diff --git a/UMLS_Extractor/extractor.py b/UMLS_Extractor/extractor.py new file mode 100644 index 0000000000000000000000000000000000000000..cd00b4889eaa636afbbc84478ff9d81f45446347 --- /dev/null +++ b/UMLS_Extractor/extractor.py @@ -0,0 +1,47 @@ +# -*- coding: utf-8 -* +""" +Created on Thu Jan 28 09:38:20 2021 +""" +import jkesExtractor +import sys, os, json + +#FunciĆ³n que dado un path, un listado de datos y el modo de apertura del archivo, +#escribe los datos en el archivo correspondiente +def write(pathOutput,dataList,mode): + + #Output in the second route + if os.path.exists(pathOutput): + with open(pathOutput, mode) as outputfile: + for i in range(0,len(dataList)): + strOutput = "" + for j in range(0,len(dataList[i])): + strOutput += str(dataList[i][j])+"\t" + outputfile.write(strOutput.strip()+"\n") + + print("Completed") + + else: + print("Output file doesn't exist") + + +#Funcion principal: recibe el path del archivo de anotaciones y del archivo de salida +def main(): + #First arg is the route of .json BIO + + inputRoute = sys.argv[1] + outputRoute = sys.argv[2] + if os.path.exists(inputRoute): + + #Extraction of CUIS + print("Checking JKES CUIS") + listConceptsJkes, listConceptsUmls = jkesExtractor.jkes_concept_extractor(inputRoute) + write(outputRoute,listConceptsJkes,"w") + write(outputRoute,listConceptsUmls,"a") + + else: + print("Input file doesn't exist") + + +if __name__ == "__main__": + main() +