cicv_evaluate_master-single_eval/config_parser.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
##################################################################
#
# Copyright (c) 2023 CICV, Inc. All Rights Reserved
#
##################################################################
"""
@Authors:           yangzihao(yangzihao@china-icv.cn)
@Data:              2023/11/02
@Last Modified:     2023/11/02
@Summary:           This module provides the function to parse the config json file.
"""

import pandas as pd
import numpy as np

from common import json2dict
from collections import OrderedDict


class ConfigParse(object):
    """

    """

    def __init__(self, json_file):
        # weight info
        self.scoreModel = ""

        self.dimension_weight = {}
        self.dimension_list = []
        self.dimension_name = {}

        self.type_weight = {}
        self.type_list = []
        self.type_name = {}

        self.metric_list = []
        self.metric_dict = {}
        self.name_dict = {}

        self.unit_dict = {}

        self.builtinDimensionList = ["safe", "function", "compliance", "comfort", "efficient"]
        self.builtinTypeDict = {
            "safe": ['safeTime', 'safeDistance', 'safeAcceleration', 'safeProbability'],
            "function": ['functionACC', 'functionLKA'],
            "compliance": ['functionACC', 'functionLKA', 'safeAcceleration', 'safeProbability'],
            "comfort": ['comfortLat', 'comfortLon'],
            "efficient": ['efficientDrive', 'efficientStop']
        }
        self.builtinMetricList = ["TTC", "MTTC", "THW", "LatSD", "LonSD", "DRAC", "BTN", "STN", "collisionRisk",
                                  "collisionSeverity", "followSpeedDeviation", "followDistanceDeviation",
                                  "followStopDistance", "followResponseTime", 'laneDistance',
                                  'centerDistanceExpectation', 'centerDistanceStandardDeviation',
                                  'centerDistanceMax', 'centerDistanceMin', 'centerDistanceFrequency',
                                  'centerDistanceRange',
                                  "zigzag", "shake", "cadence", "slamBrake", "slamAccelerate",
                                  "pressSolidLine", "runRedLight", "overspeed20_50", "overspeed50", "overspeed10",
                                  "overspeed10_20",
                                  "averageSpeed", "stopDuration", "stopCount"]

        # score info
        self.config = {}

        # initialization
        self.config_dict = json2dict(json_file)
        self.parse_dict = self._config_parse(self.config_dict)

    def _config_parse(self, config_dict):
        # score
        if 'scoreModel' in list(config_dict.keys()):
            self.scoreModel = config_dict['scoreModel']
        else:
            self.scoreModel = "builtin"

        # dimension info
        dimensionWeight = config_dict['dimensionWeight']
        self.dimension_weight = dimensionWeight
        self.dimension_name = config_dict['dimensionName']
        self.dimension_list = list(self.dimension_weight.keys())

        # type info
        typeWeight = config_dict['typeWeight']
        self.type_weight = typeWeight
        self.type_name = config_dict["typeName"]

        for dimension in self.dimension_list:
            if dimension == 'compliance':
                self.config[dimension] = self._compliance_config_parse(config_dict, typeWeight, dimensionWeight)
            else:
                self.config[dimension] = self._dimension_config_parse(dimension, config_dict, typeWeight,
                                                                      dimensionWeight)

            # self.name_dict[dimension] = self.config[dimension]['name']
            # self.unit_dict[dimension] = self.config[dimension]['unit']
            self.name_dict.update(self.config[dimension]['name'])
            self.unit_dict.update(self.config[dimension]['unit'])

            self.metric_dict[dimension] = self.config[dimension]['typeMetricDict']
            self.metric_list.extend(self.config[dimension]['metric'])
            self.type_list.extend(self.config[dimension]['type'])

        print()

    def _dimension_config_parse(self, dimension, config_dict, typeWeight, dimensionWeight):

