citygml_give_unique_building_ids.py

#!/usr/bin/env python
# coding: utf-8

# python citygml_give_unique_building_ids.py filename1.gml filename2.gml ...

import re
import argparse

parser = argparse.ArgumentParser(description='Process CityGML files and give unique building IDs if needed.')
parser.add_argument("citygml_filename", nargs='+')
args = parser.parse_args()

BUILDING_ID_PATTERN =  re.compile("(?<=bldg:Building gml:id=\")\w+")

input_filenames = args.citygml_filename

for input_filename in input_filenames:
    print(f'Processing {input_filename}')
    output_filename = input_filename.replace('.gml', '_unique_ids.gml')

    seen_ids = {}

    with open(input_filename) as gml:
        with open(output_filename, 'w+') as new_gml:
            for line in gml:
                m = re.search(BUILDING_ID_PATTERN, line)
                if m:
                    # Building ID has been found
                    building_id = m.group(0)
                    if building_id in seen_ids:
                        # Building ID has already been used
                        seen_ids[building_id] += 1
                        new_id = f'{building_id}_{seen_ids[building_id]}'
                        print(new_id)
                        line = re.sub(BUILDING_ID_PATTERN, new_id, line)
                    else:
                        # Building ID.
                        seen_ids[building_id] = 1
                    
                new_gml.write(line)