def main():
fs = FS()
args = get_arguments()
#print args
src_files = []
if args.recursive:
for dirpath, dirnames, filenames in os.walk(args.source_dir):
for filename in filenames:
if args.file_type is None:
src_files.append(os.path.join(dirpath, filename))
elif filename.endswith(args.file_type):
src_files.append(os.path.join(dirpath, filename))
else:
for filename in next(os.walk(args.source_dir))[2]:
if args.file_type is None:
src_files.append(os.path.join(args.source_dir, filename))
elif filename.endswith(args.file_type):
src_files.append(os.path.join(args.source_dir, filename))
cmds = []
dest_files = []
if not args.mapping_file is None:
src2dest = {}
content = fs.read2list(args.mapping_file)
fs.close()
#if args.mapping_header:
for i in xrange(0, len(content)):
try:
(src_filename, dest_filename) = content[i].split()
src2dest[src_filename] = dest_filename
except Exception as ex:
logging.warning(
"mapping src file to dest file [empty line?]: " + str(ex))
src_file2dest_file = {}
for src_file in src_files:
src_filename = os.path.basename(src_file)
dest_filename = src_filename
if not args.re_list is None:
for regular_expression in args.re_list:
(find_str, replace_str) = regular_expression.split(":")
print find_str, "be replaced with:", replace_str
dest_filename = re.sub(find_str, replace_str, dest_filename)
dest_file = os.path.join(args.out_dir, dest_filename)
src_file2dest_file[src_file] = dest_file
elif not args.mapping_file is None:
if not src_filename in src2dest:
logging.warning(
"\"%s\" is NOT listed in the mapping file!!!\n" %
src_filename)
else:
dest_file = os.path.join(args.out_dir, src2dest[src_filename])
src_file2dest_file[src_file] = dest_file
else:
dest_file = os.path.join(args.out_dir, src_filename)
src_file2dest_file[src_file] = dest_file
for (src_file, dest_file) in sorted(src_file2dest_file.items()):
if not os.path.exists(src_file):
logging.error("\"%s\" does not exist!!!\n" % src_file)
sys.exit()
if args.hard_copy:
cmd_line = "cp '%s' '%s'" % (src_file, dest_file)
else:
cmd_line = "ln -s %s %s" % (src_file, dest_file)
logging.info(cmd_line)
#print "\n"
os.system(cmd_line)
def main():
fs = FS()
z = ZIP()
args = get_arguments()
src_dir = args.src_dir
next_seq_merge_dir = os.path.join(src_dir, "next_seq_merge")
if not os.path.exists(next_seq_merge_dir):
print next_seq_merge_dir, "is created!\n"
os.makedirs(next_seq_merge_dir)
next_seq_merge_command_sh = os.path.join(next_seq_merge_dir,
"next_seq_merge_commands.sh")
gz_file_list = []
for root, dirs, files in os.walk(src_dir, topdown=False):
for filename in files:
if filename.endswith(r".gz"):
gz_file_list.append(os.path.join(root, filename))
gz_file_list = sorted(gz_file_list)
if args.check_gz:
z.test_gz(gz_file_list)
#print gz_file_list
#sys.exit()
#z.test_gz(gz_file_list)
#for gz_file in gz_file_list:
# test_gz_cmd = "gunzip -t " + gz_file
# fs.write(next_seq_merge_command_sh, test_gz_cmd + "\n")
#fs.write(next_seq_merge_command_sh, test_gz_cmd + "\n\n")
out_gz_list = []
for gz_file in sorted(gz_file_list):
if "_L001_" in gz_file:
all_lane_file_found = True
out_gz = os.path.join(
next_seq_merge_dir,
os.path.basename(gz_file.replace("_L001_", "_")))
to_cat_list = [gz_file]
to_cat_list.append(gz_file.replace("_L001_", "_L002_"))
to_cat_list.append(gz_file.replace("_L001_", "_L003_"))
to_cat_list.append(gz_file.replace("_L001_", "_L004_"))
for to_cat_gz in to_cat_list:
if to_cat_gz not in gz_file_list:
print "CAN'T find", to_cat_gz
all_lane_file_found = False
if all_lane_file_found:
fs.write(next_seq_merge_command_sh,
z.cat_gz(out_gz, to_cat_list) + "\n")
out_gz_list.append(out_gz)
fs.write(next_seq_merge_command_sh, "\n\n")
if args.fastq:
for out_gz in out_gz_list:
fs.write(next_seq_merge_command_sh, "gunzip " + out_gz + "\n")
fs.write(next_seq_merge_command_sh, "\n\n")
qsub_cmd = "nohup qsub_cmd.py " + next_seq_merge_command_sh + " >/dev/null 2>&1 &"
#print cmd3
fs.write(next_seq_merge_command_sh, "\n#run this: " + qsub_cmd + "\n")
fs.close()
os.system(qsub_cmd)
