def __init__(self,
fms="vs",
systemBase="http://livevista.caregraf.info/",
k3Base="http://datasets.caregraf.org/"):
self.fms = fms
self.systemBase = systemBase
self.k3Base = k3Base
self.rdfb = RDFBuilder(
["%s" % fms, "http://datasets.caregraf.org/%s/" % fms],
"http://datasets.caregraf.org/%s/" % fms,
extraNSInfos=[["fms", "http://datasets.caregraf.org/fms/"],
["xsd", "http://www.w3.org/2001/XMLSchema#"],
["dc", "http://purl.org/dc/elements/1.1/"]])
self.fileTypes = set()
self.outsideReferences = set()
self.codedValueReferences = set()
self.described = set() # recheck outside refs as build graph
def __init__(self, fms="vs", systemBase="http://livevista.caregraf.info/", k3Base="http://schemes.caregraf.info/"):
self.fms = fms
self.systemBase = systemBase
self.k3Base = k3Base
self.rdfb = RDFBuilder(["%s" % fms, "http://datasets.caregraf.org/%s/" % fms], "http://datasets.caregraf.org/%s/" % fms, extraNSInfos=[["fms", "http://datasets.caregraf.org/fms/"], ["xsd", "http://www.w3.org/2001/XMLSchema#"], ["dc", "http://purl.org/dc/elements/1.1/"]])
self.fileTypes = set()
self.outsideReferences = set()
self.described = set()
class DescribeRepliesToSGraph:
"""
This builds on DescribeReplyToRDF to produce a self contained (all URI's labeled) graph from one or more Describe Replies. These replies may be about a Patient or a Ward or System information.
TODO:
- FIX TO DescribeReply: needs to count Enum values in outside references
- add support for SAMEAS described things ie/ even though described, still sameas
- add a Dataset/Graph header
- SUPPORT exposure of outside refs ie/ to non same as types
- DO DUMMY/SIMPLE rdfb
"""
def __init__(self,
fms="vs",
systemBase="http://livevista.caregraf.info/",
k3Base="http://datasets.caregraf.org/"):
self.fms = fms
self.systemBase = systemBase
self.k3Base = k3Base
self.rdfb = RDFBuilder(
["%s" % fms, "http://datasets.caregraf.org/%s/" % fms],
"http://datasets.caregraf.org/%s/" % fms,
extraNSInfos=[["fms", "http://datasets.caregraf.org/fms/"],
["xsd", "http://www.w3.org/2001/XMLSchema#"],
["dc", "http://purl.org/dc/elements/1.1/"]])
self.fileTypes = set()
self.outsideReferences = set()
self.codedValueReferences = set()
self.described = set() # recheck outside refs as build graph
def processReply(self, describeReply):
drrdf = DescribeReplyToRDF(self.rdfb, self.fms, self.systemBase)
drrdf.processReply(describeReply)
self.fileTypes |= describeReply.fileTypes()
# Track outside references as go, removing if completely define a record
self.outsideReferences |= describeReply.outsideReferences()
self.codedValueReferences |= describeReply.codedValueReferences(
) # outside by defn
self.described |= set(record.asReference()
for record in describeReply.records())
def done(self):
self.outsideReferences -= self.described
sameAss = set()
sameAss2 = defaultdict(list)
for reference in self.outsideReferences:
if not self.rdfb.startNode(self.systemBase + reference.id, ""):
raise Exception(
