comparison galaxyp-galaxyp-toolshed-msconvert-9663ae998499/msconvert_wrapper.py @ 0:942cec8d04c7 draft

Initial commit.
author jmchilton
date Wed, 19 Sep 2012 00:07:16 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:942cec8d04c7
1 #!/usr/bin/env python
2 import optparse
3 import os
4 import sys
5 import tempfile
6 import shutil
7 import subprocess
8 import re
9 from os.path import basename
10 import logging
11
12 assert sys.version_info[:2] >= ( 2, 6 )
13
14 log = logging.getLogger(__name__)
15 working_directory = os.getcwd()
16 tmp_stderr_name = tempfile.NamedTemporaryFile(dir = working_directory, suffix = '.stderr').name
17 tmp_stdout_name = tempfile.NamedTemporaryFile(dir = working_directory, suffix = '.stdout').name
18
19 def stop_err( msg ):
20 sys.stderr.write( "%s\n" % msg )
21 sys.exit()
22
23 def read_stderr():
24 stderr = ''
25 if(os.path.exists(tmp_stderr_name)):
26 with open(tmp_stderr_name, 'rb') as tmp_stderr:
27 buffsize = 1048576
28 try:
29 while True:
30 stderr += tmp_stderr.read(buffsize)
31 if not stderr or len(stderr) % buffsize != 0:
32 break
33 except OverflowError:
34 pass
35 return stderr
36
37 def execute(command, stdin=None):
38 try:
39 with open(tmp_stderr_name, 'wb') as tmp_stderr:
40 with open(tmp_stdout_name, 'wb') as tmp_stdout:
41 proc = subprocess.Popen(args=command, shell=True, stderr=tmp_stderr.fileno(), stdout=tmp_stdout.fileno(), stdin=stdin, env=os.environ)
42 returncode = proc.wait()
43 if returncode != 0:
44 raise Exception, "Program returned with non-zero exit code %d. stderr: %s" % (returncode, read_stderr())
45 finally:
46 print open(tmp_stderr_name, "r").read()
47 print open(tmp_stdout_name, "r").read()
48
49
50 def delete_file(path):
51 if os.path.exists(path):
52 try:
53 os.remove(path)
54 except:
55 pass
56
57 def delete_directory(directory):
58 if os.path.exists(directory):
59 try:
60 shutil.rmtree(directory)
61 except:
62 pass
63
64 def symlink(source, link_name):
65 import platform
66 if platform.system() == 'Windows':
67 import win32file
68 win32file.CreateSymbolicLink(source, link_name, 1)
69 else:
70 os.symlink(source, link_name)
71
72
73 def copy_to_working_directory(data_file, relative_path):
74 if os.path.abspath(data_file) != os.path.abspath(relative_path):
75 shutil.copy(data_file, relative_path)
76 return relative_path
77
78 def __main__():
79 run_script()
80
81 #ENDTEMPLATE
82
83 to_extensions = ['mzML', 'mzXML', 'unindexed_mzML', 'unindexed_mzXML', 'mgf', 'txt', 'ms2', 'cms2']
84
85
86 def str_to_bool(v):
87 """ From http://stackoverflow.com/questions/715417/converting-from-a-string-to-boolean-in-python """
88 return v.lower() in ["yes", "true", "t", "1"]
89
90
91 def _add_filter(filters_file, contents):
92 filters_file.write("filter=\"%s\"\n" % contents)
93
94
95 def _read_table_numbers(path):
96 unique_numbers = set([])
97 input = open(path, "r")
98 first_line = True
99 for line in input:
100 if not line:
101 continue
102 line = line.strip()
103 if line.startswith("#"):
104 first_line = False
105 continue
106 match = re.match("\d+", line)
107 if match:
108 unique_numbers.add(int(match.group()))
109 first_line = False
110 return unique_numbers
111
112
113 def _add_filter_line_from_file(file, filter_file, filter_prefix):
114 if not file:
115 return
116 numbers = _read_table_numbers(file)
117 msconvert_int_set = " ".join([str(number) for number in numbers])
118 _add_filter(filter_file, "%s %s" % (filter_prefix, msconvert_int_set))
119
120
121 def _create_filters_file(options):
122 filters_file_path = "filters"
123 filters_file = open(filters_file_path, "w")
124 if options.filters_file:
125 filters_file.write(open(options.filters_file, "r").read())
126 for filter in options.filter:
127 _add_filter(filters_file, filter)
128 _add_filter_line_from_file(options.filter_indices_table, filters_file, "index")
129 _add_filter_line_from_file(options.filter_numbers_table, filters_file, "scanNumber")
130
131 filters_file.close()
132 print open(filters_file_path, "r").read()
133 return filters_file_path
134
135
136 def run_script():
137 parser = optparse.OptionParser()
138 parser.add_option('--input', dest='input')
139 parser.add_option('--output', dest='output')
140 parser.add_option('--fromextension', dest='fromextension')
141 parser.add_option('--toextension', dest='toextension', default='mzML', choices=to_extensions)
142 parser.add_option('--binaryencoding', dest='binaryencoding', choices=['32', '64'])
143 parser.add_option('--mzencoding', dest='mzencoding', choices=['32', '64'])
144 parser.add_option('--intensityencoding', dest='intensityencoding', choices=['32', '64'])
145 parser.add_option('--zlib', dest='zlib', default="false")
146 parser.add_option('--filter', dest='filter', action='append', default=[])
147 parser.add_option('--filters_file', dest='filters_file', default=None)
148 parser.add_option('--filter_indices_table', default=None)
149 parser.add_option('--filter_numbers_table', default=None)
150
151 (options, args) = parser.parse_args()
152
153 input_file = 'input.%s' % options.fromextension
154 copy_to_working_directory(options.input, input_file)
155 os.mkdir('output')
156 to_extension = options.toextension
157 if to_extension.startswith("unindexed_"):
158 to_extension = to_extension[len("unindexed_"):]
159 to_params = "--noindex"
160 else:
161 to_params = ""
162 cmd = "msconvert --%s %s -o output" % (to_extension, to_params)
163 if str_to_bool(options.zlib):
164 cmd = "%s %s" % (cmd, "--zlib")
165 if options.binaryencoding:
166 cmd = "%s --%s" % (cmd, options.binaryencoding)
167 if options.mzencoding:
168 cmd = "%s --mz%s" % (cmd, options.mzencoding)
169 if options.intensityencoding:
170 cmd = "%s --inten%s" % (cmd, options.intensityencoding)
171 cmd = "%s %s" % (cmd, input_file)
172 filters_file_path = _create_filters_file(options)
173 cmd = "%s -c %s" % (cmd, filters_file_path)
174 print cmd
175 execute(cmd)
176 output_files = os.listdir('output')
177 assert len(output_files) == 1
178 output_file = output_files[0]
179 shutil.copy(os.path.join('output', output_file), options.output)
180
181
182 if __name__ == '__main__': __main__()