def getXML(self, doc, elementName):
""" return an xml node that contains label information """
node = doc.createElement(elementName)
node.setAttribute('type',self.type)
for attrib, value in self.attributes.items():
node.appendChild(xmlutil.createNodeWithTextChild(doc, attrib, value))
return node
def getXML(self, doc):
""" return an xml element containing all token information """
node = doc.createElement('token')
node.setAttribute('id', str(self.index))
node.setAttribute('text', self.text)
if len(self.lemma) > 0 and self.lemma != self.text:
node.setAttribute('lemma', self.lemma)
if len(self.pos) > 0:
node.setAttribute('pos', self.pos)
# add governors
for gov in self.governors:
node.appendChild(gov.getXML(doc, 'gov'))
# add dependents
for dep in self.dependents:
node.appendChild(dep.getXML(doc, 'dep'))
# add semantic tags
for tag in self.semanticTags:
sNode = xmlutil.createNodeWithTextChild(doc, 'semantic', tag)
node.appendChild(sNode)
# add true (original) annotations
for annotation in self.annotations:
node.appendChild(annotation.getXML(doc, 'annotation'))
# add detected labels
for label in self.labels:
node.appendChild(label.getXML(doc, 'label'))
for uc in self.umlsConcepts:
node.appendChild(uc.getXML(doc))
return node
def getXML(self, doc, elementName):
""" return an xml node that contains label information """
node = doc.createElement(elementName)
node.setAttribute('type', self.type)
for attrib, value in self.attributes.items():
node.appendChild(
xmlutil.createNodeWithTextChild(doc, attrib, value))
return node
def getXML(self, doc):
node = doc.createElement("umls")
node.setAttribute("id", self.id)
if len(self.snomed) > 0:
node.setAttribute("snomed", self.snomed)
node.setAttribute("score", str(self.score))
if self.isNegated:
node.setAttribute("negated", "true")
else:
node.setAttribute("negated", "false")
for type in self.types:
node.appendChild(xmlutil.createNodeWithTextChild(doc, "type", type))
if self.inSnomed:
node.appendChild(xmlutil.createNodeWithTextChild(doc, "source", "SNOMEDCT"))
if self.inRxnorm:
node.appendChild(xmlutil.createNodeWithTextChild(doc, "source", "RXNORM"))
return node
def getXML(self, doc):
""" return an xml element containing information for a clinical report element"""
node = doc.createElement('report')
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'id', self.id))
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'gender', self.gender))
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'minAge', self.minAge))
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'maxAge', self.maxAge))
if len(self.locations) > 0:
lcNode = doc.createElement('location_countries')
node.appendChild(lcNode)
for country in self.locations:
lcNode.appendChild(xmlutil.createNodeWithTextChild(doc, 'country', country))
for condition in self.conditions:
node.appendChild(condition.getXML(doc))
if len(self.eligibilityCriteria) > 0:
ecNode = doc.createElement('eligibility')
node.appendChild(ecNode)
for ec in self.eligibilityCriteria:
ecNode.appendChild(ec.getXML(doc))
if len(self.inclusionCriteria) > 0:
icNode = doc.createElement('inclusion')
node.appendChild(icNode)
for ic in self.inclusionCriteria:
icNode.appendChild(ic.getXML(doc))
if len(self.exclusionCriteria) > 0:
ecNode = doc.createElement('exclusion')
node.appendChild(ecNode)
for ec in self.exclusionCriteria:
ecNode.appendChild(ec.getXML(doc))
for intervention in self.interventions:
node.appendChild(intervention.getXML(doc))
for outcome in self.outcomes:
node.appendChild(outcome.getXML(doc))
return node
def getXML(self, doc):
""" return xml node contain all information relevant to this age value"""
node = xmlutil.createNodeWithTextChild(doc, 'AgeValue', str(self.value))
if self.type != None:
node.setAttribute('type',self.type)
if self.units != None:
node.setAttribute('units', self.units)
if self.bounds != None:
node.setAttribute('bounds', str(self.bounds))
node.setAttribute('source', self.source)
return node
def getXML(self, doc):
""" return xml node contain all information relevant to this age value"""
node = xmlutil.createNodeWithTextChild(doc, 'AgeValue',
str(self.value))
if self.type != None:
node.setAttribute('type', self.type)
if self.units != None:
node.setAttribute('units', self.units)
if self.bounds != None:
node.setAttribute('bounds', str(self.bounds))
node.setAttribute('source', self.source)
return node
def getXML(self, doc):
"""
Create an XML element with publication information
"""
node = doc.createElement('PublicationInformation')
if self._journalNode is not None:
node.appendChild(self._journalNode)
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'Country', self._country))
if self._authorListNode is not None:
node.appendChild(self._authorListNode)
if self._publicationTypeListNode is not None:
node.appendChild(self._publicationTypeListNode)
xmlutil.normalizeXMLTree(node)
return node
def getGroupXML(self, doc, omTemplate, idPrefix):
""" return XML node containing outcome results for group.
