ParseCMR42Reports.py

#!/usr/bin/python3
# JJM -- parse cvi42 / cmr42 xml reports and extract relevant cardiac parameters 
# Plonk them into a csv file for later dealings (in R) 
# Supposedly biologist friendly. 
import sys
import csv
from bs4 import BeautifulSoup

multiplyFlag = True

if __name__ == "__main__":
    if ((len(sys.argv) ==1) or (len(sys.argv) >3)) :
        print("\nusage: input_file.xml output_file.csv")
        print("Alternately, just supply input_file.xml and the output is stdout\n")
        print("Starting gui...\n")

        from Tkinter import *
        import Tkinter, Tkconstants, tkFileDialog
        window = Tk()
        window.title("Cmr42 XML to CSV script")
        lbl = Label(window, text="CMR42 Converter script")
        lbl.grid(column=0,row=0)
        def clicked():
            xmlNames = tkFileDialog.askopenfilenames(title="Select XML files",filetypes = (("XML files","*.xml"),))
            csv_file_path = tkFileDialog.asksaveasfilename(title="CSV File Path",filetypes = (("CSV files","*.csv"),),defaultextension="csv")
            #Open CSV file, write header

            csv_file = csv.writer(open(csv_file_path, "wb+"),quoting=csv.QUOTE_MINIMAL)

            for ix in range(len(xmlNames)):
                #Write contents
                xfp = open(xmlNames[ix],'r')
                y = BeautifulSoup(xfp,features="xml")
                try:
                    value=y.RV.EDV["val"];
                    doRV = True;
                except:
                    doRV = False;

                if (ix == 0):
                    if (doRV == True):
                        csv_file.writerow(["Name",
                            "Custom.ID",
                            "History",
                            "EDV.ml",
                            "ESV.ml",
                            "SV.ml",
                            "EF.percent",
                            "CO.ml.min",
                            "HeartRate.min",
                            "MyoMassDiastole.g",
                            "MyoMassSystole.g",
                            "RVEDV.ml",
                            "RVESV.ml",
                            "RVSV.ml",
                            "RVEF.percent",
                            "RVCO.ml.min"])
                    elif (doRV == False):
                        csv_file.writerow(["Name",
                            "Custom.ID",
                            "History",
                            "EDV.ml",
                            "ESV.ml",
                            "SV.ml",
                            "EF.percent",
                            "CO.ml.min",
                            "HeartRate.min",
                            "MyoMassDiastole.g",
                            "MyoMassSystole.g"])



                if (doRV == True):
                    csv_file.writerow([y.Patient.Name["val"],
                            y.Patient.CustomPatientID["val"],
                            y.Study.History["val"],
                            y.LV.EDV["val"],
                            y.LV.ESV["val"],
                            y.LV.SV["val"],
                            y.LV.EF["val"],
                            str(5*float(y.LV.CO["val"])),
                            str(5*float(y.LV.HR["val"])),
                            y.LV.MyoMass_diast["val"],
                            y.LV.MyoMass_syst["val"],
                            y.RV.EDV["val"],
                            y.RV.ESV["val"],
                            y.RV.SV["val"],
                            y.RV.EF["val"],
                            y.RV.CO["val"]\
                            ])
                elif (doRV == False):

                    csv_file.writerow([y.Patient.Name["val"],
                            y.Patient.CustomPatientID["val"],
                            y.Study.History["val"],
                            y.LV.EDV["val"],
                            y.LV.ESV["val"],
                            y.LV.SV["val"],
                            y.LV.EF["val"],
                            str(5*float(y.LV.CO["val"])),
                            str(5*float(y.LV.HR["val"])),
                            y.LV.MyoMass_diast["val"],
                            y.LV.MyoMass_syst["val"],
                            ])
                else:
                    print("This should never have occurred Jack!");
            quit()
        btn=Button(window,text="Run",command=clicked)
        btn.grid(column=0,row=1)

        window.mainloop()
        quit()
    elif len(sys.argv) == 2:
        #Put output to stdout
        doStdOutFlag=1
    elif len(sys.argv) == 3:
        doStdOutFlag=0

    #Read arguments
    xml_file_path = sys.argv[1]
    xfp = open(xml_file_path,'r')
    y = BeautifulSoup(xfp,features="xml")

    if doStdOutFlag:
        csv_file = csv.writer(sys.stdout,quoting=csv.QUOTE_MINIMAL)
    else:
        csv_file_path = sys.argv[2]
        csv_file = csv.writer(open(csv_file_path, "wb+"),quoting=csv.QUOTE_MINIMAL)

    csv_file.writerow(["Name",
        "Custom.ID",
        "History",
        "EDV.ml",
        "ESV.ml",
        "SV.ml",
        "EF.percent",
        "CO.ml.min",
        "HeartRate.min",
        "MyoMassDiastole.g",
        "MyoMassSystole.g",
        "RVEDV.ml",
        "RVESV.ml",
        "RVSV.ml",
        "RVEF.percent",
        "RVCO.ml.min"])
    csv_file.writerow([y.Patient.Name["val"],
            y.Patient.CustomPatientID["val"],
            y.Study.History["val"],
            y.LV.EDV["val"],
            y.LV.ESV["val"],
            y.LV.SV["val"],
            y.LV.EF["val"],
            y.LV.CO["val"],
            y.LV.HR["val"],
            y.LV.MyoMass_diast["val"],
            y.LV.MyoMass_syst["val"],
            y.RV.EDV["val"],
            y.RV.ESV["val"],
            y.RV.SV["val"],
            y.RV.EF["val"],
            y.RV.CO["val"]\
            ])