CloudHost/WooBC
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

perf: filter by BC category before fetching instead of per-product lookups

Browse Source 
When wbc_category_filter is set, sync now pages through BC's /items
endpoint filtered server-side by itemCategoryCode (or-chained eq - BC's
classic OData rejects the `in` operator) and matches each returned item to
a WooCommerce product locally via wc_get_product_id_by_sku(), instead of
querying BC once per WooCommerce product only to discard most of them.
Confirmed against live data that BC's `in` filter isn't supported here but
chained `eq ... or ...` is.

Also reorders find_bc_item_by_sku() to try GTIN/number before Item
Reference (ReferenciasArticulo). Empirically, on this catalog Reference_No
is always identical to Item_No, so the reference lookup never matched
anything the GTIN/number checks didn't already find - it was costing an
extra BC round trip on every single product for no benefit. Kept as a
last-resort fallback both directions (SKU->item and item->SKU) in case
that doesn't hold for some future item.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Malin
2026-07-09 14:30:51 +02:00
parent 8e093f4c76
commit 0a08086272
3 changed files with 315 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

174
woo-business-central/includes/class-wbc-batch-sync.php
View File

@@ -27,6 +27,11 @@ class WBC_Batch_Sync {
*/
const BATCH_SIZE = 25;
/**
* Number of BC items fetched per page in category-filtered mode
*/
const CATEGORY_PAGE_SIZE = 100;
/**
* Option name storing the current/last run's progress
*/
@@ -38,28 +43,42 @@ class WBC_Batch_Sync {
const LOCK_TRANSIENT = 'wbc_sync_lock';
/**
* Action Scheduler hook processed per batch
* Action Scheduler hook processed per batch (WooCommerce-first mode)
*/
const ACTION_HOOK = 'wbc_process_sync_batch';
/**
* Action Scheduler hook processed per page (category-filtered, BC-first mode)
*/
const CATEGORY_ACTION_HOOK = 'wbc_process_category_page';
/**
* Action Scheduler group name
*/
const ACTION_GROUP = 'wbc_product_sync';
/**
* Register the Action Scheduler batch handler
* Register the Action Scheduler batch handlers
*/
public function register_hooks() {
add_action( self::ACTION_HOOK, array( $this, 'process_batch' ), 10, 2 );
add_action( self::CATEGORY_ACTION_HOOK, array( $this, 'process_category_page' ), 10, 2 );
}
/**
* Kick off a full catalog sync in the background
*
* Gathers WooCommerce product IDs with a SKU (ID-only query - never
* loads full product objects for the whole catalog into memory), splits
* them into batches, and enqueues one background action per batch.
* If a category filter is configured, syncs BC-first: pages through
* /items filtered server-side by itemCategoryCode and matches each item
* to a WooCommerce product locally. This avoids querying BC once per
* WooCommerce product only to discard most of them - with a filter set,
* the vast majority of products aren't in an allowed category, so doing
* the filtering on BC's side first is far fewer API calls.
*
* Without a filter, falls back to the WooCommerce-first approach (ID-only
* query, batched, never loads the whole catalog as full objects), which
* doesn't have BC-side filtering available to narrow the scope.
*
* Returns immediately; actual work happens asynchronously.
*
* @return array Result with 'success' and 'message'.
@@ -99,6 +118,13 @@ class WBC_Batch_Sync {
);
}
$category_filter = get_option( 'wbc_category_filter', '' );
$categories = array_filter( array_map( 'trim', explode( ',', $category_filter ) ) );
if ( ! empty( $categories ) ) {
return self::queue_category_filtered_sync( array_values( $categories ) );
}
// ID-only query with a meta filter, so 60k+ catalogs don't get loaded
// as full WC_Product objects just to check for a SKU.
$product_ids = wc_get_products( array(
@@ -127,6 +153,7 @@ class WBC_Batch_Sync {
update_option( self::RUN_OPTION, array(
'run_id' => $run_id,
'mode' => 'by_id',
'status' => 'running',
'total' => count( $product_ids ),
'total_batches' => count( $batches ),
@@ -166,6 +193,143 @@ class WBC_Batch_Sync {
);
}
/**
* Kick off a BC-first, category-filtered sync
*
* Pages through BC's /items endpoint filtered server-side by
