ReporterAdapter

ReporterAdapter handles result output and report generation, wrapping the Reporter module to provide a unified pipeline interface.

Class Architecture

graph TB
    subgraph ReporterAdapter["ReporterAdapter"]
        RA[ReporterAdapter<br/>Wrapper Class]
        Config[ReporterConfig<br/>Configuration Class]
    end

    subgraph Reporter["Reporter Module"]
        Reporter[Reporter<br/>Factory Class]
        SaveData[ReporterSaveData]
        SaveReport[ReporterSaveReport]
        SaveTiming[ReporterSaveTiming]
        SaveValidation[ReporterSaveValidation]
    end

    RA --> Config
    RA ..> Reporter
    Reporter --> SaveData
    Reporter --> SaveReport
    Reporter --> SaveTiming
    Reporter --> SaveValidation

    style RA fill:#e6d5f0,stroke:#9966cc,stroke-width:2px
    style Config fill:#ffe6f0,stroke:#ff66b2,stroke-width:1px
    style Reporter fill:#d5e6ff,stroke:#4a90e2,stroke-width:2px
    style SaveData fill:#d5f0e6,stroke:#66cc99,stroke-width:1px
    style SaveReport fill:#d5f0e6,stroke:#66cc99,stroke-width:1px
    style SaveTiming fill:#d5f0e6,stroke:#66cc99,stroke-width:1px
    style SaveValidation fill:#d5f0e6,stroke:#66cc99,stroke-width:1px

Legend:

• Light purple box: ReporterAdapter main class
• Blue box: Reporter factory class
• Green boxes: Concrete reporter implementations
• Light pink box: Configuration class
• -->: Ownership relationship
• ..>: Dependency relationship

Main Features

• Unified report generation interface
• Support for multiple report methods (data saving, evaluation reports, timing information)
• Automatic filename generation with naming strategies
• Pipeline state integration for automatic report data collection
• Multi-granularity report generation support

Method Reference

__init__(config: dict)

Initialize ReporterAdapter instance.

Parameters:

• config: dict, required
  • Configuration parameter dictionary
  • Must include method key
  • Different methods require different additional parameters:
    • save_data: Requires source parameter
    • save_report: Requires granularity parameter
    • save_timing: Optional time_unit and module parameters
    • save_validation: No additional required parameters
  • Optional parameters:
    • output: Output filename prefix (default: ‘petsard’)
    • naming_strategy: Naming strategy (’traditional’ or ‘compact’)

run(input: dict)

Execute report generation operation.

Parameters:

• input: dict, required
  • Data collected from pipeline state
  • Usually set automatically by set_input() method
  • Contains experiment results to report

Returns: No direct return value. Report files are saved directly to disk.

set_input(status)

Automatically collect data to report from pipeline state.

Parameters:

• status: Status object
  • Pipeline execution state
  • Contains execution results from all modules

Returns:

• dict: Data dictionary prepared for reporting

get_result()

Get result data from report generation.

Returns:

• dict | pd.DataFrame | None: Processed report data
  • save_data: DataFrame dictionary
  • save_report: Granularity result dictionary
  • save_timing: Timing information DataFrame
  • save_validation: Validation result dictionary

Usage Examples

Save Synthetic Data

from petsard.adapter import ReporterAdapter

# Save synthetic data
adapter = ReporterAdapter({
    "method": "save_data",
    "source": "Synthesizer",
    "output": "my_experiment"
})

# Set input from state
input_data = adapter.set_input(status)

# Execute report generation
adapter.run(input_data)
# Output: my_experiment_Synthesizer[exp1].csv

Generate Evaluation Report (Single Granularity)

from petsard.adapter import ReporterAdapter

# Generate global evaluation report
adapter = ReporterAdapter({
    "method": "save_report",
    "granularity": "global",
    "output": "evaluation_results"
})

# Execute
input_data = adapter.set_input(status)
adapter.run(input_data)
# Output: evaluation_results[Report]_eval1_[global].csv

Generate Evaluation Report (Multiple Granularities)

from petsard.adapter import ReporterAdapter

# Generate multi-granularity evaluation report
adapter = ReporterAdapter({
    "method": "save_report",
    "granularity": ["global", "columnwise", "details"],
    "naming_strategy": "compact"
})

input_data = adapter.set_input(status)
adapter.run(input_data)
# Output:
# - petsard_eval1_global.csv
# - petsard_eval1_columnwise.csv
# - petsard_eval1_details.csv

Save Timing Information

from petsard.adapter import ReporterAdapter

# Save timing information for specific modules
adapter = ReporterAdapter({
    "method": "save_timing",
    "time_unit": "minutes",
    "module": ["Loader", "Synthesizer"],
    "output": "timing_analysis"
})

input_data = adapter.set_input(status)
adapter.run(input_data)
# Output: timing_analysis_timing_report.csv

Use Compact Naming Strategy

from petsard.adapter import ReporterAdapter

# Use compact naming
adapter = ReporterAdapter({
    "method": "save_data",
    "source": ["Synthesizer"],
    "naming_strategy": "compact",
    "output": "results"
})

input_data = adapter.set_input(status)
adapter.run(input_data)
# Output: results_Synthesizer_exp1.csv (more concise filename)

Configuration Parameter Details

Common Parameters

save_data Method Specific

save_report Method Specific

save_timing Method Specific

Workflow

1. Initialize: Create ReporterAdapter and configure settings
2. Collect Data: set_input() collects relevant data from pipeline state
3. Process Data: Reporter internally processes and formats data
4. Generate Report: Write data to CSV files
5. Return Result: Provide processed data via get_result()

Naming Strategies

Traditional Strategy (Default)

Maintains backward compatibility with detailed naming format:

• petsard_Synthesizer[exp1].csv
• petsard[Report]_eval1_[global].csv

Compact Strategy

Uses simplified naming for better readability:

• petsard_Synthesizer_exp1.csv
• petsard_eval1_global.csv

Pipeline Integration

ReporterAdapter typically serves as the final stage of the pipeline, collecting and outputting results:

from petsard.config import Config
from petsard.executor import Executor

config_dict = {
    "Loader": {
        "load_data": {"filepath": "data.csv"}
    },
    "Synthesizer": {
        "synth": {"method": "default"}
    },
    "Evaluator": {
        "eval": {"method": "default"}
    },
    "Reporter": {
        # Save synthetic data
        "save_synthetic": {
            "method": "save_data",
            "source": "Synthesizer"
        },
        # Generate evaluation report
        "save_evaluation": {
            "method": "save_report",
            "granularity": ["global", "columnwise"]
        },
        # Save timing information
        "save_timing": {
            "method": "save_timing",
            "time_unit": "minutes"
        }
    }
}

config = Config(config_dict)
executor = Executor(config)
executor.run()

Notes

• This is an internal API; direct usage is not recommended
• Use YAML configuration files and Executor instead
• Report files are saved in the current working directory
• Files with the same name will be overwritten
• Multi-granularity reports generate multiple files
• set_input() automatically collects data from pipeline state
• CSV files use UTF-8 encoding

Related Documentation

• Reporter Python API: Detailed Reporter module description
• Reporter YAML: YAML configuration explanation
• Adapter Overview: Overview of all Adapters