-- ================================================
-- Repository Analyzer Memory System Database Migration
-- Version: 1.0
-- Description: Complete database setup for AI memory system
-- ================================================

-- Enable required extensions
CREATE EXTENSION IF NOT EXISTS vector;
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";

-- ================================================
-- CORE TABLES
-- ================================================

-- Code embeddings table for semantic search of analyzed code
CREATE TABLE IF NOT EXISTS code_embeddings (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    repo_id VARCHAR(255) NOT NULL,
    file_path TEXT NOT NULL,
    content_hash VARCHAR(64) NOT NULL,
    embedding vector(384) NOT NULL,
    metadata JSONB DEFAULT '{}',
    created_at TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    last_accessed TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    access_count INTEGER DEFAULT 0,
    
    -- Ensure uniqueness per repo/file/hash combination
    CONSTRAINT unique_code_analysis UNIQUE(repo_id, file_path, content_hash)
);

-- Query embeddings for episodic memory (user interactions)
CREATE TABLE IF NOT EXISTS query_embeddings (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    session_id VARCHAR(255) NOT NULL,
    query_text TEXT NOT NULL,
    query_embedding vector(384) NOT NULL,
    response_embedding vector(384),
    repo_context VARCHAR(255),
    timestamp TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    metadata JSONB DEFAULT '{}',
    
    -- Index for session-based queries
    CONSTRAINT valid_session_id CHECK (LENGTH(session_id) > 0)
);

-- Persistent knowledge embeddings for long-term learning
CREATE TABLE IF NOT EXISTS knowledge_embeddings (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    fact_id VARCHAR(255) UNIQUE NOT NULL,
    content TEXT NOT NULL,
    category VARCHAR(100) NOT NULL,
    embedding vector(384) NOT NULL,
    confidence REAL DEFAULT 1.0 CHECK (confidence >= 0.0 AND confidence <= 1.0),
    source_repos TEXT[] DEFAULT '{}',
    created_at TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    last_accessed TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    access_frequency INTEGER DEFAULT 0,
    
    -- Ensure valid categories
    CONSTRAINT valid_category CHECK (category IN ('code_pattern', 'best_practice', 'vulnerability', 'architecture', 'security_vulnerability', 'performance'))
);

-- Repository metadata for tracking analyzed repositories
CREATE TABLE IF NOT EXISTS repository_metadata (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    repo_id VARCHAR(255) UNIQUE NOT NULL,
    repo_path TEXT NOT NULL,
    repo_name VARCHAR(500),
    primary_language VARCHAR(100),
    total_files INTEGER DEFAULT 0,
    total_lines INTEGER DEFAULT 0,
    last_analyzed TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    analysis_count INTEGER DEFAULT 0,
    quality_score REAL DEFAULT 5.0 CHECK (quality_score >= 0.0 AND quality_score <= 10.0),
    metadata JSONB DEFAULT '{}'
);

-- Session tracking for episodic memory correlation
CREATE TABLE IF NOT EXISTS analysis_sessions (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    session_id VARCHAR(255) UNIQUE NOT NULL,
    user_identifier VARCHAR(255),
    start_time TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    end_time TIMESTAMP WITH TIME ZONE,
    total_queries INTEGER DEFAULT 0,
    repositories_analyzed TEXT[] DEFAULT '{}',
    session_metadata JSONB DEFAULT '{}'
);

-- File analysis history for change tracking
CREATE TABLE IF NOT EXISTS file_analysis_history (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    repo_id VARCHAR(255) NOT NULL,
    file_path TEXT NOT NULL,
    content_hash VARCHAR(64) NOT NULL,
    language VARCHAR(100),
    lines_of_code INTEGER DEFAULT 0,
    complexity_score REAL DEFAULT 0.0,
    severity_score REAL DEFAULT 5.0 CHECK (severity_score >= 0.0 AND severity_score <= 10.0),
    issues_count INTEGER DEFAULT 0,
    analyzed_at TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    analysis_version VARCHAR(50) DEFAULT '1.0'
);

