from easylink.utilities import validation_utils
rule all:
    message: 'Grabbing final output'
    localrule: True    
    input:
        final_output=['intermediate/step_4_python_pandas/result.parquet'],
        validation='input_validations/final_validator',
    output: ['result.parquet']
    run:
        import os
        for input_path, output_path in zip(input.final_output, output):
            os.symlink(input_path, output_path)
rule:
    name: "results_main_input_validator"
    input: ['intermediate/step_4_python_pandas/result.parquet']
    output: touch("input_validations/final_validator")
    localrule: True         
    message: "Validating results input slot main_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_1_python_pandas_step_1_main_input_validator"
    input: ['{test_dir}/input_data1/file1.csv', '{test_dir}/input_data2/file2.csv']
    output: touch("input_validations/step_1_python_pandas/step_1_main_input_validator")
    localrule: True         
    message: "Validating step_1_python_pandas input slot step_1_main_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_1_python_pandas"
    message: "Running step_1 implementation: step_1_python_pandas"
    input:
        dummy_container_main_input_file_paths=['{test_dir}/input_data1/file1.csv', '{test_dir}/input_data2/file2.csv'],
        validations=['input_validations/step_1_python_pandas/step_1_main_input_validator'],
    output: ['intermediate/step_1_python_pandas/result.parquet']
    log: "diagnostics/step_1_python_pandas/step_1_python_pandas-output.log"
    container: "/mnt/team/simulation_science/priv/engineering/er_ecosystem/images/python_pandas.sif"
    shell:
        '''
        export DUMMY_CONTAINER_OUTPUT_PATHS=intermediate/step_1_python_pandas/result.parquet
        export DUMMY_CONTAINER_DIAGNOSTICS_DIRECTORY=diagnostics/step_1_python_pandas
        export DUMMY_CONTAINER_MAIN_INPUT_FILE_PATHS={test_dir}/input_data1/file1.csv,{test_dir}/input_data2/file2.csv
        python /dummy_step.py > {log} 2>&1
        '''
rule:
    name: "step_2_python_pandas_step_2_main_input_validator"
    input: ['intermediate/step_1_python_pandas/result.parquet']
    output: touch("input_validations/step_2_python_pandas/step_2_main_input_validator")
    localrule: True         
    message: "Validating step_2_python_pandas input slot step_2_main_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_2_python_pandas"
    message: "Running step_2 implementation: step_2_python_pandas"
    input:
        dummy_container_main_input_file_paths=['intermediate/step_1_python_pandas/result.parquet'],
        validations=['input_validations/step_2_python_pandas/step_2_main_input_validator'],
    output: ['intermediate/step_2_python_pandas/result.parquet']
    log: "diagnostics/step_2_python_pandas/step_2_python_pandas-output.log"
    container: "/mnt/team/simulation_science/priv/engineering/er_ecosystem/images/python_pandas.sif"
    shell:
        '''
        export DUMMY_CONTAINER_OUTPUT_PATHS=intermediate/step_2_python_pandas/result.parquet
        export DUMMY_CONTAINER_DIAGNOSTICS_DIRECTORY=diagnostics/step_2_python_pandas
        export DUMMY_CONTAINER_MAIN_INPUT_FILE_PATHS=intermediate/step_1_python_pandas/result.parquet
        python /dummy_step.py > {log} 2>&1
        '''
rule:
    name: "step_3_python_pandas_step_3_main_input_validator"
    input: ['intermediate/step_2_python_pandas/result.parquet']
    output: touch("input_validations/step_3_python_pandas/step_3_main_input_validator")
    localrule: True         
    message: "Validating step_3_python_pandas input slot step_3_main_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_3_python_pandas"
    message: "Running step_3 implementation: step_3_python_pandas"
    input:
        dummy_container_main_input_file_paths=['intermediate/step_2_python_pandas/result.parquet'],
        validations=['input_validations/step_3_python_pandas/step_3_main_input_validator'],
    output: ['intermediate/step_3_python_pandas/result.parquet']
    log: "diagnostics/step_3_python_pandas/step_3_python_pandas-output.log"
    container: "/mnt/team/simulation_science/priv/engineering/er_ecosystem/images/python_pandas.sif"
    shell:
        '''
        export DUMMY_CONTAINER_OUTPUT_PATHS=intermediate/step_3_python_pandas/result.parquet
        export DUMMY_CONTAINER_DIAGNOSTICS_DIRECTORY=diagnostics/step_3_python_pandas
        export DUMMY_CONTAINER_MAIN_INPUT_FILE_PATHS=intermediate/step_2_python_pandas/result.parquet
        python /dummy_step.py > {log} 2>&1
        '''
rule:
    name: "step_4_python_pandas_step_4_secondary_input_validator"
    input: ['{test_dir}/input_data1/file1.csv', '{test_dir}/input_data2/file2.csv']
    output: touch("input_validations/step_4_python_pandas/step_4_secondary_input_validator")
    localrule: True         
    message: "Validating step_4_python_pandas input slot step_4_secondary_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_4_python_pandas_step_4_main_input_validator"
    input: ['intermediate/step_3_python_pandas/result.parquet']
    output: touch("input_validations/step_4_python_pandas/step_4_main_input_validator")
    localrule: True         
    message: "Validating step_4_python_pandas input slot step_4_main_input"
    run:
        for f in input:
            validation_utils.validate_input_file_dummy(f)
rule:
    name: "step_4_python_pandas"
    message: "Running step_4 implementation: step_4_python_pandas"
    input:
        dummy_container_secondary_input_file_paths=['{test_dir}/input_data1/file1.csv', '{test_dir}/input_data2/file2.csv'],
        dummy_container_main_input_file_paths=['intermediate/step_3_python_pandas/result.parquet'],
        validations=['input_validations/step_4_python_pandas/step_4_secondary_input_validator', 'input_validations/step_4_python_pandas/step_4_main_input_validator'],
    output: ['intermediate/step_4_python_pandas/result.parquet']
    log: "diagnostics/step_4_python_pandas/step_4_python_pandas-output.log"
    container: "/mnt/team/simulation_science/priv/engineering/er_ecosystem/images/python_pandas.sif"
    shell:
        '''
        export DUMMY_CONTAINER_OUTPUT_PATHS=intermediate/step_4_python_pandas/result.parquet
        export DUMMY_CONTAINER_DIAGNOSTICS_DIRECTORY=diagnostics/step_4_python_pandas
        export DUMMY_CONTAINER_SECONDARY_INPUT_FILE_PATHS={test_dir}/input_data1/file1.csv,{test_dir}/input_data2/file2.csv
        export DUMMY_CONTAINER_MAIN_INPUT_FILE_PATHS=intermediate/step_3_python_pandas/result.parquet
        export INPUT_ENV_VARS=DUMMY_CONTAINER_MAIN_INPUT_FILE_PATHS,DUMMY_CONTAINER_SECONDARY_INPUT_FILE_PATHS
        python /dummy_step.py > {log} 2>&1
        '''