evaluation/run_evaluation.py

# -*- coding: utf-8 -*-
"""
Run Evaluation - Execute full sentiment analysis evaluation

Runs all tests and generates comprehensive reports
"""

import sys
import os
import time

# Add parent to path
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from avatar import SentimentAnalyzer, EmojiMapper
from evaluation.emotion_test_suite import EmotionTestSuite
from evaluation.accuracy_benchmark import AccuracyBenchmark
from evaluation.live_stream_test import LiveStreamTest
from evaluation.report_generator import ReportGenerator


def run_full_evaluation():
    """Run complete evaluation and generate reports"""
    print("=" * 70)
    print("EMOJI AI AVATAR - SENTIMENT ANALYSIS EVALUATION")
    print("=" * 70)
    print()
    
    # Initialize components
    print("[1/6] Initializing components...")
    start_time = time.time()
    
    analyzer = SentimentAnalyzer()
    mapper = EmojiMapper()
    test_suite = EmotionTestSuite()
    benchmark = AccuracyBenchmark(analyzer, mapper)
    stream_test = LiveStreamTest(analyzer, mapper)
    report_gen = ReportGenerator()
    
    init_time = time.time() - start_time
    print(f"      Components initialized in {init_time:.2f}s")
    print(f"      Emotions to test: {test_suite.get_emotion_count()}")
    print(f"      Test cases: {test_suite.get_test_count()}")
    print()
    
    # Run accuracy benchmark
    print("[2/6] Running accuracy benchmark...")
    benchmark_start = time.time()
    benchmark_results = benchmark.run_benchmark(test_suite.EMOTION_TEST_DATA)
    benchmark_time = time.time() - benchmark_start
    print(f"      Benchmark completed in {benchmark_time:.2f}s")
    print(f"      Accuracy: {benchmark_results.accuracy:.1%}")
    print()
    
    # Run live stream tests
    print("[3/6] Running live stream transition tests...")
    stream_start = time.time()
    transition_results = stream_test.run_all_transitions()
    stream_time = time.time() - stream_start
    
    correct_transitions = sum(1 for r in transition_results if r.transition_correct)
    trans_accuracy = correct_transitions / len(transition_results) if transition_results else 0
    
    print(f"      Transitions completed in {stream_time:.2f}s")
    print(f"      Transition accuracy: {trans_accuracy:.1%}")
    print()
    
    # Generate console summary
    print("[4/6] Generating console summary...")
    summary = report_gen.generate_summary_report(benchmark_results)
    print()
    print(summary)
    print()
    
    # Generate markdown report
    print("[5/6] Generating markdown report...")
    md_path = report_gen.generate_markdown_report(benchmark_results, transition_results)
    print(f"      Saved to: {md_path}")
    print()
    
    # Generate JSON report
    print("[6/6] Generating JSON report...")
    json_path = report_gen.generate_json_report(benchmark_results, transition_results)
    print(f"      Saved to: {json_path}")
    print()
    
    # Print live stream report
    print(stream_test.get_transition_report(transition_results))
    print()
    
    # Final summary
    total_time = time.time() - start_time
    print("=" * 70)
    print("EVALUATION COMPLETE")
    print("=" * 70)
    print(f"Total time:           {total_time:.2f}s")
    print(f"Overall accuracy:     {benchmark_results.accuracy:.1%}")
    print(f"Transition accuracy:  {trans_accuracy:.1%}")
    print(f"Avg inference time:   {benchmark_results.avg_inference_time_ms:.2f}ms")
    print()
    print("Reports saved to:")
    print(f"  - {md_path}")
    print(f"  - {json_path}")
    print("=" * 70)
    
    # Return pass/fail for CI/CD
    passing_threshold = 0.5  # 50% minimum accuracy
    if benchmark_results.accuracy >= passing_threshold:
        print("\n✅ EVALUATION PASSED")
        return 0
    else:
        print(f"\n❌ EVALUATION FAILED (accuracy below {passing_threshold:.0%})")
        return 1


if __name__ == "__main__":
    exit_code = run_full_evaluation()
    sys.exit(exit_code)