"""
Comprehensive integration tests for Labeling Pipeline.
"""

import pytest
import asyncio
import json
import tempfile
import os
from unittest.mock import AsyncMock, MagicMock, patch

from labeling_pipeline import (
    LabelingPipeline, LabelingPipelineRunner,
    OnChainVerifier, NewsCrossReferenceVerifier, MarketDataVerifier,
    FactVerificationEngine
)
from sentiment_engine.schemas.processed import SentimentScores, EmotionScores, EventType


class TestLabelingPipeline:
    """Tests for LabelingPipeline"""

    @pytest.fixture
    def pipeline(self):
        return LabelingPipeline()

    @pytest.mark.asyncio
    async def test_label_text_returns_complete_structure(self, pipeline):
        """label_text should return complete labeled structure"""
        result = await pipeline.label_text("Bitcoin surges to $100k!")
        
        assert "labels" in result
        assert "confidence" in result
        assert "verified" in result
        assert "verification_details" in result
        assert "labeled_at" in result
        
        assert result["labels"]["sentiment"] in ["Bearish", "Bullish", "Neutral"]
        assert result["labels"]["event_type"] in [e.value for e in EventType]

    @pytest.mark.asyncio
    async def test_label_text_with_context(self, pipeline):
        """label_text should use context"""
        result = await pipeline.label_text(
            "Bitcoin surges!",
            context={"source_id": "coindesk", "source_type": "news", "source_credibility": 0.9}
        )
        
        assert result["confidence"]["verification"] > 0.0

    @pytest.mark.asyncio
    async def test_label_batch(self, pipeline):
        """Should label multiple texts"""
        texts = [
            "Bitcoin surges to $100k!",
            "Major hack on exchange!",
            "SEC approves ETF!"
        ]
        
        results = await pipeline.label_batch(texts)
        
        assert len(results) == 3
        assert all("labels" in r for r in results)

    def test_label_text_sync(self, pipeline):
        """Synchronous label_text should work"""
        result = pipeline.label_text_sync("Bitcoin surges!")
        
        assert "labels" in result


class TestOnChainVerifier:
    """Tests for OnChainVerifier"""

    @pytest.fixture
    def verifier(self):
        return OnChainVerifier()

    @pytest.mark.asyncio
    async def test_verify_transaction(self, verifier):
        """Should verify transaction exists"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={"status": "confirmed"})
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.verify("tx_hash_123")
            
            assert result["verified"] is True

    @pytest.mark.asyncio
    async def test_verify_address_balance(self, verifier):
        """Should verify address balance"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={"balance": "1000000000000000000"})
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.verify_address("0x123...", min_balance=1e18)
            
            assert result["verified"] is True

    @pytest.mark.asyncio
    async def test_verify_contract_interaction(self, verifier):
        """Should verify contract interaction"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={"logs": [{"address": "0x..."}]})
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.verify_contract_interaction("0x123...", "0x456...")
            
            assert result["verified"] is True


class TestNewsCrossReferenceVerifier:
    """Tests for NewsCrossReferenceVerifier"""

    @pytest.fixture
    def verifier(self):
        return NewsCrossReferenceVerifier()

    @pytest.mark.asyncio
    async def test_cross_reference_single_source(self, verifier):
        """Should find matching articles"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={
                "articles": [
                    {"title": "Bitcoin surges", "source": "Reuters"},
                    {"title": "BTC hits new high", "source": "Bloomberg"}
                ]
            })
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.cross_reference("Bitcoin surges to new high")
            
            assert result["verified"] is True
            assert result["matches"] >= 2

    @pytest.mark.asyncio
    async def test_cross_reference_no_matches(self, verifier):
        """Should return unverified when no matches"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={"articles": []})
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.cross_reference("Completely unique obscure news")
            
            assert result["verified"] is False

    @pytest.mark.asyncio
    async def test_cross_reference_confidence(self, verifier):
        """Should calculate confidence based on match count"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={
                "articles": [
                    {"title": "Bitcoin surges", "source": "Reuters"},
                    {"title": "BTC new high", "source": "Bloomberg"},
                    {"title": "Bitcoin record", "source": "CNBC"},
                    {"title": "Crypto surges", "source": "WSJ"},
                    {"title": "Digital gold", "source": "FT"}
                ]
            })
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.cross_reference("Bitcoin surges to new high")
            
            assert result["confidence"] > 0.8


class TestMarketDataVerifier:
    """Tests for MarketDataVerifier"""

