How to write and read from collections

[alexander-schranz]

I'm trying to index documents in newly created collections. I know that I can define a specific connection at create a client, but I need to read and write into multiple collection but I'm not sure how to achieve this.

<?php

use Solarium\Client;
use Solarium\Core\Client\Adapter\Curl;

// create Client
$adapter = new Curl();
$eventDispatcher = new EventDispatcher();
$options = [
    'endpoint' => array(
        'localhost' => array(
            'host' => $host,
            'port' => $port,
        ),
    )
];

$client = new Client($adapter, $eventDispatcher, $options);

// create collection test_simple
$collectionQuery = $this->client->createCollections();

$action = $collectionQuery->createDelete(['name' => 'test_simple']);
$collectionQuery->setAction($action);

$this->client->collections($collectionQuery);

// create collection test_complex
$collectionQuery = $this->client->createCollections();

$action = $collectionQuery->createDelete(['name' => 'test_complex']);
$collectionQuery->setAction($action);

$this->client->collections($collectionQuery);

// now I try to index different documents into different collections but it seems to fail:
$update = $this->client->createUpdate([
    'collection' => 'test_simple',
]);


$indexDocument = $update->createDocument(['id' => 'c1ee078b-289e-42e6-8040-d5ea3288e804', 'title' => 'Test']);

$update->addDocuments([$indexDocument], true);

$update->addCommit();

$result = $this->client->update($update);

But it fails with:

Solarium\Exception\HttpException: Solr HTTP error: Neither collection nor core set. (404)

I'm setting the collectino as option so I'm not sure how to achieve this corectly?

[thomascorthals]

You're calling the wrong method to create collection (twice). This should be createCreate() instead of createDelete().

$action = $collectionQuery->createDelete(['name' => 'test_simple']);

After that, you have to add an endpoint that points to your new collection to the client instance.

$endpoint = [
    'key' => 'test_simple', // this is how you want to identify the endpoint, doesn't have to be the collection name
    'host' => $host,
    'port' => $port,
    'collection' => 'test_simple',
];

$this->client->addEndpoint($endpoint);

You can now use the endpoint key when executing an update (or any other) query.

$update = $this->client->createUpdate(); // no options needed here for your use case

// add documents, commit

$result = $this->client->update($update, 'test_simple'); // execute against endpoint identified as 'test_simple'

[alexander-schranz]

Oh that was a typo. The hint with the collection did work. Really thank you for your fast reponse. But as the collection is in my case dynamic and I and the collections are not known when the client is created I did go with:

$this->client->getEndpoint()
    ->setCollection('test_simple');

[thomascorthals]

From your question regarding schemas, I assume you're trying to configure which schema to use for your new collection? You can specify the name of a configuration to use for the collection.

$action = $collectionQuery->createCreate();
$action->setName('test_simple')
       ->setCollectionConfigName('techproducts');

(I prefer the above syntax, but $collectionQuery->createCreate(['name' => 'test_simple', 'collection.configName' => 'techproducts']) also works.)

If you're not sure what "name of a configuration" means here and how it relates to the schema, start with reading configsets in the ref guide.

[thomascorthals]

You can use the Configsets API to upload a configset to Zookeeper.

solarium/examples/execute_all.php

Lines 37 to 44 in 3e0eb01

	$configsetsQuery = $client->createConfigsets();
	$UploadAction = $configsetsQuery->createUpload();
	$UploadAction
	->setFile(__DIR__.DIRECTORY_SEPARATOR.'..'.DIRECTORY_SEPARATOR.'tests'.DIRECTORY_SEPARATOR.'Integration'.DIRECTORY_SEPARATOR.'Fixtures'.DIRECTORY_SEPARATOR.'techproducts.zip')
	->setName('techproducts')
	->setOverwrite(true);
	$configsetsQuery->setAction($UploadAction);
	$client->configsets($configsetsQuery);

An existing schema can be updated through the Schema API.

solarium/examples/execute_all.php

Lines 101 to 115 in 3e0eb01

	$query = $client->createApi([
	'version' => Request::API_V1,
	'handler' => $collection_or_core_name.'/schema',
	'method' => Request::METHOD_POST,
	'rawdata' => json_encode([
	'add-dynamic-field' => [
	'name' => 'attr_*',
	'type' => 'text_general',
	'indexed' => true,
	'stored' => true,
	'multiValued' => true,
	],
	]),
	]);
	$client->execute($query);

[thomascorthals]

@alexander-schranz I've updated my comment above with example code to upload a configset. Does this answer your question?

[alexander-schranz]

Thx for the response 👍, This looks what I was searching for. I think it should also work for nested objects (docs) but think that should be able to findout in the schema api docs.

Will give it a try this week, and give you an update here then. Thank you again for your help!

[alexander-schranz]

@thomascorthals Thank you your mention method over the configset and raw createApi did work for me. For later stumble over same problem the code can be found here: https://github.com/schranz-search/schranz-search/pull/73/files#diff-c873e4216e21cddd561c38583e6f9bdfe50253a8b1c9c9d9ffd5ac1ffcd39352 in the createIndex method. I just needed to add SOLR_OPTS: '-Dsolr.disableConfigSetsCreateAuthChecks=true' to my docker container else I was not able to create configssets

Next step for me will be to findout how nested docs / objects works in solr over the schema api.

[alexander-schranz]

@thomascorthals Thx for your support. I did flatten all the objects now like elasticsearch is doing it internal, just on the PHP side to get nested json work also.

So I was able to merge: schranz-search/schranz-search#73

And so the adapter is now available here: https://github.com/schranz-search/seal-solr-adapter