Mercurial > repos > jdv > albacore

comparison albacore_1D.py @ 4:8a9f61d08201 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/jvolkening/galaxy-tools/tree/master/tools/albacore commit 9a7fa5134243f712595bef72121b8006be8a318b
author jdv
date Fri, 13 Oct 2017 16:13:22 -0400
parents d561e3f9ccbb
children 3870c546db18
comparison
equal deleted inserted replaced
3:d561e3f9ccbb 4:8a9f61d08201
29 "--files_per_batch_folder", "0", 29 "--files_per_batch_folder", "0",
30 "--output_format", out_fmt, 30 "--output_format", out_fmt,
31 "--reads_per_fastq_batch", "999999999" ] + 31 "--reads_per_fastq_batch", "999999999" ] +
32 ["--barcoding"] * demux ) 32 ["--barcoding"] * demux )
33 33
34 out_path = "out_dir/workspace"
35 pass_path = os.path.join( out_path, "pass" )
36 if os.path.exists( pass_path ):
37 out_path = pass_path
34 if demux: 38 if demux:
35 #check for demuxed albacore output and copy to Galaxy output 39 #check for demuxed albacore output and copy to Galaxy output
36 final_dir = "final" 40 final_dir = "final"
37 if not os.path.exists(final_dir): 41 if not os.path.exists(final_dir):
38 os.makedirs(final_dir) 42 os.makedirs(final_dir)
39 dirs = glob.glob("out_dir/workspace/*") 43 dirs = glob.glob( os.path.join(out_path, "*") )
40 for d in dirs: 44 for d in dirs:
41 45
42 if out_fmt == 'fastq': 46 if out_fmt == 'fastq':
43 bc = os.path.basename( os.path.normpath( d ) ) + ".fastq" 47 bc = os.path.basename( os.path.normpath( d ) ) + ".fastq"
44 print(d) 48 print(d)
49 raise ValueError('No or multiple FASTQ output files found') 53 raise ValueError('No or multiple FASTQ output files found')
50 found_file = files[0] 54 found_file = files[0]
51 shutil.copy(found_file, out) 55 shutil.copy(found_file, out)
52 56
53 elif out_fmt == 'fast5': 57 elif out_fmt == 'fast5':
54 bc = os.path.basename( os.path.normpath( d ) ) + ".fast5.tar.gz" 58 if (os.path.isfile(d)):
55 print(d) 59 if (d.endswith('.fast5')):
56 print(bc) 60 bc = os.path.basename( os.path.normpath(d) ) + ".tar.gz"
61 files = [d]
62 else:
63 continue
64 else:
65 bc = os.path.basename( os.path.normpath( d ) ) + ".fast5.tar.gz"
66 files = glob.glob( os.path.join( d, "**", "*.fast5"), recursive=True)
57 out = os.path.join( final_dir, bc ) 67 out = os.path.join( final_dir, bc )
58 files = glob.glob( os.path.join( d, "**", "*.fast5"), recursive=True)
59 if len(files) < 1: 68 if len(files) < 1:
60 raise ValueError('No FAST5 output files found') 69 raise ValueError('No FAST5 output files found')
61 tar = tarfile.open(out, 'w:gz') 70 tar = tarfile.open(out, 'w:gz')
62 tar.add( d ) 71 tar.add( d )
63 tar.close() 72 tar.close()
66 raise ValueError('Bad output format specified') 75 raise ValueError('Bad output format specified')
67 76
68 else: 77 else:
69 if out_fmt == 'fastq': 78 if out_fmt == 'fastq':
70 #check for single albacore output and copy to Galaxy output 79 #check for single albacore output and copy to Galaxy output
71 files = glob.glob("out_dir/workspace/*.fastq") 80 files = glob.glob( os.path.join(out_path, "*.fastq") )
72 if len(files) != 1: 81 if len(files) != 1:
73 raise ValueError('No or multiple FASTQ output files found') 82 raise ValueError('No or multiple FASTQ output files found')
74 found_file = files[0] 83 found_file = files[0]
75 shutil.copy(found_file, out_file) 84 shutil.copy(found_file, out_file)
76 elif out_fmt == 'fast5': 85 elif out_fmt == 'fast5':
77 #check for single albacore output and copy to Galaxy output 86 #check for single albacore output and copy to Galaxy output
78 files = glob.glob("out_dir/workspace/**/*.fast5", recursive=True) 87 files = glob.glob( os.path.join(out_path,"**","*.fast5"), recursive=True )
79 if len(files) < 1: 88 if len(files) < 1:
80 raise ValueError('No FAST5 output files found') 89 raise ValueError('No FAST5 output files found')
81 tar = tarfile.open(out_file, 'w:gz') 90 tar = tarfile.open(out_file, 'w:gz')
82 tar.add("out_dir/workspace") 91 tar.add(out_path)
83 tar.close() 92 tar.close()
84 else: 93 else:
85 raise ValueError('Bad output format specified') 94 raise ValueError('Bad output format specified')
86 95
87 96
95 # python's tarfile interface does not sanitize file paths within 104 # python's tarfile interface does not sanitize file paths within
96 # tarballs, which can be a big security risk. GNU tar does sanitize by 105 # tarballs, which can be a big security risk. GNU tar does sanitize by
97 # default, so it's easier/safer here just to call the system tar 106 # default, so it's easier/safer here just to call the system tar
98 subprocess.call([ 107 subprocess.call([
99 "tar", 108 "tar",
109 "--warning=no-unknown-keyword",
100 "-xf", 110 "-xf",
101 fn, 111 fn,
102 "-C", 112 "-C",
103 in_dir]) 113 in_dir])
104 114