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