utils/block_var_setter.py

import json

def update_scratch_project_data(project_data):
    """
    Updates variable and broadcast definitions in a Scratch project JSON,
    populating the 'variables' and 'broadcasts' sections of the Stage target
    and extracting initial values for variables.

    Args:
        project_data (dict): The loaded JSON data of the Scratch project.

    Returns:
        dict: The updated project JSON data.
    """

    stage_target = None
    for target in project_data['targets']:
        if target.get('isStage'):
            stage_target = target
            break

    if stage_target is None:
        print("Error: Stage target not found in the project data.")
        return project_data

    # Ensure 'variables' and 'broadcasts' exist in the Stage target
    if "variables" not in stage_target:
        stage_target["variables"] = {}
    if "broadcasts" not in stage_target:
        stage_target["broadcasts"] = {}

    # Helper function to recursively find and update variable/broadcast fields
    def process_dict(obj):
        if isinstance(obj, dict):
            # Check for "data_setvariableto" opcode to extract initial values
            if obj.get("opcode") == "data_setvariableto":
                variable_field = obj.get("fields", {}).get("VARIABLE")
                value_input = obj.get("inputs", {}).get("VALUE")

                if variable_field and isinstance(variable_field, list) and len(variable_field) == 2:
                    var_name = variable_field[0]
                    var_id = variable_field[1]

                    initial_value = ""
                    if value_input and isinstance(value_input, list) and len(value_input) > 1 and \
                       isinstance(value_input[1], list) and len(value_input[1]) > 1:
                        # Extract value from various formats, e.g., [1, [10, "0"]] or [3, [12, "score", "id"], [10, "0"]]
                        if value_input[1][0] == 10: # Direct value like [10, "0"]
                            initial_value = str(value_input[1][1])
                        elif value_input[1][0] == 12 and len(value_input) > 2 and isinstance(value_input[2], list) and value_input[2][0] == 10: # Variable reference with initial value block
                             initial_value = str(value_input[2][1])
                        elif isinstance(value_input[1], (str, int, float)): # For direct number/string inputs
                            initial_value = str(value_input[1])


                    # Add/update the variable in the Stage's 'variables' with its initial value
                    stage_target["variables"][var_id] = [var_name, initial_value]


            for key, value in obj.items():
                # Process variable definitions in 'fields' (for blocks that define variables like 'show variable')
                if key == "VARIABLE" and isinstance(value, list) and len(value) == 2:
                    var_name = value[0]
                    var_id = value[1]
                    # Only add if not already defined with an initial value from set_variableto
                    if var_id not in stage_target["variables"]:
                        stage_target["variables"][var_id] = [var_name, ""] # Default to empty string if no initial value found yet
                    elif stage_target["variables"][var_id][0] != var_name: # Update name if ID exists but name is different
                        stage_target["variables"][var_id][0] = var_name


                # Process broadcast definitions in 'inputs' (BROADCAST_INPUT)
                elif key == "BROADCAST_INPUT" and isinstance(value, list) and len(value) == 2 and \
                     isinstance(value[1], list) and len(value[1]) == 3 and value[1][0] == 11:
                    broadcast_name = value[1][1]
                    broadcast_id = value[1][2]
                    # Add/update the broadcast in the Stage's 'broadcasts'
                    stage_target["broadcasts"][broadcast_id] = broadcast_name

                # Process broadcast definitions in 'fields' (BROADCAST_OPTION)
                elif key == "BROADCAST_OPTION" and isinstance(value, list) and len(value) == 2:
                    broadcast_name = value[0]
                    broadcast_id = value[1]
                    # Add/update the broadcast in the Stage's 'broadcasts'
                    stage_target["broadcasts"][broadcast_id] = broadcast_name

                # Recursively call for nested dictionaries or lists
                process_dict(value)
        elif isinstance(obj, list):
            for i, item in enumerate(obj):
                # Process variable references in 'inputs' (like [12, "score", "id"])
                if isinstance(item, list) and len(item) == 3 and item[0] == 12:
                    var_name = item[1]
                    var_id = item[2]
                    # Only add if not already defined with an initial value from set_variableto
                    if var_id not in stage_target["variables"]:
                        stage_target["variables"][var_id] = [var_name, ""] # Default to empty string if no initial value found yet
                    elif stage_target["variables"][var_id][0] != var_name: # Update name if ID exists but name is different
                        stage_target["variables"][var_id][0] = var_name

                process_dict(item)

    # Iterate through all targets to process their blocks
    for target in project_data['targets']:
        if "blocks" in target:
            for block_id, block_data in target["blocks"].items():
                process_dict(block_data)

    return project_data

def deduplicate_variables(project_data):
    """
    Removes duplicate variable entries in the 'variables' dictionary of the Stage target,
    prioritizing entries with non-empty values.