-- Memory consolidation log for tracking knowledge extraction
CREATE TABLE IF NOT EXISTS memory_consolidation_log (
    id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
    source_type VARCHAR(50) NOT NULL, -- 'episodic', 'code_analysis', 'manual'
    source_id VARCHAR(255) NOT NULL,
    target_memory_type VARCHAR(50) NOT NULL, -- 'persistent', 'working'
    target_id VARCHAR(255),
    consolidation_confidence REAL DEFAULT 0.5,
    consolidation_timestamp TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    consolidation_metadata JSONB DEFAULT '{}'
);

-- ================================================
-- PERFORMANCE INDEXES
-- ================================================

-- Code embeddings indexes
CREATE INDEX IF NOT EXISTS idx_code_embeddings_repo_id ON code_embeddings(repo_id);
CREATE INDEX IF NOT EXISTS idx_code_embeddings_file_path ON code_embeddings(file_path);
CREATE INDEX IF NOT EXISTS idx_code_embeddings_accessed ON code_embeddings(last_accessed DESC);
CREATE INDEX IF NOT EXISTS idx_code_embeddings_metadata ON code_embeddings USING gin(metadata);

-- Vector similarity indexes (using IVFFlat for better performance)
CREATE INDEX IF NOT EXISTS idx_code_embeddings_vector 
ON code_embeddings USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100);

-- Query embeddings indexes
CREATE INDEX IF NOT EXISTS idx_query_embeddings_session ON query_embeddings(session_id);
CREATE INDEX IF NOT EXISTS idx_query_embeddings_timestamp ON query_embeddings(timestamp DESC);
CREATE INDEX IF NOT EXISTS idx_query_embeddings_repo_context ON query_embeddings(repo_context);
CREATE INDEX IF NOT EXISTS idx_query_embeddings_vector 
ON query_embeddings USING ivfflat (query_embedding vector_cosine_ops) WITH (lists = 100);

-- Knowledge embeddings indexes
CREATE INDEX IF NOT EXISTS idx_knowledge_embeddings_category ON knowledge_embeddings(category);
CREATE INDEX IF NOT EXISTS idx_knowledge_embeddings_confidence ON knowledge_embeddings(confidence DESC);
CREATE INDEX IF NOT EXISTS idx_knowledge_embeddings_access_freq ON knowledge_embeddings(access_frequency DESC);
CREATE INDEX IF NOT EXISTS idx_knowledge_embeddings_vector 
ON knowledge_embeddings USING ivfflat (embedding vector_cosine_ops) WITH (lists = 100);
CREATE INDEX IF NOT EXISTS idx_knowledge_source_repos ON knowledge_embeddings USING gin(source_repos);

-- Repository metadata indexes
CREATE INDEX IF NOT EXISTS idx_repository_metadata_repo_id ON repository_metadata(repo_id);
CREATE INDEX IF NOT EXISTS idx_repository_metadata_analyzed ON repository_metadata(last_analyzed DESC);
CREATE INDEX IF NOT EXISTS idx_repository_metadata_language ON repository_metadata(primary_language);

-- File history indexes
CREATE INDEX IF NOT EXISTS idx_file_history_repo_file ON file_analysis_history(repo_id, file_path);
CREATE INDEX IF NOT EXISTS idx_file_history_analyzed ON file_analysis_history(analyzed_at DESC);
CREATE INDEX IF NOT EXISTS idx_file_history_severity ON file_analysis_history(severity_score);

-- ================================================
-- MATERIALIZED VIEWS FOR COMMON QUERIES
-- ================================================

-- High confidence knowledge view
CREATE MATERIALIZED VIEW IF NOT EXISTS high_confidence_knowledge AS
SELECT 
    fact_id,
    content,
    category,
    confidence,
    source_repos,
    created_at,
    last_accessed,
    access_frequency
FROM knowledge_embeddings 
WHERE confidence > 0.8
ORDER BY confidence DESC, access_frequency DESC;

CREATE INDEX ON high_confidence_knowledge (category);
CREATE INDEX ON high_confidence_knowledge (confidence DESC);

