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

package processing

import (
	"context"
	"fmt"

	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
	"github.com/descrybe/descrybe-v2/apps/api/internal/catalog"
	"github.com/google/uuid"
	"github.com/jackc/pgx/v5/pgxpool"
)

// FixCompanyCatalogOpts controls optional Fix A1 / catalog hygiene steps.
type FixCompanyCatalogOpts struct {
	// BackfillCategories copies mapped_data category unique_ids onto empty
	// processed_products.category.
	BackfillCategories bool
	// ReprocessSampleLimit caps recommended raw_product_ids in the response
	// (0 = counts only, no sample ids).
	ReprocessSampleLimit int
	// AIPrompts optionally re-applies BuiltInDefaults product_enhance per content language.
	AIPrompts *aiprompts.Service
}

// FixCompanyCatalogResult is the idempotent admin Fix A1 report.
type FixCompanyCatalogResult struct {
	CompanyID   uuid.UUID `json:"company_id"`
	CompanyName string    `json:"company_name"`
	A1Cohort    bool      `json:"a1_cohort"`

	CategoryAttributeOrphansRemoved int `json:"category_attribute_orphans_removed"`
	CategoryAttributeLinks          int `json:"category_attribute_links"`

	CategoryPromptsUpdated   int `json:"category_prompts_updated"`
	CategoryPromptsAlreadyOK int `json:"category_prompts_already_ok"`
	CategoryPromptsEmpty     int `json:"category_prompts_empty_skipped"`
	// Prompts aliases category_prompts_updated for flash.admin.fixA1Success {prompts}.
	Prompts int `json:"prompts"`

	ProductEnhanceLanguages int `json:"product_enhance_languages"`

	WeakHashesCleared    int `json:"weak_hashes_cleared"`
	CategoriesBackfilled int `json:"categories_backfilled"`
	// MappedCategoriesBackfilled is processed → mapped_data.category copies.
	MappedCategoriesBackfilled int `json:"mapped_categories_backfilled"`
	// Coverage after Fix (Products UI reads mapped_data.category for raw rows).
	TaxonomyCategories    int64 `json:"taxonomy_categories"`
	MappedWithCategory    int64 `json:"mapped_with_category"`
	MappedWithoutCategory int64 `json:"mapped_without_category"`
	ProcessedWithCategory    int64 `json:"processed_with_category"`
	ProcessedWithoutCategory int64 `json:"processed_without_category"`
	// Hashes / Categories alias the long names for flash.admin.fixA1Success.
	Hashes     int `json:"hashes"`
	Categories int `json:"categories"`
	// MappedBackfilled aliases mapped_categories_backfilled for flash {mapped_backfilled}.
	MappedBackfilled int `json:"mapped_backfilled"`

	DescriptionsNormalized int   `json:"descriptions_normalized"`
	DescriptionsBackfilled int   `json:"descriptions_backfilled"`
	MappedDescriptionsFlat int   `json:"mapped_descriptions_flattened"`
	MetaBackfilled         int   `json:"meta_backfilled"`
	AttributesSanitized    int64 `json:"attributes_sanitized"`
	ProductsScanned        int   `json:"products_scanned"`

	ReprocessNeededCount         int         `json:"reprocess_needed_count"`
	ReprocessSampleRawProductIDs []uuid.UUID `json:"reprocess_sample_raw_product_ids,omitempty"`
}

// FixCompanyCatalog runs the admin Fix A1 / catalog hygiene sequence in place.
// Never clears the catalog or uses A1 as a clone destination.
// Step 2 applies RepairCompanyCategoryEnhancePrompts (same repaired map as
// RepairA1DemoCategoryEnhancePrompts / cmd/repair-category-prompts -apply).
// Legacy converters + weak-hash clear live in FixCatalogHygieneWithIDs.
func FixCompanyCatalog(ctx context.Context, pool *pgxpool.Pool, companyID uuid.UUID, companyName string, a1Cohort bool, opts FixCompanyCatalogOpts) (FixCompanyCatalogResult, error) {
	out := FixCompanyCatalogResult{
		CompanyID:   companyID,
		CompanyName: companyName,
		A1Cohort:    a1Cohort,
	}
	if pool == nil {
		return out, fmt.Errorf("nil pool")
	}

	orphans, links, err := catalog.EnsureCategoryAttributeLinks(ctx, pool, companyID)
	if err != nil {
		return out, err
	}
	out.CategoryAttributeOrphansRemoved = orphans
	out.CategoryAttributeLinks = links

	updated, alreadyOK, emptySkipped, err := catalog.RepairCompanyCategoryEnhancePrompts(ctx, pool, companyID)
	if err != nil {
		return out, err
	}
	out.CategoryPromptsUpdated = updated
	out.CategoryPromptsAlreadyOK = alreadyOK
	out.CategoryPromptsEmpty = emptySkipped

	if opts.AIPrompts != nil {
		n, err := opts.AIPrompts.ApplyBuiltInProductEnhance(ctx, companyID)
		if err != nil {
			return out, fmt.Errorf("product_enhance templates: %w", err)
		}
		out.ProductEnhanceLanguages = n
	}

	hygiene, clearedIDs, err := FixCatalogHygieneWithIDs(ctx, pool, companyID, opts.BackfillCategories)
	if err != nil {
		return out, err
	}
	out.WeakHashesCleared = hygiene.WeakHashesCleared
	out.CategoriesBackfilled = hygiene.CategoriesBackfilled
	out.DescriptionsNormalized = hygiene.DescriptionsNormalized
	out.DescriptionsBackfilled = hygiene.DescriptionsBackfilled
	out.MappedDescriptionsFlat = hygiene.MappedDescriptionsFlat
	out.MetaBackfilled = hygiene.MetaBackfilled
	out.ProductsScanned = hygiene.ProductsScanned

	if opts.BackfillCategories {
		mappedN, err := BackfillMappedCategoriesFromProcessed(ctx, pool, companyID)
		if err != nil {
			return out, fmt.Errorf("mapped category backfill: %w", err)
		}
		out.MappedCategoriesBackfilled = int(mappedN)
	}

	sanitized, err := BackfillCompanyProcessedAttributes(ctx, pool, companyID)
	if err != nil {
		return out, fmt.Errorf("sanitize attributes: %w", err)
	}
	out.AttributesSanitized = sanitized

	needed, sample, err := RecommendReprocessRawProductIDs(ctx, pool, companyID, clearedIDs, opts.ReprocessSampleLimit)
	if err != nil {
		return out, err
	}
	out.ReprocessNeededCount = needed
	out.ReprocessSampleRawProductIDs = sample

	mWith, mWithout, taxonomy, err := CountMappedCategoryCoverage(ctx, pool, companyID)
	if err != nil {
		return out, err
	}
	out.MappedWithCategory = mWith
	out.MappedWithoutCategory = mWithout
	out.TaxonomyCategories = taxonomy
	pWith, pWithout, err := CountProcessedCategoryCoverage(ctx, pool, companyID)
	if err != nil {
		return out, err
	}
	out.ProcessedWithCategory = pWith
	out.ProcessedWithoutCategory = pWithout

	// Short keys for flash.admin.fixA1Success {prompts,hashes,categories,mapped_backfilled}.
	out.Prompts = out.CategoryPromptsUpdated
	out.Hashes = out.WeakHashesCleared
	out.Categories = out.CategoriesBackfilled
	out.MappedBackfilled = out.MappedCategoriesBackfilled
	return out, nil
}