def __init__(self):
print("Init")
if self.instance is not None:
raise ValueError(
"looks like you forgot to call getbot function ? ")
self.stemmer = LancasterStemmer()
data = pickle.load(open(paths.getPath('trained_data'), "rb"))
self.words = data['words']
self.classes = data['classes']
train_x = data['train_x']
train_y = data['train_y']
with open(paths.getJsonPath()) as json_data:
self.intents = json.load(json_data)
net = tflearn.input_data(shape=[None, len(train_x[0])])
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(net,
len(train_y[0]),
activation='softmax')
net = tflearn.regression(net)
self.model = tflearn.DNN(net,
tensorboard_dir=paths.getPath('train_logs'))
self.model.load(paths.getPath('model.tflearn'))
def saveGoals( self, fileName='goals.txt' ):
'''Saves the goals to the specified file.
@param fileName The name to write the goals to.
'''
Goals.DIGITS = 2
path = paths.getPath( fileName, False )
print "Writing goals to:", path
f = open( path, 'w' )
f.write ('''<?xml version="1.0"?>
<Population >
''')
for gs in self.goalEditor:
node = gs.xmlElement()
if ( node ):
node.writexml( f, indent=' ', addindent=' ', newl='\n' )
f.write( '\n</Population>' )
f.close()
def handleKey(event, context, view, theGraph, obstacles, agents, field):
global editState
result = ContextResult()
if (context):
result = context.handleKeyboard(event, view)
if (result.isHandled()):
return result
mods = pygame.key.get_mods()
hasShift = mods & pygame.KMOD_SHIFT
hasCtrl = mods & pygame.KMOD_CTRL
hasAlt = mods & pygame.KMOD_ALT
if (event.type == pygame.KEYDOWN):
if (event.key == pygame.K_p):
print theGraph
elif (event.key == pygame.K_s and hasCtrl):
if (editState == GRAPH_EDIT):
fileName = paths.getPath('graph.txt', False)
f = open(fileName, 'w')
f.write('%s\n' % theGraph.newAscii())
f.close()
print "Graph saved!", fileName
elif (event.key == pygame.K_e):
editState = (editState + 1) % EDIT_STATE_COUNT
result.setNeedsRedraw(True)
elif (event.key == pygame.K_DELETE):
if (editState == GRAPH_EDIT):
if (theGraph.fromID != None):
theGraph.deleteVertex(theGraph.fromID)
theGraph.fromID = None
result.setNeedsRedraw(True)
elif (theGraph.activeEdge != None):
theGraph.deleteEdge(theGraph.activeEdge)
theGraph.activeEdge = None
result.setNeedsRedraw(True)
elif (hasAlt and event.key == pygame.K_F4):
raise Exception
return result
def main():
import optparse
parser = optparse.OptionParser()
parser.add_option("-r",
"--roadmap",
help="Optional roadmap file to load",
action="store",
dest="roadmapName",
default='')
parser.add_option("-g",
"--goals",
help="Optional goal definition file to load",
action="store",
dest="goalsName",
default='')
parser.add_option("-b",
"--obstacle",
help="Optional obstacle file to load.",
action="store",
dest='obstName',
default='')
parser.add_option("-a",
"--agent",
help="Optional agent position file to load.",
action="store",
dest='agtName',
default='')
parser.add_option("-f",
"--field",
help="Optional vector field file to load.",
action="store",
dest='fieldName',
default='')
parser.add_option(
"-c",
"--createField",
help=
'Creates a field based on the domain of the obstacles - ignored if --field(-f) is specified',
action='store_true',
dest='createField',
default=False)
parser.add_option("-s",
"--scb",
help="Optional scb file to load.",
action="store",
dest='scbName',
default='')
parser.add_option(
"-i",
"--inDir",
help=
"Optional directory to find input files - only applied to file names with relative paths",
action="store",
dest='inDir',
default='.')
parser.add_option(
"-o",
"--outDir",
help=
"Optional directory to write output files - only applied to file names with relative paths",
action="store",
dest='outDir',
default='.')
parser.add_option(
'--region',
help=
'Specify the bounding region of the action. If provided, it will set the initial camera properties. Format is minX minY maxX maxY',
nargs=4,
type='float',
dest='boundRegion',
default=None)
options, args = parser.parse_args()
paths.setInDir(options.inDir)
paths.setOutDir(options.outDir)
obstName = paths.getPath(options.obstName)
agtName = paths.getPath(options.agtName)
roadmapName = paths.getPath(options.roadmapName)
fieldName = paths.getPath(options.fieldName)
scbName = paths.getPath(options.scbName)
goalsName = paths.getPath(options.goalsName)
print "Arguments:"
print "\tInput dir: ", options.inDir
print "\tOutput dir:", options.outDir
print "\tobstacles: ", obstName
print "\tagents: ", agtName
print "\troad map: ", roadmapName
print "\tfield: ", fieldName
print "\tscbName: ", scbName
print "\tGoals: ", goalsName
if (obstName):
obstacles, bb = readObstacles(obstName)
else:
obstacles = ObstacleSet()
bb = AABB()
if (not options.boundRegion is None):
bb.min = Vector3(options.boundRegion[0], options.boundRegion[1], 0)
bb.max = Vector3(options.boundRegion[2], options.boundRegion[3], 0)
print bb
else:
bb.min = Vector3(-100, -100, 0)
bb.max = Vector3(100, 100, 0)
agents = AgentSet(0.23)
if (agtName):
agents.initFromFile(agtName)
if (scbName != ''):
