src/wikibase_helpers.py

import requests
import logging
from wikibaseintegrator import WikibaseIntegrator, datatypes,  wbi_helpers
from wikibaseintegrator.wbi_config import config
from wikibaseintegrator.wbi_exceptions import MWApiError
from wikibaseintegrator.wbi_helpers import mediawiki_api_call_helper, SearchError
from wikibaseintegrator import WikibaseIntegrator, wbi_login
login_wikibase = wbi_login.Login(user="Traopia", password="diqfiz-wodnI4-jafwax", mediawiki_api_url="https://fashionwiki.wikibase.cloud/w/api.php")

wikibase_api_url = 'https://fashionwiki.wikibase.cloud/w/api.php'
config = {
"SPARQL_ENDPOINT_URL": "https://fashionwiki.wikibase.cloud/query/sparql",
'USER_AGENT':  'YourBotName/1.0 (https://yourwebsite.org/bot-info)',
'WIKIBASE_URL': wikibase_api_url,
}




# List of valid language codes (can be expanded)
VALID_LANGUAGE_CODES = ['en']

def get_property_id_by_label(property_label, api_url):
    """
    Resolve the property label to its corresponding property ID from Wikibase.
    
    Args:
        property_label (str): The label of the property to search.
        api_url (str): The API URL of the target Wikibase or Wikidata.
    
    Returns:
        str: The property ID if found, otherwise None.
    """
    url = f'{api_url}/w/api.php?action=wbsearchentities&search={property_label}&language=en&type=property&format=json'
    response = requests.get(url)
    
    if response.status_code == 200:
        search_results = response.json()
        if 'search' in search_results and search_results['search']:
            # Return the first matching property ID
            return search_results['search'][0]['id']
        else:
            logging.info(f"No property found for label: {property_label}")
            return None
    else:
        logging.error(f"Failed to search for property by label in the target Wikibase. HTTP Status Code: {response.status_code}")
        return None
    

def get_entity_id_by_label(search_string,wiki, dict_result=False) -> list:
    """
    Performs a search for entities in the Wikibase instance using labels and aliases.
    You can have more information on the parameters in the MediaWiki API help (https://www.wikidata.org/w/api.php?action=help&modules=wbsearchentities)

    :param search_string: A string which should be searched for in the Wikibase instance (labels and aliases)
    :param wiki: The wiki to search in. It can be "wikidata" or "wikibase"
    :param dict_result: If True, the result will be a list of dictionaries with the keys 'id', 'label', 'match', 'description' and 'aliases'. If False, the result will be a list of strings with the entity IDs.
    :return: A list of dictionaries or strings with the search results
    """

    if wiki == "wikibase":
        login = login_wikibase
        mediawiki_api_url = wikibase_api_url

    language = "en"
    strict_language = False

    params = {
        'action': 'wbsearchentities',
        'search': search_string,
        'language': language,
        'type': "item",
        'limit': 50,
        'format': 'json',
    }

    if strict_language:
        params.update({'strict_language': ''})

    cont_count = 0
    results = []

    while True:
        params.update({'continue': cont_count})
        search_results = mediawiki_api_call_helper(data=params, login = login, mediawiki_api_url=mediawiki_api_url, user_agent = config['USER_AGENT'])
        if search_results['success'] != 1:
            raise SearchError('Wikibase API wbsearchentities failed')

        for i in search_results['search']:
            if dict_result:
                description = i['description'] if 'description' in i else None
                aliases = i['aliases'] if 'aliases' in i else None
                results.append({
                    'id': i['id'],
                    'label': i['label'],
                    'match': i['match'],
                    'description': description,
                    'aliases': aliases
                })
            else:
                results.append(i['id'])

        if 'search-continue' not in search_results:
            break
        cont_count = search_results['search-continue']
        if cont_count >= 50:
            break
    return results