        # get weight type
        typeWeightDimension = typeWeight[dimension]
        typeDimensionList = list(typeWeightDimension.keys())
        if dimension in self.builtinDimensionList:
            typeWeightDimension_builtin = OrderedDict(
                (key, typeWeightDimension[key]) for key in self.builtinTypeDict[dimension] if key in typeDimensionList)
            typeWeightDimension_builtin = list(typeWeightDimension_builtin.keys())
            typeDimension = typeWeightDimension_builtin + [x for x in typeDimensionList if
                                                           x not in self.builtinTypeDict[dimension]]
        else:
            typeDimension = typeDimensionList

        typeWeightDimensionList = list(typeWeightDimension.values())
        flagCustomDimension = not all(x is None for x in typeWeightDimensionList)

        # get type name
        typeNameDimension = self.type_name[dimension]

        # Dimension
        dimension_dict = config_dict[dimension]
        dimension_value_dict = {}
        typeMetricDict = {}

        for type in typeDimension:
            dimension_value_dict.update(dimension_dict[type])
            typeMetricDict[type] = list(dimension_dict[type].keys())

        df_dimension_value = pd.DataFrame(dimension_value_dict).T

        # get metric list
        metricDimension = df_dimension_value.index.tolist()

        # get name list
        nameDimension = df_dimension_value['name'].to_dict()
        # nameDimensionList = list(nameDimension.values())

        # get unit list
        unitDimension = df_dimension_value['unit'].to_dict()
        unitDimension = {key: (value if value is not None else "") for key, value in unitDimension.items()}
        # unitDimensionList = list(unitDimension.values())

        # get weight list
        weightDimension = df_dimension_value['weight'].astype(float).to_dict()
        weightDimensionList = list(weightDimension.values())

        # get priority list
        priorityDimension = df_dimension_value['priority'].astype(int).to_dict()
        priorityDimensionList = list(priorityDimension.values())

        # get paramList
        paramDimension = df_dimension_value['paramList'].to_dict()

        bulitin_first_key = next((key for key in paramDimension if key in self.builtinMetricList), None)
        first_key = next(iter(paramDimension)) if not bulitin_first_key else bulitin_first_key
        paramNum = len(paramDimension[first_key])

        kindDimension = [{} for _ in range(paramNum)]
        optimalDimension = [{} for _ in range(paramNum)]
        multipleDimension = [{} for _ in range(paramNum)]
        spareDimension = [{} for _ in range(paramNum)]
        # spare1Dimension = [{} for _ in range(paramNum)]
        # spare2Dimension = [{} for _ in range(paramNum)]

        customMetricParam = {}  # custiom metric paramList

        for key, value_list in paramDimension.items():
            if key in self.builtinMetricList:
                for i in range(len(value_list)):
                    kindDimension[i][key] = int(value_list[i]['kind'])
                    optimalDimension[i][key] = float(value_list[i]['optimal'])
                    multipleDimension[i][key] = [float(x) for x in value_list[i]['multiple']]
                    spareDimension[i][key] = [item["param"] for item in value_list[i]["spare"]]
                    # spareDimension[i][key] = [float(item["param"]) for item in value_list[i]["spare"]]
                    # spare1Dimension[i][key] = (value_list[i]['spare1'])
                    # spare2Dimension[i][key] = (value_list[i]['spare2'])
            else:
                customMetricParam[key] = value_list

        kindDimensionList = [value for dict_val in kindDimension for value in dict_val.values()]
        optimalDimensionList = [value for dict_val in optimalDimension for value in dict_val.values()]
        multipleDimensionList = [value for dict_val in multipleDimension for value in dict_val.values()]
        spareDimensionList = [value for dict_val in spareDimension for value in dict_val.values()]
        # spare1DimensionList = [value for dict_val in spare1Dimension for value in dict_val.values()]
        # spare2DimensionList = [value for dict_val in spare2Dimension for value in dict_val.values()]

        if paramNum == 1:
            kindDimension = kindDimension[0]
            optimalDimension = optimalDimension[0]
            multipleDimension = multipleDimension[0]
            spareDimension = spareDimension[0]
            # spare1Dimension = spare1Dimension[0]
            # spare2Dimension = spare2Dimension[0]

