Mercurial > repos > miller-lab > genome_diversity
comparison reorder.py @ 24:248b06e86022
Added gd_genotype datatype. Modified tools to support new datatype.
| author | Richard Burhans <burhans@bx.psu.edu> |
|---|---|
| date | Tue, 28 May 2013 16:24:19 -0400 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| 23:66a183c44dd5 | 24:248b06e86022 |
|---|---|
| 1 #!/usr/bin/env python | |
| 2 | |
| 3 import sys | |
| 4 | |
| 5 def parse_rangelist(string): | |
| 6 rv = [] | |
| 7 | |
| 8 tokens = strip_split(string, ',') | |
| 9 for token in tokens: | |
| 10 int_list = parse_token(token) | |
| 11 for int_val in int_list: | |
| 12 int_val -= 1 | |
| 13 if int_val not in rv: | |
| 14 rv.append(int_val) | |
| 15 | |
| 16 return rv | |
| 17 | |
| 18 def parse_token(token): | |
| 19 values = strip_split(token, '-') | |
| 20 num_values = len(values) | |
| 21 | |
| 22 if num_values not in [1, 2]: | |
| 23 print >> sys.stderr, 'Error: "%s" is not a valid range' % token | |
| 24 sys.exit(1) | |
| 25 | |
| 26 int_list = [] | |
| 27 for value in values: | |
| 28 if value: | |
| 29 int_val = as_int(value) | |
| 30 | |
| 31 if int_val < 1: | |
| 32 print >> sys.stderr, 'Error: "%s" is not >= 1' % value | |
| 33 sys.exit(1) | |
| 34 | |
| 35 int_list.append(int_val) | |
| 36 else: | |
| 37 print >> sys.stderr, 'Error: "%s" is not a valid range' % token | |
| 38 sys.exit(1) | |
| 39 | |
| 40 if num_values == 1: | |
| 41 return int_list | |
| 42 | |
| 43 a, b = int_list | |
| 44 | |
| 45 if a <= b: | |
| 46 return range(a, b+1) | |
| 47 else: | |
| 48 return range(a, b-1, -1) | |
| 49 | |
| 50 def strip_split(string, delim): | |
| 51 return [elem.strip() for elem in string.split(delim)] | |
| 52 | |
| 53 def as_int(string): | |
| 54 try: | |
| 55 val = int(string) | |
| 56 except: | |
| 57 print >> sys.stderr, 'Error: "%s" does not appear to be an integer' % string | |
| 58 sys.exit(1) | |
| 59 return val | |
| 60 | |
| 61 def get_lines(filename): | |
| 62 rv = [] | |
| 63 | |
| 64 fh = open(filename) | |
| 65 for line in fh: | |
| 66 line = line.rstrip('\r\n') | |
| 67 rv.append(line) | |
| 68 fh.close() | |
| 69 | |
| 70 return rv | |
| 71 | |
| 72 def reorder(old_lines, new_order, filename): | |
| 73 max_index = len(old_lines) - 1 | |
| 74 | |
| 75 fh = open(filename, 'w') | |
| 76 | |
| 77 for index in new_order: | |
| 78 if index <= max_index: | |
| 79 print >> fh, old_lines[index] | |
| 80 old_lines[index] = None | |
| 81 | |
| 82 for line in old_lines: | |
| 83 if line is not None: | |
| 84 print >> fh, line | |
| 85 | |
| 86 fh.close() | |
| 87 | |
| 88 if len(sys.argv) != 4: | |
| 89 print >> sys.stderr, "Usage" | |
| 90 sys.exit(1) | |
| 91 | |
| 92 input, output, order_string = sys.argv[1:] | |
| 93 | |
| 94 new_order = parse_rangelist(order_string) | |
| 95 old_lines = get_lines(input) | |
| 96 reorder(old_lines, new_order, output) | |
| 97 | |
| 98 sys.exit(0) |