* itemCategoryCode (or-chained eq - BC's classic OData rejects the `in`
* operator here) and self-chains one background action per page, since
* the total item count isn't known upfront. Each page is matched against
* WooCommerce locally (no further BC calls in the common case).
*
* @param string[] $categories Allowed BC item category codes.
* @return array Result with 'success' and 'message'.
*/
private static function queue_category_filtered_sync( array $categories ) {
$run_id = uniqid( 'wbc_', true );
update_option( self::RUN_OPTION, array(
'run_id' => $run_id,
'mode' => 'category',
'status' => 'running',
'categories' => $categories,
'skip' => 0,
'pages_done' => 0,
'total' => 0,
'success' => 0,
'failed' => 0,
'skipped' => 0,
'started_at' => current_time( 'mysql' ),
'finished_at' => '',
), false );
set_transient( self::LOCK_TRANSIENT, $run_id, 4 * HOUR_IN_SECONDS );
as_enqueue_async_action( self::CATEGORY_ACTION_HOOK, array( $run_id, 0 ), self::ACTION_GROUP );
WBC_Logger::info( 'BatchSync', 'Queued category-filtered background sync', array(
'run_id' => $run_id,
'categories' => $categories,
) );
return array(
'success' => true,
'started' => true,
'message' => sprintf(
/* translators: %s: comma-separated category codes */
__( 'Sync started in the background, filtered to categories: %s.', 'woo-business-central' ),
implode( ', ', $categories )
),
);
}
/**
* Process one page of BC items filtered by category (called by Action Scheduler)
*
* Self-chains: enqueues the next page if this one came back full, or marks
* the run complete otherwise. Positional args, not a combined array - see
* the note on process_batch().
*
* @param string $run_id Run ID this page belongs to.
* @param int $skip OData $skip offset for this page.
*/
public function process_category_page( $run_id, $skip = 0 ) {
$run = get_option( self::RUN_OPTION, array() );
if ( empty( $run ) || ( $run['run_id'] ?? '' ) !== $run_id ) {
return;
}
$categories = $run['categories'] ?? array();
if ( empty( $categories ) ) {
return;
}
$filter_parts = array_map( function ( $code ) {
return "itemCategoryCode eq '" . WBC_API_Client::escape_odata_string( $code ) . "'";
}, $categories );
$result = WBC_API_Client::get( '/items', array(
'$filter' => '(' . implode( ' or ', $filter_parts ) . ')',
'$select' => WBC_Product_Sync::SELECT_FIELDS,
'$top' => self::CATEGORY_PAGE_SIZE,
'$skip' => $skip,
) );
if ( is_wp_error( $result ) ) {
WBC_Logger::error( 'BatchSync', 'Failed to fetch category page from BC, stopping run', array(
'run_id' => $run_id,
'skip' => $skip,
'error' => $result->get_error_message(),
) );
$run['status'] = 'failed';
$run['finished_at'] = current_time( 'mysql' );
update_option( self::RUN_OPTION, $run, false );
delete_transient( self::LOCK_TRANSIENT );
return;
}
$items = isset( $result['value'] ) ? $result['value'] : array();
$sync = new WBC_Product_Sync();
foreach ( $items as $item ) {
$sync->process_bc_item( $item );
usleep( 50000 ); // 50ms, matches the rate limiting used elsewhere
}
$page_results = $sync->get_results();
// Best-effort counters - see note on process_batch().
$run = get_option( self::RUN_OPTION, $run );
if ( ( $run['run_id'] ?? '' ) !== $run_id ) {
return;
}
$run['pages_done'] = (int) ( $run['pages_done'] ?? 0 ) + 1;
$run['total'] = (int) ( $run['total'] ?? 0 ) + count( $items );
$run['success'] = (int) ( $run['success'] ?? 0 ) + $page_results['success'];
$run['failed'] = (int) ( $run['failed'] ?? 0 ) + $page_results['failed'];
$run['skipped'] = (int) ( $run['skipped'] ?? 0 ) + $page_results['skipped'];
$has_more_pages = count( $items ) === self::CATEGORY_PAGE_SIZE;
if ( $has_more_pages ) {
$next_skip = $skip + self::CATEGORY_PAGE_SIZE;
$run['skip'] = $next_skip;
update_option( self::RUN_OPTION, $run, false );
as_enqueue_async_action( self::CATEGORY_ACTION_HOOK, array( $run_id, $next_skip ), self::ACTION_GROUP );
} else {
$run['status'] = 'completed';
$run['finished_at'] = current_time( 'mysql' );
update_option( self::RUN_OPTION, $run, false );
WBC_Cron::update_last_sync_time();
delete_transient( self::LOCK_TRANSIENT );
WBC_Logger::info( 'BatchSync', 'Category-filtered background sync completed', $run );
}
}
/**
* Process a single batch (called by Action Scheduler)
*