samara.py

#!/usr/bin/env python3

from pprint import pprint
from time import sleep
import xml.etree.ElementTree as ET

def write_header(f):
    f.write('''<?xml version="1.0" encoding="UTF-8"?>
<odoo>

''')

def write_footer(f):
    f.write("</odoo>\n")


def convert_fias_id_to_record_id(fias_id):
    return "fias_" + fias_id.replace('-', '_')


def write_record(f, model, record_id, vals, refs=None):
#    print("record", record_id)
    if not refs:
        refs = {}
    f.write('    <record id="{}" model="{}">\n'.format(record_id, model))

    for k, v in vals.items():
        f.write('        <field name="{}">{}</field>\n'.format(k, v))

    for k, v in refs.items():
        f.write('        <field name="{}" ref="{}" />\n'.format(k, v))

    f.write('    </record>\n\n')
    
def make_vals(attrib, extra_vals=None):
    data = {
        'name': attrib['OFFNAME'],
        'fias_aoguid': attrib['AOGUID'],
        'fias_okato': attrib['OKATO'],
    }
    if 'OKTMO' in attrib:
        data['fias_oktmo'] = attrib['OKTMO']
    if extra_vals:
        data.update(extra_vals)
    return data
    
        
def write_region(f, attrib):
    print('region', attrib['OFFNAME'])
    write_record(f, "res.country.state", convert_fias_id_to_record_id(attrib['AOGUID']), make_vals(attrib, {
        "code": attrib['REGIONCODE'],
    }), {
        "country_id": "base.ru",
    })

    
def write_city(f, attrib):
    print('city', attrib['OFFNAME'])
    write_record(f, "res.city", convert_fias_id_to_record_id(attrib['AOGUID']), make_vals(attrib), {
        "country_id": "base.ru",
    })

def write_state_district(f, attrib):
    print('state_disctrict', attrib)
    write_record(f, "res.state.district",  convert_fias_id_to_record_id(attrib['AOGUID']), make_vals(attrib), {
        'state_id': convert_fias_id_to_record_id(attrib['PARENTGUID']),
    })
                 
if __name__ == '__main__':
    gen = ET.iterparse("AS_ADDROBJ_20191226_9985b7cf-05ef-4113-8b40-1dcf76ce6363.XML")
    in_region = False

    cities = open("res.city.xml", "w")
    region_f = open("res.country.state.xml", "w")

    
    state_districts = open("res.state.district.xml", "w")
    fs = (cities, region_f,state_districts)

    for f in fs:
        write_header(f)

    
    for event, el in gen:
        # сбрасываем неактуальные записи
        if el.attrib['CURRSTATUS'] != '0':
            el.clear()
            continue

        if el.attrib['AOLEVEL'] == '1':
            if in_region:
                break
            in_region = el.attrib['OFFNAME'] == 'Самарская'
            if in_region:
                pprint(el.attrib)
                write_region(region_f, el.attrib)
            print(el.attrib['OFFNAME'])
            #pprint(el.attrib)

        is_state_district = el.attrib['AOLEVEL'] == '3'
        is_city = el.attrib['AOLEVEL'] in ('4', '6')

        if in_region:
            if is_city:
                write_city(cities, el.attrib)
            elif is_state_district:
                write_state_district(state_districts, el.attrib)
 #               write_region(region_f, el.attrib)
                pass
            
        el.clear()

    
    for f in fs:
        write_footer(f)
        f.close()