def get_multi_term_patterns(dataframe, col_with_prefixes, prefix_col):
prefixes = dataframe[prefix_col]
unique_prefixes = list(set(prefixes))
unique_prefixes.sort()
inner_id_patterns = {}
for one_unique in unique_prefixes:
temp = list(
dataframe[col_with_prefixes][dataframe[prefix_col] == one_unique])
extracted = rexpy.extract(temp)
extracted = extracted[0]
extracted = re.sub('^\\^', '', extracted)
extracted = re.sub('\\$$', '', extracted)
inner_id_patterns[one_unique] = extracted
return inner_id_patterns
def pattern_discovery(dataset, file_name):
# function to discover patterns from the input dataset and
# generate the corresponding file
from tdda import rexpy
import collections
from functools import reduce
# dataset = dataset.select_dtypes(['object'])
patterns = collections.defaultdict(list)
list_pattern = list()
listp = []
fn = './save/' + file_name + '_patterns.txt'
# file = open(fn,"w")
for c in dataset.columns.values:
corpus = dataset[c].unique().astype('str').tolist()
results = rexpy.extract(corpus)
patterns[c].append(results)
for i in range(0, len(patterns)):
lp = list(patterns.items())[i][1]
lp = reduce(lambda x, y: x + y, lp)
p = reduce(lambda x, y: x + y, lp)
list_pattern = (c, i, "'" + str(p) + "'")
listp.append(list_pattern)
# file.writelines(str(list_pattern)+ '\n')
p = pd.DataFrame(listp, columns=['col', 'num', 'pattern'])
p.to_csv(fn, header=('col', 'num', 'pattern'), index=False, sep=';')
return p
def discover_id_pattern(example_ids):
extracted = rexpy.extract(example_ids)
extracted = extracted[0]
extracted = re.sub('^\\^', '', extracted)
extracted = re.sub('\\$$', '', extracted)
return extracted
def find_rexes(self, colname, values=None):
if values is None:
return rexpy.pdextract(self.df[colname])
else:
return rexpy.extract(values)
def find_rexes(self, colname, values=None):
if not values:
values = self.get_database_unique_values(self.tablename, colname)
return rexpy.extract(values)
def find_rexes(self, colname, values=None, seed=None):
if not values:
values = self.get_database_unique_values(self.tablename, colname)
return rexpy.extract(values, seed=seed)
from __future__ import print_function
from tdda.rexpy import extract
from tdda.rexpy.seq import common_string_sequence
from tdda.rexpy.relib import re
x = extract(['Roger', 'Coger', 'Doger'], tag=True, as_object=True)
print(x)
patternToExamples = x.pattern_matches()
sequences = []
for j, (pattern, examples) in enumerate(patternToExamples.items()):
N = len(examples)
if N < 1:
print('%s:%s' % (pattern, examples))
else:
eparts = [re.match(x.results.rex[j], e).groups() for e in examples]
nparts = len(eparts[0])
for i in range(nparts):
(L, R) = (eparts[0][i], eparts[1][i])
n = 2
s = common_string_sequence(L, R)
while n < N and s != '':
s = common_string_sequence(s, eparts[n][i])
n += 1
sequences.append(s)
print(sequences)
from tdda import rexpy
corpus = ['123-AA-971', '12-DQ-802', '198-AA-045', '1-BA-834']
results = rexpy.extract(corpus)
print('Number of regular expressions found: %d' % len(results))
for rex in results:
print(' ' + rex)
def find_rexes(self, colname, values=None, seed=None):
if values is None:
return rexpy.pdextract(self.df[colname])
else:
return rexpy.extract(values, seed=None)