    Args:
        project_data (dict): The loaded JSON data of the Scratch project.

    Returns:
        dict: The updated project JSON data with deduplicated variables.
    """

    stage_target = None
    for target in project_data['targets']:
        if target.get('isStage'):
            stage_target = target
            break

    if stage_target is None:
        print("Error: Stage target not found in the project data.")
        return project_data

    if "variables" not in stage_target:
        return project_data # No variables to deduplicate

    # Use a temporary dictionary to store the preferred variable entry by name
    # Format: {variable_name: [variable_id, variable_name, variable_value]}
    resolved_variables = {}

    for var_id, var_info in stage_target["variables"].items():
        var_name = var_info[0]
        var_value = var_info[1]

        if var_name not in resolved_variables:
            # If the variable name is not yet seen, add it
            resolved_variables[var_name] = [var_id, var_name, var_value]
        else:
            # If the variable name is already seen, decide which one to keep
            existing_id, existing_name, existing_value = resolved_variables[var_name]

            # Prioritize the entry with a non-empty value
            if var_value != "" and existing_value == "":
                resolved_variables[var_name] = [var_id, var_name, var_value]
            # If both have non-empty values, or both are empty, keep the current one (arbitrary choice, but consistent)
            # The current logic will effectively keep the last one encountered that has a value,
            # or the very last one if all are empty.
            elif var_value != "" and existing_value != "":
                 # If there are multiple non-empty values for the same variable name
                 # this keeps the one from the most recent iteration.
                 # For the given example, this will correctly keep "5".
                resolved_variables[var_name] = [var_id, var_name, var_value]
            elif var_value == "" and existing_value == "":
                # If both are empty, just keep the current one (arbitrary)
                resolved_variables[var_name] = [var_id, var_name, var_value]


    # Reconstruct the 'variables' dictionary using the resolved entries
    new_variables_dict = {}
    for var_name, var_data in resolved_variables.items():
        var_id_to_keep = var_data[0]
        var_name_to_keep = var_data[1]
        var_value_to_keep = var_data[2]
        new_variables_dict[var_id_to_keep] = [var_name_to_keep, var_value_to_keep]

