def main(args):
tsrc_distribution = pkg_resources.get_distribution("tsrc")
# pylint: disable=no-member
version = tsrc_distribution.version
location = path.Path(tsrc_distribution.location)
dirty = False
short_hash = None
rc, out = tsrc.git.run_git(location,
"rev-parse",
"--short",
"HEAD",
raises=False)
if rc == 0:
short_hash = out
dirty = tsrc.git.is_dirty(location)
message = "tsrc version %s" % version
if short_hash:
message += " - git: %s" % short_hash
if dirty:
message += " (dirty)"
ui.info(message)
def run_from_config(config_dir, gpu_devices, script_name):
"""Call a script using parameters from an input JSON config file.
CUDA_VISIBLE_DEVICES is set using the gpu_devices parameter passed to this
Python script.
"""
config_dir = path.Path(config_dir)
with open(config_dir / 'config.json', 'r') as f:
config = json.load(f)
config_string = ""
for option in config:
# NOTE(brendan): Booleans are passed as --no-<option-name> for False,
# and --<option-name> for True.
if isinstance(config[option], bool):
yes_no = '' if config[option] else 'no-'
config_string += ' --' + yes_no + option
elif isinstance(config[option], list):
config_string += ' --' + option
for thing in config[option]:
config_string += ' ' + str(thing)
elif config[option] is not None:
config_string += ' --' + option + ' ' + str(config[option])
cuda_visible_devices = 'CUDA_VISIBLE_DEVICES='
if gpu_devices == 'CPU':
gpu_devices = ''
else:
gpu_warn = 'gpu_devices should be of the format <int>(,<int>)*'
assert (re.match(r'^\d(,\d)*$', gpu_devices) is not None), gpu_warn
cuda_visible_devices += gpu_devices
if script_name.endswith('.py'):
cmd = f'{cuda_visible_devices} python3 {script_name} {config_string}'
else:
cmd = f'{cuda_visible_devices} python3 -m {script_name} {config_string}'
print(cmd)
os.system(cmd)
def main():
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter, )
parser.add_argument("--nogui", action="store_true", help="no gui")
args = parser.parse_args()
if args.nogui:
connection_method = pybullet.DIRECT
else:
connection_method = pybullet.GUI
now = datetime.datetime.utcnow()
timestamp = now.strftime("%Y%m%d_%H%M%S.%f")
root_dir = morefusion.utils.get_data_path(
f"wkentaro/morefusion/ycb_video/synthetic_data/{timestamp}")
root_dir = path.Path(root_dir)
n_video = 1200
for index in range(1, n_video + 1):
video_dir = root_dir / f"{index:08d}"
random_state = np.random.RandomState(index)
generate_a_video(video_dir, random_state, connection_method)
def _prepare_upload(self):
session = self.account.http_settings.session
proxy = self.account.http_settings.proxy
if not self.account.is_logined():
raise NotLogined('YouPorn account is not logined')
upload_page = session.get('http://www.youporn.com/upload',
proxies=proxy)
doc = etree.fromstring(upload_page.content, HTMLParser())
callback_url = doc.xpath('//input[@name="callbackUrl"]/@value')[0]
session.headers.update({
"X-Requested-With":
"XMLHttpRequest",
"Content-Type":
"application/x-www-form-urlencoded; charset=UTF-8"
})
video_size = os.path.getsize(self.video_upload_request.video_file)
video = path.Path(self.video_upload_request.video_file)
post = {'file': video.name, 'size': video_size}
create_upload_request = session.post(
'http://www.youporn.com/upload/create-videos/',
data=post,
proxies=proxy)
response = json.loads(create_upload_request.content)
if 'success' in response:
if not response['success']:
raise FailedUpload(
'Failed to upload video reason:{reason}'.format(
reason=response['reason']))
del session.headers['X-Requested-With']
response['callback_url'] = callback_url
return response
def test_filter_daophot_out(self):
with path.Path(self.test_dir):
print("Files before: ", glob.glob("*"))
prefix = "20180611T055249_053"
file_in = prefix + ".coo"
file_out = prefix + ".lst"
xmin, xmax, ymin, ymax = 5, 45, 5, 45
daophot.filter_daophot_out(file_in, file_out, xmin, xmax, ymin,
ymax)
print("Files after: ", glob.glob("*"))
# Make sure output files exists.
self.assertTrue(os.path.exists(file_out))
