encode_submission/scan_extension.py

   1 from __future__ import print_function, unicode_literals
   2
   3 from optparse import OptionParser
   4 import os
   5 import sys
   6 from pprint import pprint
   7
   8 def main(cmdline=None):
   9     parser = make_parser()
  10     opts, args = parser.parse_args(cmdline)
  11
  12     extensions = scan(args)
  13     common_extensions = find_common_suffix(extensions)
  14
  15     if opts.rdf:
  16         print_rdf(common_extensions)
  17     else:
  18         print(common_extensions)
  19
  20 def make_parser():
  21     parser = OptionParser("%prog: directory [directory...]")
  22     parser.add_option('--rdf', action="store_true", default=False,
  23                       help="Produce rdf configuration file for ucsc_gather")
  24     return parser
  25
  26 def scan(toscan):
  27     index = {}
  28     for cur_scan_dir in toscan:
  29         for path, dirnames, filenames in os.walk(cur_scan_dir):
  30             for filename in filenames:
  31                 base, ext = os.path.splitext(filename)
  32                 if ext in ('.daf', 'ddf'):
  33                     continue
  34                 next_index = index
  35                 for c in filename[::-1]:
  36                     next_index = next_index.setdefault(c, {})
  37     return index
  38
  39 def find_common_suffix(index, tail=[]):
  40     if len(tail) > 0 and len(index) > 1:
  41         return "".join(tail[::-1])
  42
  43     results = []
  44     for key, choice in index.items():
  45         r = find_common_suffix(choice, tail+[key])
  46         if r is not None:
  47             results.append (r)
  48
  49     if len(results) == 0:
  50         return None
  51     elif len(results) == 1:
  52         return results[0]
  53     else:
  54         return results
  55
  56 def print_rdf(common_extensions):
  57     import RDF
  58     from htsworkflow.util import rdfhelp
  59     model = rdfhelp.get_model()
  60
  61     viewName = 'http://jumpgate.caltech.edu/wiki/SubmissionsLog/NAME/view/'
  62     subView = RDF.NS(viewName)
  63     fileReTerm = rdfhelp.dafTermOntology['filename_re']
  64
  65     count = 1
  66     for ext in common_extensions:
  67         s = RDF.Statement(subView['VIEW{0}'.format(count)],
  68                           fileReTerm,
  69                           '.*{0}$'.format(ext.replace('.', '\\.')))
  70         model.add_statement(s)
  71         count += 1
  72
  73     writer = rdfhelp.get_serializer()
  74     writer.set_namespace('thisSubmissionView', subView._prefix)
  75     print(writer.serialize_model_to_string(model))
  76
  77 if __name__ == "__main__":
  78     main()