#
# Platform specific calls used the add-on.
import os
import shlex
import subprocess
import sys
import xbmc
import xbmcgui
import xbmcvfs
import xbmcaddon
from utility import debugTrace, errorTrace, infoTrace, newPrint, infoPrint, enum, getID, isCustom, getCustom
from sys import platform
# **** ADD MORE PLATFORMS HERE ****
platforms = enum(UNKNOWN=0, WINDOWS=1, LINUX=2, RPI=3, ANDROID=4, MAC=5)
platforms_str = ("Unknown", "Windows", "Linux, openvpn installed",
"Linux, openvpn plugin", "Android", "Apple")
fake_name = "FAKECONNECTION.txt"
def fakeConnection():
# Return True to fake out any calls to openVPN to change the network.
# This is governed by the existance of 'FAKECONNECTION.txt' in the userdata directory.
return xbmcvfs.exists(getUserDataPath(fake_name))
def fakeItTillYouMakeIt(fake):
try:
if fake:
if not fakeConnection():
CHAR_INFO_WIDTH = 50
#INFO_POPUP_WIDTH = 30
LIMIT_FPS = 20 # 20 frames-per-second maximum
# size of the map
MAP_WIDTH = 80
MAP_HEIGHT = 43
def exp_xp_curve(level):
# exponential
from math import exp
return exp(level) * 5
# character generation values
DEFAULT_SPEED = 10
DEFAULT_XP_CURVE = exp_xp_curve
AIR = 0
WATER = 1
EARTH = 2
FIRE = 3
aspects = ["AIR", "WATER", "EARTH", "FIRE"] # clockwise from air
# (hot_cold, wet_dry), where hot and wet are positive
factors = [(1,1), (1,-1), (-1,-1), (-1,1)]
tile_types = enum("CAVE_FLOOR", "CAVE_WALL", "GRASS",
"SAND", "SHALLOW_WATER", "DEEP_WATER")
import cmd
import gevent
import config
from config import MSG, DBQUERY
import network
from utility import enum
import partition
STATE = enum('WAITING_FOR_DATABASE',
'WAITING_FOR_SERVERS',
'ACCEPTING_PLAYERS')
class System:
def __init__(self, system_id=-1, server=None, connections=[]):
self.id = system_id
self.server = server
self.connections = connections
class Server:
def __init__(self, server_id=-1, server_socket=None, ip="", client_port=-1,
server_port=-1, systems=None, neighbours=None):
self.id = server_id
self.socket = server_socket
self.ip = ip
self.client_port = client_port # address for clients to connect on
self.server_port = server_port # address for servers to connect on
self.systems = systems if systems is not None else [] # systems we're responsible for
self.edge_systems = {}
self.neighbours = neighbours if neighbours is not None else [] # neighbouring servers
# Platform specific calls used by VPN Manager for OpenVPN add-on.
import os
import shlex
import subprocess
import sys
import xbmc
import xbmcgui
import xbmcvfs
import xbmcaddon
from utility import debugTrace, errorTrace, infoTrace, newPrint, infoPrint, enum
from sys import platform
# **** ADD MORE PLATFORMS HERE ****
platforms = enum(UNKNOWN=0, WINDOWS=1, LINUX=2, RPI=3, ANDROID=4, MAC=5)
platforms_str = ("Unknown", "Windows", "Linux, openvpn installed", "Linux, openvpn plugin", "Android", "Apple")
fake_name = "FAKECONNECTION.txt"
def fakeConnection():