    stage_target["variables"] = new_variables_dict

    return project_data

# Example usage with your provided JSON:
if __name__ == "__main__":
    json_data = {
        "targets": [
            {"isStage": True, "name": "Stage", "variables": {},"lists": {}, "broadcasts": {}, "blocks": {
        "event_whenflagclicked_1": {
            "opcode": "event_whenflagclicked",
            "inputs": {},
            "fields": {},
            "shadow": False,
            "topLevel": True,
            "parent": None,
            "next": "data_setvariableto_1"
        },
        "data_setvariableto_1": {
            "opcode": "data_setvariableto",
            "inputs": {
                "VALUE": [
                    1,
                    [
                        4,
                        "0"
                    ]
                ]
            },
            "fields": {
                "VARIABLE": [
                    "score",
                    "$ELgBAKpb[&l+DX$EMK4"
                ]
            },
            "shadow": False,
            "topLevel": False,
            "parent": "event_whenflagclicked_1",
            "next": "data_setvariableto_2"
        },
        "data_setvariableto_2": {
            "opcode": "data_setvariableto",
            "inputs": {
                "VALUE": [
                    1,
                    [
                        4,
                        "3"
                    ]
                ]
            },
            "fields": {
                "VARIABLE": [
                    "lives",
                    "Gb]1jA+H{h_6z1^Fn!-a"
                ]
            },
            "shadow": False,
            "topLevel": False,
            "parent": "data_setvariableto_1",
            "next": "data_showvariable_1"
        },
        "data_showvariable_1": {
            "opcode": "data_showvariable",
            "inputs": {},
            "fields": {
                "VARIABLE": [
                    "score",
                    "$ELgBAKpb[&l+DX$EMK4"
                ]
            },
            "shadow": False,
            "topLevel": False,
            "parent": "data_setvariableto_2",
            "next": "data_showvariable_2"
        },
        "data_showvariable_2": {
            "opcode": "data_showvariable",
            "inputs": {},
            "fields": {
                "VARIABLE": [
                    "lives",
                    "Gb]1jA+H{h_6z1^Fn!-a"
                ]
            },
            "shadow": False,
            "topLevel": False,
            "parent": "data_showvariable_1",
            "next": "event_broadcast_1"
        },
        "event_broadcast_1": {
            "opcode": "event_broadcast",
            "inputs": {
                "BROADCAST_INPUT": [
                    1,
                    [
                        11,
                        "Game Start",
                        "hN1d@^S}e)H~8(qp)rGN"
                    ]
                ]
            },
            "fields": {},
            "shadow": False,
            "topLevel": False,
            "parent": "data_showvariable_2",
            "next": None
        }
    }, "comments": {}, "currentCostume": 1, "costumes": [{"name": "backdrop1", "dataFormat": "svg", "assetId": "cd21514d0531fdffb22204e0ec5ed84a", "md5ext": "cd21514d0531fdffb22204e0ec5ed84a.svg", "rotationCenterX": 240, "rotationCenterY": 180}, {"name": "Blue Sky", "bitmapResolution": 1, "dataFormat": "svg", "assetId": "e7c147730f19d284bcd7b3f00af19bb6", "rotationCenterX": 240, "rotationCenterY": 180}], "sounds": [{"name": "pop", "assetId": "83a9787d4cb6f3b7632b4ddfebf74367", "dataFormat": "wav", "format": "", "rate": 48000, "sampleCount": 1123, "md5ext": "83a9787d4cb6f3b7632b4ddfebf74367.wav"}], "volume": 100, "layerOrder": 0, "tempo": 60, "videoTransparency": 50, "videoState": "on", "textToSpeechLanguage": None},
            {"isStage": False, "name": "Sprite1", "variables": {}, "lists": {}, "broadcasts": {}, "blocks": {}, "comments": {}, "currentCostume": 0, "costumes": [{"name": "costume1", "bitmapResolution": 1, "dataFormat": "svg", "assetId": "bcf454acf82e4504149f7ffe07081dbc", "md5ext": "bcf454acf82e4504149f7ffe07081dbc.svg", "rotationCenterX": 48, "rotationCenterY": 50}, {"name": "costume2", "bitmapResolution": 1, "dataFormat": "svg", "assetId": "0fb9be3e8397c983338cb71dc84d0b25", "md5ext": "0fb9be3e8397c983338cb71dc84d0b25.svg", "rotationCenterX": 46, "rotationCenterY": 53}], "sounds": [{"name": "Meow", "assetId": "83c36d806dc92327b9e7049a565c6bff", "dataFormat": "wav", "format": "", "rate": 48000, "sampleCount": 40681, "md5ext": "83c36d806dc92327b9e7049a565c6bff.wav"}], "volume": 100, "layerOrder": 1, "visible": True, "x": 0, "y": -120, "size": 100, "direction": 90, "draggable": False, "rotationStyle": "all around"},
            {"isStage": False, "name": "Soccer Ball", "variables": {}, "lists": {}, "broadcasts": {}, "blocks": {}, "comments": {}, "currentCostume": 0, "costumes": [{"name": "soccer ball", "bitmapResolution": 1, "dataFormat": "svg", "assetId": "5d973d7a3a8be3f3bd6e1cd0f73c32b5", "md5ext": "5d973d7a3a8be3f3bd6e1cd0f73c32b5.svg", "rotationCenterX": 23, "rotationCenterY": 22}], "sounds": [{"name": "basketball bounce", "assetId": "1727f65b5f22d151685b8e5917456a60", "dataFormat": "wav", "format": "adpcm", "rate": 22050, "sampleCount": 8129, "md5ext": "1727f65b5f22d151685b8e5917456a60.wav"}], "volume": 100, "layerOrder": 2, "visible": True, "x": -130, "y": -60, "size": 100, "direction": 90, "draggable": False, "rotationStyle": "all around"}], "monitors": [{"id": "76*2udMupx@8!=)9Uqvj", "mode": "default", "opcode": "data_variable", "params": {"VARIABLE": "score"}, "spriteName": None, "value": "0", "width": 0, "height": 0, "x": 5, "y": 5, "visible": True, "sliderMin": 0, "sliderMax": 100, "isDiscrete": True}, {"id": "YiV;2%+lgjnJ$|*Jy.{H", "mode": "default", "opcode": "data_variable", "params": {"VARIABLE": "cloud Var"}, "spriteName": None, "value": 0, "width": 0, "height": 0, "x": 5, "y": 32, "visible": True, "sliderMin": 0, "sliderMax": 100, "isDiscrete": True}, {"id": "@D/}fdt{0XaJ`kRE0t~F", "mode": "default", "opcode": "data_variable", "params": {"VARIABLE": "lives"}, "spriteName": None, "value": "3", "width": 0, "height": 0, "x": 5, "y": 59, "visible": True, "sliderMin": 0, "sliderMax": 100, "isDiscrete": True}, {"id": "^R,uz7yRjtK`=uP~6SN4", "mode": "default", "opcode": "data_variable", "params": {"VARIABLE": "speed"}, "spriteName": None, "value": "5", "width": 0, "height": 0, "x": 5, "y": 86, "visible": True, "sliderMin": 0, "sliderMax": 100, "isDiscrete": True}], "extensions": [], "meta": {"semver": "3.0.0", "vm": "11.3.0", "agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36"}}


    updated_json_data = update_scratch_project_data(json_data)
    updated_json_data = deduplicate_variables(json_data)
    print(json.dumps(updated_json_data, indent=1))