Mercurial > repos > brenninc > xdirectory_reader_limited_by_data_table

comparison directory_copier.py @ 0:ccabef3f7d5f draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded first version
author brenninc
date Sun, 08 May 2016 11:01:03 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:ccabef3f7d5f
1 import gzip
2 import optparse # using optparse as hydra still python 2.6
3 import os.path
4 import shutil
5 import sys
6
7 def report_error(*args):
8 sys.stderr.write(' '.join(map(str,args)) + '\n')
9 sys.stderr.flush()
10 sys.exit(1)
11
12
13 def get_tool_data(name):
14 root_dir = os.path.dirname((os.path.realpath(__file__)))
15 path = os.path.join(root_dir,"tool-data",name)
16 if not(os.path.isfile(path)):
17 report_error(name,"file not found in tool's tool-data folder. Please ask you galaxy admin to add it back")
18 return path
19
20
21 def check_white_list(path_to_check):
22 white_list = get_tool_data("white-list.ini")
23 with open(white_list, 'r') as white_list_file:
24 for line in white_list_file:
25 line = line.strip()
26 if len(line) >= 1 and path_to_check.startswith(line):
27 return True
28 report_error(path_to_check,"has not been included in the white list. Please contact the local galaxy admin to add it.")
29
30
31 def check_black_list(path_to_check):
32 black_list = get_tool_data("black-list.ini")
33 with open(black_list, 'r') as black_list_file:
34 for line in black_list_file:
35 line = line.strip()
36 if len(line) >= 1 and line in path_to_check:
37 report_error(line,"has been black list so",path_to_check,"is not allowed. Please contact the local galaxy admin to change that, or add a symlink.")
38 return True
39
40
41 def check_pattern_get_new_name(a_file, ending, options):
42 if options.start:
43 if not(a_file.startswith(options.start)):
44 return None
45 if options.last:
46 if ending[0] == ".":
47 last = options.last + ending
48 else:
49 if options.last[-1] == ".":
50 last = options.last + ending
51 else:
52 last = options.last + "." + ending
53 if not(a_file.endswith(last)):
54 return None
55 if options.new_ending:
56 name = a_file[:-len(ending)]
57 if options.new_ending[0] ==".":
58 if name[-1] == ".":
59 name = name[:-1]
60 return name + options.new_ending
61 if options.decompress:
62 if a_file.endswith(".gz"):
63 return a_file[:-3]
64 return a_file
65
66
67 def check_and_get_new_name(a_file, options):
68 for ending in options.endings:
69 if a_file.endswith(ending):
70 return check_pattern_get_new_name (a_file, ending, options)
71 return None
72
73
74 def link(a_file, new_name, path):
75 file_path = os.path.join(os.path.realpath(path), a_file)
76 sym_path = os.path.join(os.path.realpath("output"), new_name)
77 #if not(os.path.exists(sym_path)):
78 os.link(file_path, sym_path)
79
80
81 def decompress(a_file, new_name, path):
82 file_path = os.path.join(os.path.realpath(path), a_file)
83 target_path = os.path.join(os.path.realpath("output"), new_name)
84 with gzip.open(file_path, 'rb') as f_in, open(target_path, 'wb') as f_out:
85 shutil.copyfileobj(f_in, f_out)
86
87
88 def copy_and_link(path, options):
89 os.mkdir("output")
90 with open(options.list, 'w') as list_file:
91 files = os.listdir(path)
92 files.sort()
93 for a_file in files:
94 new_name = check_and_get_new_name(a_file, options)
95 if new_name:
96 list_file.write(new_name)
97 list_file.write("\n")
98 if options.decompress:
99 if a_file.endswith(".gz"):
100 decompress(a_file, new_name,path)
101 else:
102 link(a_file, new_name, path)
103 elif options.link:
104 link(a_file, new_name, path)
105
106
107 if __name__ == '__main__':
108 parser = optparse.OptionParser()
109 parser.add_option("--path", action="store", type="string",
110 help="Path of directory to check. ")
111 parser.add_option("--ending", action="append", type="string", dest="endings",
112 help="Ending that can be listed and if requested linked or decompressed. ")
113 parser.add_option("--start", action="store", type="string",
114 help="String that must be at the start of the file name ")
115 parser.add_option("--last", action="store", type="string",
116 help="String that must be the last bit of the file name before the endings")
117 parser.add_option("--new_ending", action="store", type="string",
118 help="New ending to replace any previous ending in list and if required links or decompressions. Note: If not set decompression will auto remove the compressioned part of the ending")
119 #parser.add_option("--regex", action="store", type="string",
120 # help="Regex pattern the file name (less . ending) must match before the endings")
121 parser.add_option("--list", action="store", type="string",
122 help="Path to where all files should be listed. ")
123 parser.add_option("--link", action="store_true", default=False,
124 help="If set will cause links to be added in output directory. ")
125 parser.add_option("--decompress", action="store_true", default=False,
126 help="If set will cause gz files to be decompressed or if not a supported decompression ending linked.")
127 (options, args) = parser.parse_args()
128
129
130 path = options.path.strip()
131 if path[-1] != '/':
132 path = path + "/"
133 check_white_list(path)
134 print path, "white listed"
135 check_black_list(path)
136 print path, "not black listed"
137 copy_and_link(path, options)
138