omTemplate = outcome measurement template containing outcome info for
the group"""
groupNode = self.createGroupNode(doc, omTemplate.getGroup(), idPrefix)
bad = omTemplate.getOutcomes()
if bad >= 0:
badNode = xmlutil.createNodeWithTextChild(doc, 'Bad', str(bad))
groupNode.appendChild(badNode)
size = omTemplate.getGroupSize()
if size > 0:
gsNode = xmlutil.createNodeWithTextChild(doc, 'GroupSize', str(size))
groupNode.appendChild(gsNode)
er = omTemplate.eventRateString()
if len(er) > 0:
er = omTemplate.eventRateString()
erNode = xmlutil.createNodeWithTextChild(doc, 'EventRate', er)
groupNode.appendChild(erNode)
scoreNode = xmlutil.createNodeWithTextChild(doc, 'Score',
str(omTemplate.getConfidence()))
groupNode.appendChild(scoreNode)
return groupNode
def getXML(self, doc, idPrefix):
""" return an xml node that contains information about the outcomes
measured in the study. """
locationList = self.getLocations()
if len(locationList) == 0:
return None
locationListNode = doc.createElement('Locations')
lCount = 0
for location in locationList:
locationNode = xmlutil.createNodeWithTextChild(doc, 'Location', location)
locationNode.setAttribute('source', self.source)
locationNode.setAttribute('id', idPrefix+'loc'+str(lCount))
lCount += 1
# setUMLSAttribute(nameNode, oTemplate)
locationListNode.appendChild(locationNode)
return locationListNode
def getXML(self, doc, idPrefix):
""" return an xml node that contains information about the outcomes
measured in the study. """
locationList = self.getLocations()
if len(locationList) == 0:
return None
locationListNode = doc.createElement('Locations')
lCount = 0
for location in locationList:
locationNode = xmlutil.createNodeWithTextChild(
doc, 'Location', location)
locationNode.setAttribute('source', self.source)
locationNode.setAttribute('id', idPrefix + 'loc' + str(lCount))
lCount += 1
# setUMLSAttribute(nameNode, oTemplate)
locationListNode.appendChild(locationNode)
return locationListNode
def getXML(self, doc):
""" return an xml element containing sentence information """
node = doc.createElement('sentence')
if len(self.section) > 0:
node.setAttribute('section', self.section)
if len(self.nlmCategory) > 0:
node.setAttribute('nlmCategory', self.nlmCategory)
# if self.templates.noTemplates() == False:
# node.appendChild(self.templates.getXML(doc))
for token in self.tokens:
node.appendChild(token.getXML(doc))
for umlsChunk in self.umlsChunks:
node.appendChild(umlsChunk.getXML(doc))
if self.parseTree != None:
s = self.parseTree.treebankString()
# s = self.parseString
node.appendChild(xmlutil.createNodeWithTextChild(doc, 'parse', s))
return node
def getOutcomeXML(self, doc, oTemplate, idPrefix):
""" write given outcome template to XML file stream """
outcomeNode = doc.createElement('Outcome')
if oTemplate.isPrimary():
oType = 'primary'
else:
oType = 'unknown'
poNode = xmlutil.createNodeWithTextChild(doc, 'Type', oType)
id = idPrefix+oTemplate.id+'oType'
poNode.setAttribute('Id', id)
oTemplate.primaryOutcomeEvaluation.id = id
outcomeNode.appendChild(poNode)
id = idPrefix+oTemplate.id
outcomeNode.setAttribute('Id', id)
oTemplate.evaluation.id = id
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', oTemplate.getCanonicalName())
# setUMLSAttribute(nameNode, oTemplate)
outcomeNode.appendChild(nameNode)
# write Cost values
cvCount = 0
for cvTemplate in oTemplate.costValues:
cvNode = doc.createElement('CostValue')
outcomeNode.appendChild(cvNode)
id = idPrefix + oTemplate.id + 'cv' + str(cvCount)
cvNode.setAttribute('id', id)
cvCount += 1
if cvTemplate.group.name is not None:
cvNode.appendChild(self.createGroupNode(doc, cvTemplate.group, idPrefix))
valueNode = xmlutil.createNodeWithTextChild(doc, 'Value', cvTemplate.token.text)
valueNode.setAttribute('units', cvTemplate.token.getUnits())
cvNode.appendChild(valueNode)
# write ARR/NNT
epCount = 0
for ssTemplate in oTemplate.summaryStats:
epNode = doc.createElement('Endpoint')
id = idPrefix+oTemplate.id+'ep'+str(epCount)
