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