# This reads the agent radius from the file, but no longer visualizes the agents
# This is a bit of a hack. It would be better to simply disable drawing the
# agents when the scb playback context is active. But I don't have the time
# to figure out an elegant solution to this.
agents.clear()
if (fieldName):
field = GLVectorField((0, 0), (1, 1), 1)
field.read(fieldName)
elif (options.createField):
print "Instantiate vector field from geometry:", bb
bbSize = bb.max - bb.min
field = GLVectorField((bb.min.x, bb.min.y), (bbSize.y, bbSize.x), 2.0)
else:
field = None
graph = Graph()
if (roadmapName):
graph.initFromFile(roadmapName)
# create viewer
pygame.init()
fontname = pygame.font.get_default_font()
font = pygame.font.Font(fontname, 12)
w = bb.max.x - bb.min.x
h = bb.max.y - bb.min.y
view = View((w, h), (bb.min.x, bb.min.y), (w, h), (bb.min.x, bb.min.y),
(800, 600), font)
view.HELP_TEXT = 'View controls:' + \
'\n\tpan - Ctrl + left mouse button' + \
'\n\tzoom in - mouse wheel up' + \
'\n\tzoom out - mouse wheel down' + \
'\n\tzoom - Shift + left mouse button (up and down)'
view.initWindow('Create roadmap')
pygame.key.set_repeat(250, 10)
view.initGL()
if (field):
field.newGLContext()
## field = None
# load goals
if (goalsName):
try:
goalSets = Goals.readGoals(goalsName)
except ValueError as e:
print "Error parsing goals:", str(e)
print "\tUsing empty GoalSet"
goalSets = [Goals.GoalSet()]
else:
goalSets = [Goals.GoalSet()]
goalVis = GoalEditor(goalSets)
context = ContextSwitcher()
context.addContext(PositionContext(), pygame.K_q)
context.addContext(GoalContext(goalVis), pygame.K_g)
context.addContext(AgentContext(agents), pygame.K_a)
context.addContext(ObstacleContext(obstacles), pygame.K_o)
if (field):
context.addContext(FieldEditContext(field), pygame.K_f)
if (scbName != ''):
context.addContext(SCBContext(scbName, obstacles, agents.defRadius),
pygame.K_s)
context.newGLContext()
redraw = True
running = True
while running:
for event in pygame.event.get():
if event.type == pygame.QUIT:
running = False
break
elif (event.type == pygame.MOUSEMOTION
or event.type == pygame.MOUSEBUTTONUP
or event.type == pygame.MOUSEBUTTONDOWN):
result = handleMouse(event, context, view, graph, obstacles,
agents, field)
redraw |= result.needsRedraw()
elif (event.type == pygame.KEYDOWN or event.type == pygame.KEYUP):
try:
result = handleKey(event, context, view, graph, obstacles,
agents, field)
redraw |= result.needsRedraw()
except Exception as e:
print "Error with keyboard event"
print "\t", e
running = False
elif (event.type == pygame.VIDEORESIZE):
view.resizeGL(event.size)
if (field):
field.newGLContext()
context.newGLContext()
redraw |= True
elif (event.type == pygame.VIDEOEXPOSE):
redraw |= True
if (redraw):
drawGL(view, context, obstacles, graph, agents, field, goalVis)
message(view, updateMsg(agents.count()))
pygame.display.flip()
if (context):
name = context.exportDisplay()
if (name):
pygame.image.save(view.surface, name)
redraw = False
writeRoadmap()
training.append([bag, output_row])
random.shuffle(training)
training = np.array(training)
train_x = list(training[:, 0])
train_y = list(training[:, 1])
tf.reset_default_graph()
net = tflearn.input_data(shape=[None, len(train_x[0])])
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(net, 8)
net = tflearn.fully_connected(net, len(train_y[0]), activation='softmax')
net = tflearn.regression(net)
model = tflearn.DNN(net, tensorboard_dir=paths.getPath('train_logs'))
model.fit(train_x, train_y, n_epoch=20000, batch_size=500, show_metric=True)
model.save(paths.getPath('model.tflearn'))
def clean_up_sentence(sentence):
sentence_words = nltk.word_tokenize(sentence)
sentence_words = [stemmer.stem(word.lower()) for word in sentence_words]
return sentence_words
def bow(sentence, words, show_details=False):
sentence_words = clean_up_sentence(sentence)
bag = [0] * len(words)
for s in sentence_words:
for i, w in enumerate(words):