Mercurial > repos > mvdbeek > sra_tools
annotate sra.py @ 5:6f68b86da01a draft default tip
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit d754d6a6fc019236513d2228542fe1a2757d55cf-dirty
author | mvdbeek |
---|---|
date | Wed, 30 Sep 2015 11:47:04 -0400 |
parents | 68527898a9ea |
children |
rev | line source |
---|---|
0
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
1 """ |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
2 NCBI sra class |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
3 """ |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
4 import logging |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
5 import binascii |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
6 from galaxy.datatypes.data import nice_size |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
7 from galaxy.datatypes.binary import Binary |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
8 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
9 log = logging.getLogger(__name__) |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
10 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
11 class Sra(Binary): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
12 """ Sequence Read Archive (SRA) """ |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
13 file_ext = 'sra' |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
14 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
15 def __init__( self, **kwd ): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
16 Binary.__init__( self, **kwd ) |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
17 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
18 def sniff( self, filename ): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
19 """ The first 8 bytes of any NCBI sra file is 'NCBI.sra', and the file is binary. |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
20 For details about the format, see http://www.ncbi.nlm.nih.gov/books/n/helpsra/SRA_Overview_BK/#SRA_Overview_BK.4_SRA_Data_Structure |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
21 """ |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
22 try: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
23 header = open(filename).read(8) |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
24 if binascii.b2a_hex(header) == binascii.hexlify('NCBI.sra'): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
25 return True |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
26 else: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
27 return False |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
28 except: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
29 return False |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
30 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
31 def set_peek(self, dataset, is_multi_byte=False): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
32 if not dataset.dataset.purged: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
33 dataset.peek = 'Binary sra file' |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
34 dataset.blurb = nice_size(dataset.get_size()) |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
35 else: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
36 dataset.peek = 'file does not exist' |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
37 dataset.blurb = 'file purged from disk' |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
38 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
39 def display_peek(self, dataset): |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
40 try: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
41 return dataset.peek |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
42 except: |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
43 return 'Binary sra file (%s)' % (nice_size(dataset.get_size())) |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
44 |
68527898a9ea
planemo upload for repository https://bitbucket.org/drosofff/gedtools/ commit 92d32947393757598262fbaed175b2c6fd84c181
mvdbeek
parents:
diff
changeset
|
45 Binary.register_sniffable_binary_format('sra', 'sra', Sra) |