# Check that only x y ramin that fall inside of xmin,xmax and ymin,ymax.
with open(file_out) as f:
ll = f.readlines()
for l in ll[3:]:
tt = l.split()
x, y = float(tt[1]), float(tt[2])
self.assertTrue(x > xmin)
self.assertTrue(y < xmax)
self.assertTrue(y > ymin)
self.assertTrue(y < ymax)
def test():
print("lane detector test")
img_files = path.Path("./data/cut").glob("*")
num_imgs = len(img_files)
print("Total %d imgs" % num_imgs)
imgs = list(
map(lambda x: cv2.imread(str(x), cv2.COLOR_RGB2GRAY), img_files))
laneDetector = LaneDetector(config)
for idx, img in enumerate(imgs):
type_result = laneDetector.pcr_detect(img)
print(idx, "번째 pcr:", type_result)
if type_result[0][0] == 3:
print(idx, " th : no parking lot")
continue
# type_result = [[0], [0]]
result, sess = laneDetector.psd_detect(type_result, img)
print(idx, "번째 psd:", result)
sess.close()
tf.reset_default_graph()
def __init__(self, img_file, spawningpoint):
"""
Descr: __init__ sets initial variables and image
Params: self
img_file is a string containing file name
spawningpoint is a tuple containing coordinates
Returns: none """
pygame.sprite.Sprite.__init__(self)
#Puts image into the invader
image = pygame.image.load('assets/' + img_file).convert_alpha()
self.image = pygame.transform.scale(image, (20, 20))
#Maintains reference to rectangle/allows movement
self.rect = self.image.get_rect()
self.rect.x = spawningpoint[0]
self.rect.y = spawningpoint[1]
#Stats of the Invaders
self.health = 100
self.speed = 30
self.location = 1
self.p = path.Path()
self.whichimage = 1
def execute(self, msg):
"""
程序本体逻辑，processor之间存在依赖关系，通过在另一个类中数据传递处理这种依赖(数据传递的耦合相对较低)，
避免了被依赖类的对象向另一个类对象传递的高度耦合。
"""
if self.input.condition and not self.output(None).listdir():
print(self.__class__.__name__)
p2s = "====".join(self.input.path2.listdir())
for f1_full_name in self.input.path1.listdir():
f1name = f1_full_name.name.splitext()[0]
f1_index = p2s.index(f1name)
f2_full_name = self.input.path2.joinpath(p2s[f1_index:(
f1_index + f1name.__len__())] + "_result_1.csv")
print(f1_full_name, f1_full_name)
df = pd.read_csv(f1_full_name).merge(pd.read_csv(f2_full_name),
on="time10",
how="inner")
gsm = GaitStatusMarkerProcessor(ProjParaShop.STATUS_DEFINITION)
df = gsm.process(df)
df.to_csv(path.Path(
ProjParaShop.FORMED_SIGNAL_MARK_PATH).joinpath(
f1_full_name.name),
index=False)
def generate(self, m=None):
if m is None:
m = self.matrix
object_path = path.Path()
svg_parser.parse_svg_path(object_path, self.path)
self.box = object_path.bbox()
if VARIABLE_NAME_SPEED in self.properties:
speed = self.properties.get(VARIABLE_NAME_SPEED)
yield COMMAND_SET_SPEED, speed
if VARIABLE_NAME_POWER in self.properties:
power = self.properties.get(VARIABLE_NAME_POWER)
yield COMMAND_SET_POWER, power
if VARIABLE_NAME_DRATIO in self.properties:
d_ratio = self.properties.get(VARIABLE_NAME_DRATIO)
yield COMMAND_SET_D_RATIO, d_ratio
plot = object_path * m
first_point = plot.first_point
yield COMMAND_RAPID_MOVE, first_point
yield COMMAND_SET_STEP, 0
yield COMMAND_MODE_COMPACT, 0
yield COMMAND_PLOT, plot
yield COMMAND_MODE_DEFAULT, 0
yield COMMAND_SET_SPEED, None
yield COMMAND_SET_D_RATIO, None
def Compute(self, x, y,currPath, desPath, theta, r, d):
del self.path.route
del self.path.point
self.path = path.Path(MAXSTEER)
p1 = path.Point(x,y)
self.currentPath = currPath
self.destPath = desPath
p2 = path.Point(x + 14*math.cos(theta+ d*math.pi/16.0), y + 14*math.sin(theta+ d*math.pi/16.0))
self.computed = False
for j in range(0, len(self.destPath.point)-1):
p3 = self.destPath.point[j]
p4 = self.destPath.point[j+1]
if p4.id == r:
if self.intersect(p1, p2, p3, p4) :
#print '{0} {1} {2} {3}'.format(1*d, math.pi/7.8, -1*d, math.pi/7.8)
self.path.append(1*d, math.pi/7.8)
self.path.append(-1*d, math.pi/7.8)
self.path.computePath(p1.x, p1.y)
self.computed = True
break;
self.p1 = p1
self.p2 = p2
self.theta = theta
def eventCalculator(cls):