    @pytest.fixture
    def verifier(self):
        return MarketDataVerifier()

    @pytest.mark.asyncio
    async def test_verify_price_movement(self, verifier):
        """Should verify price movement matches sentiment"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={
                "price_change_24h": 0.15,  # 15% increase
                "volume_24h": 1000000000
            })
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.verify_price_movement("BTC", "bullish")
            
            assert result["verified"] is True

    @pytest.mark.asyncio
    async def test_verify_price_movement_contradiction(self, verifier):
        """Should detect contradiction"""
        with patch('aiohttp.ClientSession.get') as mock_get:
            mock_response = AsyncMock()
            mock_response.status = 200
            mock_response.json = AsyncMock(return_value={
                "price_change_24h": -0.20,  # 20% decrease
                "volume_24h": 1000000000
            })
            mock_get.return_value.__aenter__.return_value = mock_response
            
            result = await verifier.verify_price_movement("BTC", "bullish")
            
            assert result["verified"] is False


class TestFactVerificationEngine:
    """Tests for FactVerificationEngine"""

    @pytest.fixture
    def engine(self):
        return FactVerificationEngine()

    @pytest.mark.asyncio
    async def test_verify_combines_all_sources(self, engine):
        """Should combine all verification sources"""
        text = "Bitcoin surges to $100k as institutional inflows surge"
        
        with patch.object(engine.on_chain_verifier, 'verify', new_callable=AsyncMock) as mock_onchain, \
             patch.object(engine.news_verifier, 'cross_reference', new_callable=AsyncMock) as mock_news, \
             patch.object(engine.market_verifier, 'verify_price_movement', new_callable=AsyncMock) as mock_market:
            
            mock_onchain.return_value = {"verified": True, "details": "tx confirmed"}
            mock_news.return_value = {"verified": True, "matches": 3, "confidence": 0.8}
            mock_market.return_value = {"verified": True, "price_change": 0.15}
            
            result = await engine.verify(text, "BTC", "bullish")
            
            assert result["verified"] is True
            assert result["confidence"] > 0.7
            assert len(result["evidence"]) >= 3

    @pytest.mark.asyncio
    async def test_verify_partial_sources(self, engine):
        """Should handle partial verification"""
        text = "Some obscure news"
        
        with patch.object(engine.on_chain_verifier, 'verify', new_callable=AsyncMock) as mock_onchain, \
             patch.object(engine.news_verifier, 'cross_reference', new_callable=AsyncMock) as mock_news, \
             patch.object(engine.market_verifier, 'verify_price_movement', new_callable=AsyncMock) as mock_market:
            
            mock_onchain.return_value = {"verified": False, "details": "no tx"}
            mock_news.return_value = {"verified": False, "matches": 0, "confidence": 0.0}
            mock_market.return_value = {"verified": False, "price_change": 0.0}
            
            result = await engine.verify(text, "UNKNOWN", "neutral")
            
            assert result["verified"] is False
            assert result["confidence"] < 0.3


class TestLabelingPipelineRunner:
    """Tests for LabelingPipelineRunner"""

    @pytest.fixture
    def runner(self):
        return LabelingPipelineRunner()

    @pytest.mark.asyncio
    async def test_run_on_dataset(self, runner):
        """Should process dataset file"""
        with tempfile.NamedTemporaryFile(mode='w', suffix='.jsonl', delete=False) as f:
            for i in range(5):
                f.write(json.dumps({
                    "id": f"sample_{i}",
                    "raw_text": f"Bitcoin news {i}",
                    "source_id": "test",
                    "source_type": "news",
                    "credibility": 0.8
                }) + '\n')
            input_file = f.name
        
        output_file = input_file.replace('.jsonl', '_labeled.jsonl')
        
        try:
            results = await runner.run_on_dataset(input_file, output_file)
            
            assert len(results) == 5
            assert all("labels" in r for r in results)
            assert os.path.exists(output_file)
        finally:
            os.unlink(input_file)
            if os.path.exists(output_file):
                os.unlink(output_file)

