find_matches.py

from pathlib import Path
import json
from queries import query_formularios, query_usuarios, get_macros_differences
                     
formularios_weight_difference_path = 'anonymized_formularios_weight_difference'
usuarios_macros_difference_path = 'usuarios_macros_difference'

# Find users dates that match the query
def find_user_dates_matches(query, debug=False):
    # Create a dictionary to store the matches
    matches_dict = {}

    # Get all the files in the formularios_path
    files = Path(formularios_weight_difference_path).glob('*.json')
    files = list(files)
    files.sort()

    # Iterate over the user data
    for i, file in enumerate(files):
        with open(file, 'r') as f:
            data = json.load(f)

        if file.name == '[email protected]':
            dates = query_formularios(data, query, debug=False, file_name=file.name)
        else:
            dates = query_formularios(data, query, debug=debug)
        if len(dates) > 0:
            file_name = file.name
            matches_dict[file_name] = dates
            if debug:
                if i > 0:
                    print("")
                print(f"{file_name} has {len(dates)} dates that match the query:")
                for date in dates:
                    print(f"\t{date}")
                    for query_item in query:
                        key = list(query_item.keys())[0]
                        data_value = data[date][key]
                        query_value = query_item[key]['value']
                        operator = query_item[key]['operator']
                        if type(data_value) == int or type(data_value) == float:
                            data_value = f"{data_value:.2f}"
                        if operator == 'in' or operator == 'contains':
                            print(f"\t\t{key} data: \"{data_value}\", query: \"{query_value}\"")
                        else:
                            print(f"\t\t{key} data: \"{data_value}\" \"{operator}\" query: \"{query_value}\"")
    
    return matches_dict

def find_macros_that_match_dates_of_users(matches_dict, debug=False):
    # Create a list to store the macros differences
    macros_differences_list = []

    # Iterate over the matches dictionary
    for match_user in matches_dict:
        if debug: print(f"match_user: {match_user}")

        # Get dates list
        dates_list_from_user = matches_dict[match_user]

        # Get user data
        user_data = usuarios_macros_difference_path + '/' + match_user
        user_data = json.load(open(user_data, 'r'))

        # Query usuarios
        dates_that_match = query_usuarios(user_data, dates_list_from_user, debug=False, limit_days=31)
        if len(dates_that_match) > 0:
            if debug: print(f"\tdates that match: {dates_that_match}")

            # Get macros differences
            macros_differences = get_macros_differences(user_data, dates_that_match)
            if type(macros_differences) == list:
                if len(macros_differences) > 0:
                    for macros_difference in macros_differences:
                        macros_differences_list.append(macros_difference)
                        if debug: print(f"\tmacros_differences: {macros_difference}")
                    if debug: print("")
            else:
                macros_differences_list.append(macros_differences)
                if debug: print(f"\tmacros_differences: {macros_differences}\n")
    
    return macros_differences_list