"Unexpected to try to reference define the fully defined: "
+ str(reference))
self.rdfb.addLiteralAssertion("rdfs:label",
{"value": reference.label})
self.rdfb.addURIAssertion(
"rdf:type", {
"value": ("http://datasets.caregraf.org/%s/" % self.fms) +
reference.fileType
})
if reference.sameAs:
eSameAs = self.__expandSameAsURI(reference.sameAs)
self.rdfb.addURIAssertion("owl:sameAs", {"value": eSameAs})
sameAss.add((eSameAs, reference.sameAsLabel))
sameAss2[eSameAs].append(reference.sameAsLabel)
self.rdfb.endNode()
# These used to be tagged with rdf:type fms:OutsideConcept
for sameAs in sameAss:
if not self.rdfb.startNode(sameAs[0], ""):
continue # for now skip second definition of same sameas ex/ > 1 80 to an ICD. Should fix TODO intercept before this point
# raise Exception("Unexpected to try to reference define the defined: " + str(sameAs))
self.rdfb.addLiteralAssertion("rdfs:label", {"value": sameAs[1]})
self.rdfb.addURIAssertion(
"rdf:type",
{"value": "http://datasets.caregraf.org/fms/CommonConcept"})
self.rdfb.endNode()
for codedValueReference in self.codedValueReferences:
self.rdfb.startNode(
("http://datasets.caregraf.org/%s/" % self.fms) +
codedValueReference.id, "")
self.rdfb.addLiteralAssertion("rdfs:label", {
"type": "literal",
"value": codedValueReference.label
})
# Rem: in schema, Enum == instance of Class defined with owl:oneOf
# ex/ http://datasets.caregraf.org/chcss/2__02_E has
# http://datasets.caregraf.org/chcss/2__02_E-M with label MALE
self.rdfb.addURIAssertion(
"rdf:type", {
"type":
"uri",
"value":
self.fms + ":" + codedValueReference.id.split("_E-")[0] +
"_E"
})
self.rdfb.endNode()
for fileType in self.fileTypes:
self.rdfb.startNode(
("http://datasets.caregraf.org/%s/" % self.fms) + fileType[0],
"")
self.rdfb.addLiteralAssertion("rdfs:label", {
"type": "literal",
"value": fileType[1]
})
self.rdfb.addURIAssertion("rdf:type", {
"type": "uri",
"value": "owl:Class"
})
self.rdfb.endNode()
return self.rdfb.done()
SCHEMEMNMAP = {"ICD9": "ICD9CM", "PROVIDER": "HPTC"}
def __expandSameAsURI(self, sameAsURI):
# in mixed files (local only and sameas'ed terminologies), the local only entries are marked "LOCAL" in their same as fields
if sameAsURI == "LOCAL":
return ""
uriMatch = re.match(r'([^:]+):(.+)$', sameAsURI)
if not uriMatch:
return ""
# TODO: what if define. Should process this earlier
if uriMatch.group(1) == "LOCAL":
return self.systemBase + uriMatch.group(2)
# Tmp map til FMQL v1
schemeMN = self.SCHEMEMNMAP[uriMatch.group(1)] if uriMatch.group(
1) in self.SCHEMEMNMAP else uriMatch.group(1)
id = uriMatch.group(2)
if schemeMN == "ICD9CM":
id = re.sub(r'\_$', '',
re.sub(r'\.', '_',
id)) # code form to kGraf form, no trailing .'s
return self.k3Base + schemeMN.lower() + "/" + id
class DescribeRepliesToSGraph:
"""
This builds on DescribeReplyToRDF to produce a self contained (all URI's labeled) graph from one or more Describe Replies. These replies may be about a Patient or a Ward or System information.