-- Repository quality summary view
CREATE MATERIALIZED VIEW IF NOT EXISTS repository_quality_summary AS
SELECT 
    rm.repo_id,
    rm.repo_path,
    rm.repo_name,
    rm.primary_language,
    rm.total_files,
    rm.total_lines,
    rm.quality_score,
    rm.last_analyzed,
    COUNT(ce.id) as total_embeddings,
    AVG(fah.severity_score) as avg_file_quality,
    COUNT(DISTINCT fah.file_path) as analyzed_files_count
FROM repository_metadata rm
LEFT JOIN code_embeddings ce ON rm.repo_id = ce.repo_id
LEFT JOIN file_analysis_history fah ON rm.repo_id = fah.repo_id
GROUP BY rm.repo_id, rm.repo_path, rm.repo_name, rm.primary_language, 
         rm.total_files, rm.total_lines, rm.quality_score, rm.last_analyzed;

CREATE INDEX ON repository_quality_summary (quality_score DESC);
CREATE INDEX ON repository_quality_summary (last_analyzed DESC);

-- Recent activity view
CREATE MATERIALIZED VIEW IF NOT EXISTS recent_activity AS
SELECT 
    'query' as activity_type,
    session_id as identifier,
    query_text as description,
    timestamp as activity_time,
    repo_context
FROM query_embeddings
WHERE timestamp >= CURRENT_TIMESTAMP - INTERVAL '7 days'
UNION ALL
SELECT 
    'analysis' as activity_type,
    repo_id as identifier,
    file_path as description,
    analyzed_at as activity_time,
    repo_id as repo_context
FROM file_analysis_history
WHERE analyzed_at >= CURRENT_TIMESTAMP - INTERVAL '7 days'
ORDER BY activity_time DESC;

CREATE INDEX ON recent_activity (activity_time DESC);
CREATE INDEX ON recent_activity (activity_type);

-- ================================================
-- STORED FUNCTIONS AND PROCEDURES
-- ================================================

-- Function to refresh all materialized views
CREATE OR REPLACE FUNCTION refresh_memory_views()
RETURNS void AS $$
BEGIN
    REFRESH MATERIALIZED VIEW CONCURRENTLY high_confidence_knowledge;
    REFRESH MATERIALIZED VIEW CONCURRENTLY repository_quality_summary;
    REFRESH MATERIALIZED VIEW CONCURRENTLY recent_activity;
    
    -- Log the refresh
    INSERT INTO memory_consolidation_log (
        source_type, source_id, target_memory_type, target_id, 
        consolidation_confidence, consolidation_metadata
    ) VALUES (
        'system', 'materialized_views', 'system', 'view_refresh',
        1.0, '{"refresh_time": "' || CURRENT_TIMESTAMP || '"}'::jsonb
    );
END;
$$ LANGUAGE plpgsql;

-- Function to calculate semantic similarity between texts
CREATE OR REPLACE FUNCTION calculate_similarity(embedding1 vector(384), embedding2 vector(384))
RETURNS real AS $$
BEGIN
    RETURN 1 - (embedding1 <=> embedding2);
END;
$$ LANGUAGE plpgsql IMMUTABLE STRICT;

-- Function to update access patterns
CREATE OR REPLACE FUNCTION update_access_pattern(table_name text, id_column text, id_value text)
RETURNS void AS $$
BEGIN
    CASE table_name
        WHEN 'knowledge_embeddings' THEN
            EXECUTE 'UPDATE knowledge_embeddings SET last_accessed = CURRENT_TIMESTAMP, access_frequency = access_frequency + 1 WHERE fact_id = $1'
            USING id_value;
        WHEN 'code_embeddings' THEN
            EXECUTE 'UPDATE code_embeddings SET last_accessed = CURRENT_TIMESTAMP, access_count = access_count + 1 WHERE id = $1::uuid'
            USING id_value;
        ELSE
            RAISE EXCEPTION 'Unsupported table: %', table_name;
    END CASE;
END;
$$ LANGUAGE plpgsql;

-- Function to cleanup old memories
CREATE OR REPLACE FUNCTION cleanup_old_memories(retention_days integer DEFAULT 365)
RETURNS integer AS $$
DECLARE
    deleted_count integer := 0;
    cutoff_date timestamp;
BEGIN
    cutoff_date := CURRENT_TIMESTAMP - (retention_days || ' days')::interval;
    
    -- Delete old query embeddings (episodic memories)
    DELETE FROM query_embeddings WHERE timestamp < cutoff_date;
    GET DIAGNOSTICS deleted_count = ROW_COUNT;
    
