diff directory_copier.py @ 0:ccabef3f7d5f draft

Uploaded first version
author brenninc
date Sun, 08 May 2016 11:01:03 -0400
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/directory_copier.py	Sun May 08 11:01:03 2016 -0400
@@ -0,0 +1,138 @@
+import gzip
+import optparse  # using optparse as hydra still python 2.6
+import os.path
+import shutil
+import sys
+
+def report_error(*args):
+    sys.stderr.write(' '.join(map(str,args)) + '\n')
+    sys.stderr.flush()
+    sys.exit(1)
+
+
+def get_tool_data(name):
+    root_dir = os.path.dirname((os.path.realpath(__file__)))
+    path = os.path.join(root_dir,"tool-data",name)
+    if not(os.path.isfile(path)): 
+        report_error(name,"file not found in tool's tool-data folder. Please ask you galaxy admin to add it back")
+    return path
+
+
+def check_white_list(path_to_check):
+    white_list = get_tool_data("white-list.ini")
+    with open(white_list, 'r') as white_list_file:
+        for line in white_list_file:
+            line = line.strip()
+            if len(line) >= 1 and path_to_check.startswith(line):
+                return True
+    report_error(path_to_check,"has not been included in the white list. Please contact the local galaxy admin to add it.")
+
+
+def check_black_list(path_to_check):
+    black_list = get_tool_data("black-list.ini")
+    with open(black_list, 'r') as black_list_file:
+        for line in black_list_file:
+            line = line.strip()
+            if len(line) >= 1 and line in path_to_check:
+                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.")
+    return True
+
+
+def check_pattern_get_new_name(a_file, ending, options):
+    if options.start:
+        if not(a_file.startswith(options.start)):
+            return None
+    if options.last:
+        if ending[0] == ".":
+            last = options.last + ending
+        else:
+            if options.last[-1] == ".":
+                last = options.last + ending
+            else:
+                last = options.last + "." + ending
+        if not(a_file.endswith(last)):
+            return None
+    if options.new_ending:
+        name = a_file[:-len(ending)]
+        if options.new_ending[0] ==".":
+            if name[-1] == ".":
+                name = name[:-1]
+        return name + options.new_ending
+    if options.decompress:
+        if a_file.endswith(".gz"):
+            return a_file[:-3]
+    return a_file
+
+
+def check_and_get_new_name(a_file, options):
+    for ending in options.endings:
+        if a_file.endswith(ending):
+            return check_pattern_get_new_name (a_file, ending, options)
+    return None
+
+
+def link(a_file, new_name, path):
+    file_path = os.path.join(os.path.realpath(path), a_file)
+    sym_path = os.path.join(os.path.realpath("output"), new_name)
+    #if not(os.path.exists(sym_path)):
+    os.link(file_path, sym_path)
+
+
+def decompress(a_file, new_name, path):
+    file_path = os.path.join(os.path.realpath(path), a_file)
+    target_path = os.path.join(os.path.realpath("output"), new_name)
+    with gzip.open(file_path, 'rb') as f_in, open(target_path, 'wb') as f_out:
+        shutil.copyfileobj(f_in, f_out)
+
+
+def copy_and_link(path, options):
+    os.mkdir("output")
+    with open(options.list, 'w') as list_file:
+        files = os.listdir(path)
+        files.sort()
+        for a_file in files:
+            new_name = check_and_get_new_name(a_file, options)
+            if new_name:
+                list_file.write(new_name)
+                list_file.write("\n")
+                if options.decompress:
+                    if a_file.endswith(".gz"):
+                        decompress(a_file, new_name,path)
+                    else:
+                        link(a_file, new_name, path)
+                elif options.link:
+                    link(a_file, new_name, path)
+
+
+if __name__ == '__main__':
+    parser = optparse.OptionParser()
+    parser.add_option("--path", action="store", type="string",
+                      help="Path of directory to check. ")
+    parser.add_option("--ending", action="append", type="string", dest="endings",
+                      help="Ending that can be listed and if requested linked or decompressed. ")
+    parser.add_option("--start", action="store", type="string",
+                      help="String that must be at the start of the file name ")
+    parser.add_option("--last", action="store", type="string",
+                      help="String that must be the last bit of the file name before the endings")
+    parser.add_option("--new_ending", action="store", type="string", 
+                      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")
+    #parser.add_option("--regex", action="store", type="string",
+    #                  help="Regex pattern the file name (less . ending) must match before the endings")
+    parser.add_option("--list", action="store", type="string",
+                      help="Path to where all files should be listed. ")
+    parser.add_option("--link", action="store_true", default=False,
+                      help="If set will cause links to be added in output directory. ")
+    parser.add_option("--decompress", action="store_true", default=False,
+                      help="If set will cause gz files to be decompressed or if not a supported decompression ending linked.")
+    (options, args) = parser.parse_args()
+
+
+    path = options.path.strip()
+    if path[-1] != '/':
+        path = path + "/"
+    check_white_list(path)
+    print path, "white listed"
+    check_black_list(path)
+    print path, "not black listed"
+    copy_and_link(path, options)
+