在magento2中使用persisted-query

什么是 persisted query

scandipwa/persisted-query 是一个用于提升 magento2 graphql 响应速度的库。大致原理是，把请求的解释缓存在 redis ，把响应的结果缓存在 full page cache (通常是 varnish) 。只用于 query ，不用于 mutation

persisted-query 依赖

Redis
Varnish
PHP ext-phpredis

安装

通过 composer 安装

composer require scandipwa/persisted-query:^2.2

修改配置文件 app/etc/env.php ，在 cache中加上这一段

    'persisted-query' => [
        'redis' => [
            'host' => 'localhost',
            'scheme' => 'tcp',
            'port' => '6379',
            'database' => '5'
        ]
    ],

修改配置文件 app/etc/config.php

'ScandiPWA_PersistedQuery' => 1,

安装完后记得运行构建命令。大致例子，仅供参考

php bin/magento setup:upgrade && \
php bin/magento setup:di:compile && \
php bin/magento setup:static-content:deploy -f && \
php bin/magento indexer:reindex && \
php bin/magento cache:flush;

使用

大致的使用流程

先发送一次 PUT 请求

url 里需要带上一个 hash 参数， hash 的值是一个随机数，这个值不能重复

https://localhost/graphql?hash=2443957263

PUT 请求成功后会返回 201

HTTP/1.1 201 Created
Content-Type: application/json
Content-Length: 55

{"error":false,"code":"1","message":"Query registered"}

后续发送 GET 请求， GET 请求中也需要带上 hash 参数
- 后续的 GET 请求才会有具体的数据

关键代码的位置 vendor/scandipwa/persisted-query/src/Plugin/PersistedQuery.php

可以通过这个 http 头设置过期时间 SW-cache-age ，这是设置查询的过期时间

如果没有特别的设置，缓存在刷新时才会失效

刷新查询
```
bin/magento scandipwa:pq:flush
```

刷新响应

bin/magento cache:flush persisted_query_response

缓存无效时会返回 401

HTTP/1.1 410 Gone
Content-Type: application/json
Content-Length: 61

{"error":true,"code":"410","message":"Query hash is unknown"}

不带参数的请求

先发送一次 PUT 请求

curl 'https://localhost/graphql?hash=2443957263' \
  -X 'PUT' \
  -H 'accept: */*' \
  -H 'content-type: application/json' \
  --data-raw ''

后续发送 GET 请求

curl -X 'GET' 'https://localhost/graphql?hash=2443957263'

带参数的请求

先发送一次 PUT 请求

curl 'https://localhost/graphql?hash=2443957282' \
-X 'PUT' \
-H 'accept: */*' \
-H 'content-type: application/json' \
-H 'Cookie: XDEBUG_SESSION=vscode;' \
-d '{
    "query": "query GetProductsBySkus($_filter:ProductAttributeFilterInput!, $_pageSize:Int) { products(search: \"*\", filter: $_filter, pageSize: $_pageSize ) { items { id uid sku name image { url } url_key url_suffix stock_status categories { uid name } } } }"
}'

后续发送 GET 请求

curl -X 'GET' 'https://localhost/graphql?hash=2443957282&_filter=%257B%2522sku%2522%253A%257B%2522in%2522%253A%255B%2522sku1%2522%252C%2522sku2%2522%252C%2522sku3%2522%255D%257D%257D&_pageSize=1'

如何构造一个带参数的请求

假设这是原本的请求

curl 'https://localhost/graphql' \
  -X 'POST' \
  -H 'accept: */*' \
  -H 'content-type: application/json' \
  -H 'Cookie: XDEBUG_SESSION=vscode;' \
  -d '{
    "query": "query GetProductsBySkus($_filter:ProductAttributeFilterInput!, $_pageSize:Int) { products(search: \"*\", filter: $_filter, pageSize: $_pageSize ) { items { id uid sku name image { url } url_key url_suffix stock_status categories { uid name } } } }",
    "variables":{"_filter":{"sku":{"in":["sku1","sku2","sku3"],"_pageSize":1}}}
}'

删除请求体中的 variables
把请求的方法改成 PUT
url 中加上 hash 参数
这是最后的结果

curl 'https://localhost/graphql?hash=2443957282' \
  -X 'PUT' \
  -H 'accept: */*' \
  -H 'content-type: application/json' \
  -H 'Cookie: XDEBUG_SESSION=vscode;' \
  -d '{
    "query": "query GetProductsBySkus($_filter:ProductAttributeFilterInput!, $_pageSize:Int) { products(search: \"*\", filter: $_filter, pageSize: $_pageSize ) { items { id uid sku name image { url } url_key url_suffix stock_status categories { uid name } } } }",
}'

后续的请求中，把 variables 转换为 url编码，再直接加在 url 里

curl -X 'GET' 'https://localhost/graphql?hash=2443957282&_filter=%257B%2522sku%2522%253A%257B%2522in%2522%253A%255B%2522sku1%2522%252C%2522sku2%2522%252C%2522sku3%2522%255D%257D%257D&_pageSize=1'

这是把 variables 转换为 url编码的伪代码

$variables = '{"_filter":{"sku":{"in":["sku1","sku2","sku3"]}},"_pageSize":1}';

$variables = json_decode($variables, true);

foreach ($variables as $key => $value) {
    $variables[$key] = urlencode(json_encode($value));
}

echo http_build_query($variables);