htsworkflow/pipelines/firecrest.py

   1 """
   2 Extract information about the Firecrest run
   3
   4 Firecrest
   5   class holding the properties we found
   6 firecrest
   7   Firecrest factory function initalized from a directory name
   8 fromxml
   9   Firecrest factory function initalized from an xml dump from
  10   the Firecrest object.
  11 """
  12
  13 from datetime import date
  14 from glob import glob
  15 import logging
  16 import os
  17 import re
  18 import time
  19
  20 from htsworkflow.pipelines.runfolder import \
  21    ElementTree, \
  22    VERSION_RE, \
  23    EUROPEAN_STRPTIME
  24
  25 LOGGER = logging.getLogger(__name__)
  26
  27 __docformat__ = "restructuredtext en"
  28
  29 class Firecrest(object):
  30     XML_VERSION=1
  31
  32     # xml tag names
  33     FIRECREST = 'Firecrest'
  34     SOFTWARE_VERSION = 'version'
  35     START = 'FirstCycle'
  36     STOP = 'LastCycle'
  37     DATE = 'run_time'
  38     USER = 'user'
  39     MATRIX = 'matrix'
  40
  41     def __init__(self, xml=None):
  42         self.start = None
  43         self.stop = None
  44         self.version = None
  45         self.date = date.today()
  46         self.user = None
  47         self.matrix = None
  48
  49         if xml is not None:
  50             self.set_elements(xml)
  51
  52     def _get_time(self):
  53         return time.mktime(self.date.timetuple())
  54     time = property(_get_time, doc='return run time as seconds since epoch')
  55
  56     def dump(self):
  57         print "Starting cycle:", self.start
  58         print "Ending cycle:", self.stop
  59         print "Firecrest version:", self.version
  60         print "Run date:", self.date
  61         print "user:", self.user
  62
  63     def get_elements(self):
  64         attribs = {'version': str(Firecrest.XML_VERSION) }
  65         root = ElementTree.Element(Firecrest.FIRECREST, attrib=attribs)
  66         version = ElementTree.SubElement(root, Firecrest.SOFTWARE_VERSION)
  67         version.text = self.version
  68         start_cycle = ElementTree.SubElement(root, Firecrest.START)
  69         start_cycle.text = str(self.start)
  70         stop_cycle = ElementTree.SubElement(root, Firecrest.STOP)
  71         stop_cycle.text = str(self.stop)
  72         run_date = ElementTree.SubElement(root, Firecrest.DATE)
  73         run_date.text = str(self.time)
  74         user = ElementTree.SubElement(root, Firecrest.USER)
  75         user.text = self.user
  76         if self.matrix is not None:
  77             matrix = ElementTree.SubElement(root, Firecrest.MATRIX)
  78             matrix.text = self.matrix
  79         return root
  80
  81     def set_elements(self, tree):
  82         if tree.tag != Firecrest.FIRECREST:
  83             raise ValueError('Expected "Firecrest" SubElements')
  84         xml_version = int(tree.attrib.get('version', 0))
  85         if xml_version > Firecrest.XML_VERSION:
  86             LOGGER.warn('Firecrest XML tree is a higher version than this class')
  87         for element in list(tree):
  88             if element.tag == Firecrest.SOFTWARE_VERSION:
  89                 self.version = element.text
  90             elif element.tag == Firecrest.START:
  91                 self.start = int(element.text)
  92             elif element.tag == Firecrest.STOP:
  93                 self.stop = int(element.text)
  94             elif element.tag == Firecrest.DATE:
  95                 self.date = date.fromtimestamp(float(element.text))
  96             elif element.tag == Firecrest.USER:
  97                 self.user = element.text
  98             elif element.tag == Firecrest.MATRIX:
  99                 self.matrix = element.text
 100             else:
 101                 raise ValueError("Unrecognized tag: %s" % (element.tag,))
 102
 103 def firecrest(pathname):
 104     """
 105     Examine the directory at pathname and initalize a Firecrest object
 106     """
 107     f = Firecrest()
 108     f.pathname = pathname
 109
 110     # parse firecrest directory name
 111     path, name = os.path.split(pathname)
 112     groups = name.split('_')
 113     # grab the start/stop cycle information
 114     cycle = re.match("C([0-9]+)-([0-9]+)", groups[0])
 115     f.start = int(cycle.group(1))
 116     f.stop = int(cycle.group(2))
 117     # firecrest version
 118     version = re.search(VERSION_RE, groups[1])
 119     f.version = (version.group(1))
 120     # datetime
 121     t = time.strptime(groups[2], EUROPEAN_STRPTIME)
 122     f.date = date(*t[0:3])
 123     # username
 124     f.user = groups[3]
 125
 126     bustard_pattern = os.path.join(pathname, 'Bustard*')
 127     # should I parse this deeper than just stashing the
 128     # contents of the matrix file?
 129     matrix_pathname = os.path.join(pathname, 'Matrix', 's_matrix.txt')
 130     if os.path.exists(matrix_pathname):
 131         # this is for firecrest < 1.3.2
 132         f.matrix = open(matrix_pathname, 'r').read()
 133     elif glob(bustard_pattern) > 0:
 134         f.matrix = None
 135         # there are runs here. Bustard should save the matrix.
 136     else:
 137         return None
 138
 139     return f
 140
 141 def fromxml(tree):
 142     """
 143     Initialize a Firecrest object from an element tree node
 144     """
 145     f = Firecrest()
 146     f.set_elements(tree)
 147     return f