paratext_directory.py

import os
import xml.etree.ElementTree as ET
import json

def parse_xml_file(file_path):
    """ Parse an XML file and return needed metadata as a dictionary. """
    tree = ET.parse(file_path)
    root = tree.getroot()

    data = {
        'Language': root.find('Language').text if root.find('Language') is not None else '',
        'FullName': root.find('FullName').text if root.find('FullName') is not None else '',
        'Copyright': root.find('Copyright').text if root.find('Copyright') is not None else '',
        'Name': root.find('Name').text if root.find('Name') is not None else '',
        'subdirectory': os.path.dirname(file_path)
    }
    return data

def create_indexed_dictionaries(base_path):
    """ Create dictionaries indexed by language, fullname, and name from Settings.xml files. """
    language_dict = {}
    fullname_dict = {}
    name_dict = {}

    # Walk through the directory and find all Settings.xml files
    for root, dirs, files in os.walk(base_path):
        for file in files:
            if file == 'Settings.xml':
                full_path = os.path.join(root, file)
                metadata = parse_xml_file(full_path)

                language_dict[metadata['Language']] = metadata
                fullname_dict[metadata['FullName']] = metadata
                name_dict[metadata['Name']] = metadata

    return {
        'ByLanguage': language_dict,
        'ByFullName': fullname_dict,
        'ByName': name_dict
    }

# Directory containing the subdirectories with Settings.xml
base_directory = '.'

# Create dictionaries and print them as JSON
indexed_data = create_indexed_dictionaries(base_directory)
print(json.dumps(indexed_data, indent=4, ensure_ascii=False))