# Return True to fake out any calls to openVPN to change the network.
# This is governed by the existance of 'FAKECONNECTION.txt' in the userdata directory.
return xbmcvfs.exists(getUserDataPath(fake_name))
def fakeItTillYouMakeIt(fake):
try:
if fake:
if not fakeConnection():
f = open(getUserDataPath(fake_name),'w')
import controls
import data
import utility
ManifestType = utility.enum(
FIELD_DESCRIPTION = 3,
FIELD_DESCRIPTION_NAME = 'Field Description',
KEY_VALUE_LIST = 44,
KEY_VALUE_LIST_NAME = 'Key-Value pairs list',
OPERATIONAL_STATE = 27,
OPERATIONAL_STATE_NAME = 'Operational State',
STANDARD_CONFIGURATION_CONTROLS = 6,
STANDARD_CONFIGURATION_CONTROLS_NAME = 'Standard UI Controls',
STANDARD_FR8_TERMINAL = 23,
STANDARD_FR8_TERMINAL_NAME = 'Standard Fr8 Terminal',
STANDARD_PAYLOAD_DATA = 5,
STANDARD_PAYLOAD_DATA_NAME = 'Standard Payload Data'
)
class Manifest(object):
def __init__(self, **kwargs):
self.manifest_type = kwargs.get('manifest_type')
self.manifest_type_name = kwargs.get('manifest_type_name')
class OperationalStateCM(Manifest):
def __init__(self, **kwargs):
super(OperationalStateCM, self).__init__(
manifest_type=ManifestType.OPERATIONAL_STATE,
from utility import enum __doc__ = """Yeah, globals might be bad, but here's a file full of them anyway""" target0 = 200 target1 = 824 target2 = 512 MOTORS = [1,2,3,4,5,6,7,8,9] LEFT_FRONT = MOTORS[0] RIGHT_FRONT = MOTORS[1] LEFT_BACK = MOTORS[2] RIGHT_BACK = MOTORS[3] LEFT_FRONT_E = MOTORS[4] RIGHT_FRONT_E = MOTORS[5] LEFT_BACK_E = MOTORS[6] RIGHT_BACK_E = MOTORS[7] IR_X = 1 IR_Y = 2 COMMANDS = enum(GetSensorValue='GetSensorValue', GetMotorTargetPosition='GetMotorTargetPosition', GetMotorCurrentPosition='GetMotorCurrentPosition', SetMotorTargetPosition='SetMotorTargetPosition')
import uuid
import manifests
import utility
def extract_fr8_crate_contents(contents, manifest_id):
manifest_id_str = str(manifest_id)
if manifest_id_str not in manifests.manifest_extractors:
raise 'Unsupported manifest_id, data extraction failed.'
return manifests.manifest_extractors[manifest_id_str](contents)
TerminalStatus = utility.enum(ACTIVE=1, INACTIVE=0)
AuthenticationType = utility.enum(NONE=1, INTERNAL=2, EXTERNAL=3, INTERNAL_WITH_DOMAIN=4)
ActivityType = utility.enum(STANDARD=1, LOOP=2, SOLUTION=3)
AvailabilityType = utility.enum(NOTSET=0, CONFIGURATION=1, RUNTIME=2, ALWAYS=3)
class ActivityDTO(object):
def __init__(self, **kwargs):
self.name = kwargs.get('name')
self.label = kwargs.get('label')
self.activity_template = kwargs.get('activity_template')
self.root_plan_node_id = kwargs.get('root_plan_node_id')
self.parent_plan_node_id = kwargs.get('parent_plan_node_id')
self.current_view = kwargs.get('current_view')
self.ordering = kwargs.get('ordering')
# network values
SOCKET_SERVER_MAX_QUEUE = 512
SOCKET_CLIENT_MAX_QUEUE = 512
SOCKET_DATABASE_MAX_QUEUE = 512
SOCKET_RECV_MAX_BYTES = 4096
GATEWAY_SERVER_COUNT = 2
MSG_END_CHAR = chr(1)
MSG_TOP_DELIMITER = 2
NEGATIVE_CHAR = 126
MSG_TIME_ACCURACY = 3 # how many digits after decimal point
# database query types
DBQUERY = enum('GW_STARTINFO', 'GS_SYSTEMSINFO', 'GS_UPDATE', 'GW_NEWPLAYER')
# Message types
# Type ID format: '[SENDER]_[RECEIVER]_[DESCRIPTION]'
class MSG:
OK = 0
# CLIENT to GATEWAY
CL_GW_LOGIN = 1
CL_GW_LOGOUT = 2
CL_GW_INPUT = 5
CL_GW_REGISTER = 7
# GATEWAY to CLIENT
GW_CL_LOGIN_SUCCESSFUL = 3
GW_CL_LOGIN_FAILED = 4
GW_CL_FRAME = 6
GW_CL_REGISTRATION_SUCCESSFUL = 8
from utility import enum, Rect
FRAME_RATE = 1.0/100 # length of a frame in seconds