    -- Update knowledge confidence based on access patterns
    UPDATE knowledge_embeddings 
    SET confidence = LEAST(confidence * (
        CASE 
            WHEN EXTRACT(EPOCH FROM (CURRENT_TIMESTAMP - last_accessed)) / 86400 < 30 
            THEN 1.05 
            ELSE 0.98 
        END * 
        (1.0 + LOG(access_frequency + 1) / 20.0)
    ), 1.0);
    
    -- Log cleanup activity
    INSERT INTO memory_consolidation_log (
        source_type, source_id, target_memory_type, target_id, 
        consolidation_confidence, consolidation_metadata
    ) VALUES (
        'system', 'cleanup_function', 'system', 'memory_cleanup',
        1.0, ('{"deleted_records": ' || deleted_count || ', "cutoff_date": "' || cutoff_date || '"}')::jsonb
    );
    
    RETURN deleted_count;
END;
$$ LANGUAGE plpgsql;

-- Function to find similar code patterns
CREATE OR REPLACE FUNCTION find_similar_code(
    query_embedding vector(384),
    repo_filter text DEFAULT NULL,
    similarity_threshold real DEFAULT 0.7,
    max_results integer DEFAULT 10
)
RETURNS TABLE (
    id uuid,
    repo_id varchar(255),
    file_path text,
    similarity real,
    metadata jsonb
) AS $$
BEGIN
    RETURN QUERY
    SELECT 
        ce.id,
        ce.repo_id,
        ce.file_path,
        (1 - (ce.embedding <=> query_embedding))::real as similarity,
        ce.metadata
    FROM code_embeddings ce
    WHERE (repo_filter IS NULL OR ce.repo_id = repo_filter)
      AND (1 - (ce.embedding <=> query_embedding)) > similarity_threshold
    ORDER BY similarity DESC
    LIMIT max_results;
END;
$$ LANGUAGE plpgsql;

-- Function to get knowledge by category
CREATE OR REPLACE FUNCTION get_knowledge_by_category(
    category_filter varchar(100),
    min_confidence real DEFAULT 0.5,
    max_results integer DEFAULT 20
)
RETURNS TABLE (
    fact_id varchar(255),
    content text,
    confidence real,
    access_frequency integer,
    source_repos text[]
) AS $$
BEGIN
    RETURN QUERY
    SELECT 
        ke.fact_id,
        ke.content,
        ke.confidence,
        ke.access_frequency,
        ke.source_repos
    FROM knowledge_embeddings ke
    WHERE ke.category = category_filter
      AND ke.confidence >= min_confidence
    ORDER BY ke.confidence DESC, ke.access_frequency DESC
    LIMIT max_results;
END;
$$ LANGUAGE plpgsql;

-- ================================================
-- TRIGGERS FOR AUTOMATIC MAINTENANCE
-- ================================================

-- Trigger function to update repository metadata when embeddings are added
CREATE OR REPLACE FUNCTION update_repository_stats()
RETURNS trigger AS $$
BEGIN
    -- Update or insert repository metadata
    INSERT INTO repository_metadata (repo_id, repo_path, analysis_count, last_analyzed)
    VALUES (NEW.repo_id, NEW.repo_id, 1, CURRENT_TIMESTAMP)
    ON CONFLICT (repo_id) 
    DO UPDATE SET 
        analysis_count = repository_metadata.analysis_count + 1,
        last_analyzed = CURRENT_TIMESTAMP;
    
    RETURN NEW;
END;
$$ LANGUAGE plpgsql;

-- Create triggers
DROP TRIGGER IF EXISTS trigger_update_repo_stats ON code_embeddings;
CREATE TRIGGER trigger_update_repo_stats
    AFTER INSERT ON code_embeddings
    FOR EACH ROW
    EXECUTE FUNCTION update_repository_stats();

-- Trigger to automatically update access patterns
CREATE OR REPLACE FUNCTION auto_update_access()
RETURNS trigger AS $$
BEGIN
    NEW.last_accessed = CURRENT_TIMESTAMP;
    NEW.access_count = COALESCE(OLD.access_count, 0) + 1;
    RETURN NEW;
END;
$$ LANGUAGE plpgsql;

