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 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_software(self):
  53         return "Firecrest"
  54     software = property(_get_software)
  55
  56     def _get_time(self):
  57         return time.mktime(self.date.timetuple())
  58     time = property(_get_time, doc='return run time as seconds since epoch')
  59
  60     def dump(self):
  61         print "Starting cycle:", self.start
  62         print "Ending cycle:", self.stop
  63         print "Firecrest version:", self.version
  64         print "Run date:", self.date
  65         print "user:", self.user
  66
  67     def get_elements(self):
  68         attribs = {'version': str(Firecrest.XML_VERSION) }
  69         root = ElementTree.Element(Firecrest.FIRECREST, attrib=attribs)
  70         version = ElementTree.SubElement(root, Firecrest.SOFTWARE_VERSION)
  71         version.text = self.version
  72         start_cycle = ElementTree.SubElement(root, Firecrest.START)
  73         start_cycle.text = str(self.start)
  74         stop_cycle = ElementTree.SubElement(root, Firecrest.STOP)
  75         stop_cycle.text = str(self.stop)
  76         run_date = ElementTree.SubElement(root, Firecrest.DATE)
  77         run_date.text = str(self.time)
  78         user = ElementTree.SubElement(root, Firecrest.USER)
  79         user.text = self.user
  80         if self.matrix is not None:
  81             matrix = ElementTree.SubElement(root, Firecrest.MATRIX)
  82             matrix.text = self.matrix
  83         return root
  84
  85     def set_elements(self, tree):
  86         if tree.tag != Firecrest.FIRECREST:
  87             raise ValueError('Expected "Firecrest" SubElements')
  88         xml_version = int(tree.attrib.get('version', 0))
  89         if xml_version > Firecrest.XML_VERSION:
  90             LOGGER.warn('Firecrest XML tree is a higher version than this class')
  91         for element in list(tree):
  92             if element.tag == Firecrest.SOFTWARE_VERSION:
  93                 self.version = element.text
  94             elif element.tag == Firecrest.START:
  95                 self.start = int(element.text)
  96             elif element.tag == Firecrest.STOP:
  97                 self.stop = int(element.text)
  98             elif element.tag == Firecrest.DATE:
  99                 self.date = date.fromtimestamp(float(element.text))
 100             elif element.tag == Firecrest.USER:
 101                 self.user = element.text
 102             elif element.tag == Firecrest.MATRIX:
 103                 self.matrix = element.text
 104             else:
 105                 raise ValueError("Unrecognized tag: %s" % (element.tag,))
 106
 107 def firecrest(pathname):
 108     """
 109     Examine the directory at pathname and initalize a Firecrest object
 110     """
 111     f = Firecrest()
 112     f.pathname = pathname
 113
 114     # parse firecrest directory name
 115     path, name = os.path.split(pathname)
 116     groups = name.split('_')
 117     # grab the start/stop cycle information
 118     cycle = re.match("C([0-9]+)-([0-9]+)", groups[0])
 119     f.start = int(cycle.group(1))
 120     f.stop = int(cycle.group(2))
 121     # firecrest version
 122     version = re.search(VERSION_RE, groups[1])
 123     f.version = (version.group(1))
 124     # datetime
 125     t = time.strptime(groups[2], EUROPEAN_STRPTIME)
 126     f.date = date(*t[0:3])
 127     # username
 128     f.user = groups[3]
 129
 130     bustard_pattern = os.path.join(pathname, 'Bustard*')
 131     # should I parse this deeper than just stashing the
 132     # contents of the matrix file?
 133     matrix_pathname = os.path.join(pathname, 'Matrix', 's_matrix.txt')
 134     if os.path.exists(matrix_pathname):
 135         # this is for firecrest < 1.3.2
 136         f.matrix = open(matrix_pathname, 'r').read()
 137     elif glob(bustard_pattern) > 0:
 138         f.matrix = None
 139         # there are runs here. Bustard should save the matrix.
 140     else:
 141         return None
 142
 143     return f
 144
 145 def fromxml(tree):
 146     """
 147     Initialize a Firecrest object from an element tree node
 148     """
 149     f = Firecrest()
 150     f.set_elements(tree)
 151     return f