"""对数据源的数据执行不同决策策略"""
from .services import EventCalculator
from .core_algo import StrategyResult2
calculator = EventCalculator()
# 待计算的数据位置
root_p = path.Path(
r'E:\my_proj\fog_recognition\ExtendFoGData\fixed_data\Tuned4Model')
data = joblib.load(root_p.joinpath('ModelParaSelector/15'))
for k in data:
data[k]["predict_result"].to_csv(
root_p.joinpath(
'ModelParaSelector/SelectedParaResult').joinpath(k))
data_path = root_p.joinpath('ModelParaSelector/SelectedParaResult')
calculator.set_para_with_prop({
"strategy":
StrategyResult2(),
"data_path":
data_path,
"save_path":
root_p.joinpath('EventParaSelector')
})
return calculator
import path import vtk #print(dir(path)) ## create a Path . # cpt1 = [2.0, 0.0, 0.0] cpt2 = [3.0, 0.0, 0.0] cpt3 = [4.0, 0.0, 0.0] cpt4 = [5.0, 0.0, 0.0] path = path.Path() path.add_control_point(cpt1) path.add_control_point(cpt2) path.add_control_point(cpt3) path.add_control_point(cpt4) path_geom = path.get_geometry()
import cv2
import path
import numpy as np
from PreProcessamento import PreProcessamento
# Obtem o camihho para o arquivo do haarcascade
ARQUIVO_HAARCASCADE_FRONTALFACE = path.Path(
"resources\\haarcascade_frontalface_default.xml")
class CapturaCamera:
def __init__(self, preProc, classificadorFaces):
self.PreProc = preProc
# Cria um classificador com base no Haarcascade
self.classificadorFaces = classificadorFaces
def CapturarFrames(self):
# cria a camera a partir do opencv
camera = cv2.VideoCapture(0)
while (True):
# Lê da camera
conectado, imagem = camera.read()
# Cria uma imagem cinza
imagemCinza = cv2.cvtColor(imagem, cv2.COLOR_BGR2GRAY)
# Obtem as faces detectadas pelo haarcascade
# facesDetectadas = self.classificadorFaces.detectMultiScale(imagemCinza, scaleFactor=1.5, minSize=(100, 100))
facesDetectadas = self.classificadorFaces.CapturarRosto(
imagemCinza)
# Desenha os retangulos de acordo com as faces obtidas pelo haarcascade
for (x, y, l, a) in facesDetectadas:
os.remove(file_name)
def check(p, KOM):
keywords = [re.sub(r'[ -]', '_', kw.name[1:]) for kw in KOM.keywords]
keywords = sorted(set(keywords))
# print(keywords)
print('Total {} keywords'.format(len(keywords)))
pages = [fn for fn in os.listdir(p.doc) if fn.endswith('.html')]
pages = sorted(pages)
print('Total {} HTML pages'.format(len(pages)))
# print(pages)
if len(keywords) > len(pages):
for page in pages:
if page[:-5] in keywords:
keywords.remove(page[:-5])
print('Those keywords have no HTML pages:')
print(keywords)
# Run test
if __name__ == '__main__':
clean.screen()
p = path.Path()
s = settings.Settings(p)
KOM = kom.KOM(p, s)
# regenerate_documentation(p, KOM)
# remove_html_trash(p, KOM)
# remove_png_trash()
# check(p, KOM)
def approximate_path_by_pathcount(args, result_path, source_path, ktest_tool_path):
if(not "=" in args):
print("Usage: python find_approx.py --approximate-path-by-pathcount=<N>")
sys.exit()
pathcount_threshold = int(args.split('=')[1])
if(pathcount_threshold > 100):
print("Path count threshold should be less than 100")
sys.exit()
#print
print("Source: " + source_path)
print("Output: " + result_path)
print("Selected path count threshold: %.2f%%" % pathcount_threshold)
# create all path objects
paths = []
input_error_repeat = 100
scaling = 1.0
probability_sum = 0.0
for root, dirs, files in os.walk(result_path):
for filename in files:
if filename.endswith(".prob"):
with open(result_path + "/" + filename, 'r') as fin:
idx = int(fin.readline().split(",")[2].strip())
prob = float(fin.readline().split(",")[1])
new_path = path.Path(idx, prob)
paths.append(new_path)
probability_sum += prob
# Get the input variables and their types and mark those for which error is tracked
# TODO: Handle floats converted to ints (we only need to do this handling if the conversion happened in the input)
source = open(source_path, "r")
input_variables = []
for line in source:
if re.match("(.*)klee_make_symbolic(.*)", line):