DROP TRIGGER IF EXISTS trigger_auto_access_update ON code_embeddings;
CREATE TRIGGER trigger_auto_access_update
    BEFORE UPDATE ON code_embeddings
    FOR EACH ROW
    EXECUTE FUNCTION auto_update_access();

-- ================================================
-- SECURITY AND PERMISSIONS
-- ================================================

-- Create roles for different access levels
DO $$
BEGIN
    IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = 'repo_analyzer_read') THEN
        CREATE ROLE repo_analyzer_read;
    END IF;
    
    IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = 'repo_analyzer_write') THEN
        CREATE ROLE repo_analyzer_write;
    END IF;
    
    IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = 'repo_analyzer_admin') THEN
        CREATE ROLE repo_analyzer_admin;
    END IF;
END
$$;

-- Grant permissions
GRANT SELECT ON ALL TABLES IN SCHEMA public TO repo_analyzer_read;
GRANT SELECT ON high_confidence_knowledge TO repo_analyzer_read;
GRANT SELECT ON repository_quality_summary TO repo_analyzer_read;
GRANT SELECT ON recent_activity TO repo_analyzer_read;

GRANT SELECT, INSERT, UPDATE ON ALL TABLES IN SCHEMA public TO repo_analyzer_write;
GRANT SELECT ON high_confidence_knowledge TO repo_analyzer_write;
GRANT SELECT ON repository_quality_summary TO repo_analyzer_write;
GRANT SELECT ON recent_activity TO repo_analyzer_write;
GRANT USAGE, SELECT ON ALL SEQUENCES IN SCHEMA public TO repo_analyzer_write;

GRANT ALL PRIVILEGES ON ALL TABLES IN SCHEMA public TO repo_analyzer_admin;
GRANT ALL PRIVILEGES ON high_confidence_knowledge TO repo_analyzer_admin;
GRANT ALL PRIVILEGES ON repository_quality_summary TO repo_analyzer_admin;
GRANT ALL PRIVILEGES ON recent_activity TO repo_analyzer_admin;
GRANT ALL PRIVILEGES ON ALL SEQUENCES IN SCHEMA public TO repo_analyzer_admin;
GRANT EXECUTE ON ALL FUNCTIONS IN SCHEMA public TO repo_analyzer_admin;

-- ================================================
-- DATA VALIDATION AND CONSTRAINTS
-- ================================================

-- Add check constraints for data quality
-- Note: Vector dimensions are validated at insertion time, no need for runtime checks

-- Add constraints for reasonable data ranges
DO $$
BEGIN
    IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'reasonable_lines_of_code') THEN
        ALTER TABLE file_analysis_history ADD CONSTRAINT reasonable_lines_of_code 
            CHECK (lines_of_code >= 0 AND lines_of_code <= 1000000);
    END IF;
    
    IF NOT EXISTS (SELECT 1 FROM pg_constraint WHERE conname = 'reasonable_complexity') THEN
        ALTER TABLE file_analysis_history ADD CONSTRAINT reasonable_complexity 
            CHECK (complexity_score >= 0.0 AND complexity_score <= 100.0);
    END IF;
END
$$;

-- ================================================
-- INITIAL DATA AND CONFIGURATION
-- ================================================

-- Insert initial system configuration
INSERT INTO memory_consolidation_log (
    source_type, source_id, target_memory_type, target_id, 
    consolidation_confidence, consolidation_metadata
) VALUES (
    'system', 'database_migration', 'system', 'initial_setup',
    1.0, ('{"migration_version": "1.0", "setup_time": "' || CURRENT_TIMESTAMP || '"}')::jsonb
) ON CONFLICT DO NOTHING;

-- Create initial knowledge categories
INSERT INTO knowledge_embeddings (
    fact_id, content, category, embedding, confidence, source_repos
) VALUES
(
    'init_security_001',
    'Always validate and sanitize user input to prevent injection attacks',
    'security_vulnerability',
    array_fill(0.0, ARRAY[384])::vector(384),
    0.95,
    ARRAY[]::text[]
),
(
    'init_performance_001', 
    'Use appropriate data structures and algorithms for better performance',
    'performance',
    array_fill(0.0, ARRAY[384])::vector(384),
    0.9,
    ARRAY[]::text[]
),
(
    'init_best_practice_001',
    'Follow consistent naming conventions and code formatting standards',
    'best_practice', 
    array_fill(0.0, ARRAY[384])::vector(384),
    0.85,
    ARRAY[]::text[]
)
ON CONFLICT (fact_id) DO NOTHING;

