descrybe/apps/api/internal/processing/category_backfill.go

package processing

import (
	"context"
	"fmt"

	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"
)

// CategoryBackfillResult is the outcome of BackfillProcessedCategoriesFromMapped.
type CategoryBackfillResult struct {
	Updated int64 `json:"updated"`
}

// BackfillProcessedCategoriesFromMapped fills empty processed_products.category from
// raw_products.mapped_data when a usable unique_id is present. When the company has
// categories rows, only known unique_ids are applied (same rule as processOne filter).
// Safe for admin "Fix catalog" and migrator reuse. Does not invent categories via vector.
func BackfillProcessedCategoriesFromMapped(ctx context.Context, pool *pgxpool.Pool, companyID uuid.UUID) (CategoryBackfillResult, error) {
	var out CategoryBackfillResult
	if pool == nil {
		return out, fmt.Errorf("category backfill: nil pool")
	}
	if companyID == uuid.Nil {
		return out, fmt.Errorf("category backfill: empty company id")
	}

	ct, err := pool.Exec(ctx, `
		WITH src AS (
			SELECT
				p.id AS processed_id,
				BTRIM(COALESCE(
					CASE
						WHEN jsonb_typeof(r.mapped_data->'category') IN ('string', 'number')
							THEN NULLIF(BTRIM(r.mapped_data->>'category'), '')
						ELSE NULL
					END,
					NULLIF(BTRIM(r.mapped_data->>'category_unique_id'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'unique_id'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'category_unique_id'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'#text'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'id'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'value'), ''),
					NULLIF(BTRIM(r.mapped_data->'category'->>'code'), '')
				)) AS cat_raw
			FROM processed_products p
			INNER JOIN raw_products r ON r.id = p.raw_product_id AND r.company_id = p.company_id
			WHERE p.company_id = $1
			  AND (
				COALESCE(NULLIF(BTRIM(p.category), ''), '') = ''
				OR LOWER(BTRIM(p.category)) = 'none'
			  )
		),
		mapped AS (
			SELECT
				processed_id,
				CASE
					WHEN cat_raw IS NULL OR cat_raw = '' OR LOWER(cat_raw) = 'none' THEN NULL
					ELSE cat_raw
				END AS cat_uid
			FROM src
		),
		eligible AS (
			SELECT m.processed_id, m.cat_uid
			FROM mapped m
			WHERE m.cat_uid IS NOT NULL
			  AND (
				NOT EXISTS (SELECT 1 FROM categories c WHERE c.company_id = $1)
				OR EXISTS (
					SELECT 1 FROM categories c
					WHERE c.company_id = $1
					  AND BTRIM(c.unique_id) = m.cat_uid
				)
			  )
		)
		UPDATE processed_products p
		SET category = e.cat_uid,
			field_sources = jsonb_set(
				COALESCE(p.field_sources, '{}'::jsonb),
				'{category}',
				'"mapped_backfill"'::jsonb,
				true
			),
			updated_at = now()
		FROM eligible e
		WHERE p.id = e.processed_id`, companyID)
	if err != nil {
		return out, fmt.Errorf("category backfill: %w", err)
	}
	out.Updated = ct.RowsAffected()
	return out, nil
}

// BackfillMappedCategoriesFromProcessed copies processed_products.category into
// raw_products.mapped_data.category when mapped category is empty. Legacy A1 dumps
// often store category only on processed rows; feed mappings do not map category,
// so Fix Catalog / re-seed must push processed → mapped for Products UI coverage.
// Does not invent categories.
func BackfillMappedCategoriesFromProcessed(ctx context.Context, pool *pgxpool.Pool, companyID uuid.UUID) (int64, error) {
	if pool == nil {
		return 0, fmt.Errorf("mapped category backfill: nil pool")
	}
	if companyID == uuid.Nil {
		return 0, fmt.Errorf("mapped category backfill: empty company id")
	}
	ct, err := pool.Exec(ctx, `
		UPDATE raw_products r
		SET mapped_data = jsonb_set(
			COALESCE(r.mapped_data, '{}'::jsonb),
			'{category}',
			to_jsonb(p.category),
			true
		),
		updated_at = now()
		FROM processed_products p
		WHERE p.raw_product_id = r.id
		  AND p.company_id = $1
		  AND r.company_id = $1
		  AND COALESCE(NULLIF(trim(p.category), ''), '') <> ''
		  AND lower(trim(p.category)) <> 'none'
		  AND COALESCE(NULLIF(trim(r.mapped_data->>'category'), ''), '') = ''`, companyID)
	if err != nil {
		return 0, fmt.Errorf("mapped category backfill: %w", err)
	}
	return ct.RowsAffected(), nil
}

// CountMappedCategoryCoverage returns how many raw_products rows have a non-empty
// mapped_data.category vs empty, plus taxonomy row count for the company.
func CountMappedCategoryCoverage(ctx context.Context, pool *pgxpool.Pool, companyID uuid.UUID) (withCat, withoutCat, taxonomy int64, err error) {
	if pool == nil {
		return 0, 0, 0, fmt.Errorf("mapped category coverage: nil pool")
	}
	err = pool.QueryRow(ctx, `
		SELECT
			COUNT(*) FILTER (
				WHERE COALESCE(NULLIF(trim(mapped_data->>'category'), ''), '') <> ''
			),
			COUNT(*) FILTER (
				WHERE COALESCE(NULLIF(trim(mapped_data->>'category'), ''), '') = ''
			)
		FROM raw_products
		WHERE company_id = $1`, companyID).Scan(&withCat, &withoutCat)
	if err != nil {
		return 0, 0, 0, fmt.Errorf("mapped category coverage: %w", err)
	}
	err = pool.QueryRow(ctx, `
		SELECT COUNT(*) FROM categories WHERE company_id = $1`, companyID).Scan(&taxonomy)
	if err != nil {
		return withCat, withoutCat, 0, fmt.Errorf("taxonomy count: %w", err)
	}
	return withCat, withoutCat, taxonomy, nil
}

// CountProcessedCategoryCoverage returns how many processed rows for companyID have
// a non-empty category vs empty/"none".
func CountProcessedCategoryCoverage(ctx context.Context, pool *pgxpool.Pool, companyID uuid.UUID) (withCat, withoutCat int64, err error) {
	if pool == nil {
		return 0, 0, fmt.Errorf("category coverage: nil pool")
	}
	err = pool.QueryRow(ctx, `
		SELECT
			COUNT(*) FILTER (
				WHERE COALESCE(NULLIF(BTRIM(category), ''), '') <> ''
				  AND LOWER(BTRIM(category)) <> 'none'
			),
			COUNT(*) FILTER (
				WHERE COALESCE(NULLIF(BTRIM(category), ''), '') = ''
				  OR LOWER(BTRIM(category)) = 'none'
			)
		FROM processed_products
		WHERE company_id = $1`, companyID).Scan(&withCat, &withoutCat)
	if err != nil {
		return 0, 0, fmt.Errorf("category coverage: %w", err)
	}
	return withCat, withoutCat, nil
}