tokens = re.split(r'[(|)]|\"', line)
input_variables.append((tokens[2], tokens[4]))
source.close()
# print(input_variables)
source = open(source_path, "r")
approximable_input = []
for line in source:
if re.match("(.*)klee_track_error(.*)", line):
tokens = re.split(r'[(|)]|\"|&|,', line)
approximable_input.append(tokens[2])
source.close()
# Maintain a measure of the approximability of the input
input_approximability_count = []
expression_count = 0
for var in approximable_input:
input_approximability_count.append(0)
# Get the non-approximable input
non_approximable_input = list(set([x[1] for x in input_variables]) - set(approximable_input))
#sort by path probability
paths.sort(key=lambda p: p.path_prob)
# find approximable variables in each path
running_path_count = 0.0
all_variables = set()
for p in paths:
running_path_count += 1
if(((running_path_count * 100) / len(paths)) > pathcount_threshold):
break
# Get the path condition with error
path_condition_with_error = ""
source = open(result_path + "/" + "test" + "{:0>6}".format(str(p.path_id)) + ".kquery_precision_error", "r")
for line in source:
path_condition_with_error += line.rstrip("\n\r")
path_condition_with_error += " "
source.close()
path_condition_with_error = path_condition_with_error.replace("!", "not")
path_condition_with_error = path_condition_with_error.replace(" = ", " == ")
path_condition_with_error = path_condition_with_error.replace("&&", "and")
path_condition_with_error = path_condition_with_error.replace(">> 0", "")
path_condition_with_error = path_condition_with_error.replace(">> ", "/2**")
path_condition_with_error = path_condition_with_error.replace("<< ", "*2**")
# generate an input, for which the path condition is satisfied
result = subprocess.run([ktest_tool_path, '--write-ints', result_path + "/" + "test" + "{:0>6}".format(str(p.path_id)) + '.ktest'], stdout=subprocess.PIPE)
output_string = result.stdout.decode('utf-8')
tokens = re.split(r'\n|:', output_string)
idx = 5
num_args = int(tokens[idx].strip())
for args in range(num_args):
exec("%s = %d" % (tokens[idx + 3].strip().replace("'", ""), int(tokens[idx + 9].strip())))
idx += 9
if(not os.path.isfile(result_path + "test" + "{:0>6}".format(str(p.path_id)) + '.precision_error')):
continue
with open(result_path + "/" + "test" + "{:0>6}".format(str(p.path_id)) + '.precision_error', 'r') as infile:
for line in infile:
method_name_line_tokens = line.split()
if(len(method_name_line_tokens) > 0 and method_name_line_tokens[0] == 'Line'):
method_name = method_name_line_tokens[4].rstrip(':')
# process expression line
next_line = infile.readline()
tokens = next_line.split()
if(len(tokens) > 0 and tokens[0] == 'Output'):
expression_count += 1
# if the error expression is 0, add to non-approximable list
if(tokens[5] == '0'):
p.non_approximable_var.append((tokens[3].strip(), method_name))
all_variables.add(tokens[3])
p.all_var.append(tokens[3].strip())
continue
# read and sanitize expression
exp = next_line.split(' ', 5)[5].strip("\n")
exp = exp.replace(">> 0", "")
exp = exp.replace(">> ", "/2**")
exp = exp.replace("<< ", "*2**")
is_var_approximable = 0
# For each approximable input variable
for idx, var in enumerate(approximable_input):
# assign other variable errors to zero
for temp_var in approximable_input:
var_with_err_name = temp_var + "_err"
exec("%s = %f" % (var_with_err_name, 0.0))
# for repeat
result = []
for x in range(input_error_repeat):
# Generate a random error value in (0,1) for the concerned variable
var_with_err_name = var + "_err"
input_error = random.uniform(0.0, 1.0)
exec("%s = %f" % (var_with_err_name, input_error))
# Check if path condition with error is satisfied
if(eval(path_condition_with_error)):
# If satisfied, get the output error from expression
output_error = eval(exp)
result.append((input_error, output_error))
input_approximability_count[idx] += 1
if(len(result)):
