暂存

scripts/reindex_opensearch.php

@@ -3,31 +3,59 @@
 
 use app\service\Search\ChunkSearchIndexHandler;
 use app\service\Search\ChunkSearchIndexRepository;
+use app\service\Search\OpenSearchClientFactory;
 use app\service\Search\OpenSearchChunkIndex;
-use support\Db;
 
 require __DIR__ . '/../vendor/autoload.php';
 require __DIR__ . '/../support/bootstrap.php';
-require __DIR__ . '/../vendor/webman/database/src/support/Db.php';
 
 $archiveUid = null;
+$forceReset = false;
 foreach (array_slice($argv, 1) as $argument) {
     if (str_starts_with($argument, '--archive_uid=')) {
         $archiveUid = substr($argument, strlen('--archive_uid='));
+        continue;
+    }
+
+    if ($argument === '--reset') {
+        $forceReset = true;
     }
 }
 
 $repository = new ChunkSearchIndexRepository();
 $handler = new ChunkSearchIndexHandler();
 $index = new OpenSearchChunkIndex();
+$clientFactory = new OpenSearchClientFactory();
+$bulkSize = max(1, (int) config('opensearch.bulk.chunk_size', 500));
 
 try {
+    $client = $clientFactory->make();
+    $indexName = config('opensearch.indices.chunks', 'proofdb_chunks');
+    $indexExists = (bool) $client->indices()->exists(['index' => $indexName]);
+
     $index->ensureExists();
 
-    $resetCount = $repository->resetEmbeddedChunksToPending($archiveUid);
-    $archiveCount = 0;
+    $totalChunks = $repository->countEmbeddedChunks($archiveUid);
+    if ($totalChunks === 0) {
+        echo 'OpenSearch reindex completed.' . PHP_EOL;
+        echo 'Index: ' . $indexName . PHP_EOL;
+        echo 'Archive filter: ' . ($archiveUid ?: '(all embedded archives)') . PHP_EOL;
+        echo 'Mode: nothing-to-do' . PHP_EOL;
+        echo 'Eligible embedded chunks: 0' . PHP_EOL;
+        exit(0);
+    }
+
+    $mode = $forceReset || !$indexExists ? 'reset' : 'resume';
+    $resetCount = $mode === 'reset'
+        ? $repository->resetEmbeddedChunksToPending($archiveUid)
+        : $repository->resetRecoverableChunksToPending($archiveUid);
+
+    $batchCount = 0;
     $indexedArchives = [];
-    $indexedChunks = 0;
+    $progress = $repository->countIndexedChunks($archiveUid);
+
+    echo 'Progress granularity: OpenSearch bulk batches (up to ' . $bulkSize . ' chunks each)' . PHP_EOL;
+    renderProgress($progress, $totalChunks, 'Reindexing');
 
     while (true) {
         $archiveUids = $repository->queuePendingArchiveTasks(100);
@@ -36,28 +64,44 @@ try {
         }
 
         foreach ($archiveUids as $uid) {
-            $handler->handle([
+            $processedChunkCount = $handler->handle([
                 'task_type' => 'search_index',
                 'target_type' => 'archive',
                 'target_uid' => $uid,
                 'attempt' => 1,
             ]);
-            $archiveCount++;
+            if ($processedChunkCount <= 0) {
+                continue;
+            }
+
+            $batchCount++;
             $indexedArchives[] = $uid;
+            $progress = $repository->countIndexedChunks($archiveUid);
+            renderProgress($progress, $totalChunks, 'Reindexing');
+            fwrite(STDOUT, PHP_EOL . sprintf(
+                'Batch #%d archive=%s chunks=%d progress=%d/%d%s',
+                $batchCount,
+                $uid,
+                $processedChunkCount,
+                $progress,
+                $totalChunks,
+                PHP_EOL
+            ));
         }
     }
 
-    $indexedChunksQuery = Db::table('chunks')->where('search_index_status', 3);
-    if ($archiveUid !== null && trim($archiveUid) !== '') {
-        $indexedChunksQuery->where('archive_uid', trim($archiveUid));
-    }
-    $indexedChunks = (int) $indexedChunksQuery->count();
+    $indexedChunks = $repository->countIndexedChunks($archiveUid);
+    renderProgress($indexedChunks, $totalChunks, 'Reindexing', true);
 
     echo 'OpenSearch reindex completed.' . PHP_EOL;
-    echo 'Index: ' . config('opensearch.indices.chunks', 'proofdb_chunks') . PHP_EOL;
+    echo 'Index: ' . $indexName . PHP_EOL;
     echo 'Archive filter: ' . ($archiveUid ?: '(all embedded archives)') . PHP_EOL;
+    echo 'Mode: ' . $mode . ($forceReset ? ' (--reset)' : (!$indexExists ? ' (index was missing)' : '')) . PHP_EOL;
+    echo 'Eligible embedded chunks: ' . $totalChunks . PHP_EOL;
+    echo 'OpenSearch bulk size: ' . $bulkSize . PHP_EOL;
     echo 'Reset chunks: ' . $resetCount . PHP_EOL;
-    echo 'Indexed archives: ' . $archiveCount . PHP_EOL;
+    echo 'Indexed archives: ' . count(array_unique($indexedArchives)) . PHP_EOL;
+    echo 'Processed batches: ' . $batchCount . PHP_EOL;
     echo 'Indexed chunk rows now marked indexed: ' . $indexedChunks . PHP_EOL;
     if ($indexedArchives !== []) {
         echo 'Archives: ' . implode(', ', $indexedArchives) . PHP_EOL;
@@ -66,3 +110,19 @@ try {
     fwrite(STDERR, $exception::class . ': ' . $exception->getMessage() . PHP_EOL);
     exit(1);
 }
+
+function renderProgress(int $done, int $total, string $label, bool $final = false): void
+{
+    $total = max(1, $total);
+    $done = max(0, min($done, $total));
+    $width = 32;
+    $filled = (int) floor(($done / $total) * $width);
+    $bar = str_repeat('=', $filled) . str_repeat(' ', max(0, $width - $filled));
+    $percent = str_pad(number_format(($done / $total) * 100, 1), 5, ' ', STR_PAD_LEFT);
+    $line = sprintf("\r%s [%s] %s%% (%d/%d)", $label, $bar, $percent, $done, $total);
+    fwrite(STDOUT, $line);
+
+    if ($final || $done >= $total) {
+        fwrite(STDOUT, PHP_EOL);
+    }
+}