#!/usr/bin/env python3
"""
Validation script for new MIT-licensed dataset transformers.
Checks that all new transformer methods exist and are callable.
"""

from warbler_cda.utils.hf_warbler_ingest import HFWarblerIngestor
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).parent))


def main():
    print("\n" + "=" * 60)
    print("Validating New MIT-Licensed Dataset Transformers")
    print("=" * 60)

    try:
        ingestor = HFWarblerIngestor()
        print("✓ HFWarblerIngestor initialized successfully")
    except Exception as e:
        print(f"✗ Failed to initialize HFWarblerIngestor: {e}")
        return False

    required_transformers = [
        "transform_arxiv",
        "transform_prompt_report",
        "transform_novels",
        "transform_manuals",
        "transform_enterprise",
        "transform_portuguese_education",
    ]

    required_helpers = [
        "_create_arxiv_content",
        "_create_prompt_report_content",
        "_create_novel_content",
        "_create_manual_content",
        "_create_enterprise_content",
        "_create_portuguese_content",
        "_chunk_text",
    ]

    all_good = True

    print("\n📋 Checking Transformer Methods:")
    for transformer_name in required_transformers:
        if hasattr(ingestor, transformer_name):
            method = getattr(ingestor, transformer_name)
            if callable(method):
                print(f"  ✓ {transformer_name}")
            else:
                print(f"  ✗ {transformer_name} (exists but not callable)")
                all_good = False
        else:
            print(f"  ✗ {transformer_name} (missing)")
            all_good = False

    print("\n🔧 Checking Helper Methods:")
    for helper_name in required_helpers:
        if hasattr(ingestor, helper_name):
            method = getattr(ingestor, helper_name)
            if callable(method):
                print(f"  ✓ {helper_name}")
            else:
                print(f"  ✗ {helper_name} (exists but not callable)")
                all_good = False
        else:
            print(f"  ✗ {helper_name} (missing)")
            all_good = False

    print("\n📊 Testing Basic Transformer Structure:")

    test_doc = {
        "content_id": "test/001",
        "content": "Test content for validation",
        "metadata": {
            "pack": "test-pack",
            "source_dataset": "test/dataset",
            "license": "MIT",
            "realm_type": "test",
            "realm_label": "test",
            "lifecycle_stage": "emergence",
            "activity_level": 0.5,
            "dialogue_type": "test",
        },
    }

    required_keys = ["content_id", "content", "metadata"]
    required_metadata_keys = ["pack", "source_dataset", "license", "realm_type", "realm_label"]

    for key in required_keys:
        if key in test_doc:
            print(f"  ✓ Document has '{key}'")
        else:
            print(f"  ✗ Document missing '{key}'")
            all_good = False

    for key in required_metadata_keys:
        if key in test_doc["metadata"]:
            print(f"  ✓ Metadata has '{key}'")
        else:
            print(f"  ✗ Metadata missing '{key}'")
            all_good = False

    print("\n🔍 Testing _chunk_text Helper:")
    try:
        test_text = "word " * 3000
        chunks = ingestor._chunk_text(test_text, chunk_size=100)
        print(f"  ✓ Successfully chunked text into {len(chunks)} chunks")
        if all(isinstance(chunk, str) for chunk in chunks):
            print(f"  ✓ All chunks are strings")
        else:
            print(f"  ✗ Some chunks are not strings")
            all_good = False
    except Exception as e:
        print(f"  ✗ _chunk_text failed: {e}")
        all_good = False

    print("\n" + "=" * 60)
    if all_good:
        print("✅ ALL VALIDATIONS PASSED")
        print("=" * 60)
        return True
    else:
        print("❌ SOME VALIDATIONS FAILED")
        print("=" * 60)
        return False


if __name__ == "__main__":
    success = main()
    sys.exit(0 if success else 1)