htsworkflow/submission/results.py

   1 """Help collect and process results for submission
   2 """
   3 from collections import MutableMapping
   4 import os
   5 import shutil
   6 import logging
   7
   8 from collections import namedtuple
   9
  10 LOGGER = logging.getLogger(__name__)
  11
  12 class ResultMap(MutableMapping):
  13     """Store list of results
  14     """
  15     def __init__(self):
  16         self.results_order = []
  17         self.results = {}
  18
  19     def __iter__(self):
  20         for item in self.results_order:
  21             yield item
  22
  23     def __len__(self):
  24         l = len(self.results)
  25         assert l == len(self.results_order)
  26         return l
  27
  28     def __setitem__(self, key, value):
  29         self.results_order.append(key)
  30         self.results[key] = value
  31
  32     def __getitem__(self, key):
  33         return self.results[key]
  34
  35     def __delitem__(self, key):
  36         del self.results[key]
  37         i = self.results_order.index(key)
  38         del self.results_order[i]
  39
  40     def add_results_from_file(self, filename):
  41         pathname = os.path.abspath(filename)
  42         basepath, name = os.path.split(pathname)
  43         results = read_result_list(filename)
  44         for lib_id, lib_path in results:
  45             if not os.path.isabs(lib_path):
  46                 lib_path = os.path.join(basepath, lib_path)
  47             self[lib_id] = lib_path
  48
  49     def make_tree_from(self, source_path, destpath = None, link=True):
  50         """Create a tree using data files from source path.
  51         """
  52         if destpath is None:
  53             destpath = os.getcwd()
  54
  55         LOGGER.debug("Source_path: %s", source_path)
  56         LOGGER.debug("Dest_path: %s", destpath)
  57         for lib_id in self.results_order:
  58             lib_path = self.results[lib_id]
  59             LOGGER.debug("lib_path: %s", lib_path)
  60             if os.path.isabs(lib_path):
  61                 lib_path = os.path.relpath(lib_path, destpath)
  62
  63             LOGGER.debug('lib_path: %s', lib_path)
  64             lib_destination = os.path.join(destpath, lib_path)
  65             if not os.path.exists(lib_destination):
  66                 LOGGER.info("Making dir {0}".format(lib_destination))
  67                 os.mkdir(lib_destination)
  68
  69             source_rel_dir = os.path.join(source_path, lib_path)
  70             source_lib_dir = os.path.abspath(source_rel_dir)
  71             LOGGER.debug("source_lib_dir: %s", source_lib_dir)
  72
  73             for filename in os.listdir(source_lib_dir):
  74                 source_pathname = os.path.join(source_lib_dir, filename)
  75                 target_pathname = os.path.join(lib_destination, filename)
  76                 if not os.path.exists(source_pathname):
  77                     raise IOError(
  78                         "{0} does not exist".format(source_pathname))
  79                 if not (os.path.exists(target_pathname) or os.path.isdir(source_pathname)):
  80                     if link:
  81                         os.symlink(source_pathname, target_pathname)
  82                     else:
  83                         shutil.copy(source_pathname, target_pathname)
  84                     LOGGER.info(
  85                         'LINK {0} to {1}'.format(source_pathname,
  86                                                  target_pathname))
  87
  88 def read_result_list(filename):
  89     """
  90     Read a file that maps library id to result directory.
  91     Does not support spaces in filenames.
  92
  93     For example:
  94       10000 result/foo/bar
  95     """
  96     stream = open(filename, 'r')
  97     results = parse_result_list(stream)
  98     stream.close()
  99     return results
 100
 101
 102 def parse_result_list(stream):
 103     results = []
 104     for line in stream:
 105         line = line.rstrip()
 106         if not line.startswith('#') and len(line) > 0:
 107             library_id, result_dir = line.split()
 108             results.append((library_id, result_dir))
 109     return results