Leading Whitespace Quantification Function 0

 1import numpy
 2
 3def parse_string_list(string_list,tabs_to_spaces=4):
 4    '''Takes an iterable containter of strings.
 5    
 6    Quantifies number of line-leading whitespace characters. Converts tabs to spaces.
 7    
 8    Returns three lists of equal length as a tuple.
 9    The first is a list the strings, whitespace-stripped.
10    The second is a list of numerical counts of line-leading whitespace characters clipped.
11    The third is a list of the lengths of the stripped strings.'''
12
13    rstripped = [l.rstrip() for l in string_list]
14    expanded_lines = [l.expandtabs(tabs_to_spaces) for l in rstripped]
15    both_stripped = [l.strip() for l in string_list]
16    expanded_lens = [len(l) for l in expanded_lines]
17    stripped_lens = [len(l) for l in both_stripped]
18    
19    exp_lens_array = numpy.asarray(expanded_lens)
20    stripped_lens_array = numpy.asarray(stripped_lens)
21    indentations = exp_lens_array - stripped_lens_array
22    indentations_list = list(indentations)
23    
24    return both_stripped,indentations_list,stripped_lens
25
26# Toy example
27if __name__ == "__main__":
28    sample_string_list = ['stuff',
29                          '    space_indented',
30                          '        double_indented',
31                          'not_indented',
32                          '',
33                          'after_blank',
34                          'no_blank']
35    out = lines,indentations,content_lengths = parse_string_list(sample_string_list)
36    for l in out:
37        print l
38        print '...of length', len(l)