tested on strapi v3.x
latest test: v3.4.0
This plugin has not been tested on mongodb
The purpose of developing this plugin is to use the elastic search engine in Strapi to help the application development process
📝 Table of Contents
Prerequisites
Install Elasticsearch - https://www.elastic.co/downloads/elasticsearch
Install plugin
-
Go to the project path
cd PROJECT/plugins
-
Clone the project
git clone https://github.com/marefati110/strapi-plugin-elasticsearch.git
-
Install dependencies
yarn install
🏁 Getting Started
How plugin works?
After the first run of the project, it creates a config file at PROJECT/config/elasticsearch.js
config file should look like the image
By default, syncing occurs in two ways
The answer of any request that makes a change in the model is stored in Elasticsearch this is especially true for the Strap panel admin
Or in response to any request, search for the pk of model the model in which the change was made, and after retrieving the information from the database, stores it in the elasticsearch
In the following, solutions are prepared for more complex scenarios.
After installing the plugin and running it, it creates an config file in the PROJECT/config/elasticsearch.js
In the connections section, the settings related to the connection to the elasticsearch are listed, there is also a help link
In the setting section, there are the initial settings related to the elastic plugin.
In the models section for all models in the Project/api/** path there is a model being built and you can change the initial settings
🎈 Usage
Scenario 1
For example, we want to make changes to the article model and then see the changes in the Elasticsearch.
The first step is to activate in the settings related to this model
After saving and restarting the plugin, it creates an index for this model in the elasticsearch.
Note that the name selected for the index can be changed in the settings of the model.
At the end of the settings should be as follows
{
model: 'article',
pk: 'id',
plugin: null, // changed to true
enabled: true,
index: 'article',
relations: [],
conditions: {},
supportAdminPanel: true,
fillByResponse: true,
migration: false,
urls: [],
},Now in the strapi admin panel, by making an creating , deleting or updating , you can see the changes in Elasticsearch.
Scenario 2
In this scenario, we want to make a change in the model using the rest api and see the result in Elasticsearch.
After sending a post request to /articles, changes will be applied and we will receive a response to this
{
"id": 1,
"title": "title",
"content": "content"
}and model config should change to
{
model: 'article',
pk: 'id',
plugin: null,
enabled: true,
index: 'article',
relations: [],
conditions: {},
supportAdminPanel: true,
fillByResponse: true, // default value
migration: false,
urls: ['/articles'], //changed
},If the fillByResponse settings are enabled for the model, the same data will be stored in Elasticsearch, otherwise the data will be retrieved from the database using pk and stored in Elasticsearch.
Scenario 3
This scenario is quite similar to the previous scenario with these differences being the response
{
"metaData": null,
"data": {
"articleID": 1,
"title": "title",
"content": "content"
}
}By default, the plugin looks for pk in the response or ctx.body.id
We can rewrite these settings for a specific url
config model should change to
{
model: 'article',
pk: 'id',
plugin: null,
enabled: true,
index: 'article',
relations: [],
conditions: {},
supportAdminPanel: true,
fillByResponse: true,
migration: false,
urls: [
{
'/articles':{
pk: 'data.articleID', // over write
relations: [], // over write
conditions: {}, // over write
}
}
],
},Scenario 4
In this scenario, no pk may be found in the request response
{
"success": true
}In this case, the synchronization operation can be performed on the controller
there is some functions for help
const articleData = { title: 'title', content: 'content' };
const article = await strapi.query('article').create(articleData);
strapi.elastic.createOrUpdate('article', { data: article, id: article.id });
// or
strapi.elastic.migrateById('article', { id: article.id }); // execute new queryand for delete data
const articleId = 1;
const article = await strapi.query('article').delete(articleData);
strapi.elastic.destroy('article', { id: articleID });Functions
| Command | Description | example |
|---|---|---|
strapi.elastic |
official elasticsearch package | example |
strapi.elastic.createOrUpdate |
Create to update data | example |
strapi.elastic.findOne |
Find specific data by id | example |
strapi.elastic.destroy |
delete data | example |
strapi.elastic.migrateById |
migrate data | example |
strapi.elastic.migrateModel |
migrate specific data | example |
strapi.elastic.models |
migrate all enabled models | example |
strapi.log |
log data to elasticsearch | example |
Api
| Url | Method | Description | body |
|---|---|---|---|
| /migrate-models | POST | Migrate all enabled Models | |
| /migrate-Model | POST | Migrate specific model | {model:'MODEL_NAME'} |
Examples
elastic
For use official Elasticsearch package we can use strapi.elastic, and can access builtin function
elasticsearch reference api
const count = strapi.elastic.count({ index: 'article' }); // https://www.elastic.co/guide/en/elasticsearch/client/javascript-api/current/api-reference.html#_count
const article = strapi.elastic.get({ index: 'article', id: 1 }); // https://www.elastic.co/guide/en/elasticsearch/client/javascript-api/current/api-reference.html#_getCreateOrUpdate
const result = strapi.elastic.createOrUpdate('article', {
id: 1,
data: { title: 'title', content: 'content' },
});findOne
const result = strapi.elastic.findOne('article', { id: 1 });destroy
const result_one = strapi.elastic.destroy('article', { id: 1 });
// or
const result_two = strapi.elastic.destroy('article', { id_in: [1, 2, 3] });migrateById
const result_one = strapi.elastic.migrateById('article', { id: 1 });
const result_two = strapi.elastic.migrateById('article', { id_in: [1, 2, 3] });migrateModel
const result = strapi.elastic.migrateModel('article', {
conditions, // optional
});migrateModels
const result = strapi.elastic.migrateModels({
conditions, // optional (the conditions apply on all models)
});Logging
strapi use Pino to logging but can store logs or send it to elasticsearch
at now wen can send logs to elasticsearch by strapi.elastic.log there is no difference between strapi.elastic.log with strapi.log to call functions.
strapi.log.info('log message in console');
strapi.elastic.log.info('log message console and store it to elasticsearch');
strapi.log.debug('log message');
strapi.elastic.log.debug('log message console and store it to elasticsearch');
strapi.log.warn('log message');
strapi.elastic.log.warn('log message console and store it to elasticsearch');
strapi.log.error('log message');
strapi.elastic.log.error('log message console and store it to elasticsearch');
strapi.log.fatal('log message');
strapi.elastic.log.fatal('log message console and store it to elasticsearch');Also there is some more options
// just send log to elastic and avoid to display in console
strapi.elastic.log.info('some message', { setting: { show: false } });
// just display relations, // optional ni console and avoid to save it to elastic search
strapi.elastic.log.info('some message', { setting: { saveToElastic: false } });
// send more data to elasticsearch
const logData = { description: 'description' };
strapi.elastic.log.info('some message', logData);By default strapi.log send some metaData to elasticsearch such as free memory, cpu load avg, current time, hostname ,...
Tricks
to avoid config plugin for all model or write a lot of code we can create cron job for migration
const moment = require('moment');
module.exports = {
'*/10 * * * *': async () => {
const updateTime = moment()
.subtract(10, 'minutes')
.format('YYYY-MM-DD HH:mm:ss');
// currentTime
await strapi.elastic.migrateModels({
conditions: {
updated_at_gt: updateTime,
/* to utilise Draft/Publish feature & migrate only published entities
you can add following in conditions
*/
_publicationState: 'live'
},
});
},
};