htsworkflow/submission/results.py

   1 """Help collect and process results for submission
   2 """
   3 import os
   4 import logging
   5
   6 from collections import namedtuple
   7
   8 LOGGER = logging.getLogger(__name__)
   9
  10 class ResultMap(object):
  11     """Store list of results
  12     """
  13     def __init__(self):
  14         self.results_order = []
  15         self.results = {}
  16
  17     def keys(self):
  18         return self.results_order
  19
  20     def values(self):
  21         return ( self.results[r] for r in self.results_order )
  22
  23     def items(self):
  24         return ( (r, self.results[r]) for r in self.results_order )
  25
  26     def __getitem__(self, key):
  27         return self.results[key]
  28
  29     def add_results_from_file(self, filename):
  30         pathname = os.path.abspath(filename)
  31         basepath, name = os.path.split(pathname)
  32         results = read_result_list(filename)
  33         for lib_id, lib_path in results:
  34             if not os.path.isabs(lib_path):
  35                 lib_path = os.path.join(basepath, lib_path)
  36             self.add_result(lib_id, lib_path)
  37
  38     def add_result(self, lib_id, lib_path):
  39         self.results_order.append(lib_id)
  40         self.results[lib_id] = lib_path
  41
  42     def make_tree_from(self, source_path, destpath = None):
  43         """Create a tree using data files from source path.
  44         """
  45         print source_path, destpath
  46         if destpath is None:
  47             destpath = os.getcwd()
  48
  49         for lib_id in self.results_order:
  50             lib_path = self.results[lib_id]
  51             lib_destination = os.path.join(destpath, lib_path)
  52             if not os.path.exists(lib_destination):
  53                 LOGGER.info("Making dir {0}".format(lib_destination))
  54                 os.mkdir(lib_destination)
  55
  56             source_rel_dir = os.path.join(source_path, lib_path)
  57             source_lib_dir = os.path.abspath(source_rel_dir)
  58
  59             print "source_lib_dir", source_lib_dir
  60             for filename in os.listdir(source_lib_dir):
  61                 source_pathname = os.path.join(source_lib_dir, filename)
  62                 target_pathname = os.path.join(lib_destination, filename)
  63                 if not os.path.exists(source_pathname):
  64                     raise IOError(
  65                         "{0} does not exist".format(source_pathname))
  66                 print target_pathname
  67                 if not os.path.exists(target_pathname):
  68                     os.symlink(source_pathname, target_pathname)
  69                     LOGGER.info(
  70                         'LINK {0} to {1}'.format(source_pathname,
  71                                                  target_pathname))
  72
  73 def read_result_list(filename):
  74     """
  75     Read a file that maps library id to result directory.
  76     Does not support spaces in filenames.
  77
  78     For example:
  79       10000 result/foo/bar
  80     """
  81     stream = open(filename, 'r')
  82     results = parse_result_list(stream)
  83     stream.close()
  84     return results
  85
  86
  87 def parse_result_list(stream):
  88     results = []
  89     for line in stream:
  90         line = line.rstrip()
  91         if not line.startswith('#') and len(line) > 0:
  92             library_id, result_dir = line.split()
  93             results.append((library_id, result_dir))
  94     return results