CloudSearch deleteByQuery

Question

The official Solr Java API has a deleteByQuery operation where we can delete documents that satisfy a query. The AWS CloudSearch SDK doesn't seem to have matching functionality. Am I just not seeing the deleteByQuery equivalent, or is this something we'll need to roll our own?

Something like this:

SearchRequest searchRequest = new SearchRequest();
searchRequest.setQuery(queryString);
searchRequest.setReturn("id,version");
SearchResult searchResult = awsCloudSearch.search(searchRequest);
JSONArray docs = new JSONArray();
for (Hit hit : searchResult.getHits().getHit()) {
    JSONObject doc = new JSONObject();
    doc.put("id", hit.getId());
    // is version necessary?
    doc.put("version", hit.getFields().get("version").get(0));
    doc.put("type", "delete");
    docs.put(doc);
}
UploadDocumentsRequest uploadDocumentsRequest = new UploadDocumentsRequest();
StringInputStream documents = new StringInputStream(docs.toString());
uploadDocumentsRequest.setDocuments(documents);
UploadDocumentsResult uploadResult = awsCloudSearch.uploadDocuments(uploadDocumentsRequest);

Is this reasonable? Is there an easier way?

Answer 1

You're correct that CloudSearch doesn't have an equivalent to deleteByQuery. Your approach looks like the next best thing.

And no, version is not necessary -- it was removed with the CloudSearch 01-01-2013 API (aka v2).

Answer 2

CloudSearch doesn't provide delete as query, it supports delete in a slightly different way ie build json object having only document id (to be deleted) and operation should be specified as delete. These json objects can be batched together but batch size has to be less than 5 MB.

Following class supports this functionality, you just pass its delete method the array of ids to be deleted:

class AWS_CS
{
    protected $client;

    function connect($domain)
    {
        try{
            $csClient = CloudSearchClient::factory(array(
                            'key'          => 'YOUR_KEY',
                            'secret'      => 'YOUR_SECRET',
                            'region'     =>  'us-east-1'

                        ));

            $this->client = $csClient->getDomainClient(
                        $domain,
                        array(
                            'credentials' => $csClient->getCredentials(),
                            'scheme' => 'HTTPS'
                        )
                    );
        }
        catch(Exception $ex){
            echo "Exception: ";
            echo $ex->getMessage();
        }
        //$this->client->addSubscriber(LogPlugin::getDebugPlugin());        
    }
    function search($queryStr, $domain){

        $this->connect($domain);

        $result = $this->client->search(array(
            'query' => $queryStr,
            'queryParser' => 'lucene',
            'size' => 100,
            'return' => '_score,_all_fields'
            ))->toArray();

        return json_encode($result['hits']);
        //$hitCount = $result->getPath('hits/found');
        //echo "Number of Hits: {$hitCount}\n";
    }

    function deleteDocs($idArray, $operation = 'delete'){

        $batch = array();

        foreach($idArray as $id){
            //dumpArray($song);
            $batch[] = array(
                        'type'        => $operation,
                        'id'        => $id);                       
        }
        $batch = array_filter($batch);
        $jsonObj = json_encode($batch, JSON_HEX_TAG | JSON_HEX_APOS | JSON_HEX_QUOT | JSON_HEX_AMP);

        print_r($this->client->uploadDocuments(array(
                        'documents'     => $jsonObj,
                        'contentType'     =>'application/json'
                    )));

        return $result['status'] == 'success' ? mb_strlen($jsonObj) : 0;
    }   
}