annotate tools/protein_analysis/seq_analysis_utils.py @ 29:3cb02adf4326 draft

v0.2.9 Python style improvements
author peterjc
date Wed, 01 Feb 2017 09:46:14 -0500
parents 20139cb4c844
children 6d9d7cdf00fc
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 """A few useful functions for working with FASTA files and running jobs.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3 This module was originally written to hold common code used in both the TMHMM
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 and SignalP wrappers in Galaxy.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 Given Galaxy currently supports Python 2.4+ this cannot use the Python module
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 multiprocessing so the function run_jobs instead is a simple pool approach
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8 using just the subprocess library.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
9 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
10 import sys
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
11 import os
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
12 import subprocess
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13 from time import sleep
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
15 __version__ = "0.0.2"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
16
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
17 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
18 from multiprocessing import cpu_count
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
19 except ImportError:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
20 # Must be under Python 2.5, this is copied from multiprocessing:
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
21 def cpu_count():
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
22 """Returns the number of CPUs in the system."""
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
23 if sys.platform == 'win32':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
24 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
25 num = int(os.environ['NUMBER_OF_PROCESSORS'])
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
26 except (ValueError, KeyError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
27 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
28 elif 'bsd' in sys.platform or sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
29 comm = '/sbin/sysctl -n hw.ncpu'
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
30 if sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
31 comm = '/usr' + comm
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
32 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
33 with os.popen(comm) as p:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
34 num = int(p.read())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
35 except ValueError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
36 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
37 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
38 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
39 num = os.sysconf('SC_NPROCESSORS_ONLN')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
40 except (ValueError, OSError, AttributeError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
41 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
42
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
43 if num >= 1:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
44 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
45 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
46 raise NotImplementedError('cannot determine number of cpus')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
47
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
48
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
49 def thread_count(command_line_arg, default=1):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
50 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
51 num = int(command_line_arg)
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
52 except ValueError:
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
53 num = default
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
54 if num < 1:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
55 sys.exit("Threads argument %r is not a positive integer" % command_line_arg)
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
56 # Cap this with the pysical limit of the machine,
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
57 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
58 num = min(num, cpu_count())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
59 except NotImplementedError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
60 pass
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
61 # For debugging,
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
62 # hostname = os.environ.get("HOSTNAME", "this machine")
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
63 # sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
64 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
65
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
66
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
67 def fasta_iterator(filename, max_len=None, truncate=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
68 """Simple FASTA parser yielding tuples of (title, sequence) strings."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
69 handle = open(filename)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
70 title, seq = "", ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
71 for line in handle:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 if line.startswith(">"):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 if max_len and len(seq) > max_len:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
77 raise ValueError("Sequence %s is length %i, max length %i"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
80 title = line[1:].rstrip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 seq = ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
82 elif title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
83 seq += line.strip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84 elif not line.strip() or line.startswith("#"):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
85 # Ignore blank lines, and any comment lines
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
86 # between records (starting with hash).
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
87 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
88 else:
25
41a42022f815 Uploaded v0.2.6, embedded citations
peterjc
parents: 7
diff changeset
89 handle.close()
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
90 raise ValueError("Bad FASTA line %r" % line)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
91 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
94 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
95 if max_len and len(seq) > max_len:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
96 raise ValueError("Sequence %s is length %i, max length %i"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
97 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
99 raise StopIteration
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
101
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 def split_fasta(input_filename, output_filename_base, n=500, truncate=None, keep_descr=False, max_len=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 """Split FASTA file into sub-files each of at most n sequences.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105 Returns a list of the filenames used (based on the input filename).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
106 Each sequence can also be truncated (since we only need the start for
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
107 SignalP), and have its description discarded (since we don't usually
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
108 care about it and some tools don't like very long title lines).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
109
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 If a max_len is given and any sequence exceeds it no temp files are
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111 created and an exception is raised.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
113 iterator = fasta_iterator(input_filename, max_len, truncate)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
114 files = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
115 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
116 while True:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
117 records = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
118 for i in range(n):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
119 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
120 records.append(iterator.next())
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
121 except StopIteration:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
122 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
123 if not records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
124 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
125 new_filename = "%s.%i.tmp" % (output_filename_base, len(files))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
126 handle = open(new_filename, "w")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
127 if keep_descr:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
128 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
129 handle.write(">%s\n" % title)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
130 for i in range(0, len(seq), 60):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
131 handle.write(seq[i:i + 60] + "\n")
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
132 else:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
133 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
134 handle.write(">%s\n" % title.split()[0])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
135 for i in range(0, len(seq), 60):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
136 handle.write(seq[i:i + 60] + "\n")
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
137 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
138 files.append(new_filename)
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
139 # print "%i records in %s" % (len(records), new_filename)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
140 except ValueError, err:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
141 # Max length failure from parser - clean up
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
142 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
143 handle.close()
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
144 except Exception:
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
145 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
146 for f in files:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
147 if os.path.isfile(f):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
148 os.remove(f)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
149 raise err
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
150 for f in files:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
151 assert os.path.isfile(f), "Missing split file %r (!??)" % f
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
152 return files
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
153
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
154
6
39a6e46cdda3 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository
peterjc
parents: 3
diff changeset
155 def run_jobs(jobs, threads, pause=10, verbose=False):
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
156 """Takes list of cmd strings, returns dict with error levels."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
157 pending = jobs[:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
158 running = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
159 results = {}
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
160 if threads == 1:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
161 # Special case this for speed, don't need the waits
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
162 for cmd in jobs:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
163 results[cmd] = subprocess.call(cmd, shell=True)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
164 return results
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
165 while pending or running:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
166 # See if any have finished
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
167 for (cmd, process) in running:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
168 return_code = process.poll() # non-blocking
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
169 if return_code is not None:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
170 results[cmd] = return_code
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
171 running = [(cmd, process) for (cmd, process) in running
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
172 if cmd not in results]
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
173 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
174 print "%i jobs pending, %i running, %i completed" \
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
175 % (len(pending), len(running), len(results))
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
176 # See if we can start any new threads
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
177 while pending and len(running) < threads:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
178 cmd = pending.pop(0)
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
179 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
180 print cmd
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
181 process = subprocess.Popen(cmd, shell=True)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
182 running.append((cmd, process))
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
183 # Loop...
6
39a6e46cdda3 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository
peterjc
parents: 3
diff changeset
184 sleep(pause)
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
185 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
186 print "%i jobs completed" % len(results)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
187 assert set(jobs) == set(results)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
188 return results