Files
atlas-seller-kit/apps/worker/src/platform-catalog-scheduler.ts
T

228 lines
9.3 KiB
TypeScript

import type { Database, DatabaseTransaction } from '@atlas/database';
type SourceRow = { revision: number };
type BatchItem = {
batch_id: string;
category_id: string;
source_revision: number;
priority: 'high' | 'low';
};
const platformActiveRunLimit = 6;
/** 每周只自动刷新树;类型属性仅从管理员显式创建的批次中限量领取。 */
export class PlatformCatalogScheduler {
public constructor(private readonly database: Database) {}
public async run() {
const sources = await this.database.query<{ platform_id: string }>(
`SELECT source.platform_id FROM atlas.platform_catalog_sources source
JOIN atlas.platforms platform ON platform.id = source.platform_id
WHERE platform.status = 'active' AND upper(platform.code) = 'OZON'`,
);
for (const source of sources.rows) {
await this.schedulePlatform(source.platform_id);
}
}
private async schedulePlatform(platformId: string) {
await this.database.transaction(async (tx) => {
const source = await tx.query<SourceRow>(
`SELECT revision FROM atlas.platform_catalog_sources
WHERE platform_id = $1 FOR UPDATE`,
[platformId],
);
if (!source.rows[0]) return;
await refreshBatchStatuses(tx, platformId, source.rows[0].revision);
const pending = await tx.query<{ count: string }>(
`SELECT count(*)::text AS count FROM atlas.platform_catalog_sync_runs
WHERE platform_id = $1 AND status IN ('queued', 'running')`,
[platformId],
);
let capacity = platformActiveRunLimit - Number(pending.rows[0]?.count ?? 0);
if (capacity <= 0) return;
if (await treeNeedsRefresh(tx, platformId)) {
await enqueueRun(tx, {
platformId,
scopeKey: 'tree',
sourceRevision: source.rows[0].revision,
trigger: 'weekly',
priority: 'low',
});
capacity -= 1;
}
if (capacity <= 0) return;
const items = await pendingBatchItems(tx, platformId, capacity);
for (const item of items) {
const runId = await enqueueRun(tx, {
platformId,
scopeKey: `attributes:${item.category_id}`,
categoryId: item.category_id,
sourceRevision: item.source_revision,
trigger: 'admin_batch',
priority: item.priority,
});
await tx.query(
`UPDATE atlas.platform_catalog_attribute_batch_items
SET run_id = $3, linked_at = now()
WHERE batch_id = $1 AND category_id = $2 AND run_id IS NULL`,
[item.batch_id, item.category_id, runId],
);
}
if (items.length) {
await tx.query(
`UPDATE atlas.platform_catalog_attribute_batches
SET status = 'running', started_at = COALESCE(started_at, now())
WHERE id = $1 AND status = 'queued'`,
[items[0].batch_id],
);
}
});
}
}
async function treeNeedsRefresh(tx: DatabaseTransaction, platformId: string) {
const result = await tx.query<{ due: boolean }>(
`SELECT
(state.last_success_at IS NULL
OR state.last_success_at <= now() - interval '7 days')
AND NOT EXISTS (
SELECT 1 FROM atlas.platform_catalog_sync_runs run
WHERE run.platform_id = $1 AND run.scope_key = 'tree'
AND run.status IN ('queued', 'running')
)
AND NOT EXISTS (
SELECT 1 FROM atlas.platform_catalog_sync_runs run
WHERE run.platform_id = $1 AND run.scope_key = 'tree'
AND run.created_at > now() - interval '1 hour'
) AS due
FROM (SELECT 1) marker
LEFT JOIN atlas.platform_catalog_resource_states state
ON state.platform_id = $1 AND state.scope_key = 'tree'`,
[platformId],
);
return result.rows[0]?.due === true;
}
async function pendingBatchItems(
tx: DatabaseTransaction,
platformId: string,
limit: number,
): Promise<BatchItem[]> {
const result = await tx.query<BatchItem>(
`SELECT item.batch_id, item.category_id, batch.source_revision,
CASE WHEN state.last_success_at IS NULL
OR state.last_success_at < now() -
make_interval(days => batch.attribute_expire_days)
