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 from __future__ import print_function
  13
  14 from datetime import date
  15 from glob import glob
  16 import logging
  17 import os
  18 import re
  19 import time
  20
  21 from htsworkflow.pipelines import \
  22    ElementTree, \
  23    VERSION_RE, \
  24    EUROPEAN_STRPTIME
  25
  26 LOGGER = logging.getLogger(__name__)
  27
  28 class Firecrest(object):
  29     """Gather information about older firecrest runs
  30     """
  31     XML_VERSION=1
  32
  33     # xml tag names
  34     FIRECREST = 'Firecrest'
  35     SOFTWARE_VERSION = 'version'
  36     START = 'FirstCycle'
  37     STOP = 'LastCycle'
  38     DATE = 'run_time'
  39     USER = 'user'
  40     MATRIX = 'matrix'
  41
  42     def __init__(self, xml=None):
  43         """Initialize a Firecrest object
  44
  45         consider using factory :function:firecrest
  46
  47         :param xml: xml serialzation element to initialze from [optional]
  48         """
  49         self.start = None
  50         self.stop = None
  51         self.version = None
  52         self.date = date.today()
  53         self.user = None
  54         self.matrix = None
  55
  56         if xml is not None:
  57             self.set_elements(xml)
  58
  59     def _get_software(self):
  60         return "Firecrest"
  61     software = property(_get_software)
  62
  63     def _get_time(self):
  64         return time.mktime(self.date.timetuple())
  65     time = property(_get_time, doc='return run time as seconds since epoch')
  66
  67     def dump(self):
  68         """Report debugginf information
  69         """
  70         print("Starting cycle:", self.start)
  71         print("Ending cycle:", self.stop)
  72         print("Firecrest version:", self.version)
  73         print("Run date:", self.date)
  74         print("user:", self.user)
  75
  76     def get_elements(self):
  77         """Return XML serialization structure.
  78         """
  79         attribs = {'version': str(Firecrest.XML_VERSION) }
  80         root = ElementTree.Element(Firecrest.FIRECREST, attrib=attribs)
  81         version = ElementTree.SubElement(root, Firecrest.SOFTWARE_VERSION)
  82         version.text = self.version
  83         start_cycle = ElementTree.SubElement(root, Firecrest.START)
  84         start_cycle.text = str(self.start)
  85         stop_cycle = ElementTree.SubElement(root, Firecrest.STOP)
  86         stop_cycle.text = str(self.stop)
  87         run_date = ElementTree.SubElement(root, Firecrest.DATE)
  88         run_date.text = str(self.time)
  89         user = ElementTree.SubElement(root, Firecrest.USER)
  90         user.text = self.user
  91         if self.matrix is not None:
  92             matrix = ElementTree.SubElement(root, Firecrest.MATRIX)
  93             matrix.text = self.matrix
  94         return root
  95
  96     def set_elements(self, tree):
  97         if tree.tag != Firecrest.FIRECREST:
  98             raise ValueError('Expected "Firecrest" SubElements')
  99         xml_version = int(tree.attrib.get('version', 0))
 100         if xml_version > Firecrest.XML_VERSION:
 101             LOGGER.warn('Firecrest XML tree is a higher version than this class')
 102         for element in list(tree):
 103             if element.tag == Firecrest.SOFTWARE_VERSION:
 104                 self.version = element.text
 105             elif element.tag == Firecrest.START:
 106                 self.start = int(element.text)
 107             elif element.tag == Firecrest.STOP:
 108                 self.stop = int(element.text)
 109             elif element.tag == Firecrest.DATE:
 110                 self.date = date.fromtimestamp(float(element.text))
 111             elif element.tag == Firecrest.USER:
 112                 self.user = element.text
 113             elif element.tag == Firecrest.MATRIX:
 114                 self.matrix = element.text
 115             else:
 116                 raise ValueError("Unrecognized tag: %s" % (element.tag,))
 117
 118 def firecrest(pathname):
 119     """
 120     Examine the directory at pathname and initalize a Firecrest object
 121     """
 122     f = Firecrest()
 123     f.pathname = pathname
 124
 125     # parse firecrest directory name
 126     path, name = os.path.split(pathname)
 127     groups = name.split('_')
 128     # grab the start/stop cycle information
 129     cycle = re.match("C([0-9]+)-([0-9]+)", groups[0])
 130     f.start = int(cycle.group(1))
 131     f.stop = int(cycle.group(2))
 132     # firecrest version
 133     version = re.search(VERSION_RE, groups[1])
 134     f.version = (version.group(1))
 135     # datetime
 136     t = time.strptime(groups[2], EUROPEAN_STRPTIME)
 137     f.date = date(*t[0:3])
 138     # username
 139     f.user = groups[3]
 140
 141     bustard_pattern = os.path.join(pathname, 'Bustard*')
 142     # should I parse this deeper than just stashing the
 143     # contents of the matrix file?
 144     matrix_pathname = os.path.join(pathname, 'Matrix', 's_matrix.txt')
 145     if os.path.exists(matrix_pathname):
 146         # this is for firecrest < 1.3.2
 147         f.matrix = open(matrix_pathname, 'r').read()
 148     elif glob(bustard_pattern) > 0:
 149         f.matrix = None
 150         # there are runs here. Bustard should save the matrix.
 151     else:
 152         return None
 153
 154     return f
 155
 156 def fromxml(tree):
 157     """
 158     Initialize a Firecrest object from an element tree node
 159     """
 160     f = Firecrest()
 161     f.set_elements(tree)
 162     return f