-- ================================================
-- BACKUP AND MAINTENANCE PROCEDURES
-- ================================================

-- Function to create backup of critical memory data
CREATE OR REPLACE FUNCTION backup_memory_data(backup_path text DEFAULT '/tmp/memory_backup')
RETURNS text AS $$
DECLARE
    backup_file text;
    result_message text;
BEGIN
    backup_file := backup_path || '_' || to_char(CURRENT_TIMESTAMP, 'YYYY-MM-DD_HH24-MI-SS') || '.sql';
    
    -- This would need to be implemented with actual backup logic
    -- For now, just return the intended backup file name
    result_message := 'Backup would be created at: ' || backup_file;
    
    -- Log backup activity
    INSERT INTO memory_consolidation_log (
        source_type, source_id, target_memory_type, target_id, 
        consolidation_confidence, consolidation_metadata
    ) VALUES (
        'system', 'backup_function', 'system', 'backup_created',
        1.0, ('{"backup_file": "' || backup_file || '"}')::jsonb
    );
    
    RETURN result_message;
END;
$$ LANGUAGE plpgsql;

-- ================================================
-- MONITORING AND ANALYTICS
-- ================================================

-- View for system health monitoring
CREATE OR REPLACE VIEW system_health_monitor AS
SELECT 
    'code_embeddings' as table_name,
    COUNT(*) as record_count,
    MAX(created_at) as latest_record,
    AVG(access_count) as avg_access_count
FROM code_embeddings
UNION ALL
SELECT 
    'query_embeddings' as table_name,
    COUNT(*) as record_count,
    MAX(timestamp) as latest_record,
    NULL as avg_access_count
FROM query_embeddings
UNION ALL
SELECT 
    'knowledge_embeddings' as table_name,
    COUNT(*) as record_count,
    MAX(created_at) as latest_record,
    AVG(access_frequency) as avg_access_count
FROM knowledge_embeddings;

-- Function to get comprehensive system statistics
CREATE OR REPLACE FUNCTION get_system_statistics()
RETURNS jsonb AS $$
DECLARE
    stats jsonb;
BEGIN
    SELECT jsonb_build_object(
        'total_code_embeddings', (SELECT COUNT(*) FROM code_embeddings),
        'total_query_embeddings', (SELECT COUNT(*) FROM query_embeddings),
        'total_knowledge_embeddings', (SELECT COUNT(*) FROM knowledge_embeddings),
        'unique_repositories', (SELECT COUNT(DISTINCT repo_id) FROM code_embeddings),
        'high_confidence_knowledge', (SELECT COUNT(*) FROM knowledge_embeddings WHERE confidence > 0.8),
        'recent_activity_7d', (SELECT COUNT(*) FROM query_embeddings WHERE timestamp >= CURRENT_TIMESTAMP - INTERVAL '7 days'),
        'average_code_quality', (SELECT AVG(quality_score) FROM repository_metadata),
        'last_updated', CURRENT_TIMESTAMP
    ) INTO stats;
    
    RETURN stats;
END;
$$ LANGUAGE plpgsql;

-- ================================================
-- COMPLETION MESSAGE
-- ================================================

DO $$
BEGIN
    RAISE NOTICE '================================================';
    RAISE NOTICE 'Repository Analyzer Memory System Database Setup Complete';
    RAISE NOTICE '================================================';
    RAISE NOTICE 'Tables created: code_embeddings, query_embeddings, knowledge_embeddings';
    RAISE NOTICE 'Indexes created: Vector similarity indexes with IVFFlat';
    RAISE NOTICE 'Functions created: Similarity search, cleanup, statistics';
    RAISE NOTICE 'Materialized views created: High confidence knowledge, repository summary';
    RAISE NOTICE 'Triggers created: Auto-update repository stats and access patterns';
    RAISE NOTICE '================================================';
    RAISE NOTICE 'Ready for AI-enhanced repository analysis with persistent memory';
    RAISE NOTICE '================================================';
END
$$;