DATABASE_FRAME_RATE = 4.0 # how often in seconds to commit to DB
PLAYER_SPEED = 100 # pixels/second
PLAYER_TURN_SPEED = math.radians(180) # radians/second
PLAYER_FIRE_RATE = 1.0/5 # per second
WORMHOLE_SIZE = 30
PROJECTILE_SIZE = 5
INTERACTION_AREA = 100
PROJECTILE_SPEED = 200 #pixels/second
INPUTS = enum('FORWARD', 'LEFT', 'RIGHT', 'FIRE', 'INTERACT')
PLAYER_SIZE = 5;
class Client:
def __init__(self, client_socket=None, address=None, player=None):
self.socket = client_socket
self.address = address
self.player = player
class OfflinePlayer:
def __init__(self, player_id, name, x, y, solar_system):
self.id = player_id
self.name = name
self.x = x
self.y = y
#coding=utf-8
# ycat 2014/09/28 create
import pytest
import json
import pytest
import session
import utility
import sqlite3
import traceback
import sys,os,bottle,datetime
sex_type = utility.enum(Male=1,Female=0)
class ctrl_user_manager:
@staticmethod
def register(nick,pwd,sex,age):
assert int(sex) <= 1
age = int(age)
try:
now = utility.now()
db = utility.get_db()
c = db.cursor()
c.execute("INSERT INTO u_user (NickName,Sex,Password,CreateDate,BirthdayYear,CertfState)VALUES(?,?,?,?,?,?)",
(nick,int(sex),pwd,now.strftime("%Y-%m-%d %H:%M:%S"),now.year - age,0))
db.commit() #不这样做取不到user_id
user = session.login(nick,pwd)
c.execute("INSERT INTO u_profile(ID,EditDate)VALUES(?,?)",(user.user_id,now))
utility.write_log(user.user_id,"注册成功",1,False)
db.commit()
return user
import data
import manifests
import utility
ControlTypes = utility.enum(
DROP_DOWN_LIST = 'DropDownList',
TEXT_SOURCE = 'TextSource'
)
class ControlEvent(object):
def __init__(self, **kwargs):
self.name = kwargs.get('name')
self.handler = kwargs.get('handler')
def to_fr8_json(self):
return {
'name': self.name,
'handler': self.handler
}
@staticmethod
def from_fr8_json(json_data):
return ControlEvent(name=json_data.get('name'), handler=json_data.get('handler'))
class ControlSource(object):
def __init__(self, **kwargs):
self.manifest_type = kwargs.get('manifest_type')
self.label = kwargs.get('label')
class Model(object):
Modes = enum(ONLINE=1, OFFLINE=2)
# Data split by percentage for
# [training, validation, testing]
Split = [0.8, 0.10, 0.05]
TrainMax = 1024 * 5
ValidMax = 1024 * 2
BestValidationInterval = 4
def __init__(
self,
rng, # random number generator use to initialize weights
input, # a theano.tensor.dmatrix of shape (n_examples, n_in)
batch_size, # size of mini batch
patch_size, # size of feature map
train_time, # batch training time before resampling
path, # where to load or save best model
type, # type of model (CNN or MLP)
id # the unique identifier of the model (used for persistence)
):
self.done = False
self.id = id
self.type = type
self.path = path
self.train_time = train_time
self.batchSize = batch_size
self.patchSize = patch_size
self.input = input
self.rng = rng
self.params = []
self.best_loss = np.inf
self.i_train = []
self.i_valid = []
self.i_test = []
self.i = []
self.n_data = 0
self.iteration = 0
self.initialized = False
self.x = input
self.index = T.lscalar()
self.label_dist = []
def setTrainingData(
self,
x, #
y, #
p, #
l, #
learning_rate, #
momentum):
# save the training set and params
self.x_data = x
self.y_data = y
self.p_data = p
self.i_labels = l
self.n_data = len(y)
self.learning_rate = learning_rate
self.momentum = momentum
self.iteration = 0
# spli the training set into training, validation, and test
self.n_train = int(math.floor(self.n_data * Model.Split[0]))
self.n_valid = int(math.floor(self.n_data * Model.Split[1]))