ssTemplate.evaluation.id = id
epNode.setAttribute('id', id)
epCount += 1
outcomeNode.appendChild(epNode)
epNode.appendChild(self.getGroupXML(doc, ssTemplate.lessEffective, idPrefix))
epNode.appendChild(self.getGroupXML(doc, ssTemplate.moreEffective, idPrefix))
if ssTemplate.time != None:
tNode = xmlutil.createNodeWithTextChild(doc, 'time', ssTemplate.time.name)
if len(ssTemplate.time.units) > 0:
tNode.setAttribute('units', ssTemplate.time.units)
if ssTemplate.time.value > 0:
tNode.setAttribute('value', str(ssTemplate.time.value))
epNode.appendChild(tNode)
ssNode = doc.createElement('SummaryStatistics')
epNode.appendChild(ssNode)
sNode = doc.createElement('Statistic')
sNode.setAttribute('Worse', idPrefix+ssTemplate.lessEffective.getGroup().id)
sNode.setAttribute('Better', idPrefix+ssTemplate.moreEffective.getGroup().id)
ssNode.appendChild(sNode)
arNode = doc.createElement('AbsoluteRisk')
sNode.appendChild(arNode)
arNode.setAttribute('Type', ssTemplate.riskType())
vNode = xmlutil.createNodeWithTextChild(doc, 'Value', ssTemplate.arrString())
arNode.appendChild(vNode)
if ssTemplate.hasConfidenceIntervals():
iNode = doc.createElement('Interval')
iNode.setAttribute('lower', ssTemplate.arrLowerBound())
iNode.setAttribute('upper', ssTemplate.arrUpperBound())
arNode.appendChild(iNode)
if ssTemplate.infiniteNumberNeeded() == False:
nnNode = doc.createElement('NumberNeeded')
sNode.appendChild(nnNode)
nnNode.setAttribute('Type', ssTemplate.numberNeededType())
vNode = xmlutil.createNodeWithTextChild(doc, 'Value', ssTemplate.nntString())
nnNode.appendChild(vNode)
if ssTemplate.hasConfidenceIntervals():
iNode = doc.createElement('Interval')
iNode.setAttribute('lower', ssTemplate.nntLowerBound())
iNode.setAttribute('upper', ssTemplate.nntUpperBound())
nnNode.appendChild(iNode)
# for omTemplate in oTemplate.unusedNumbers:
# epNode = doc.createElement('Endpoint')
# epNode.setAttribute('id', idPrefix+oTemplate.id+'ep'+str(epCount))
# epCount += 1
# outcomeNode.appendChild(epNode)
# if omTemplate.getGroup() != None:
# if omTemplate.getTime() != None:
# tNode = xmlutil.createNodeWithTextChild(doc, 'time', omTemplate.getTime().name)
# if len(omTemplate.getTime().units) > 0:
# tNode.setAttribute('units', omTemplate.getTime().units)
# if omTemplate.getTime().value > 0:
# tNode.setAttribute('value', str(omTemplate.getTime().value))
#
# epNode.appendChild(tNode)
# outcomeNode.appendChild(epNode)
# epNode.appendChild(self.getGroupXML(doc, omTemplate, idPrefix))
return outcomeNode
def getXML(self, doc, idPrefix):
node = xmlutil.createNodeWithTextChild(doc, 'Gender', self.value)
node.setAttribute('source', self.source)
node.setAttribute('id', idPrefix + 'gen0')
return node
def writeXML(self, path, version):
""" Write summary to XML file in the destination directory specified
by path. """
idPrefix = self.abstract.id + 'v' + version
# build XML summary
doc = xml.dom.minidom.Document()
articleNode = doc.createElement('Study')
if self.randomized:
articleNode.setAttribute('RandomizationPresent', 'true')
else:
articleNode.setAttribute('RandomizationPresent', 'false')
articleNode.setAttribute('Version', version)
gmTime = time.gmtime()
cNode = doc.createElement('Created')
articleNode.appendChild(cNode)
cNode.appendChild(xmlutil.createNodeWithTextChild(doc, 'Month', str(gmTime.tm_mon)))
cNode.appendChild(xmlutil.createNodeWithTextChild(doc, 'Day', str(gmTime.tm_mday)))
cNode.appendChild(xmlutil.createNodeWithTextChild(doc, 'Year', str(gmTime.tm_year)))
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', self.abstract.id)
articleNode.appendChild(nameNode)
titleString = ''
for s in self.abstract.titleSentences:
titleString += s.toString()
titleNode = xmlutil.createNodeWithTextChild(doc, 'Title', titleString)
articleNode.appendChild(titleNode)
link = 'http://www.ncbi.nlm.nih.gov/pubmed/' + self.abstract.id
linkNode = xmlutil.createNodeWithTextChild(doc, 'AbstractLink', link)
