暂存

2026-05-08 00:05:51 +08:00
parent 549b706fcc
commit ed70a140a2
40 changed files with 5590 additions and 36 deletions
@@ -0,0 +1,117 @@
+#!/usr/bin/env php
+<?php
+
+use app\service\ArticleImportService;
+use support\Db;
+
+require __DIR__ . '/../vendor/autoload.php';
+require __DIR__ . '/../support/bootstrap.php';
+require __DIR__ . '/../vendor/webman/database/src/support/Db.php';
+
+$archiveUid = null;
+$force = false;
+$dryRun = false;
+
+foreach (array_slice($argv, 1) as $argument) {
+    if (str_starts_with($argument, '--archive_uid=')) {
+        $archiveUid = substr($argument, strlen('--archive_uid='));
+        continue;
+    }
+
+    if ($argument === '--force') {
+        $force = true;
+        continue;
+    }
+
+    if ($argument === '--dry-run') {
+        $dryRun = true;
+    }
+}
+
+$query = Db::table('archives')->orderBy('id');
+if ($archiveUid !== null && trim($archiveUid) !== '') {
+    $query->where('archive_uid', trim($archiveUid));
+}
+
+if (!$force) {
+    $query->where(function ($builder) {
+        $builder->whereNull('content')->orWhere('content', '');
+    });
+}
+
+$archives = $query->get(['archive_uid', 'title', 'content', 'raw'])->all();
+$normalizer = new ArticleImportService();
+
+$scanned = 0;
+$updated = 0;
+$fromRaw = 0;
+$fromChunks = 0;
+$skipped = 0;
+
+foreach ($archives as $archive) {
+    $scanned++;
+    $archiveUidValue = (string) $archive->archive_uid;
+    $raw = is_string($archive->raw ?? null) ? $archive->raw : null;
+    $content = null;
+    $source = 'none';
+
+    if (is_string($raw) && trim($raw) !== '') {
+        $content = $normalizer->normalizeArchiveContentString($raw);
+        $source = 'raw';
+    } else {
+        $chunks = Db::table('chunks')
+            ->where('archive_uid', $archiveUidValue)
+            ->orderBy('chunk_index')
+            ->pluck('text')
+            ->all();
+
+        $chunks = array_values(array_filter(array_map(
+            static fn ($value): string => trim((string) $value),
+            $chunks
+        ), static fn (string $value): bool => $value !== ''));
+
+        if ($chunks !== []) {
+            $content = trim(implode("\n\n", $chunks));
+            $source = 'chunks';
+        }
+    }
+
+    if ($content === null || $content === '') {
+        $skipped++;
+        echo "[skip] {$archiveUidValue} no usable raw/chunks" . PHP_EOL;
+        continue;
+    }
+
+    if ($dryRun) {
+        echo "[dry-run] {$archiveUidValue} source={$source} content_length=" . mb_strlen($content) . PHP_EOL;
+        if ($source === 'raw') {
+            $fromRaw++;
+        } else {
+            $fromChunks++;
+        }
+        continue;
+    }
+
+    Db::table('archives')
+        ->where('archive_uid', $archiveUidValue)
+        ->update(['content' => $content]);
+
+    $updated++;
+    if ($source === 'raw') {
+        $fromRaw++;
+    } else {
+        $fromChunks++;
+    }
+
+    echo "[updated] {$archiveUidValue} source={$source} content_length=" . mb_strlen($content) . PHP_EOL;
+}
+
+echo 'Archive content backfill completed.' . PHP_EOL;
+echo 'Archive filter: ' . ($archiveUid ?: 'auto') . PHP_EOL;
+echo 'Force mode: ' . ($force ? 'yes' : 'no') . PHP_EOL;
+echo 'Dry run: ' . ($dryRun ? 'yes' : 'no') . PHP_EOL;
+echo 'Scanned: ' . $scanned . PHP_EOL;
+echo 'Updated: ' . $updated . PHP_EOL;
+echo 'From raw: ' . $fromRaw . PHP_EOL;
+echo 'From chunks: ' . $fromChunks . PHP_EOL;
+echo 'Skipped: ' . $skipped . PHP_EOL;
@@ -0,0 +1,68 @@
+#!/usr/bin/env php
+<?php
+
+use app\service\Search\ChunkSearchIndexHandler;
+use app\service\Search\ChunkSearchIndexRepository;
+use app\service\Search\OpenSearchChunkIndex;
+use support\Db;
+
+require __DIR__ . '/../vendor/autoload.php';
+require __DIR__ . '/../support/bootstrap.php';
+require __DIR__ . '/../vendor/webman/database/src/support/Db.php';
+
+$archiveUid = null;
+foreach (array_slice($argv, 1) as $argument) {
+    if (str_starts_with($argument, '--archive_uid=')) {
+        $archiveUid = substr($argument, strlen('--archive_uid='));
+    }
+}
+
+$repository = new ChunkSearchIndexRepository();
+$handler = new ChunkSearchIndexHandler();
+$index = new OpenSearchChunkIndex();
+
+try {
+    $index->ensureExists();
+
+    $resetCount = $repository->resetEmbeddedChunksToPending($archiveUid);
+    $archiveCount = 0;