# Check for monotonicity of output error. If not monotonous continue to evaluate other inputs.
result = sorted(result, key=lambda x: x[0])
monotonous_count = 0
for index, item in enumerate(result):
if(index < (len(result) - 1) and item[1] <= result[index + 1][1]):
monotonous_count += 1
# If at least 90% monotonous, get the linear regression gradient
if((monotonous_count / (len(result) - 1)) >= 0.8):
list_x, list_y = zip(*result)
# linear reqression code from https://www.geeksforgeeks.org/linear-regression-python-implementation/
xdata = np.array(list_x)
ydata = np.array(list_y)
n = np.size(xdata)
m_x, m_y = np.mean(xdata), np.mean(ydata)
SS_xy = np.sum(ydata * xdata - n * m_y * m_x)
SS_xx = np.sum(xdata * xdata - n * m_x * m_x)
b_1 = SS_xy / SS_xx
# If gradient > 50% mark as non-approximable, else continue for other variables in the expression
if(b_1 <= 0.5):
is_var_approximable = 1
# If for at least one variable in the expression, the output is approximable, then add to approximable list.
# Else add to the non-approximable list
all_variables.add(tokens[3].strip())
p.all_var.append(tokens[3].strip())
if(is_var_approximable):
p.approximable_var.append((tokens[3].strip(), method_name))
else:
p.non_approximable_var.append((tokens[3].strip(), method_name))
else:
continue
approximability_result = []
for var in all_variables:
path_score = 0.0
prob_score = 0.0
number_of_paths_present_count = 0
approximable_paths_count = 0
for p in paths:
# if variable appears in that path
if(var in p.all_var):
number_of_paths_present_count += 1
# if in approximable list
if(len(p.approximable_var) > 0):
approximable_var_in_path = list(zip(*p.approximable_var))[0]
if(var in approximable_var_in_path):
approximable_paths_count += 1
prob_score += p.path_prob
path_score = approximable_paths_count * 100 / number_of_paths_present_count
prob_score = prob_score * 100 / probability_sum
approximability_result.append((var, path_score, prob_score))
print("\nApproximability of program variables\n================================")
print("var_name\tpathscore\tprobability score")
for result in approximability_result:
print("%s\t\t%.2f\t\t%e" % (result[0], result[1], result[2]))
# for p in paths:
# print("%d %.2f" %(p.path_id,(p.path_prob * 100 / probability_sum)))
print("\nApproximability of input variables\n================================")
for idx, var in enumerate(approximable_input):
print(var + ' : %d%%' % ((input_approximability_count[idx] / (expression_count * input_error_repeat)) * 100))
def collect_sources(ignore_func):
top_path = path.Path(".")
for py_path in top_path.walkfiles("*.py"):
py_path = py_path.normpath() # get rid of the leading '.'
if not ignore_func(py_path):
yield py_path
#!/usr/bin/env python3
from cffi import FFI
import json
import path
ffi_builder = FFI()
libs = ["../.build/lib/libchuck-norris.a"]
includes = ["../include"]
conan_info = json.loads(path.Path("../.build/conanbuildinfo.json").text())
for dep in conan_info["dependencies"]:
for lib_name in dep["libs"]:
lib_file = "lib{}.a".format(lib_name)
for lib_path in dep["lib_paths"]:
candidate = path.Path(lib_path).joinpath(lib_file)
if candidate.exists():
libs.append(candidate)
else:
libs.append(lib_name)
for include_path in dep["include_paths"]:
includes.append(include_path)
print(libs)
print(includes)
ffi_builder.set_source(
"_chuck_fact",
"""
Django settings for calculator project. Generated by 'django-admin startproject' using Django 1.11.2. For more information on this file, see https://docs.djangoproject.com/en/1.11/topics/settings/ For the full list of settings and their values, see https://docs.djangoproject.com/en/1.11/ref/settings/ """ import os import path # Build paths inside the project like this: os.path.join(BASE_DIR, ...) BASE_DIR = path.Path(__file__).dirname().dirname().dirname() # Quick-start development settings - unsuitable for production # See https://docs.djangoproject.com/en/1.11/howto/deployment/checklist/ # SECURITY WARNING: keep the secret key used in production secret! SECRET_KEY = '(3xt92tw&nztwo1w1coe&9(kz$5=zl1bj824u5pe$*ax$qu-mc' # SECURITY WARNING: don't run with debug turned on in production! DEBUG = True ALLOWED_HOSTS = [] # Application definition INSTALLED_APPS = [
#!/usr/bin/env python
import sys
import path
import morefusion
here = path.Path(__file__).abspath().parent
sys.path.insert(0, here / "../ycb_video_checks")
from check_dataset import get_scene # NOQA
if __name__ == "__main__":
dataset = (
morefusion.datasets.YCBVideoPoseCNNResultsRGBDPoseEstimationDataset()
) # NOQA
morefusion.extra.trimesh.display_scenes(get_scene(dataset),
height=int(320 * 0.7),
width=int(480 * 0.7))
def _run_image(outfolder,
image_data,
cmd=None,
user='******',
make_tarball=False):
# download image
image_path = path.Path('img').realpath()
print("Ensuring presence of " + image_data.base_image_url)
ensure_image(
image_data.base_image_name,
image_data.base_image_url,
IMAGES_ROOT,
image_data.base_image_md5,
untar_to=image_path,
)
lxc_file_path = write_lxc_config(image_path)
lxc_name = 'build_image-' + image_data.new_image_name
script_path = None
if cmd is None:
