response.read()
log(self.__class__.__name__+ ": Got %i bytes from %s." % (len(data), url))
except:
log(self.__class__.__name__ + ": Error while reading data.")
return(False, response_type)
if not len(data) > 0:
log(self.__class__.__name__+ ": Data size to small (%i bytes) from %s." % (len(data), url))
return(False, response_type)
else:
log(self.__class__.__name__ + ": Did not get a 200 ok response, got %i." % (response.getcode()))
return(False, response_type)
if response_type == "xml":
try:
data = xml_fromstring(data)
log(self.__class__.__name__ + ": Converted raw data to xmletree object.")
walker=data.iter()
for item in walker:
print(item, item.tag, item.text)
except:
log(self.__class__.__name__ + ": Error while converting raw data to xml.")
return(False, response_type)
if response_type == "feed":
try:
data = feedparser.parse(data)
try:
data["feed"].pop("updated_parsed")
except:
pass
log(self.__class__.__name__ + ": Converted raw data to feed object.")
#walker=data.iter()
#for item in walker:
# print(item, item.tag, item.text)
except:
log(self.__class__.__name__ + ": Error while converting raw data to feed.")
return(False, response_type)
if response_type == "json":
try:
data = loads(data)
log(self.__class__.__name__ + ": Converted raw data to json.")
except:
log(self.__class__.__name__ + ": Error while converting raw data to json.")
return(False, response_type)
return(data, response_type)