self.n_test = int(math.floor(self.n_data * Model.Split[2]))
self.n_train = min(self.n_train, Model.TrainMax)
self.n_valid = Model.ValidMax
self.n_train = self.n_data - self.n_valid
self.i = np.arange(self.n_data)
self.i = np.random.choice(self.i, self.n_data, replace=False)
self.i_valid = self.i[self.n_train:self.n_train + self.n_valid]
self.i_train = []
self.l_draws = []
l_percentages = [float(len(l_data)) / self.n_data for l_data in l]
self.n_draws = [int(self.n_train * l) for l in l_percentages]
total = np.sum(self.n_draws)
if (total < self.n_train):
i_max = np.argmax(self.n_draws)
self.n_draws[i_max] += (self.n_train - total)
self.i_train_labels = [[] for l in l_percentages]
self.i_valid_labels = [[] for l in l_percentages]
#self.l_dist = [ len(l_data) for l_data in l]
'''
print 'n_valid:', self.n_valid
print '#valid:', len(self.i_valid)
print '==setTrainingData'
print 'p_draws:',l_percentages
print 'n_draws:', self.n_draws
print 'total:', np.sum(self.n_draws)
'''
def drawRandomizedSamples(self):
print '------randomized sampling-------'
self.i_train = []
for i, n_draw in enumerate(self.n_draws):
self.i_labels[i] = np.hstack(
(self.i_labels[i], self.i_train_labels[i]))
n_ldata = len(self.i_labels[i])
self.i_labels[i] = np.random.choice(self.i_labels[i],
n_ldata,
replace=False)
self.i_train_labels[i] = self.i_labels[i][:n_draw]
self.i_labels[i] = self.i_labels[i][n_draw:]
self.i_train = np.concatenate(
(self.i_train, self.i_train_labels[i]))
def drawStratifiedSamples(self):
print '------stratified sampling-------'
n_train = len(self.i_train)
self.i_train = []
print 'n_train:', n_train
for i, n_draw in enumerate(self.n_draws):
if n_train > 0:
# determine number of good and bad
# samples based on the training results
indices = self.i_train_labels[i]
n_indices = len(indices)
p_train = self.p_data[indices]
i_sorted = np.argsort(p_train, axis=0)
n_good = len(np.where(p_train[i_sorted] == 0)[0])
n_bad = len(indices) - n_good
# keep at most 50% of the bad samples for
# retraining, and resample the rest
n_good = max(n_good, n_draw / 2)
n_bad = n_indices - n_good
i_good = i_sorted[:n_good]
i_bad = i_sorted[n_good:]
print '-------', i, '---------'
print 'n_indices:', n_indices
print 'n_good:', n_good
print 'n_bad:', n_bad
print 'i_sorted:', i_sorted
print 'indices:', indices
print 'p_train:', len(p_train), p_train
print 'i_sorted:', len(i_sorted), i_sorted
# return the good indices back to the pool
self.i_labels[i] = np.hstack((self.i_labels[i], i_good))
# draw replacement samples for the good indices
i_new = self.i_labels[i][:n_good]
self.i_labels[i] = self.i_labels[i][n_good:]
# combine with the bad indices to comprise the new
# training batch
self.i_train_labels[i] = np.hstack((i_new, i_bad))
else:
self.i_train_labels[i] = self.i_labels[i][:n_draw]
self.i_labels[i] = self.i_labels[i][n_draw:]
# add the label indices to the training batch
self.i_train = np.concatenate(
(self.i_train, self.i_train_labels[i]))
def rotateSamples(self):
print 'rotateSamples....'
# randomly rotate samples
n_half = len(self.i_train) / 2
indices = np.random.choice(self.i_train, n_half, replace=False)
for index in indices:
#print 'itrain:', self.i_train
#print 'index:', index
patch = self.x_data[index]
#print 'patch.a:',patch
patch = np.reshape(patch, (self.patchSize, self.patchSize))
if random.random() < 0.5:
patch = np.fliplr(patch)
else:
patch = np.fliplr(patch)
patch = patch.flatten()
#print 'patch.b:',patch
self.x_data[index] = patch
#exit(1)
def justDraw(self):
print 'just draw...'