wikibase_properties_id = {"instance of": get_property_id_by_label("instance of", wikibase_api_url),
                    "reference URL": get_property_id_by_label("reference URL", wikibase_api_url),
                    "start time": get_property_id_by_label("start time", wikibase_api_url),
                    "end time": get_property_id_by_label("end time", wikibase_api_url),
                    "occupation title": get_property_id_by_label("occupation title", wikibase_api_url),
                    "educated at": get_property_id_by_label("educated at", wikibase_api_url),
                    "employer": get_property_id_by_label("employer", wikibase_api_url),
                    "work location": get_property_id_by_label("work location", wikibase_api_url),
                    "award received": get_property_id_by_label("award received", wikibase_api_url),
                    "point in time": get_property_id_by_label("point in time", wikibase_api_url),
                    "exact match": get_property_id_by_label("exact match", wikibase_api_url),
                    "date of birth": get_property_id_by_label("date of birth", wikibase_api_url),
                    "place of birth": get_property_id_by_label("place of birth", wikibase_api_url),
                    "date of death": get_property_id_by_label("date of death", wikibase_api_url),
                    "country of citizenship": get_property_id_by_label("country of citizenship", wikibase_api_url),
                    "occupation": get_property_id_by_label("occupation", wikibase_api_url),
                    "sex or gender": get_property_id_by_label("sex or gender", wikibase_api_url),
                    "official website": get_property_id_by_label("official website", wikibase_api_url),
                    "perfumes": get_property_id_by_label("perfumes", wikibase_api_url),
                    "who wears it": get_property_id_by_label("who wears it", wikibase_api_url),
                    "inception": get_property_id_by_label("inception", wikibase_api_url),
                    "headquarters location": get_property_id_by_label("headquarters location", wikibase_api_url),
                    "parent organization": get_property_id_by_label("parent organization", wikibase_api_url),
                    "founded by": get_property_id_by_label("founded by", wikibase_api_url),
                    "owned by": get_property_id_by_label("owned by", wikibase_api_url),
                    "industry": get_property_id_by_label("industry", wikibase_api_url),
                    "country": get_property_id_by_label("country", wikibase_api_url),
                    "total revenue": get_property_id_by_label("total revenue", wikibase_api_url),
                    "designer employed": get_property_id_by_label("designer employed", wikibase_api_url),
                    "country of origin": get_property_id_by_label("country of origin", wikibase_api_url),
                    "fashion collection": get_property_id_by_label("fashion collection", wikibase_api_url),
                    "fashion season": get_property_id_by_label("fashion season", wikibase_api_url),
                    "fashion show location": get_property_id_by_label("fashion show location", wikibase_api_url),
                    "description of fashion collection": get_property_id_by_label("description of fashion collection", wikibase_api_url),
                    "image of fashion collection": get_property_id_by_label("image of fashion collection", wikibase_api_url),
                    "editor of fashion collection description": get_property_id_by_label("editor of fashion collection description", wikibase_api_url),
                    "date of fashion collection": get_property_id_by_label("date of fashion collection", wikibase_api_url),
                    "fashion show category": get_property_id_by_label("fashion show category", wikibase_api_url),
                    "fashion house X fashion collection": get_property_id_by_label("fashion house X fashion collection", wikibase_api_url),
                    "designer of collection": get_property_id_by_label("designer of collection", wikibase_api_url)}


classes_wikibase = {"fashion designer": get_entity_id_by_label("fashion designer", "wikibase")[0],
                        "fashion house": get_entity_id_by_label("fashion house", "wikibase")[0],
                        "business": get_entity_id_by_label("business", "wikibase")[0],
                        "academic institution": get_entity_id_by_label("academic institution", "wikibase")[0],
                        "geographic location": get_entity_id_by_label("geographic location", "wikibase")[0],
                        "fashion award": get_entity_id_by_label("fashion award","wikibase")[0],
                        "gender":get_entity_id_by_label("gender","wikibase")[0] , 
                        "occupation": get_entity_id_by_label("occupation","wikibase")[0],
                        "human": get_entity_id_by_label("human","wikibase")[0],
                        "organization": get_entity_id_by_label("organization","wikibase")[0],
                        "brand": get_entity_id_by_label("brand","wikibase")[0],
                        "lifestyle brand": get_entity_id_by_label("lifestyle brand","wikibase")[0],
                        "privately held company": get_entity_id_by_label("privately held company","wikibase")[0],
                        "fashion award": get_entity_id_by_label("fashion award", "wikibase")[0],
                        "fashion season": get_entity_id_by_label("fashion season", "wikibase")[0],
                        "fashion show category": get_entity_id_by_label("fashion show category", "wikibase")[0],
                        "fashion season collection": get_entity_id_by_label("fashion season collection", "wikibase")[0],
                        "fashion journalist": get_entity_id_by_label("fashion journalist", "wikibase")[0],
                        }