        result = {
            "weightDimension": float(dimensionWeight[dimension]),
            "weightCustom": flagCustomDimension,
            "type": typeDimension,
            "typeWeight": typeWeightDimension,
            "typeWeightList": typeWeightDimensionList,
            "typeName": typeNameDimension,
            "customMetricParam": customMetricParam,
            "metric": metricDimension,
            "typeMetricDict": typeMetricDict,
            "name": nameDimension,
            # "nameList": nameDimensionList,
            "unit": unitDimension,
            # "unitList": unitDimensionList,
            "weight": weightDimension,
            "weightList": weightDimensionList,
            "priority": priorityDimension,
            "priorityList": priorityDimensionList,
            "kind": kindDimension,
            "kindList": kindDimensionList,
            "optimal": optimalDimension,
            "optimalList": optimalDimensionList,
            "multiple": multipleDimension,
            "multipleList": multipleDimensionList,
            "spare": spareDimension,
            "spareList": spareDimensionList,
            # "spare1": spare1Dimension,
            # "spare1List": spare1DimensionList,
            # "spare2": spare2Dimension,
            # "spare2List": spare2DimensionList
        }
        return result

    def _safe_config_parse(self, config_dict, typeWeight, dimensionWeight):

        # get weight type
        typeWeightSafe = typeWeight['safe']
        # typeSafe = [key for key, value in typeWeightSafe.items() if value != 0]
        typeSafe = list(typeWeightSafe.keys())  # get type list
        # typeWeightSafeDict = {key: value for key, value in typeWeightSafe.items() if value != 0}
        # typeWeightSafeList = list(typeWeightSafeDict.values())
        typeWeightSafeList = list(typeWeightSafe.values())
        flagCustomSafe = not all(x is None for x in typeWeightSafeList)

        # safe
        safe_dict = config_dict['safe']
        safe_value_dict = {}

        for type in typeSafe:
            safe_value_dict.update(safe_dict[type])
        df_safe_value = pd.DataFrame(safe_value_dict).T

        # safe_value_dict.update(safe_dict['safeTime'])
        # safe_value_dict.update(safe_dict['safeDistance'])
        # safe_value_dict.update(safe_dict['safeAcceleration'])
        # safe_value_dict.update(safe_dict['safeProbability'])
        # df_safe_value = pd.DataFrame(safe_value_dict).T
        # df_safe_value = df_safe_value[df_safe_value['enable'] == True]

        # get metric list
        metricSafe = df_safe_value.index.tolist()

        # get weight list
        weightSafe = df_safe_value['weight'].astype(float).to_dict()
        weightSafeList = list(weightSafe.values())

        # get priority list
        prioritySafe = df_safe_value['priority'].astype(int).to_dict()
        prioritySafeList = list(prioritySafe.values())

        # get paramList
        paramSafe = df_safe_value['paramList'].to_dict()

        first_key = next(iter(paramSafe))
        paramNum = len(paramSafe[first_key])

        kindSafe = [{} for _ in range(paramNum)]
        optimalSafe = [{} for _ in range(paramNum)]
        multipleSafe = [{} for _ in range(paramNum)]
        spare1Safe = [{} for _ in range(paramNum)]
        spare2Safe = [{} for _ in range(paramNum)]

        for key, value_list in paramSafe.items():
            for i in range(len(value_list)):
                kindSafe[i][key] = int(value_list[i]['kind'])
                optimalSafe[i][key] = float(value_list[i]['optimal'])
                multipleSafe[i][key] = [float(x) for x in value_list[i]['multiple']]

                spare1Safe[i][key] = (value_list[i]['spare1'])
                spare2Safe[i][key] = (value_list[i]['spare2'])

        kindSafeList = [value for dict_val in kindSafe for value in dict_val.values()]
        optimalSafeList = [value for dict_val in optimalSafe for value in dict_val.values()]
        multipleSafeList = [value for dict_val in multipleSafe for value in dict_val.values()]
        spare1SafeList = [value for dict_val in spare1Safe for value in dict_val.values()]
        spare2SafeList = [value for dict_val in spare2Safe for value in dict_val.values()]

