htsworkflow/submission/results.py

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