parseXML.py

parseXML.py

import arcpy
import xml.etree.ElementTree as etree
from os.path import join
from dateutil.parser import parse


xml = r'Y:\Mineralsoutput.xml'
output_folder = r'C:\Temp'
gdb_name = 'XMLOutput.gdb'
gdb = join(output_folder, gdb_name)
table_name = 'Minerals'
table = join(gdb, table_name)

print('deleting old gdb')
if arcpy.Exists(gdb):
    arcpy.management.Delete(gdb)

print('building gdb and table')
arcpy.management.CreateFileGDB(output_folder, gdb_name)
arcpy.management.CreateTable(gdb, table_name)
arcpy.management.AddField(table, 'utmx', 'Long')
arcpy.management.AddField(table, 'utmy', 'Long')
arcpy.management.AddField(table, 'operator', 'Text', field_length=100)
arcpy.management.AddField(table, 'name', 'Text', field_length=100)
arcpy.management.AddField(table, 'status', 'Text', field_length=10)
arcpy.management.AddField(table, 'inspector', 'Text', field_length=100)
arcpy.management.AddField(table, 'app_acr', 'Long')
arcpy.management.AddField(table, 'permittee', 'Text', field_length=100)
arcpy.management.AddField(table, 'mineid', 'Text', field_length=12)
arcpy.management.AddField(table, 'minerals', 'Text', field_length=100)
arcpy.management.AddField(table, 'inspdate', 'Date')

all_xml_fields = ['utmx', 'utmy', 'Operator', 'SiteName', 'Status', 'inspector', 'TotalPermitArea', 'Permittee', 'Permit', 'minerals', 'inspdate']
int_fields = ['utmx', 'utmy']
float_fields = ['app_acr']
date_fields = ['inspdate']

print('parsing xml')
with open(xml, 'r+', encoding='utf-8') as file:
    xml_text = file.read()
    if not xml_text.startswith('<root>'):
        xml_text = xml_text.replace('\n', '')
        file.seek(0)
        file.truncate()
        file.write('<root>{}</root>'.format(xml_text))

tree = etree.fromstring(xml_text)

print('loading data into table')
with arcpy.da.InsertCursor(table, ['*']) as cursor:
    for record in tree:
        values = []
        for field in all_xml_fields:
            value = record.get(field)
            if field in int_fields:
                values.append(int(value))
            elif field in float_fields:
                values.append(float(value))
            elif field in date_fields:
                values.append(parse(value))
            elif value is not None:
                values.append(value.strip())
            else:
                values.append(value)
        cursor.insertRow([1] + values)

print('making xy event layer')
layer = arcpy.management.MakeXYEventLayer(table, 'utmx', 'utmy', 'MinesLayer', arcpy.SpatialReference(26912))

print('copying points')
utm = arcpy.management.CopyFeatures(layer, table + '_UTM')

print('projecting to web mercator')
projected = arcpy.management.Project(utm, table + '_WGS', arcpy.SpatialReference(3857), 'NAD_1983_To_WGS_1984_5')

print('done')