def test_multiple_intersections_1(self):
text = ' C AAA C D '
modifiers = [
(r'(C) (AAA) (C) (D) ', {
1: 'GG',
2: 'BB',
3: 'GG',
4: 'DD'
}),
(r'( GG BB G)G', {
1: 'HJK'
}),
(r'(HJK)G', {
1: 'FF'
}),
]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'FFG DD ')
self.assertEqual(procesor.span_map, [((0, 8), (0, 3)),
((9, 10), (4, 6))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '00000000 1 ')
self.assertEqual(decorated_processed_text, '000 11 ')
def test_matching_2(self):
text = ' AAA BBB CCC DDD '
modifiers = [
(r'(AAA) (BBB) (CCC)', {
1: 'ZZZZ',
2: 'YYYYY',
3: 'XXXXXX'
}),
(r'((YYYYY)|(ZZZZ))', {
1: 'WWWWWW'
}),
(r'(WWWWWW)', {
1: 'QQQQQQQ'
}),
]
ref_span_map = [((1, 4), (1, 8)), ((5, 8), (9, 16)),
((9, 12), (17, 23))]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text,
' QQQQQQQ QQQQQQQ XXXXXX DDD ')
self.assertEqual(procesor.span_map, ref_span_map)
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_processed_text,
' 0000000 1111111 222222 DDD ')
self.assertEqual(decorated_text, ' 000 111 222 DDD ')
def test_7(self):
text = 'ab'
pattern, replacement_map = r'((a)(b))', {1: 'c', 2: 'd', 3: 'e'}
with self.assertRaises(ValueError):
with Processor(text) as procesor:
procesor.process(pattern, replacement_map)
def test_intersection_5(self):
text = ' C AAA C '
modifiers = [
(r' (AAA) C', {
1: 'BBEBB'
}),
(r'C (BBEBB)', {
1: 'DD'
}),
(r'(C D)D', {
1: 'FF'
}),
]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' FFD C ')
self.assertEqual(procesor.span_map, [((1, 6), (1, 4))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 00000 C ')
self.assertEqual(decorated_processed_text, ' 000 C ')
def test_matching_1(self):
text = ' BBB AAA AAA BBB '
modifiers = [
(r'(AAA)', {
1: 'BBB'
}),
(r'(BBB)', {
1: 'YYY'
}),
]
ref_span_map = [((1, 4), (1, 4)), ((5, 8), (5, 8)), ((9, 12), (9, 12)),
((13, 16), (13, 16))]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' YYY YYY YYY YYY ')
self.assertEqual(procesor.span_map, ref_span_map)
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 111 222 333 ')
self.assertEqual(decorated_processed_text, ' 000 111 222 333 ')
def test_matching_6(self):
text = ' AAA D AAA D '
modifiers = [
(r'(AAA) (D)', {
1: 'BBBBB',
2: 'EE'
}),
(r'(BBBBB)', {
1: 'CC'
}),
(r'(EE)', {
1: 'FFFF'
}),
]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' CC FFFF CC FFFF ')
self.assertEqual(procesor.span_map, [((1, 4), (1, 3)),
((5, 6), (4, 8)),
((7, 10), (9, 11)),
((11, 12), (12, 16))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 1 222 3 ')
self.assertEqual(decorated_processed_text, ' 00 1111 22 3333 ')
def test_2(self):
with Processor(self.text_1) as procesor:
for pattern, replacement_map in self.modifiers_1:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' BBB BBB BBB BBB ')
self.assertEqual(procesor.span_map, self.span_map_1_1)
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 BBB 111 BBB ')
self.assertEqual(decorated_processed_text, ' 000 BBB 111 BBB ')
def test_5(self):
with Processor(self.text_0) as procesor:
for pattern, replacement_map in self.modifiers_3:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' CCC CCC CCC CCC ')
self.assertEqual(procesor.span_map, self.span_map_2)
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 111 CCC 222 ')
self.assertEqual(decorated_processed_text, ' 000 111 CCC 222 ')
def test_1(self):
with Processor(self.text_1) as procesor:
for pattern, replacement_map in self.modifiers_0:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' ZZZ YYY ZZZ YYY ')
self.assertEqual(procesor.span_map, self.span_map)
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 111 222 333 ')
self.assertEqual(decorated_processed_text, ' 000 111 222 333 ')
def test_single_1(self):
text = 'ABAB'
pattern, replacement_map = r'(A)(B)', {1: 'CC', 2: 'D'}
with Processor(text) as procesor:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'CCDCCD')