articleNode.appendChild(linkNode)
if self.abstract.report is not None and self.abstract.report.id[0:3] == 'NCT':
link = 'http://clinicaltrials.gov/ct2/show/study/' + self.abstract.report.id
linkNode = xmlutil.createNodeWithTextChild(doc, 'TrialRegistryLink', link)
articleNode.appendChild(linkNode)
node = self.locationList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
if self.useTrialReports and self.abstract.report is not None \
and len(self.abstract.report.conditions) > 0:
cListNode = doc.createElement('ConditionsOfInterest')
articleNode.appendChild(cListNode)
for condition in self.abstract.report.conditions:
cNode = doc.createElement('Condition')
cNode.appendChild(xmlutil.createNodeWithTextChild(doc, 'Name',
condition.sentences[0].toString()))
cNode.setAttribute('source', 'trial_registry')
cListNode.appendChild(cNode)
node = self.subjectList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
node = self.outcomeList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
if self.abstract.meshHeadingList is not None:
articleNode.appendChild(self.abstract.meshHeadingList.getXML(doc))
htmlString = StringIO.StringIO()
self.writeHTML(htmlString, showError=False)
htmlNode = xmlutil.createNodeWithTextChild(doc, 'HTMLData', htmlString.getvalue())
articleNode.appendChild(htmlNode)
htmlString.close()
# write summary to XML file
filename = path + self.abstract.id + '.summary.xml'
out = open(filename, 'w')
out.write('<?xml version="1.0" encoding="utf-8"?>\n')
# out.write('<?xml-stylesheet href="abstract.xsl" type="text/xsl"?>\n')
xmlutil.writexml(articleNode, out)
out.close()
def writeXML(self, path, version):
""" Write summary to XML file in the destination directory specified
by path. """
idPrefix = self.abstract.id + 'v' + version
# build XML summary
doc = xml.dom.minidom.Document()
articleNode = doc.createElement('Study')
if self.randomized:
articleNode.setAttribute('RandomizationPresent', 'true')
else:
articleNode.setAttribute('RandomizationPresent', 'false')
articleNode.setAttribute('Version', version)
gmTime = time.gmtime()
cNode = doc.createElement('Created')
articleNode.appendChild(cNode)
cNode.appendChild(
xmlutil.createNodeWithTextChild(doc, 'Month', str(gmTime.tm_mon)))
cNode.appendChild(
xmlutil.createNodeWithTextChild(doc, 'Day', str(gmTime.tm_mday)))
cNode.appendChild(
xmlutil.createNodeWithTextChild(doc, 'Year', str(gmTime.tm_year)))
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name',
self.abstract.id)
articleNode.appendChild(nameNode)
titleString = ''
for s in self.abstract.titleSentences:
titleString += s.toString()
titleNode = xmlutil.createNodeWithTextChild(doc, 'Title', titleString)
articleNode.appendChild(titleNode)
link = 'http://www.ncbi.nlm.nih.gov/pubmed/' + self.abstract.id
linkNode = xmlutil.createNodeWithTextChild(doc, 'AbstractLink', link)
articleNode.appendChild(linkNode)
if self.abstract.report is not None and self.abstract.report.id[
0:3] == 'NCT':
link = 'http://clinicaltrials.gov/ct2/show/study/' + self.abstract.report.id
linkNode = xmlutil.createNodeWithTextChild(doc,
'TrialRegistryLink',
link)
articleNode.appendChild(linkNode)
node = self.locationList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
if self.useTrialReports and self.abstract.report is not None \
and len(self.abstract.report.conditions) > 0:
cListNode = doc.createElement('ConditionsOfInterest')
articleNode.appendChild(cListNode)
for condition in self.abstract.report.conditions:
cNode = doc.createElement('Condition')
cNode.appendChild(
xmlutil.createNodeWithTextChild(
doc, 'Name', condition.sentences[0].toString()))
cNode.setAttribute('source', 'trial_registry')
cListNode.appendChild(cNode)
node = self.subjectList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
node = self.outcomeList.getXML(doc, idPrefix)
if node is not None:
articleNode.appendChild(node)
if self.abstract.meshHeadingList is not None:
articleNode.appendChild(self.abstract.meshHeadingList.getXML(doc))