        safe = {
            "weightSafe": float(dimensionWeight['safe']),
            "weightCustom": flagCustomSafe,
            "type": typeSafe,
            "typeWeight": typeWeightSafe,
            "typeWeightList": typeWeightSafeList,
            "metric": metricSafe,
            "weight": weightSafe,
            "weightList": weightSafeList,
            "priority": prioritySafe,
            "priorityList": prioritySafeList,
            "kind": kindSafe,
            "kindList": kindSafeList,
            "optimal": optimalSafe,
            "optimalList": optimalSafeList,
            "multiple": multipleSafe,
            "multipleList": multipleSafeList,
            "spare1": spare1Safe,
            "spare1List": spare1SafeList,
            "spare2": spare2Safe,
            "spare2List": spare2SafeList
        }
        return safe

    def _function_config_parse(self, config_dict, typeWeight, dimensionWeight):
        # get weight type
        typeWeightFunction = typeWeight['function']
        typeFunction = [key for key, value in typeWeightFunction.items() if value != 0]

        typeWeightFunctionDict = {key: value for key, value in typeWeightFunction.items() if value != 0}
        typeWeightFunctionList = list(typeWeightFunctionDict.values())
        flagCustomFunction = not all(x is None for x in typeWeightFunctionList)

        # function
        function_dict = config_dict['function']
        function_value_dict = {}

        for type in typeFunction:
            function_value_dict.update(function_dict[type])
        df_function_value = pd.DataFrame(function_value_dict).T

        # get metric list
        metricFunction = df_function_value.index.tolist()

        # get weight list
        weightFunction = df_function_value['weight'].astype(float).to_dict()
        weightFunctionList = list(weightFunction.values())

        # get priority list
        priorityFunction = df_function_value['priority'].astype(int).to_dict()
        priorityFunctionList = list(priorityFunction.values())

        # get kind list
        kindFunction = df_function_value['kind'].astype(int).to_dict()
        kindFunctionList = list(kindFunction.values())

        # get optimal list
        optimalFunction = df_function_value['optimal'].astype(float).to_dict()
        optimalFunctionList = list(optimalFunction.values())

        # get multiple list
        multipleFunction = df_function_value['multiple'].astype(float).to_dict()
        multipleFunctionList = list(multipleFunction.values())

        function = {
            "weightFunction": float(dimensionWeight['function']),
            "weightCustom": flagCustomFunction,
            "type": typeFunction,
            "typeWeight": typeWeightFunction,
            "typeWeightList": typeWeightFunctionList,
            "metric": metricFunction,
            "weight": weightFunction,
            "weightList": weightFunctionList,
            "priority": priorityFunction,
            "priorityList": priorityFunctionList,
            "kind": kindFunction,
            "kindList": kindFunctionList,
            "optimal": optimalFunction,
            "optimalList": optimalFunctionList,
            "multiple": multipleFunction,
            "multipleList": multipleFunctionList
        }
        return function

    def _compliance_config_parse(self, config_dict, typeWeight, dimensionWeight):
        # get weight type
        typeWeightCompliance = typeWeight['compliance']
        typeCompliance = list(typeWeightCompliance.keys())
        typeWeightComplianceList = list(typeWeightCompliance.values())
        flagCustomCompliance = not all(x is None for x in typeWeightComplianceList)

        # get type name
        typeNameCompliance = self.type_name["compliance"]

        # compliance
        compliance_dict = config_dict['compliance']
        compliance_value_dict = {}
        typeMetricDict = {}

        for type in typeCompliance:
            compliance_value_dict.update(compliance_dict[type])
            typeMetricDict[type] = list(compliance_dict[type].keys())

        df_compliance_value = pd.DataFrame(compliance_value_dict).T

        # get metric list
        metricCompliance = df_compliance_value.index.tolist()

        # get weight type
        typeWeightCompliance = typeWeight['compliance']
        typeWeightComplianceList = list(typeWeightCompliance.values())