self.assertEqual(procesor.span_map,
[((0, 1), (0, 2)), ((1, 2), (2, 3)), ((2, 3), (3, 5)),
((3, 4), (5, 6))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0123')
self.assertEqual(decorated_processed_text, '001223')
def test_matching_3(self):
text = 'AZA'
modifiers = [(r'(A)', {1: 'BB'}), (r'(BB)', {1: 'DD'})]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'DDZDD')
self.assertEqual(procesor.span_map, [((0, 1), (0, 2)),
((2, 3), (3, 5))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0Z1')
self.assertEqual(decorated_processed_text, '00Z11')
def test_bundled_1(self):
text = 'ABAB'
modifiers = [(r'(A)', {1: 'CC'}), (r'(B)', {1: 'D'})]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'CCDCCD')
self.assertEqual(procesor.span_map,
[((0, 1), (0, 2)), ((1, 2), (2, 3)), ((2, 3), (3, 5)),
((3, 4), (5, 6))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0123')
self.assertEqual(decorated_processed_text, '001223')
def test_bundled_4(self):
text = 'BAAB'
modifiers = [(r'(AA)', {1: 'CC'}), (r'(B)', {1: ''})]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'CC')
self.assertEqual(procesor.span_map,
[((0, 1), (0, 0)), ((1, 3), (0, 2)),
((3, 4), (2, 2))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0112')
self.assertEqual(decorated_processed_text, '11')
def test_9(self):
text = 'AA BBBB& CC&CCCC'
pattern_1, replacement_map_1 = r'([A-Za-z&]+)', {
1: lambda x: x.replace('&', '')
}
pattern_2, replacement_map_2 = r'(AA) ', {1: 'DDD DDD'}
with Processor(text) as procesor:
procesor.process(pattern_1, replacement_map_1)
procesor.swap()
procesor.process(pattern_2, replacement_map_2)
self.assertEqual(procesor.text, 'AA BBBB CCCCCC')
self.assertEqual(procesor.processed_text, 'DDD DDD BBBB& CC&CCCC')
self.assertEqual(procesor.span_map, [((0, 2), (0, 7)),
((3, 7), (8, 13)),
((8, 14), (14, 21))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '00 1111 222222')
self.assertEqual(decorated_processed_text, '0000000 11111 2222222')
def test_intersection_6(self):
modifiers = [
(r' (AAA)B', {
1: 'CCC CCC'
}),
(r'(CCCB)', {
1: lambda x: x
}),
]
text = ' AAAB'
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' CCC CCCB')
self.assertEqual(procesor.span_map, [((1, 5), (1, 9))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 0000')
self.assertEqual(decorated_processed_text, ' 00000000')
def test_matching_4(self):
text = ' AAA '
modifiers = [
(r'(AAA)', {
1: 'BBBBB'
}),
(r'(BBBBB)', {
1: 'CC'
}),
]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' CC ')
self.assertEqual(procesor.span_map, [((1, 4), (1, 3))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 000 ')
self.assertEqual(decorated_processed_text, ' 00 ')
def test_intersection_7(self):
modifiers = [
(r'B(AAA) ', {
1: 'CCC CCC'
}),
(r'(BCCC)', {
1: lambda x: x
}),
]
text = 'BAAA '
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'BCCC CCC ')
self.assertEqual(procesor.span_map, [((0, 4), (0, 8))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0000 ')
self.assertEqual(decorated_processed_text, '00000000 ')
def test_new(self):
modifiers = [
(r' (etc)\.', {
1: 'et cetera'
}),
(r'([^ ]+)', {
1: lambda x: x
}),
]
text = ' etc.'
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, ' et cetera.')
self.assertEqual(procesor.span_map, [((1, 5), (1, 11))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, ' 0000')
self.assertEqual(decorated_processed_text, ' 0000000000')
def test_intersection_1(self):
text = 'C AAA C'
modifiers = [
(r'(AAA)', {
1: 'BBBBB'
}),
(r'(C BBBBB C)', {
1: 'DD'
}),
]
with Processor(text) as procesor:
for pattern, replacement_map in modifiers:
procesor.process(pattern, replacement_map)
self.assertEqual(procesor.processed_text, 'DD')
self.assertEqual(procesor.span_map, [((0, 7), (0, 2))])
decorated_text, decorated_processed_text = procesor.decorate()
self.assertEqual(decorated_text, '0000000')
self.assertEqual(decorated_processed_text, '00')