# copy bootstrap script into place and ensure it's executable.
script = os.path.basename(image_data.script_url)
script_path = image_path / script
if os.path.exists(image_data.script_url):
shutil.copy(image_data.script_url, script_path)
else:
ensure_file(image_data.script_url, script_path)
script_path.chmod('a+x')
real_cmd = '/' + script
else:
real_cmd = cmd
# Call lxc-start, passing in our LXC config file and telling it to run
# our build script inside the container.
lxc_args = [
'lxc-start',
'--name',
lxc_name,
'--rcfile',
lxc_file_path,
'--',
real_cmd,
]
path_items = (
'/usr/local/sbin',
'/usr/local/bin',
'/usr/sbin',
'/usr/bin',
'/sbin',
'/bin',
'/usr/games',
)
env = {
'PATH': os.pathsep.join(path_items),
'HOME': '/root',
}
env.update(image_data.env or {})
if 'TERM' in os.environ:
env['TERM'] = os.environ['TERM']
logpath = os.path.join(outfolder, '%s.log' % image_data.new_image_name)
print("LOGPATH", logpath)
with open(logpath, 'w') as logfile:
tee(lxc_args, env, logfile)
# remove build script if we used one.
if cmd is not None:
os.remove(script_path)
if make_tarball:
img_name = '{image_data.new_image_name}.tar.gz'.format(**locals())
tardest = os.path.join(outfolder, img_name)
print("Compressing image to " + tardest)
with tarfile.open(tardest, 'w:gz') as tar:
tar.add(image_path, arcname='')
import path
a = path.Path('G:\python\BaiVeNha\log.txt','log.txt')
b = path.Path('G:\python\dcl\.gitignore','.gitignore')
c = path.Path('G:\python\BaiVeNha\README.md','README.md')
a.write_to_file('path.json')
b.write_to_file('path.json')
c.write_to_file('path.json')
import path
import sh
for project in path.Path(".").dirs():
with project:
sh.git.clean(force=True)
sh.git.reset(hard=True)
sh.make()
logger = logging.getLogger('cx')
logger.info('Start to Load Biological Data')
CX_Neuropils = ['PB', 'EB', 'NO', 'no', 'BU', 'bu', 'LAL', 'lal']
neuropil_name_to_node = {}
for neuropil in CX_Neuropils:
node = graph.Neuropils.create(name=neuropil, version=cx_version)
neuropil_name_to_node[neuropil] = node
logger.info('Created Neuropils')
# File names grouped by neuropil in which neurons' presynaptic terminals
# arborize:
data = path.Path('neurons')
# LAL subregions list
LAL_subregions_list = ['RGT', 'RDG', 'RVG', 'RHB']
# lal subregions list
lal_subregions_list = ['LGT', 'LDG', 'LVG', 'LHB']
# NO subregions list
NO_subregions_list = [
'(1,R)', '(2,RD)', '(2,RV)', '(3,RP)', '(3,RM)', '(3,RA)'
]
# no subregions list
no_subregions_list = [
'(1,L)', '(2,LD)', '(2,LV)', '(3,LP)', '(3,LM)', '(3,LA)'
]
def main(args):
vehicle_amount = 1
if len(args) > 1:
vehicle_amount = int(args[1])
vehicle_ids = []
for i in range(vehicle_amount):
vehicle_ids.append('v{}'.format(i + 1))
# PID parameters for path tracking.
k_p = 0.5
k_i = -0.02
k_d = 3
horizon = 15
delta_t = 0.1
Ad = numpy.matrix([[1., 0.], [delta_t, 1.]])
Bd = numpy.matrix([[delta_t], [0.]])
zeta = 0.5 # z = 1 -> full timegap tracking.
