I am trying to write a code to extract patterns from a each list in a list of lists. I search for patterns with specified lengths, such as 'B-' followed by 'I-'. For example, I want to keep patterns with lengths of two and replace others with a specified string from the following list:
list = ['O', 'B-', 'I-', 'I-', 'O', 'B-', 'I-', 'B-']
Expected output should be as follows:
expected_list_2 = ['O', 'O', 'O', 'O', 'O', 'B-', 'I-', 'O']
As can be seen only length of two pattern 'B-', 'I-' is kept and others changed with 'O' label.
If I want to keep patterns with lengths of three, the output should be as follows:
expected_list_3 = ['O', 'B-', 'I-', 'I-', 'O', 'O', 'O', 'O']
Considering each elements of my list of lists includes such kind of lists and I try to implement this task for each list, I asked for, is there any efficient or tricky way to do this instead of defining some if-else conditions and looping over the each element?
This solution should (please test with more relevant cases before deploying to production) find all positions of pattern 'B-', n-1 x 'I-' in a list. I extended the example list1 to cover more cases, like pattern at start and end of the list and consecutive patterns.
list1 = ['B-', 'I-', 'I-', 'O', 'B-', 'I-', 'I-', 'B-', 'I-', 'B-', 'I-', 'O', 'B-', 'I-', 'I-']
#n = 2: ^^^^^^^^^ ^^^^^^^^^
#n = 3: ^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^
def find_pattern(list1, n):
pattern = ['B-'] + ['I-'] * (n-1)
first = pattern[0]
# find starting indices of matching patterns
idx = [e for e, i in enumerate(list1[:-n+1 or None])
if i == first # optimization for long pattern
and list1[e:e+n] == pattern
and list1[e+n:e+n+1] != ['I-']]
# insert pattern at those indices
res = ['O'] * len(list1)
for i in idx:
res[i:i+n] = pattern
return res
print(find_pattern(list1, 2))
print(find_pattern(list1, 3))
Output
['O', 'O', 'O', 'O', 'O', 'O', 'O', 'B-', 'I-', 'B-', 'I-', 'O', 'O', 'O', 'O']
['B-', 'I-', 'I-', 'O', 'B-', 'I-', 'I-', 'O', 'O', 'O', 'O', 'O', 'B-', 'I-', 'I-']