        # get name list
        nameCompliance = df_compliance_value['name'].to_dict()

        # get unit list
        unitCompliance = df_compliance_value['unit'].to_dict()

        # get weight list
        weightCompliance = df_compliance_value['weight'].astype(float).to_dict()
        weightComplianceList = list(weightCompliance.values())

        compliance = {
            "weightDimension": float(dimensionWeight['compliance']),
            "weightCustom": flagCustomCompliance,
            "type": typeCompliance,
            "typeWeight": typeWeightCompliance,
            "typeWeightList": typeWeightComplianceList,
            "typeName": typeNameCompliance,
            "typeMetricDict": typeMetricDict,
            "metric": metricCompliance,
            "name": nameCompliance,
            "unit": unitCompliance,
            "weight": weightCompliance,
            "weightList": weightComplianceList
        }
        return compliance

    def _comfort_config_parse(self, config_dict, typeWeight, dimensionWeight):
        # get weight type
        typeWeightComfort = typeWeight['comfort']
        typeComfort = [key for key, value in typeWeightComfort.items() if value != 0]
        typeWeightComfortDict = {key: value for key, value in typeWeightComfort.items() if value != 0}
        typeWeightComfortList = list(typeWeightComfortDict.values())
        flagCustomComfort = not all(x is None for x in typeWeightComfortList)

        # comfort
        comfort_dict = config_dict['comfort']
        comfort_value_dict = {}

        for type in typeComfort:
            comfort_value_dict.update(comfort_dict[type])
        df_comfort_value = pd.DataFrame(comfort_value_dict).T

        # comfort_value_dict.update(comfort_dict['comfortLat'])
        # comfort_value_dict.update(comfort_dict['comfortLon'])
        # df_comfort_value = pd.DataFrame(comfort_value_dict).T
        # df_comfort_value = df_comfort_value[df_comfort_value['enable'] == True]

        # get metric list
        metricComfort = df_comfort_value.index.tolist()

        # get weight type
        typeWeightComfort = typeWeight['comfort']
        typeWeightComfortList = list(typeWeightComfort.values())

        # get weight list
        weightComfort = df_comfort_value['weight'].astype(float).to_dict()
        weightComfortList = list(weightComfort.values())

        # get priority list
        priorityComfort = df_comfort_value['priority'].astype(int).to_dict()
        priorityComfortList = list(priorityComfort.values())

        # get kind list
        kind1 = df_comfort_value['kind1'].astype(int).to_dict()
        kind2 = df_comfort_value['kind2'].astype(int).to_dict()
        kind3 = df_comfort_value['kind3'].astype(int).to_dict()
        kindComfort = {
            "kind1": kind1,
            "kind2": kind2,
            "kind3": kind3
        }
        kindComfortList = list(self._merge_dict(kind1, kind2, kind3).values())

        # get optimal list
        optimal1 = df_comfort_value['optimal1'].astype(float).to_dict()
        optimal2 = df_comfort_value['optimal2'].astype(float).to_dict()
        optimal3 = df_comfort_value['optimal3'].astype(float).to_dict()
        optimalComfort = {
            "optimal1": optimal1,
            "optimal2": optimal2,
            "optimal3": optimal3
        }
        optimalComfortList = list(self._merge_dict(optimal1, optimal2, optimal3).values())