i = self.i[:self.n_train]
self.i_train = np.random.choice(i, self.n_train, replace=False)
if (self.iteration % 2 == 0):
self.rotateSamples()
def sampleTrainingData(self):
self.iteration += 1
'''
if self.iteration == Model.BestValidationInterval:
self.drawRandomizedSamples()
else:
self.drawStratifiedSamples()
#self.rotateSamples()
'''
self.justDraw()
self.i_train = self.i_train.astype(dtype=int)
lens = [len(l) for l in self.i_labels]
print 'remaining sizes:', lens
lens = [len(l) for l in self.i_train_labels]
print 'ampleed sizes:', lens
print 'train indices:', len(self.i_train)
print self.i_train
#t = self.i[:self.n_train]
#self.i_train = np.random.choice( t, self.n_train, replace=False)
#self.i_valid = np.random.choice( self.i, self.n_valid, replace=False)
train_x = self.x_data[self.i_train]
train_y = self.y_data[self.i_train]
valid_x = self.x_data[self.i_valid]
valid_y = self.y_data[self.i_valid]
test_x = self.x_data[self.i_test]
test_y = self.y_data[self.i_test]
print 'tx:', np.shape(train_x)
print 'ty:', np.shape(train_y)
print 'vx:', np.shape(valid_x)
print 'vy:', np.shape(valid_y)
if (self.iteration == Model.BestValidationInterval):
self.best_loss = np.inf
self.iteration = 0
if self.initialized:
self.lr_shared.set_value(np.float32(self.learning_rate))
self.m_shared.set_value(np.float32(self.momentum))
self.train_x.set_value(np.float32(train_x))
self.valid_x.set_value(np.float32(valid_x))
self.test_x.set_value(np.float32(test_x))
self.train_y.owner.inputs[0].set_value(np.int32(train_y))
self.valid_y.owner.inputs[0].set_value(np.int32(valid_y))
self.test_y.owner.inputs[0].set_value(np.int32(test_y))
else:
self.y = T.ivector(
'y') # the labels are presented as 1D vector of [int] labels
self.lr = T.scalar('learning_rate')
self.m = T.scalar('momentum')
self.lr_shared = theano.shared(np.float32(self.learning_rate))
self.m_shared = theano.shared(np.float32(self.momentum))
self.train_x = theano.shared(train_x, borrow=True)
self.valid_x = theano.shared(valid_x, borrow=True)
self.test_x = theano.shared(test_x, borrow=True)
self.train_y = theano.shared(train_y, borrow=True)
self.valid_y = theano.shared(valid_y, borrow=True)
self.test_y = theano.shared(test_y, borrow=True)
self.train_y = T.cast(self.train_y, 'int32')
self.valid_y = T.cast(self.valid_y, 'int32')
self.test_y = T.cast(self.test_y, 'int32')
self.intialized = True
def sampleTrainingDataold(self):
# iteration counter
self.iteration += 1
if self.iteration == Model.BestValidationInterval:
self.drawRandomizedSamples()
else:
self.drawStratifiedSamples()
if len(self.i_train) > 0:
print 'second....'
self.i_train = []
if self.iteration == Model.BestValidationInterval:
self.drawRandomizedSamples()
else:
self.drawStratifiedSamples()
for i, n_draw in enumerate(self.n_draws):
indices = self.i_train_labels[i]
p_train = self.p_data[indices]
i_sorted = np.argsort(p_train, axis=0)
n_good = len(np.where(p_train[i_sorted] == 0)[0])
n_bad = len(indices) - n_good
#n_good = max( n_good, n_draw/2)
#i_good = i_sorted[ : n_good ]
#i_bad = i_sorted[ n_good: ]
n_threshold = int(len(indices) * 0.30)
print 'n_good:', n_good
print 'n_bad:', n_bad
print 'min:', n_threshold
# if not enough bad samples, just pick random samples
if self.iteration == Model.BestValidationInterval:
print '--->random sampoing<----'
self.i_labels[i] = np.hstack(
(self.i_labels[i], self.i_train_labels[i]))
n_ldata = len(self.i_labels[i])
print 'n_ldata:', n_ldata
self.i_labels[i] = np.random.choice(self.i_labels[i],
n_ldata,
replace=False)
self.i_train_labels[i] = self.i_labels[i][:n_draw]
self.i_labels[i] = self.i_labels[i][n_draw:]
self.i_train = np.concatenate(
(self.i_train, self.i_train_labels[i]))
self.drawRandomizedSamples()
else:
self.drawStratifiedSamples()
print '--->5050 sampoing<-----'
# keep 50% of the bad samples and replace the rest
n_good = max(n_good, n_draw / 2)