    @pytest.mark.asyncio
    async def test_run_on_dataset_tracks_stats(self, runner):
        """Should track verification stats"""
        with tempfile.NamedTemporaryFile(mode='w', suffix='.jsonl', delete=False) as f:
            f.write(json.dumps({"id": "1", "raw_text": "Bitcoin surges!", "source_id": "test", "source_type": "news", "credibility": 0.9}) + '\n')
            f.write(json.dumps({"id": "2", "raw_text": "Random text without crypto", "source_id": "test", "source_type": "news", "credibility": 0.5}) + '\n')
            input_file = f.name
        
        output_file = input_file.replace('.jsonl', '_labeled.jsonl')
        
        try:
            results = await runner.run_on_dataset(input_file, output_file)
            
            assert runner.total_count == 2
            assert runner.verified_count <= 2
        finally:
            os.unlink(input_file)
            if os.path.exists(output_file):
                os.unlink(output_file)


class TestLabelingPipelineIntegration:
    """Integration tests for labeling pipeline"""

    @pytest.mark.asyncio
    async def test_full_labeling_workflow(self):
        """Complete labeling workflow"""
        pipeline = LabelingPipeline()
        
        # Test various crypto events
        test_cases = [
            ("Bitcoin surges to $100k!", "Bullish", "earnings"),
            ("Major hack on exchange!", "Bearish", "hack"),
            ("SEC approves ETF!", "Bullish", "regulatory"),
            ("Ethereum upgrade activates!", "Neutral", "upgrade"),
            ("Whale moves 10000 BTC!", "Bullish", "whale"),
        ]
        
        for text, expected_sentiment, expected_event in test_cases:
            result = await pipeline.label_text(text)
            
            assert result["labels"]["sentiment"] == expected_sentiment
            assert result["labels"]["event_type"] == expected_event
            assert result["confidence"]["overall"] > 0.0

    @pytest.mark.asyncio
    async def test_labeling_with_verification(self):
        """Labeling with fact verification"""
        pipeline = LabelingPipeline()
        
        # High credibility source
        result = await pipeline.label_text(
            "Bitcoin surges to $100k!",
            context={"source_id": "bloomberg", "source_type": "news", "source_credibility": 0.95}
        )
        
        assert result["confidence"]["verification"] > 0.5

    @pytest.mark.asyncio
    async def test_batch_labeling_performance(self):
        """Batch labeling should be efficient"""
        pipeline = LabelingPipeline()
        
        texts = [f"Bitcoin news {i}" for i in range(20)]
        
        import time
        start = time.time()
        results = await pipeline.label_batch(texts)
        elapsed = time.time() - start
        
        assert len(results) == 20
        assert elapsed < 30  # Should complete within 30 seconds


class TestLabelingPipelineEdgeCases:
    """Edge case tests for labeling pipeline"""

    @pytest.fixture
    def pipeline(self):
        return LabelingPipeline()

    @pytest.mark.asyncio
    async def test_empty_text(self, pipeline):
        """Should handle empty text"""
        result = await pipeline.label_text("")
        
        assert result["labels"]["sentiment"] == "Neutral"

    @pytest.mark.asyncio
    async def test_very_long_text(self, pipeline):
        """Should handle very long text"""
        text = "Bitcoin surges! " * 1000
        
        result = await pipeline.label_text(text)
        
        assert result["labels"]["sentiment"] in ["Bullish", "Bearish", "Neutral"]

    @pytest.mark.asyncio
    async def test_unicode_text(self, pipeline):
        """Should handle unicode"""
        text = "Bitcoin 🚀 surges to $100k 💎"
        
        result = await pipeline.label_text(text)
        
        assert result["labels"]["sentiment"] == "Bullish"

    @pytest.mark.asyncio
    async def test_special_characters(self, pipeline):
        """Should handle special characters"""
        text = "BTC/USD: $50,000.00 (24h: +5.2%)"
        
        result = await pipeline.label_text(text)
        
        assert result["labels"]["sentiment"] in ["Bullish", "Bearish", "Neutral"]

    @pytest.mark.asyncio
    async def test_no_crypto_entities(self, pipeline):
        """Should handle text without crypto entities"""
        text = "The weather is nice today"
        
        result = await pipeline.label_text(text)
        
        assert result["labels"]["sentiment"] == "Neutral"
        assert result["labels"]["event_type"] == "listing"  # default


if __name__ == "__main__":
    pytest.main([__file__, "-v"])