Q_v = 1 # Part of Q matrix for velocity tracking.
Q_s = 1 # Part of Q matrix for position tracking.
Q = numpy.array([Q_v, 0, 0, Q_s]).reshape(2, 2) # State tracking.
R_acc = 0.1
R = numpy.array([1]) * R_acc # Input tracking.
velocity_min = 0.
velocity_max = 2.
position_min = -100000.
position_max = 1000000.
acceleration_min = -1.5
acceleration_max = 1.5
truck_length = 0.3
safety_distance = 0.2
timegap = 1.
delay = 0.0
simulation_length = 40 # How many seconds to simulate.
xmin = numpy.array([velocity_min, position_min])
xmax = numpy.array([velocity_max, position_max])
umin = numpy.array([acceleration_min])
umax = numpy.array([acceleration_max])
# Reference speed profile.
opt_v_pts = 400 # How many points.
opt_v_max = 1.2
opt_v_min = 0.8
opt_v_period_length = 60 # Period in meters.
speed_ref = speed_profile.Speed()
speed_ref.generate_sin(opt_v_min, opt_v_max, opt_v_period_length, opt_v_pts)
speed_ref.repeating = True
# vopt = speed_profile.Speed([1], [1])
# Controller reference path.
x_radius = 1.4
y_radius = 1.2
center = [0.2, -y_radius / 2]
pts = 400
variance = 0.
plot_data = True
save_data = True
filename = 'measurements/sim_dmpc' + '_' + '_'.join(vehicle_ids) + '_'
pt = path.Path()
pt.gen_circle_path([x_radius, y_radius], points=pts, center=center)
start_distance = 0.5
path_len = pt.get_path_length()
vehicles = []
for i, vehicle_id in enumerate(vehicle_ids):
theta = (len(vehicle_ids) - i - 1)*2*math.pi*start_distance/path_len + 0.1
xcoord = center[0] + x_radius*math.cos(theta)
ycoord = center[1] + y_radius*math.sin(theta)
x = [xcoord, ycoord, theta + math.pi/2, 0]
# x = [center[0], center[1] + y_radius, math.pi, 0]
# # x = [0, 0, math.pi, 0]
vehicles.append(trxmodel.Trx(x=x, ID=vehicle_id))
mpc = DistributedMPC(vehicles, pt, Ad, Bd, delta_t, horizon, zeta, Q, R, truck_length,
safety_distance, timegap, k_p, k_i, k_d, simulation_length,
xmin=xmin, xmax=xmax, umin=umin, umax=umax, speed_ref=speed_ref,
delay=delay, variance=variance)
mpc.run()
if save_data:
mpc.save_data_as_rosbag(filename)
if plot_data:
mpc.plot_stuff()
def main(args):
# ID of the vehicle.
if len(args) < 2:
print('Need to enter at least one vehicle ID.')
sys.exit()
vehicle_id = args[1] # First argument is the ID of the vehicle.
if len(args) > 2:
preceding_id = args[2] # Second argument is, if entered, the ID of the preceding vehicle.
is_leader = False
else:
preceding_id = 'None' # If no second argument, the vehicle is the leader.
is_leader = True
# Topic name for subscribing to truck positions.
position_topic_name = 'mocap_state'
# Topic name for publishing vehicle commands.
control_topic_name = 'pwm_commands'
# Name for starting and stopping recording.
recording_service_name = vehicle_id + '/dmpc/set_measurement'
# Filename prefix for recording data.
recording_filename = 'dmpc' + '_' + vehicle_id + '_'
# PID parameters for path tracking.
k_p = 0.5
k_i = 0
k_d = 3
# MPC information.
horizon = 20
delta_t = 0.1
Ad = numpy.matrix([[1., 0.], [delta_t, 1.]])
Bd = numpy.matrix([[delta_t], [0.]])
zeta = 0.90
s0 = 0.
v0 = 0.
Q_v = 1 # Part of Q matrix for velocity tracking.
Q_s = 0.5 # Part of Q matrix for position tracking.
Q = numpy.array([Q_v, 0, 0, Q_s]).reshape(2, 2) # State tracking.