i_good = i_sorted[:n_good]
i_bad = i_sorted[n_good:]
self.i_labels[i] = np.hstack((self.i_labels[i], i_good))
i_new = self.i_labels[i][:n_good]
self.i_train_labels[i] = np.hstack((i_new, i_bad))
self.i_train = np.concatenate(
(self.i_train, self.i_train_labels[i]))
else:
#self.drawStratifiedSamples()
self.i_train = []
self.i_train_labels = [[] for l in self.n_draws]
for i, n_draw in enumerate(self.n_draws):
self.i_train_labels[i] = self.i_labels[i][:n_draw]
self.i_labels[i] = self.i_labels[i][n_draw:]
self.i_train = np.concatenate(
(self.i_train, self.i_train_labels[i]))
#i_draw = np.random.choice( self.i_labels[i], n_draw, replace=False )
print i, n_draw, len(self.i_train_labels[i]), len(
self.i_labels[i])
self.i_train = self.i_train.astype(dtype=int)
lens = [len(l) for l in self.i_labels]
print 'remaining sizes:', lens
lens = [len(l) for l in self.i_train_labels]
print 'ampleed sizes:', lens
print 'train indices:', len(self.i_train)
print self.i_train
#self.i = self.i[ self.n_train: ]
#self.i_train = self.i[ 0 : self.n_train ]
#self.i = self.i[ self.n_train: ]
train_x = self.x_data[self.i_train]
train_y = self.y_data[self.i_train]
valid_x = self.x_data[self.i_valid]
valid_y = self.y_data[self.i_valid]
test_x = self.x_data[self.i_test]
test_y = self.y_data[self.i_test]
print 'tx:', np.shape(train_x)
print 'ty:', np.shape(train_y)
print 'vx:', np.shape(valid_x)
print 'vy:', np.shape(valid_y)
exit(1)
if (self.iteration == Model.BestValidationInterval):
self.best_loss = np.inf
self.iteration = 0
if self.initialized:
print 'exiting here....'
self.lr_shared.set_value(np.float32(self.learning_rate))
self.m_shared.set_value(np.float32(self.momentum))
self.train_x.set_value(np.float32(train_x))
self.valid_x.set_value(np.float32(valid_x))
self.test_x.set_value(np.float32(test_x))
self.train_y.owner.inputs[0].set_value(np.int32(train_y))
self.valid_y.owner.inputs[0].set_value(np.int32(valid_y))
self.test_y.owner.inputs[0].set_value(np.int32(test_y))
else:
# allocate symbolic variables for the data
#self.index = T.lscalar() # index to a [mini]batch
#self.x = T.matrix('x') # the data is presented as rasterized images
#self.x = self.input
self.y = T.ivector(
'y') # the labels are presented as 1D vector of [int] labels
self.lr = T.scalar('learning_rate')
self.m = T.scalar('momentum')
self.lr_shared = theano.shared(np.float32(self.learning_rate))
self.m_shared = theano.shared(np.float32(self.momentum))
self.train_x = theano.shared(train_x, borrow=True)
self.valid_x = theano.shared(valid_x, borrow=True)
self.test_x = theano.shared(test_x, borrow=True)
self.train_y = theano.shared(train_y, borrow=True)
self.valid_y = theano.shared(valid_y, borrow=True)
self.test_y = theano.shared(test_y, borrow=True)
self.train_y = T.cast(self.train_y, 'int32')
self.valid_y = T.cast(self.valid_y, 'int32')
self.test_y = T.cast(self.test_y, 'int32')
self.intialized = True
def setupSegmentation(self):
self.intialized = True
def train(self):
pass
def classify(self, image):
pass
def predict(self, image):
pass
def save(self):
print 'saving model...'
pass
def load(self):
pass
def reportTrainingStats(self,
elapsedTime,
batchIndex,
valLoss,
trainCost,
mode=0):
DB.storeTrainingStats(self.id, valLoss, trainCost, mode=mode)
msg = '(%0.1f) %i %f%%'%\
(
elapsedTime,
batchIndex,
valLoss
)
status = '[%f]' % (trainCost)
Utility.report_status(msg, status)
from utility import enum # a value of -1 means we don't have the information # for booleans, we use an enum which follows the convention : -1 is unknown, 0 is False, 1 is True TEAM_SIDE = enum(UNKNOWN=-1, LEFT_BOT=0, RIGHT_TOP=1) SBIRE_TYPE = enum(UNKNOWN=-1, MELEE=0, RANGE=1, GIB=2, HUGE=3) COST_TYPE = enum(UNKNOWN=-1, MANA=0, ENERGY=1, HEALTH=2, NOTHING=3) CHAMPION = enum(UNKNOWN=-1, VEIGAR=0, TRISTANA=1, AMUMU=2, LULU=3, POPPY=4) # because yordles are op... BOOL = enum(UNKNOWN=-1, FALSE=0, TRUE=1)