THEN 'high' ELSE 'low' END AS priority
FROM atlas.platform_catalog_attribute_batches batch
JOIN atlas.platform_catalog_attribute_batch_items item
ON item.batch_id = batch.id AND item.platform_id = batch.platform_id
LEFT JOIN atlas.platform_catalog_resource_states state
ON state.platform_id = batch.platform_id
AND state.scope_key = 'attributes:' || item.category_id
WHERE batch.platform_id = $1 AND batch.status IN ('queued', 'running')
AND item.run_id IS NULL
ORDER BY CASE WHEN state.last_success_at IS NULL
OR state.last_success_at < now() -
make_interval(days => batch.attribute_expire_days)
THEN 0 ELSE 1 END,
item.created_at, item.category_id
LIMIT $2
FOR UPDATE OF item SKIP LOCKED`,
[platformId, limit],
);
return result.rows;
}
async function refreshBatchStatuses(
tx: DatabaseTransaction,
platformId: string,
sourceRevision: number,
) {
await tx.query(
`UPDATE atlas.platform_catalog_attribute_batches
SET status = 'cancelled', completed_at = now()
WHERE platform_id = $1 AND status IN ('queued', 'running')
AND source_revision <> $2`,
[platformId, sourceRevision],
);
await tx.query(
`WITH progress AS (
SELECT batch.id,
count(*) FILTER (WHERE item.run_id IS NULL) AS unlinked_count,
count(*) FILTER (WHERE run.status IN ('queued', 'running')) AS active_count,
count(*) FILTER (WHERE run.status = 'failed') AS failed_count,
count(*) FILTER (WHERE run.status = 'cancelled') AS cancelled_count
FROM atlas.platform_catalog_attribute_batches batch
JOIN atlas.platform_catalog_attribute_batch_items item ON item.batch_id = batch.id
LEFT JOIN atlas.platform_catalog_sync_runs run ON run.id = item.run_id
WHERE batch.platform_id = $1 AND batch.status IN ('queued', 'running')
GROUP BY batch.id
)
UPDATE atlas.platform_catalog_attribute_batches batch
SET status = CASE
WHEN progress.unlinked_count = 0 AND progress.active_count = 0
THEN CASE WHEN progress.failed_count > 0 THEN 'failed'::atlas.operation_status
WHEN progress.cancelled_count > 0 THEN 'cancelled'::atlas.operation_status
ELSE 'succeeded'::atlas.operation_status END
ELSE batch.status
END,
completed_at = CASE
WHEN progress.unlinked_count = 0 AND progress.active_count = 0
THEN COALESCE(batch.completed_at, now())
ELSE batch.completed_at
END
FROM progress WHERE batch.id = progress.id`,
[platformId],
);
}
async function enqueueRun(
tx: DatabaseTransaction,
input: {
platformId: string;
scopeKey: string;
categoryId?: string;
sourceRevision: number;
trigger: 'weekly' | 'admin_batch';
priority: 'high' | 'low';
},
) {
const active = await tx.query<{ id: string; priority: 'high' | 'low' }>(
`SELECT id, priority FROM atlas.platform_catalog_sync_runs
WHERE platform_id = $1 AND scope_key = $2 AND status IN ('queued', 'running')
LIMIT 1`,
[input.platformId, input.scopeKey],
);
if (active.rows[0]) {
if (input.priority === 'high' && active.rows[0].priority === 'low') {
await tx.query(
`UPDATE atlas.platform_catalog_sync_runs
SET priority = 'high', promoted_at = now()
WHERE id = $1 AND status = 'queued'`,
[active.rows[0].id],
);
}
return active.rows[0].id;
}
const inserted = await tx.query<{ id: string }>(
`INSERT INTO atlas.platform_catalog_sync_runs
(platform_id, scope, scope_key, category_id, trigger_kind,
priority, source_revision)
VALUES ($1, $2, $3, $4, $5, $6, $7)
RETURNING id`,
[input.platformId, input.categoryId ? 'attributes' : 'tree', input.scopeKey,
input.categoryId ?? null, input.trigger, input.priority, input.sourceRevision],
);
await tx.query(
`INSERT INTO atlas.platform_catalog_outbox_events (run_id, platform_id)
VALUES ($1, $2)`,
[inserted.rows[0].id, input.platformId],
);
return inserted.rows[0].id;
}