1	#!/bin/sh
     2	""""exec ${PYTHON:-python} -t $0 "$@";" """
     3	# vim: filetype=python expandtab smarttab shiftwidth=4
     4	
     5	
     6	import sys
     7	import string
     8	from operator import itemgetter as nth
     9	
    10	
    11	alphanumerics = string.lowercase + string.digits
    12	
    13	def emit_top_new(paragraph, show_count, previous, keep_previous_count):
    14	    """Given some text and a list of disallowed characters, return two samples
    15	    of the most common letters.  One sample is likely to be fed back in as a
    16	    disallowed list.
    17	
    18	    >>> emit_top_new("", 3, ['1', '2'], 2)
    19	    ([], [])
    20	    >>> emit_top_new("abcdefghighighijkljklmnopppp", 3, [], 2)
    21	    (['p', 'h', 'i'], ['p', 'h'])
    22	    >>> emit_top_new("abcdefghighighijkljklmnopppp", 3, ['p', 'h'], 2)
    23	    (['i', 'g', 'l'], ['i', 'g'])
    24	    """
    25	    character_count = dict([c, 0] for c in alphanumerics)
    26	
    27	    for c in previous:
    28	        character_count.pop(c, None)
    29	
    30	    for c in paragraph.lower():
    31	        try:
    32	            character_count[c] += 1
    33	        except KeyError:
    34	            pass  # not a character we care about
    35	
    36	    sortable = character_count.items()
    37	
    38	    sortable.sort(key=nth(1))
    39	    sortable.reverse()
    40	    sortable = [k for k, v in sortable if v > 0]
    41	
    42	    return sortable[:show_count], sortable[:keep_previous_count]
    43	
    44	
    45	if __name__ == "__main__":
    46	    import doctest
    47	    doctest.testmod()
    48	
    49	    show_count = 10
    50	    keep_previous_count = 5
    51	    previous = set()
    52	
    53	    lines = []
    54	    for line in sys.stdin:
    55	        if line.strip() == "":
    56	            if lines:
    57	                new, previous = emit_top_new("".join(lines), show_count, previous, keep_previous_count)
    58	                print new
    59	                lines = []
    60	        else:
    61	            lines.append(line)
    62	
    63	    if lines:
    64	        new, previous = emit_top_new("".join(lines), show_count, previous, keep_previous_count)
    65	        print new