def driver(argv):
pathname = argv[1]
pathname = abs(pathname)
regex = re.compile(".+\.py$")
index = Indexer()
index_time_start = time.time()
for filename in iter_filenames(pathname, regex):
index.construct(filename)
index_time_end = time.time()
index_file = "index.pkl"
fh_output = open(index_file, 'wb')
dump(index.block_graph, fh_output)
fh_output.close()
index_size = os.path.getsize(index_file)
print "Total files scanned:", _TOTAL_NUM_FILES
print "Python files matched:", _TOTAL_FILES_MATCHED
print "Python files net size:", _TOTAL_MATCHED_SIZE
print "Time required to index:", index_time_end - index_time_start
print "Index size", index_size
def rel(path, parent=None, nopar=False):
"""
Like :func:`os.path.relpath`, but the *nopar* parameter can be set to return
an absolute path if the relative path would create a path element that
references a parent directory (`..`) or current directory (``.``).
"""
try:
res = os.path.relpath(path, parent)
except ValueError:
if nopar:
return abs(path)
raise
else:
if not issub(res):
return abs(path)
return res
def rel(path, parent=None, nopar=False):
"""
Like :func:`os.path.relpath`, but the *nopar* parameter can be set to return
an absolute path if the relative path would create a path element that
references a parent directory (`..`) or current directory (``.``).
"""
try:
res = os.path.relpath(path, parent)
except ValueError:
if nopar:
return abs(path)
raise
else:
if not issub(res):
return abs(path)
return res
def iter_filenames(pathname, filename_matcher):
global _TOTAL_NUM_FILES
global _TOTAL_FILES_MATCHED
global _TOTAL_MATCHED_SIZE
"""Yield every filename in the path that matches the pattern"""
for root, directories, filenames in os.walk(pathname):
for filename in filenames:
_TOTAL_NUM_FILES += 1
full = abs(join(root, filename))
if filename_matcher.match(filename):
_TOTAL_FILES_MATCHED += 1
_TOTAL_MATCHED_SIZE += os.path.getsize(full)
yield full
return True
else:
return False
PATH = os.path.dirname(__file__)
JSON_PATH = os.path.join(PATH, 'models.json')
MODELS_PATH = os.path.join(PATH, "models")
HOMOGENE = False
RANDOM = True
MIX = False
KEZEK = False
if HOMOGENE and not MIX and not KEZEK:
VALID_PATH = abs(os.path.join(PATH, 'random'))
elif RANDOM and not MIX and not KEZEK:
VALID_PATH = os.path.join(PATH, 'homogene')
elif MIX:
NPY_PATH = os.path.join(PATH, 'mix_test.npy')
elif KEZEK:
VALID_PATH = abs(os.path.join(PATH, 'fashion'))
NPY_PATH = abs(os.path.join(VALID_PATH, 'kezek.npy'))
else:
VALID_PATH = os.path.join(PATH, 'images')
if not MIX and not KEZEK:
NPY_PATH = os.path.join(VALID_PATH, 'bone_spects.npy')
TRAIN = ["all", "all"]
def validate(classify):
label_path = abs(os.path.join(PATH, "labels.json"))
imgs = Images(NPY_PATH)
# Draw on image
if DRAW_IMAGE:
labels = []
arrays = []
for i in range(VALID_IMG_NUM):
number = random.randint(0, CLASSES - 1)
img_name = "image_%s.png" % i
img_path = abs(os.path.join(VALID_PATH, img_name))
labels.append(number)
draw = Draw(imgs.arr[i], number, img_path)
arrays.append(draw.array)
arrays = np.array(arrays)
try:
with open(label_path, 'r') as file:
j_son = json.load(file)
if HOMOGENE:
j_son['homogene'] = labels
elif RANDOM:
j_son['random'] = labels
elif KEZEK:
j_son['kezek'] = labels
arrays = np.expand_dims(arrays, axis=3)
np.save(NPY_PATH, arrays)
try:
with open(label_path, 'w+') as file:
json.dump(j_son, file, indent=4)
except EOFError:
print("There is no valid .json file.")
sys.exit()
except EOFError:
print("There is no valid .json file.")
exit()
# Classifier
if classify == 'True':
images = Images(NPY_PATH)
# part = int((images.arr.shape[0]) * 3 / 4)
train_images = np.array([])
part = 300
if part != 0:
train_images = images.arr[:part]
test_images = images.arr[part:]
try:
with open(label_path, 'r') as file:
j_son = json.load(file)
label = []
if HOMOGENE:
label = j_son['homogene']
if RANDOM:
label = j_son['random']
if MIX:
label = j_son['mix']
if KEZEK:
label = j_son['kezek']
except:
print('Something went wrong.')
train_labels = list()
if part != 0:
train_labels = to_categorical(label[:part])
test_labels = to_categorical(label[part:])
classifier = Classifier(imgs.arr.shape[1:], CLASSES)
network_name = 'classifier.h5'
network_path = os.path.join(PATH, network_name)
if not find(network_name, PATH):
classifier.build_classifier(network_path)
if HOMOGENE:
trained_name = 'hom_class.h5'
trained_path = os.path.join(PATH, trained_name)
elif RANDOM:
trained_name = 'rand_class.h5'
trained_path = os.path.join(PATH, trained_name)
else:
trained_path = None
if find(trained_name, PATH):
classifier.loadModel(trained_path)
classifier.model.compile(optimizer=Adam(lr=0.001),
loss=LOSS,
metrics=['categorical_accuracy'])
elif train_images.size != 0:
classifier.loadModel(network_path)
classifier.model.compile(optimizer=Adam(lr=0.001),
loss=LOSS,
metrics=['categorical_accuracy'])
classifier.trainModel(train_images, train_labels, 2, EPOCH)
json_model = {}
try:
with open(os.path.join(PATH, 'classific.json'), 'r') as file:
j_son = json.load(file)
if HOMOGENE:
json_model = j_son['homogene']
elif RANDOM:
json_model = j_son['random']
except:
print('Something went wrong.')
if train_images.size != 0:
classifier.saveTrained(trained_path, json_model['train'])
classifier.predictModel(test_images, test_labels, 1)
print("itt a baj")
test = []
print(classifier.predicted)
for i, label in enumerate(test_labels):
test_ = dict()
test_[str(np.argmax(label))] = str(
np.argmax(classifier.predicted[i]))
test.append(test_)
json_model['test'] = test
try:
with open(os.path.join(PATH, 'classific.json'), 'w') as file:
if HOMOGENE:
j_son['homogene'] = json_model
elif RANDOM:
j_son['random'] = json_model
elif MIX:
j_son['mix'] = json_model
elif KEZEK:
j_son['kezek'] = json_model
json.dump(j_son, file, indent=4)
except EOFError:
print("Save to .json file was unsuccessful.")
else:
pass