R_acc = 0.1
R = numpy.array([1]) * R_acc # Input tracking.
velocity_min = 0.
velocity_max = 2.
position_min = -100000.
position_max = 1000000.
acceleration_min = -0.5
acceleration_max = 0.5
truck_length = 0.2
safety_distance = 0.1
timegap = 1.
delay = 1.
x0 = numpy.array([s0, v0])
xmin = numpy.array([velocity_min, position_min])
xmax = numpy.array([velocity_max, position_max])
umin = numpy.array([acceleration_min])
umax = numpy.array([acceleration_max])
# Reference speed profile.
opt_v_pts = 1000 # How many points.
opt_v_max = 1.2
opt_v_min = 0.8
opt_v_period_length = 60 # Period in meters.
vopt = speed_profile.Speed()
vopt.generate_sin(opt_v_min, opt_v_max, opt_v_period_length, opt_v_pts)
vopt.repeating = True
# Controller reference path.
x_radius = 1.4
y_radius = 1.2
center = [0.2, -y_radius/2]
pts = 400
pt = path.Path()
pt.gen_circle_path([x_radius, y_radius], points=pts, center=center)
# Initialize controller.
controller = Controller(
position_topic_name, control_topic_name, vehicle_id, preceding_id, is_leader,
pt, Ad, Bd, delta_t, horizon, zeta, Q, R, truck_length, safety_distance, timegap,
speed_ref=vopt, xmin=xmin, xmax=xmax, umin=umin, umax=umax, x0=x0,
k_p=k_p, k_i=k_i, k_d=k_d, recording_service_name=recording_service_name,
recording_filename=recording_filename, delay=delay
)
# Start controller.
controller.run()
import path, mistune, flask
a, d = flask.Flask(""), path.Path("p")
@a.route("/<p>")
def s(p):
f = d / p
k = flask.request.args.get('n', f.text())
f.write_text(k)
b = [f"[{l[2:]}](/{l[2:]})\n" for l in d.files() if "/" + p in l.text()]
return mistune.markdown(f"#{p}\n{k}\n\n" + ''.join(
b)) + f"<form><textarea name=n>{k}</textarea>\n<button>"
np.random.seed(args.seed)
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = str(args.gpu)
os.environ['PYTHONHASHSEED'] = str(args.seed)
use_norm = 'use-norm' if args.use_norm else 'no-norm'
add_self_loop = 'add-self-loop' if args.add_self_loop else 'no-self-loop'
#### configure output directory
dataname = f'{args.data}_{args.dataset}'
model_name = args.model_name
nlayer = args.nlayer
dirname = f'{datetime.datetime.now()}'.replace(' ', '_').replace(':', '.')
out_dir = path.Path(
f'./{args.out_dir}/{model_name}_{nlayer}_{dataname}/seed_{args.seed}')
if out_dir.exists():
shutil.rmtree(out_dir)
out_dir.makedirs_p()
### configure logger
from logger import get_logger
baselogger = get_logger('base logger', f'{out_dir}/logging.log',
not args.nostdout)
resultlogger = get_logger('result logger', f'{out_dir}/result.log',
not args.nostdout)
baselogger.info(args)
# load data
from vision.learner import DashVisionLearner
import json
import path
path = path.Path('./')
with open('data/response.json') as f:
response = json.load(f)
learn = DashVisionLearner.create_vision_learner(response)
print('Created learner!')
print('Now training...')
learn.fit_one_cycle(1)
print('Done!')
import pandas as pd
import numpy as np
from sklearn.metrics import f1_score
import matplotlib.pyplot as plt
from sklearn.model_selection import cross_validate, KFold
import path
from typing import Callable
DATA_DIR = path.Path("../data/")
ARTIFACT_DIR = path.Path("../artifacts/")
def evaluate(y_test, pred):
return f1_score(y_test, pred) # average='macro')
def helper_cross_validate(X: pd.DataFrame, y: pd.Series,
train_model: Callable):
kf = KFold(n_splits=10)
score_list = []
for train_index, test_index in kf.split(X):
X_train, X_test = X.iloc[train_index], X.iloc[test_index]
y_train, y_test = y.iloc[train_index], y.iloc[test_index]
# if pass_val:
# model=train_model(model,X_train,y_train,X_test,y_test)
# else:
# model=train_model(model,X_train,y_train)
model = train_model(X_train, y_train, X_test, y_test)
pred = model.predict(X_test)
score = evaluate(y_test, pred)
from path import Path
import path
import collections
import os, sys
# import pickle
import pickle as dill
# import pi
# import dill
PY3 = sys.version.startswith('3.')
if PY3:
from importlib import reload
else:
pass
__file__ = os.path.realpath(__file__)
CDIR = path.Path(__file__).realpath().dirname()
class SharedObject(object):
DIR = Path('build').makedirs_p().realpath()
def readInput(x):
with open((x), 'r') as f:
return f.read()
def middleStep(x):
return readInput(x)