+    $indexedArchives = [];
+    $indexedChunks = 0;
+
+    while (true) {
+        $archiveUids = $repository->queuePendingArchiveTasks(100);
+        if ($archiveUids === []) {
+            break;
+        }
+
+        foreach ($archiveUids as $uid) {
+            $handler->handle([
+                'task_type' => 'search_index',
+                'target_type' => 'archive',
+                'target_uid' => $uid,
+                'attempt' => 1,
+            ]);
+            $archiveCount++;
+            $indexedArchives[] = $uid;
+        }
+    }
+
+    $indexedChunksQuery = Db::table('chunks')->where('search_index_status', 3);
+    if ($archiveUid !== null && trim($archiveUid) !== '') {
+        $indexedChunksQuery->where('archive_uid', trim($archiveUid));
+    }
+    $indexedChunks = (int) $indexedChunksQuery->count();
+
+    echo 'OpenSearch reindex completed.' . PHP_EOL;
+    echo 'Index: ' . config('opensearch.indices.chunks', 'proofdb_chunks') . PHP_EOL;
+    echo 'Archive filter: ' . ($archiveUid ?: '(all embedded archives)') . PHP_EOL;
+    echo 'Reset chunks: ' . $resetCount . PHP_EOL;
+    echo 'Indexed archives: ' . $archiveCount . PHP_EOL;
+    echo 'Indexed chunk rows now marked indexed: ' . $indexedChunks . PHP_EOL;
+    if ($indexedArchives !== []) {
+        echo 'Archives: ' . implode(', ', $indexedArchives) . PHP_EOL;
+    }
+} catch (Throwable $exception) {
+    fwrite(STDERR, $exception::class . ': ' . $exception->getMessage() . PHP_EOL);
+    exit(1);
+}
@@ -0,0 +1,92 @@
+#!/usr/bin/env php
+<?php
+
+use support\Db;
+
+require __DIR__ . '/../vendor/autoload.php';
+require __DIR__ . '/../support/bootstrap.php';
+require __DIR__ . '/../vendor/webman/database/src/support/Db.php';
+
+$username = null;
+$password = null;
+$displayName = null;
+
+foreach (array_slice($argv, 1) as $argument) {
+    if (str_starts_with($argument, '--username=')) {
+        $username = substr($argument, strlen('--username='));
+        continue;
+    }
+
+    if (str_starts_with($argument, '--password=')) {
+        $password = substr($argument, strlen('--password='));
+        continue;
+    }
+
+    if (str_starts_with($argument, '--display_name=')) {
+        $displayName = substr($argument, strlen('--display_name='));
+    }
+}
+
+if (!is_string($username) || trim($username) === '' || !is_string($password) || $password === '') {
+    fwrite(STDERR, "Usage: php scripts/setup_admin_users.php --username=<username> --password=<password> [--display_name=<name>]" . PHP_EOL);
+    exit(1);
+}
+
+$username = trim($username);
+$displayName = is_string($displayName) && trim($displayName) !== '' ? trim($displayName) : $username;
+$passwordHash = password_hash($password, PASSWORD_DEFAULT);
+
+$statements = [
+    <<<SQL
+CREATE TABLE IF NOT EXISTS admin_users (
+    id BIGSERIAL PRIMARY KEY,
+    username VARCHAR(120) NOT NULL UNIQUE,
+    display_name TEXT,
+    password_hash TEXT NOT NULL,
+    is_active BOOLEAN NOT NULL DEFAULT TRUE,
+    last_login_at TIMESTAMPTZ,
+    created_time TIMESTAMPTZ NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updated_time TIMESTAMPTZ NOT NULL DEFAULT CURRENT_TIMESTAMP
+)
+SQL,
+    'CREATE INDEX IF NOT EXISTS admin_users_is_active_index ON admin_users (is_active)',
+    <<<SQL
+CREATE OR REPLACE FUNCTION set_updated_time()
+RETURNS TRIGGER AS $$
+BEGIN
+    NEW.updated_time = CURRENT_TIMESTAMP;
+    RETURN NEW;
+END;
+$$ LANGUAGE plpgsql
+SQL,
+    'DROP TRIGGER IF EXISTS admin_users_set_updated_time ON admin_users',
+    <<<SQL
+CREATE TRIGGER admin_users_set_updated_time
+BEFORE UPDATE ON admin_users
+FOR EACH ROW
+EXECUTE FUNCTION set_updated_time()
+SQL,
+];
+
+try {
+    Db::connection()->getPdo();
+    foreach ($statements as $statement) {
+        Db::statement($statement);
+    }
+
+    Db::table('admin_users')->updateOrInsert(
+        ['username' => $username],
+        [
+            'display_name' => $displayName,
+            'password_hash' => $passwordHash,
+            'is_active' => true,
+        ]
+    );
+
+    echo 'Admin users table initialized.' . PHP_EOL;
+    echo 'Seeded username: ' . $username . PHP_EOL;
+    echo 'Display name: ' . $displayName . PHP_EOL;
+} catch (Throwable $exception) {
+    fwrite(STDERR, $exception::class . ': ' . $exception->getMessage() . PHP_EOL);
+    exit(1);
+}