def TestCase(self):
kmp = KMP()
rs = kmp.search_MP("abcabcabcabc", "cabc")
self.assertListEqual(rs, [2,5,8])
rs = kmp.search_MP("ABC ABCDAB ABCDABCDABDE", "ABCDABD")
self.assertListEqual(rs, [15])
def TestCase(self):
kmp = KMP()
rs = kmp.search_MP("abcabcabcabc", "cabc")
self.assertListEqual(rs, [2, 5, 8])
rs = kmp.search_MP("ABC ABCDAB ABCDABCDABDE", "ABCDABD")
self.assertListEqual(rs, [15])
def __init__(self, p):
# p : m by m sized pattern.
self.p = list(p)
self.m = len(p)
# dr : pi[1..m] -> distinct row number.
# pp : p converted in to sequence of distinct row #s.
self.dr, self.pp = self.construct()
# ac : Aho-Corasick instance for p (row-matching)
self.ac = AC(self.dr.keys())
# kmp : KMP instance for pp (column-matching)
self.kmp = KMP(self.pp)
def do_kmp(spam_text, data_resp):
results = []
for data in data_resp:
text = data['text']
result = KMP.match_string(text, spam_text)
result['profile_img'] = data['profile_img']
result['name'] = data['name']
result['screen_name'] = data['screen_name']
results.append(result)
return results
def __init__(self, stream, pattern_len):
'''
Baker-bird 알고리즘 이니셜라이징
:param patterns: Baker-bird 알고리즘을 실행할 패턴 배열
'''
self.ac = AhoCorasick()
for _ in range(pattern_len):
line = stream.next()
self.ac.add_patterns(line)
self.ac.build()
self.r = {}
idx = 1
stream.set_seek()
for _ in range(pattern_len):
row = stream.next()
if row not in self.r.keys():
self.r[row] = str(idx)
idx += 1
stream.set_seek()
self.kmp = KMP("".join([str(self.r[stream.next()]) for _ in range(pattern_len)]))
class BakerBird(object):
def __init__(self, stream, pattern_len):
'''
Baker-bird 알고리즘 이니셜라이징
:param patterns: Baker-bird 알고리즘을 실행할 패턴 배열
'''
self.ac = AhoCorasick()
for _ in range(pattern_len):
line = stream.next()
self.ac.add_patterns(line)
self.ac.build()
self.r = {}
idx = 1
stream.set_seek()
for _ in range(pattern_len):
row = stream.next()
if row not in self.r.keys():
self.r[row] = str(idx)
idx += 1
stream.set_seek()
self.kmp = KMP("".join([str(self.r[stream.next()]) for _ in range(pattern_len)]))
def __call__(self, stream, text_len):
'''
Baker-bird 알고리즘 수행, extra space를 최적화 하기 위해 기존 Aho-corasick으로 만들어 낸 2차원 배열을 KMP를 모두 훑는 방식이 아닌
Aho-corasick 한줄을 수행 후 패턴 크기만큼의 배열만을 생성 유지하고, KMP를 step별로 계산하는 방식으로 구현
:param text: 매칭할 텍스트
:return: 매칭이 일어난 끝 좌표를 튜플 배열 형태로 반환
'''
ret = []
position = [0,] * text_len
for i in range(text_len):
row = stream.next()
row_R = ["0",] * text_len
for start, end, keyword in self.ac(row):
row_R[end-1] = self.r[keyword]
for idx, R in enumerate(row_R):
position[idx] = self.kmp.step(R, position[idx])
if position[idx] is len(self.kmp.keyword):
ret.append((i, idx))
position[idx] = self.kmp.pi[position[idx]-1]
return ret
tock = time.time()
print('re search time:', tock - tick)
tick = time.time()
for text in text_iterator():
for keyword in keywords:
result = bf_search(text.strip(), keyword)
# if result:
# print(keyword, result)
tock = time.time()
print('bf search time:', tock - tick)
tick = time.time()
for text in text_iterator():
for keyword in keywords:
result = KMP().search(text.strip(), keyword)
# if result:
# print(keyword, result)
tock = time.time()
print('kmp search time:', tock - tick)
tick = time.time()
for text in text_iterator():
result = ac_automation.search(text.strip())
tock = time.time()
# word2pos = defaultdict(list)
# for start, end in result:
# word2pos[line[start: end+1]].append((start, end))
# for word, pos in word2pos.items():
# print(word, pos)
print('ac search time:', tock - tick)
from kmp import KMP
from fa import AUTOMATA
import sys
matcher = sys.argv[1]
pattern = sys.argv[2]
file = open(sys.argv[3], "r")
text = ""
for a in file.read().splitlines():
text += a
if matcher == "KMP":
print("wyszukiwanie wzorca za pomoca KMP")
print("=================================")
kmp = KMP(text, pattern)
kmp.kmp()
else:
print("wyszukiwanie wzorca za pomoca FA")
print("================================")
finite = AUTOMATA(text, pattern)
finite.automata_matcher()
def kmp_array(fasta_list):
kmp = KMP()
kmp.build_kmp_table(fasta_list[0][1], print_table = True)
def fifthw(self):
QMessageBox.information(self, "规则","请输入 用半角逗号间隔的字符串 第一串最少6个字符最多20个 第二串最少一个 第二串字符数不超过第一串 回车完成输入",QMessageBox.Yes)
from kmp import KMP
self.demo5 = KMP()
