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