read_xml_out_csv.py

# prerequisites:
#   https://github.com/stchris/untangle/blob/master/examples.py
#   C:\temp>c:\Python27\Scripts\pip.exe install untangle
# run: C:\temp\read_xml_out_csv>c:\Python27\python.exe read_xml_out_csv.py
import untangle
import os
import time

def now():
    return time.time()*1000.0

model = {} # dictionary of dictionaries with filename -> servlet -> parameter name -> parameter value
params = [] # all params that are found
last_name = "" # last found name, so that we can add the value to the parent dictionary
start = now()

files = filter(lambda file: file.endswith(".xml"), os.listdir(".")) # all xml files in the current dir

# parse files and collect all data
for file in files:
    print "parsing file %s" % file
    model[file] = {}
    xml = untangle.parse(file)
    for servlet in xml.config.servlet:
        servlet_name = servlet.name.cdata
        model[file][servlet_name] = {}
        for child in servlet.children:
            if child._name == "param_name":
                last_name = child.cdata
                if child.cdata not in params:
                    params.append(last_name)
            elif child._name == "param_value":
                model[file][servlet_name][last_name] = child.cdata
params.sort()
                
# build header
lines = "filename,servlet name,"
for param in params:
    lines += param + ","
lines += "\n"

# build rows
for file in files:
    sorted_servlets = []
    for servlet in model[file]:
        sorted_servlets.append(servlet)
    sorted_servlets.sort()

    for servlet in sorted_servlets:
        lines += file + "," + servlet + ","
        for param in params:
            if param in model[file][servlet]:
                lines += model[file][servlet][param]
            lines += ","
        lines += "\n"

# write output
output = "output.csv"
text_file = open(output, "w")
text_file.write(lines)
text_file.close()
print "output written to %s" % output 
print "finished in %dms" % (now()-start)