biz-bud/docker/db-init/000-base-tables.sql

-- Base receipt processing schema
-- This file must run first (prefixed with 000-)

-- Create sequences first
CREATE SEQUENCE IF NOT EXISTS receipts_id_seq;
CREATE SEQUENCE IF NOT EXISTS receipt_line_items_id_seq;

-- Create base receipts table
CREATE TABLE IF NOT EXISTS rpt_receipts (
    id numeric DEFAULT nextval('receipts_id_seq'::regclass) NOT NULL,
    vendor_name varchar(255) NOT NULL,
    vendor_address text NULL,
    transaction_date date NULL,
    transaction_time time NULL,
    receipt_number varchar(100) NULL,
    customer_info text NULL,
    subtotal numeric(10, 2) NULL,
    tax_amount numeric(10, 2) NULL,
    final_total numeric(10, 2) NULL,
    total_items int4 NULL,
    payment_method varchar(50) NULL,
    card_last_four bpchar(4) NULL,
    card_type varchar(20) NULL,
    raw_receipt_text text NULL,
    created_at timestamptz DEFAULT now() NULL,
    updated_at timestamptz DEFAULT now() NULL,
    CONSTRAINT receipts_pkey PRIMARY KEY (id)
);

-- Create base indexes
CREATE INDEX IF NOT EXISTS idx_rpt_receipts_payment_method ON rpt_receipts USING btree (payment_method);
CREATE INDEX IF NOT EXISTS idx_rpt_receipts_transaction_date ON rpt_receipts USING btree (transaction_date);
CREATE INDEX IF NOT EXISTS idx_rpt_receipts_vendor_name ON rpt_receipts USING btree (vendor_name);

-- Create master products table
CREATE TABLE IF NOT EXISTS rpt_master_products (
    id serial4 NOT NULL,
    canonical_name varchar(500) NOT NULL,
    canonical_description text NULL,
    category varchar(100) NULL,
    unit_of_measure varchar(50) NULL,
    estimated_unit_price numeric(10, 2) NULL,
    is_active bool DEFAULT true NULL,
    first_seen_date timestamptz DEFAULT now() NULL,
    last_updated timestamptz DEFAULT now() NULL,
    total_occurrences int4 DEFAULT 1 NULL,
    CONSTRAINT rpt_master_products_pkey PRIMARY KEY (id)
);

CREATE INDEX IF NOT EXISTS idx_rpt_master_products_canonical_name ON rpt_master_products USING btree (canonical_name);
CREATE INDEX IF NOT EXISTS idx_rpt_master_products_category ON rpt_master_products USING btree (category);

-- Create product variations table
CREATE TABLE IF NOT EXISTS rpt_product_variations (
    id serial4 NOT NULL,
    master_product_id int4 NOT NULL,
    original_description varchar(500) NOT NULL,
    verified_description varchar(500) NULL,
    confidence_score numeric(3, 2) NULL,
    verification_source varchar(100) NULL,
    search_variations_used text[] NULL,
    successful_variation varchar(500) NULL,
    verification_notes text NULL,
    created_at timestamptz DEFAULT now() NULL,
    occurrence_count int4 DEFAULT 1 NULL,
    CONSTRAINT rpt_product_variations_pkey PRIMARY KEY (id),
    CONSTRAINT rpt_product_variations_master_product_fkey FOREIGN KEY (master_product_id) REFERENCES rpt_master_products(id)
);

CREATE INDEX IF NOT EXISTS idx_rpt_product_variations_master_id ON rpt_product_variations USING btree (master_product_id);
CREATE INDEX IF NOT EXISTS idx_rpt_product_variations_original ON rpt_product_variations USING btree (original_description);

-- Create receipt line items table
CREATE TABLE IF NOT EXISTS rpt_receipt_line_items (
    id int4 DEFAULT nextval('receipt_line_items_id_seq'::regclass) NOT NULL,
    receipt_id numeric NOT NULL,
    line_number int4 NULL,
    product_name varchar(500) NOT NULL,
    product_code varchar(100) NULL,
    quantity numeric(10, 3) NULL,
    unit_of_measure varchar(50) NULL,
    unit_price numeric(10, 2) NULL,
    total_price numeric(10, 2) NULL,
    category varchar(100) NULL,
    created_at timestamptz DEFAULT now() NULL,
    master_product_id int4 NULL,
    product_variation_id int4 NULL,
    original_ocr_text varchar(500) NULL,
    reconciliation_status varchar(50) DEFAULT 'pending'::character varying NULL,
    reconciliation_confidence numeric(3, 2) NULL,
    needs_review bool DEFAULT false NULL,
    CONSTRAINT receipt_line_items_pkey PRIMARY KEY (id),
    CONSTRAINT rpt_receipt_line_items_master_product_fkey FOREIGN KEY (master_product_id) REFERENCES rpt_master_products(id),
    CONSTRAINT rpt_receipt_line_items_receipt_id_fkey FOREIGN KEY (receipt_id) REFERENCES rpt_receipts(id),
    CONSTRAINT rpt_receipt_line_items_variation_fkey FOREIGN KEY (product_variation_id) REFERENCES rpt_product_variations(id)
);

CREATE INDEX IF NOT EXISTS idx_rpt_receipt_line_items_master_product ON rpt_receipt_line_items USING btree (master_product_id);
CREATE INDEX IF NOT EXISTS idx_rpt_receipt_line_items_receipt_id ON rpt_receipt_line_items USING btree (receipt_id);
CREATE INDEX IF NOT EXISTS idx_rpt_receipt_line_items_reconciliation_status ON rpt_receipt_line_items USING btree (reconciliation_status);

-- Create reconciliation log table
CREATE TABLE IF NOT EXISTS rpt_reconciliation_log (
    id serial4 NOT NULL,
    receipt_line_item_id int4 NOT NULL,
    original_description varchar(500) NULL,
    suggested_master_product_id int4 NULL,
    action_taken varchar(50) NULL,
    confidence_score numeric(3, 2) NULL,
    reconciled_by varchar(100) NULL,
    reconciled_at timestamptz DEFAULT now() NULL,
    notes text NULL,
    CONSTRAINT rpt_reconciliation_log_pkey PRIMARY KEY (id),
    CONSTRAINT rpt_reconciliation_log_line_item_fkey FOREIGN KEY (receipt_line_item_id) REFERENCES rpt_receipt_line_items(id),
    CONSTRAINT rpt_reconciliation_log_master_product_fkey FOREIGN KEY (suggested_master_product_id) REFERENCES rpt_master_products(id)
);

CREATE INDEX IF NOT EXISTS idx_rpt_reconciliation_log_receipt_item ON rpt_reconciliation_log USING btree (receipt_line_item_id);