-- Migration: Create item_classification table
-- This table stores classification data for items in the grocery_list table
-- Each row in grocery_list can have ONE corresponding classification row

CREATE TABLE IF NOT EXISTS item_classification (
  id INTEGER PRIMARY KEY REFERENCES grocery_list(id) ON DELETE CASCADE,
  item_type VARCHAR(50) NOT NULL,
  item_group VARCHAR(100) NOT NULL,
  zone VARCHAR(100),
  confidence DECIMAL(3,2) DEFAULT 1.0 CHECK (confidence >= 0 AND confidence <= 1),
  source VARCHAR(20) DEFAULT 'user' CHECK (source IN ('user', 'ml', 'default')),
  created_at TIMESTAMP DEFAULT NOW(),
  updated_at TIMESTAMP DEFAULT NOW()
);

-- Index for faster lookups by type
CREATE INDEX IF NOT EXISTS idx_item_classification_type ON item_classification(item_type);

-- Index for zone-based queries
CREATE INDEX IF NOT EXISTS idx_item_classification_zone ON item_classification(zone);

-- Comments
COMMENT ON TABLE item_classification IS 'Stores classification metadata for grocery list items';
COMMENT ON COLUMN item_classification.id IS 'Foreign key to grocery_list.id (one-to-one relationship)';
COMMENT ON COLUMN item_classification.item_type IS 'High-level category (produce, meat, dairy, etc.)';
COMMENT ON COLUMN item_classification.item_group IS 'Subcategory within item_type (filtered by type)';
COMMENT ON COLUMN item_classification.zone IS 'Store zone/location (optional)';
COMMENT ON COLUMN item_classification.confidence IS 'Confidence score 0-1 (1.0 for user-provided, lower for ML-predicted)';
COMMENT ON COLUMN item_classification.source IS 'Source of classification: user, ml, or default';