Mercurial > repos > miller-lab > genome_diversity
annotate LocationFile.py @ 10:9b92372de9f6
rename snp to gsSnp, sap to gdSap
| author | Richard Burhans <burhans@bx.psu.edu> |
|---|---|
| date | Tue, 04 Sep 2012 12:49:53 -0400 |
| parents | 22fe0154fa54 |
| children |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/env python |
| 2 | |
| 3 import sys | |
| 4 | |
| 5 def die( message ): | |
| 6 print >> sys.stderr, message | |
| 7 sys.exit(1) | |
| 8 | |
| 9 def open_or_die( filename, mode='r', message=None ): | |
| 10 if message is None: | |
| 11 message = 'Error opening {0}'.format( filename ) | |
| 12 try: | |
| 13 fh = open( filename, mode ) | |
| 14 except IOError, err: | |
| 15 die( '{0}: {1}'.format( message, err.strerror ) ) | |
| 16 return fh | |
| 17 | |
| 18 class LocationFile( object ): | |
| 19 def __init__( self, filename, comment_chars=None, delimiter='\t', key_column=0 ): | |
| 20 self.filename = filename | |
| 21 if comment_chars is None: | |
| 22 self.comment_chars = ( '#' ) | |
| 23 else: | |
| 24 self.comment_chars = tuple( comment_chars ) | |
| 25 self.delimiter = delimiter | |
| 26 self.key_column = key_column | |
| 27 self._map = {} | |
| 28 self._populate_map() | |
| 29 | |
| 30 def _populate_map( self ): | |
| 31 try: | |
| 32 with open( self.filename ) as fh: | |
| 33 line_number = 0 | |
| 34 for line in fh: | |
| 35 line_number += 1 | |
| 36 line = line.rstrip( '\r\n' ) | |
| 37 if not line.startswith( self.comment_chars ): | |
| 38 elems = line.split( self.delimiter ) | |
| 39 if len( elems ) <= self.key_column: | |
| 40 die( 'Location file {0} line {1}: less than {2} columns'.format( self.filename, line_number, self.key_column + 1 ) ) | |
| 41 else: | |
| 42 key = elems.pop( self.key_column ) | |
| 43 if key in self._map: | |
| 44 if self._map[key] != elems: | |
| 45 die( 'Location file {0} line {1}: duplicate key "{2}"'.format( self.filename, line_number, key ) ) | |
| 46 else: | |
| 47 self._map[key] = elems | |
| 48 except IOError, err: | |
| 49 die( 'Error opening location file {0}: {1}'.format( self.filename, err.strerror ) ) | |
| 50 | |
| 51 def get_values( self, key ): | |
| 52 if key in self._map: | |
| 53 rval = self._map[key] | |
| 54 if len( rval ) == 1: | |
| 55 return rval[0] | |
| 56 else: | |
| 57 return rval | |
| 58 else: | |
| 59 die( 'key "{0}" not found in location file {1}'.format( key, self.filename ) ) | |
|
9
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
60 |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
61 def get_values_if_exists( self, key ): |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
62 if key in self._map: |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
63 rval = self._map[key] |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
64 if len( rval ) == 1: |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
65 return rval[0] |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
66 else: |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
67 return rval |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
68 else: |
|
22fe0154fa54
added support for heterochromatic regions
Richard Burhans <burhans@bx.psu.edu>
parents:
0
diff
changeset
|
69 return None |