def main():
file_reader = FileReader(sys.argv)
if (not file_reader.is_read()):
sys.exit()
wavelet_tree = WaveletTree(file_reader.get_characters())
print wavelet_tree.track_symbol(1)
print wavelet_tree.rank_query('$', 45)
print wavelet_tree.select_query('e', 1)
def main():
file_reader = FileReader(sys.argv)
if (not file_reader.is_read()):
sys.exit()
wavelet_tree = WaveletTree(file_reader.get_characters())
print wavelet_tree.track_symbol(1)
print wavelet_tree.rank_query('$', 45)
print wavelet_tree.select_query('e', 1)
def WTquery():
script = sys.argv[0]
action = sys.argv[1]
if action == 'build':
filename = sys.argv[2]
outputfile = sys.argv[3]
data = open(filename, "r")
inputString = data.read() # type(inputString): str
inputString = list(inputString)
for i in range(len(inputString)):
if inputString[i] == ' ':
inputString[i] = '_'
wt = WaveletTree(inputString)
sys.stdout = open(outputfile, "w")
print(wt.UNIQ)
print(wt.SIZE)
print(wt.character)
print(wt.wt)
print(wt.int)
elif action == 'access':
wtfile = sys.argv[2]
index = sys.argv[3]
with open(wtfile, "r") as df:
lines = df.readlines()
s = lines[3].rstrip('\n')
wtTable = yaml.load(s, Loader=yaml.FullLoader)
#wtAccess = Access(wtTable, index)
elif action == 'rank':
wtfile = sys.argv[2]
index = int(sys.argv[3])
with open(wtfile, "r") as df:
lines = df.readlines()
a = lines[3].rstrip('\n')
wtTable = yaml.load(a, Loader=yaml.FullLoader)
b = lines[4].rstrip('\n')
wtInterval = yaml.load(b, Loader=yaml.FullLoader)
#print(wtTable)
wtRank = Rank(wtTable, wtInterval, index)
elif action == 'select':
values = data
"""
if __name__ == '__main__':
if len(sys.argv) < 4:
print 'Missing arguments: <in_file> <out_file> <task> <task_args...>'
exit(-1)
task = sys.argv[3]
fasta_file = FASTA(sys.argv[1])
fasta_file.read()
alphabet = list(set(fasta_file.data))
alphabet.sort()
tree = WaveletTree()
tree.build(alphabet, fasta_file.data)
if task == 'access':
index = int(sys.argv[4])
result = tree.access(index)
elif task == 'rank':
position = int(sys.argv[4])
character = sys.argv[5]
result = tree.rank(position, character)
elif task == 'select':
nth_occurence = int(sys.argv[4])
character = sys.argv[5]
result = tree.select(nth_occurence, character)
def report_access(tree):
print 'Access simple.fa'
for idx in xrange(len(tree.root_node.bit_vector)):
print 'access(%s) = %s' % (idx, tree.access(idx))
def report_select(tree):
print 'Select simple.fa'
for c in tree.root_node.alphabet:
for nth_occurence in xrange(10):
print 'select(%s, %s) = %s' % (nth_occurence, c, tree.select(nth_occurence, c))
if __name__ == '__main__':
fasta_file = FASTA('../simple.fa')
fasta_file.read()
alphabet = list(set(fasta_file.data))
alphabet.sort()
tree = WaveletTree()
tree.build(alphabet, fasta_file.data)
report_rank(tree)
report_access(tree)
report_select(tree)
exit(-1)
# Reading data
read_stopwatch = Utils.Stopwatch()
fasta_file = FASTA(sys.argv[1])
read_stopwatch.start()
fasta_file.read()
read_stopwatch.stop()
read_memory_kB = Utils.get_max_memory_kB()
alphabet = list(set(fasta_file.data))
alphabet.sort()
# Building tree
build_stopwatch = Utils.Stopwatch()
tree = WaveletTree()
build_stopwatch.start()
tree.build(alphabet, fasta_file.data)
build_stopwatch.stop()
build_memory_kB = Utils.get_max_memory_kB()
if sys.argv[2] == 'query':
data_size = len(fasta_file.data)
test_idx = data_size / 2
test_character = fasta_file.data[0]
test_nth_occurence = math.floor(math.sqrt(test_idx))
test_runs = int(sys.argv[3])
stopwatch = Utils.Stopwatch()
time_access = 0
time_select = 0
"""
if __name__ == '__main__':
if len(sys.argv) < 4:
print 'Missing arguments: <in_file> <out_file> <task> <task_args...>'
exit(-1)
task = sys.argv[3]
fasta_file = FASTA(sys.argv[1])
fasta_file.read()
alphabet = list(set(fasta_file.data))
alphabet.sort()
tree = WaveletTree()
tree.build(alphabet, fasta_file.data)
if task == 'access':
index = int(sys.argv[4])
result = tree.access(index)
elif task == 'rank':
position = int(sys.argv[4])
character = sys.argv[5]
result = tree.rank(position, character)
elif task == 'select':
nth_occurence = int(sys.argv[4])
character = sys.argv[5]
result = tree.select(nth_occurence, character)
exit(-1)
# Reading data
read_stopwatch = Utils.Stopwatch()
fasta_file = FASTA(sys.argv[1])
read_stopwatch.start()
fasta_file.read()
read_stopwatch.stop()
read_memory_kB = Utils.get_max_memory_kB()
alphabet = list(set(fasta_file.data))
alphabet.sort()
# Building tree
build_stopwatch = Utils.Stopwatch()
tree = WaveletTree()
build_stopwatch.start()
tree.build(alphabet, fasta_file.data)
build_stopwatch.stop()
build_memory_kB = Utils.get_max_memory_kB()
if sys.argv[2] == 'query':
data_size = len(fasta_file.data)
test_idx = data_size / 2
test_character = fasta_file.data[0]
test_nth_occurence = math.floor(math.sqrt(test_idx))
test_runs = int(sys.argv[3])
stopwatch = Utils.Stopwatch()
time_access = 0
time_select = 0