TODO:
- FIX TO DescribeReply: needs to count Enum values in outside references
- add support for SAMEAS described things ie/ even though described, still sameas
- add a Dataset/Graph header
- SUPPORT exposure of outside refs ie/ to non same as types
- DO DUMMY/SIMPLE rdfb
"""
def __init__(self, fms="vs", systemBase="http://livevista.caregraf.info/", k3Base="http://datasets.caregraf.org/"):
self.fms = fms
self.systemBase = systemBase
self.k3Base = k3Base
self.rdfb = RDFBuilder(["%s" % fms, "http://datasets.caregraf.org/%s/" % fms], "http://datasets.caregraf.org/%s/" % fms, extraNSInfos=[["fms", "http://datasets.caregraf.org/fms/"], ["xsd", "http://www.w3.org/2001/XMLSchema#"], ["dc", "http://purl.org/dc/elements/1.1/"]])
self.fileTypes = set()
self.outsideReferences = set()
self.codedValueReferences = set()
self.described = set() # recheck outside refs as build graph
def processReply(self, describeReply):
drrdf = DescribeReplyToRDF(self.rdfb, self.fms, self.systemBase)
drrdf.processReply(describeReply)
self.fileTypes |= describeReply.fileTypes()
# Track outside references as go, removing if completely define a record
self.outsideReferences |= describeReply.outsideReferences()
self.codedValueReferences |= describeReply.codedValueReferences() # outside by defn
self.described |= set(record.asReference() for record in describeReply.records())
def done(self):
self.outsideReferences -= self.described
sameAss = set()
sameAss2 = defaultdict(list)
for reference in self.outsideReferences:
if not self.rdfb.startNode(self.systemBase + reference.id, ""):
raise Exception("Unexpected to try to reference define the fully defined: " + str(reference))
self.rdfb.addLiteralAssertion("rdfs:label", {"value": reference.label})
self.rdfb.addURIAssertion("rdf:type", {"value": ("http://datasets.caregraf.org/%s/" % self.fms) + reference.fileType})
if reference.sameAs:
eSameAs = self.__expandSameAsURI(reference.sameAs)
self.rdfb.addURIAssertion("owl:sameAs", {"value": eSameAs})
sameAss.add((eSameAs, reference.sameAsLabel))
sameAss2[eSameAs].append(reference.sameAsLabel)
self.rdfb.endNode()
# These used to be tagged with rdf:type fms:OutsideConcept
for sameAs in sameAss:
if not self.rdfb.startNode(sameAs[0], ""):
continue # for now skip second definition of same sameas ex/ > 1 80 to an ICD. Should fix TODO intercept before this point
# raise Exception("Unexpected to try to reference define the defined: " + str(sameAs))
self.rdfb.addLiteralAssertion("rdfs:label", {"value": sameAs[1]})
self.rdfb.addURIAssertion("rdf:type", {"value": "http://datasets.caregraf.org/fms/CommonConcept"})
self.rdfb.endNode()
for codedValueReference in self.codedValueReferences:
self.rdfb.startNode(("http://datasets.caregraf.org/%s/" % self.fms) + codedValueReference.id, "")
self.rdfb.addLiteralAssertion("rdfs:label", {"type": "literal", "value": codedValueReference.label})
# Rem: in schema, Enum == instance of Class defined with owl:oneOf
# ex/ http://datasets.caregraf.org/chcss/2__02_E has
# http://datasets.caregraf.org/chcss/2__02_E-M with label MALE
self.rdfb.addURIAssertion("rdf:type", {"type": "uri", "value": self.fms + ":" + codedValueReference.id.split("_E-")[0] + "_E"})
self.rdfb.endNode()
for fileType in self.fileTypes:
self.rdfb.startNode(("http://datasets.caregraf.org/%s/" % self.fms) + fileType[0], "")
self.rdfb.addLiteralAssertion("rdfs:label", {"type": "literal", "value": fileType[1]})
self.rdfb.addURIAssertion("rdf:type", {"type": "uri", "value": "owl:Class"})
self.rdfb.endNode()
return self.rdfb.done()
SCHEMEMNMAP = {"ICD9": "ICD9CM", "PROVIDER": "HPTC"}
def __expandSameAsURI(self, sameAsURI):
# in mixed files (local only and sameas'ed terminologies), the local only entries are marked "LOCAL" in their same as fields
if sameAsURI == "LOCAL":
return ""
uriMatch = re.match(r'([^:]+):(.+)$', sameAsURI)
if not uriMatch:
return ""
# TODO: what if define. Should process this earlier
if uriMatch.group(1) == "LOCAL":
return self.systemBase + uriMatch.group(2)
# Tmp map til FMQL v1
schemeMN = self.SCHEMEMNMAP[uriMatch.group(1)] if uriMatch.group(1) in self.SCHEMEMNMAP else uriMatch.group(1)
id = uriMatch.group(2)
if schemeMN == "ICD9CM":
id = re.sub(r'\_$', '', re.sub(r'\.', '_', id)) # code form to kGraf form, no trailing .'s
return self.k3Base + schemeMN.lower() + "/" + id