2 Provide access to information stored in the GERALD directory.
4 from datetime import datetime, date
9 from htsworkflow.pipelines.summary import Summary
10 from htsworkflow.pipelines.eland import eland, ELAND
12 from htsworkflow.pipelines.runfolder import \
17 from htsworkflow.util.ethelp import indent, flatten
21 Capture meaning out of the GERALD directory
25 RUN_PARAMETERS='RunParameters'
28 class LaneParameters(object):
30 Make it easy to access elements of LaneSpecificRunParameters from python
32 def __init__(self, gerald, lane_id):
34 self._lane_id = lane_id
36 def __get_attribute(self, xml_tag):
37 subtree = self._gerald.tree.find('LaneSpecificRunParameters')
38 container = subtree.find(xml_tag)
41 if len(container.getchildren()) > LANES_PER_FLOWCELL:
42 raise RuntimeError('GERALD config.xml file changed')
43 lanes = [x.tag.split('_')[1] for x in container.getchildren()]
44 index = lanes.index(self._lane_id)
45 element = container[index]
47 def _get_analysis(self):
48 return self.__get_attribute('ANALYSIS')
49 analysis = property(_get_analysis)
51 def _get_eland_genome(self):
52 genome = self.__get_attribute('ELAND_GENOME')
53 # default to the chipwide parameters if there isn't an
54 # entry in the lane specific paramaters
56 subtree = self._gerald.tree.find('ChipWideRunParameters')
57 container = subtree.find('ELAND_GENOME')
58 genome = container.text
60 eland_genome = property(_get_eland_genome)
62 def _get_read_length(self):
63 return self.__get_attribute('READ_LENGTH')
64 read_length = property(_get_read_length)
66 def _get_use_bases(self):
67 return self.__get_attribute('USE_BASES')
68 use_bases = property(_get_use_bases)
70 class LaneSpecificRunParameters(object):
72 Provide access to LaneSpecificRunParameters
74 def __init__(self, gerald):
78 def _initalize_lanes(self):
80 build dictionary of LaneParameters
83 tree = self._gerald.tree
84 analysis = tree.find('LaneSpecificRunParameters/ANALYSIS')
85 # according to the pipeline specs I think their fields
86 # are sampleName_laneID, with sampleName defaulting to s
87 # since laneIDs are constant lets just try using
89 for element in analysis:
90 sample, lane_id = element.tag.split('_')
91 self._lanes[lane_id] = Gerald.LaneParameters(self._gerald, lane_id)
93 def __getitem__(self, key):
94 if self._lane is None:
95 self._initalize_lanes()
96 return self._lanes[key]
98 if self._lane is None:
99 self._initalize_lanes()
100 return self._lanes.keys()
102 if self._lane is None:
103 self._initalize_lanes()
104 return self._lanes.values()
106 if self._lane is None:
107 self._initalize_lanes()
108 return self._lanes.items()
110 if self._lane is None:
111 self._initalize_lanes()
112 return len(self._lanes)
114 def __init__(self, xml=None):
118 # parse lane parameters out of the config.xml file
119 self.lanes = Gerald.LaneSpecificRunParameters(self)
122 self.eland_results = None
125 self.set_elements(xml)
128 if self.tree is None:
129 return datetime.today()
130 timestamp = self.tree.findtext('ChipWideRunParameters/TIME_STAMP')
131 epochstamp = time.mktime(time.strptime(timestamp, '%c'))
132 return datetime.fromtimestamp(epochstamp)
133 date = property(_get_date)
136 return time.mktime(self.date.timetuple())
137 time = property(_get_time, doc='return run time as seconds since epoch')
139 def _get_version(self):
140 if self.tree is None:
142 return self.tree.findtext('ChipWideRunParameters/SOFTWARE_VERSION')
143 version = property(_get_version)
147 Debugging function, report current object
149 print 'Gerald version:', self.version
150 print 'Gerald run date:', self.date
151 print 'Gerald config.xml:', self.tree
154 def get_elements(self):
155 if self.tree is None or self.summary is None:
158 gerald = ElementTree.Element(Gerald.GERALD,
159 {'version': unicode(Gerald.XML_VERSION)})
160 gerald.append(self.tree)
161 gerald.append(self.summary.get_elements())
162 if self.eland_results:
163 gerald.append(self.eland_results.get_elements())
166 def set_elements(self, tree):
167 if tree.tag != Gerald.GERALD:
168 raise ValueError('exptected GERALD')
169 xml_version = int(tree.attrib.get('version', 0))
170 if xml_version > Gerald.XML_VERSION:
171 logging.warn('XML tree is a higher version than this class')
172 for element in list(tree):
173 tag = element.tag.lower()
174 if tag == Gerald.RUN_PARAMETERS.lower():
176 elif tag == Gerald.SUMMARY.lower():
177 self.summary = Summary(xml=element)
178 elif tag == ELAND.ELAND.lower():
179 self.eland_results = ELAND(xml=element)
181 logging.warn("Unrecognized tag %s" % (element.tag,))
184 def gerald(pathname):
186 g.pathname = pathname
187 path, name = os.path.split(pathname)
188 logging.info("Parsing gerald config.xml")
189 config_pathname = os.path.join(pathname, 'config.xml')
190 g.tree = ElementTree.parse(config_pathname).getroot()
192 # parse Summary.htm file
193 logging.info("Parsing Summary.htm")
194 summary_pathname = os.path.join(pathname, 'Summary.htm')
195 g.summary = Summary(summary_pathname)
197 g.eland_results = eland(g.pathname, g)
200 if __name__ == "__main__":
203 g = gerald(sys.argv[1])
204 #ElementTree.dump(g.get_elements())