m-ric/chunk_visualizer

1### Utilities to get overlap between strings2 3def get_overlap_length(left: str, right: str):4    good_length, overlap = 0, ""5    for i in range(min(len(left), len(right))):6        if left[-i:] == right[:i]:7            good_length = i8            overlap = left[-i:]9    return good_length, overlap10 11def get_overlap_list(strings):12    """13    Returns a list of tuples of the form (overlap_length, overlap), one tuple for each pair of strings in the input list.14    """15    overlaps = []16    for i in range(len(strings) - 1):17        overlaps.append(get_overlap_length(strings[i], strings[i+1]))18    return overlaps19 20def unoverlap_list(strings):21    """22    Returns a list of tuples of the form (content, is_overlap), where is_overlap is a boolean indicating whether the content is an overlap or not.23    """24    overlaps = get_overlap_list(strings)25    new_list = []26    for index, string in enumerate(strings):27        # Add the last overlap when needed28        if index > 0 and len(overlaps[index-1][1]) > 0:29            new_list.append((overlaps[index-1][1], True))30 31        # prune the string with left and right overlaps32        left_overlap_length, right_overlap_length = 0, 033        if index > 0:34            left_overlap_length = overlaps[index-1][0]35        if index < len(strings) - 1:36            right_overlap_length = overlaps[index][0]37 38        new_list.append((string[left_overlap_length:len(string)-right_overlap_length], False))39    return new_list