        # get multiple list
        multiple1 = df_comfort_value['multiple1'].astype(float).to_dict()
        multiple2 = df_comfort_value['multiple2'].astype(float).to_dict()
        multiple3 = df_comfort_value['multiple3'].astype(float).to_dict()
        multipleComfort = {
            "multiple1": multiple1,
            "multiple2": multiple2,
            "multiple3": multiple3
        }
        multipleComfortList = list(self._merge_dict(multiple1, multiple2, multiple3).values())

        comfort = {
            "weightComfort": float(dimensionWeight['comfort']),
            "weightCustom": flagCustomComfort,
            "type": typeComfort,
            "typeWeight": typeWeightComfort,
            "typeWeightList": typeWeightComfortList,
            "metric": metricComfort,
            "weight": weightComfort,
            "weightList": weightComfortList,
            "priority": priorityComfort,
            "priorityList": priorityComfortList,
            "kind": kindComfort,
            "kindList": kindComfortList,
            "optimal": optimalComfort,
            "optimalList": optimalComfortList,
            "multiple": multipleComfort,
            "multipleList": multipleComfortList
        }
        return comfort

    def _efficient_config_parse(self, config_dict, typeWeight, dimensionWeight):
        # get weight type
        typeWeightEfficient = typeWeight['efficient']
        typeEfficient = [key for key, value in typeWeightEfficient.items() if value != 0]
        typeWeightEfficientDict = {key: value for key, value in typeWeightEfficient.items() if value != 0}
        typeWeightEfficientList = list(typeWeightEfficientDict.values())
        flagCustomEfficient = not all(x is None for x in typeWeightEfficientList)

        # efficient
        efficient_dict = config_dict['efficient']
        efficient_value_dict = {}

        for type in typeEfficient:
            efficient_value_dict.update(efficient_dict[type])
        df_efficient_value = pd.DataFrame(efficient_value_dict).T

        # efficient_value_dict.update(efficient_dict['efficientDrive'])
        # efficient_value_dict.update(efficient_dict['efficientStop'])
        # df_efficient_value = pd.DataFrame(efficient_value_dict).T
        # df_efficient_value = df_efficient_value[df_efficient_value['enable'] == True]

        # get metric list
        metricEfficient = df_efficient_value.index.tolist()

        # get weight type
        typeWeightEfficient = typeWeight['efficient']
        typeWeightEfficientList = list(typeWeightEfficient.values())

        # get weight list
        weightEfficient = df_efficient_value['weight'].astype(float).to_dict()
        weightEfficientList = list(weightEfficient.values())

        # get priority list
        priorityEfficient = df_efficient_value['priority'].astype(int).to_dict()
        priorityEfficientList = list(priorityEfficient.values())

        # get kind list
        kindEfficient = df_efficient_value['kind'].astype(int).to_dict()
        kindEfficientList = list(kindEfficient.values())

        # get optimal list
        optimalEfficient = df_efficient_value['optimal'].astype(float).to_dict()
        optimalEfficientList = list(optimalEfficient.values())

        # get multiple list
        multipleEfficient = df_efficient_value['multiple'].astype(float).to_dict()
        multipleEfficientList = list(multipleEfficient.values())

        efficient = {
            "weightEfficient": float(dimensionWeight['efficient']),
            "weightCustom": flagCustomEfficient,
            "type": typeEfficient,
            "typeWeight": typeWeightEfficient,
            "typeWeightList": typeWeightEfficientList,
            "metric": metricEfficient,
            "weight": weightEfficient,
            "weightList": weightEfficientList,
            "priority": priorityEfficient,
            "priorityList": priorityEfficientList,
            "kind": kindEfficient,
            "kindList": kindEfficientList,
            "optimal": optimalEfficient,
            "optimalList": optimalEfficientList,
            "multiple": multipleEfficient,
            "multipleList": multipleEfficientList
        }
        return efficient

    def _merge_dict(self, dict1, dict2, dict3):
        merged_dict = {}
        # save the key and values lists
        dict1_key = list(dict1.keys())
        dict2_key = list(dict2.keys())
        dict3_key = list(dict3.keys())
        dict1_values = list(dict1.values())
        dict2_values = list(dict2.values())
        dict3_values = list(dict3.values())

        for i in range(len(dict1)):
            merged_dict[f"{dict1_key[i]}1"] = dict1_values[i]
            merged_dict[f"{dict2_key[i]}2"] = dict2_values[i]
            merged_dict[f"{dict3_key[i]}3"] = dict3_values[i]
        return merged_dict