htmlString = StringIO.StringIO()
self.writeHTML(htmlString, showError=False)
htmlNode = xmlutil.createNodeWithTextChild(doc, 'HTMLData',
htmlString.getvalue())
articleNode.appendChild(htmlNode)
htmlString.close()
# write summary to XML file
filename = path + self.abstract.id + '.summary.xml'
out = open(filename, 'w')
out.write('<?xml version="1.0" encoding="utf-8"?>\n')
# out.write('<?xml-stylesheet href="abstract.xsl" type="text/xsl"?>\n')
xmlutil.writexml(articleNode, out)
out.close()
def getXML(self, doc, idPrefix):
""" return an xml node that contains information about subjects
in the study. """
# if len(self.groupTemplates) == 0 and len(self.populationTemplates) == 0 \
# and len(self.ageTemplates) == 0 and len(self.conditionTemplates) == 0:
# return None
subjectListNode = doc.createElement('Subjects')
eligibilityNode = doc.createElement('Eligibility')
subjectListNode.appendChild(eligibilityNode)
# for pTemplate in self.populationTemplates:
# popNode = doc.createElement('Population')
# # setUMLSAttribute(popNode, pTemplate)
# nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', pTemplate.name)
# popNode.appendChild(nameNode)
# eligibilityNode.appendChild(popNode)
eligibilityNode.appendChild(self.gender.getXML(doc, idPrefix))
if len(self.ageInfo.ageValues) > 0:
eligibilityNode.appendChild(self.ageInfo.getXML(doc, idPrefix))
for cTemplate in self.conditionTemplates:
cNode = doc.createElement('Criteria')
cNode.setAttribute('source', 'abstract')
id = idPrefix+cTemplate.id
cTemplate.evaluation.id = id
cNode.setAttribute('Id', id)
# setUMLSAttribute(cNode, cTemplate)
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', cTemplate.name)
cNode.appendChild(nameNode)
type = 'unknown'
firstToken = cTemplate.mention.tokens[0]
if firstToken.text == 'with' or firstToken.lemma == 'have':
type = 'inclusion'
elif firstToken.text == 'without':
type = 'exclusion'
cNode.setAttribute('type', type)
eligibilityNode.appendChild(cNode)
if self.useTrialReports and self.abstract.report != None:
icCount = 0
ecCount = 0
for criteria in self.abstract.report.inclusionCriteria:
cNode = doc.createElement('Criteria')
cNode.setAttribute('source', 'trial_registry')
cNode.setAttribute('Id', idPrefix+'ic'+str(icCount))
icCount += 1
text = criteria.sentences[0].toString()
for i in range(1, len(criteria.sentences)):
text = text + ' ' + criteria.sentences[i].toString()
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', text)
cNode.appendChild(nameNode)
cNode.setAttribute('type', 'inclusion')
eligibilityNode.appendChild(cNode)
for criteria in self.abstract.report.exclusionCriteria:
cNode = doc.createElement('Criteria')
cNode.setAttribute('source', 'trial_registry')
cNode.setAttribute('Id', idPrefix+'ec'+str(ecCount))
ecCount += 1
text = criteria.sentences[0].toString()
for i in range(1, len(criteria.sentences)):
text = text + ' ' + criteria.sentences[i].toString()
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', text)
cNode.appendChild(nameNode)
cNode.setAttribute('type', 'exclusion')
eligibilityNode.appendChild(cNode)
for gTemplate in self.groupTemplates:
groupNode = doc.createElement('Group')
id = idPrefix+gTemplate.id
gTemplate.evaluation.id = idPrefix+gTemplate.id
groupNode.setAttribute('Id', id)
groupNode.setAttribute('Role', gTemplate.role)
gSize = gTemplate.getSize(maxSize=True)
if gSize > 0:
sNode = xmlutil.createNodeWithTextChild(doc, 'Size', str(gSize))
id = idPrefix+gTemplate.id+'size'
gTemplate.groupSizeEvaluation.id = id
sNode.setAttribute('Id', id)
groupNode.appendChild(sNode)
# setUMLSAttribute(groupNode, gTemplate)
nameNode = xmlutil.createNodeWithTextChild(doc, 'Name', gTemplate.name)
groupNode.appendChild(nameNode)
subjectListNode.appendChild(groupNode)
return subjectListNode
def getXML(self, doc, idPrefix):
node = xmlutil.createNodeWithTextChild(doc, 'Gender', self.value)
node.setAttribute('source', self.source)
node.setAttribute('id', idPrefix+'gen0')
return node
