Mercurial > repos > miller-lab > genome_diversity

annotate gd_util.py @ 32:03c22b722882

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
remove BeautifulSoup dependency
author Richard Burhans <burhans@bx.psu.edu>
date Fri, 20 Sep 2013 13:54:23 -0400
parents 8997f2ca8c7a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
27
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
1 #!/usr/bin/env python
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
2
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
3 import base64
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4 import errno
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
5 import os
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
6 import subprocess
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
7 import sys
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
8 import zlib
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
9
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
10 ################################################################################
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
11
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
12 def die(message):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
13 print >> sys.stderr, message
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14 sys.exit(1)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
15
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
16 ################################################################################
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
17
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
18 def mkdir_p(path):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19 try:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 os.makedirs(path)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
21 except OSError, e:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
22 if e.errno <> errno.EEXIST:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 raise
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
24
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
25 ################################################################################
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
27 def run_program(prog, args, stdout=subprocess.PIPE, stderr=subprocess.PIPE):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 kwargs = {
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29 "bufsize": -1,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30 "close_fds": False,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31 "creationflags": 0,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 "cwd": None,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 "env": None,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34 "executable": prog,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
35 "preexec_fn": None,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 "shell": False,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 "startupinfo": None,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38 "stderr": stderr,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39 "stdin": None,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40 "stdout": stdout,
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41 "universal_newlines": False
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 }
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44 str_args = [str(x) for x in args]
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
45
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46 p = subprocess.Popen(str_args, **kwargs)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 (stdoutdata, stderrdata) = p.communicate()
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 rc = p.returncode
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
49
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 if rc != 0:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
51 die('FAILED:\n{0}\nCOMMAND:\n{1}'.format(stderrdata, ' '.join(str_args)))
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
52
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
53 return stdoutdata, stderrdata
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
54
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 ################################################################################
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 def unwrap_string(string):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 try:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 decoded_string = base64.b64decode(string)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60 except TypeError, message:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 die('base64.b64decode: {0}: {1}'.format(message, string))
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63 try:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64 return zlib.decompress(decoded_string)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
65 except zlib.error, message:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
66 die('zlib.decompress: {0}'.format(message))
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
68 ################################################################################
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
69
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
70 def wrap_string(string, level=9):
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
71 try:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
72 compressed_string = zlib.compress(string, level)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
73 except zlib.error, message:
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
74 die('zlib.compress: {0}'.format(message))
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
75 return base64.b64encode(compressed_string)
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
76
8997f2ca8c7a Update to Miller Lab devshed revision bae0d3306d3b
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
77 ################################################################################