SearchHitIterators and SearchResponseIterator helpers revised with new version

src/Helper/Iterators/SearchHitIterator.php

@@ -0,0 +1,166 @@
+<?php
+/**
+ * Elasticsearch PHP Client
+ *
+ * @link      https://github.com/elastic/elasticsearch-php
+ * @copyright Copyright (c) Elasticsearch B.V (https://www.elastic.co)
+ * @license   https://opensource.org/licenses/MIT MIT License
+ *
+ * Licensed to Elasticsearch B.V under one or more agreements.
+ * Elasticsearch B.V licenses this file to you under the MIT License.
+ * See the LICENSE file in the project root for more information.
+ */
+declare(strict_types = 1);
+
+namespace Elastic\Elasticsearch\Helper\Iterators;
+
+use Countable;
+use Elastic\Elasticsearch\Exception\ClientResponseException;
+use Elastic\Elasticsearch\Exception\ServerResponseException;
+use Iterator;
+
+class SearchHitIterator implements Iterator, Countable
+{
+
+    /**
+     * @var SearchResponseIterator
+     */
+    private SearchResponseIterator $searchResponses;
+
+    /**
+     * @var int
+     */
+    protected int $currentKey;
+
+    /**
+     * @var int
+     */
+    protected int $currentHitIndex;
+
+    /**
+     * @var array|null
+     */
+    protected ?array $currentHitData;
+
+    /**
+     * @var int
+     */
+    protected int $count = 0;
+
+    /**
+     * Constructor
+     *
+     * @param SearchResponseIterator $searchResponses
+     */
+    public function __construct(SearchResponseIterator $searchResponses)
+    {
+        $this->searchResponses = $searchResponses;
+    }
+
+    /**
+     * Rewinds the internal SearchResponseIterator and itself
+     *
+     * @return void
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     * @see    Iterator::rewind()
+     */
+    public function rewind(): void
+    {
+        $this->currentKey = 0;
+        $this->searchResponses->rewind();
+
+        // The first page may be empty. In that case, the next page is fetched.
+        $currentPage = $this->searchResponses->current();
+        if ($this->searchResponses->valid() && empty($currentPage['hits']['hits'])) {
+            $this->searchResponses->next();
+        }
+
+        $this->count = 0;
+        if (isset($currentPage['hits']['total']['value'], $currentPage['hits']['total'])) {
+            $this->count = $currentPage['hits']['total']['value'] ?? $currentPage['hits']['total'];
+        }
+
+        $this->readPageData();
+    }
+
+    /**
+     * Advances pointer of the current hit to the next one in the current page. If there
+     * isn't a next hit in the current page, then it advances the current page and moves the
+     * pointer to the first hit in the page.
+     *
+     * @return void
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     * @see    Iterator::next()
+     */
+    public function next(): void
+    {
+        $this->currentKey++;
+        $this->currentHitIndex++;
+        $currentPage = $this->searchResponses->current();
+        if (isset($currentPage['hits']['hits'][$this->currentHitIndex])) {
+            $this->currentHitData = $currentPage['hits']['hits'][$this->currentHitIndex];
+        } else {
+            $this->searchResponses->next();
+            $this->readPageData();
+        }
+    }
+
+    /**
+     * Returns a boolean indicating whether or not the current pointer has valid data
+     *
+     * @return bool
+     * @see    Iterator::valid()
+     */
+    public function valid(): bool
+    {
+        return is_array($this->currentHitData);
+    }
+
+    /**
+     * Returns the current hit
+     *
+     * @return array
+     * @see    Iterator::current()
+     */
+    public function current(): array
+    {
+        return $this->currentHitData;
+    }
+
+    /**
+     * Returns the current hit index. The hit index spans all pages.
+     *
+     * @return int
+     * @see    Iterator::key()
+     */
+    public function key(): int
+    {
+        return $this->currentKey;
+    }
+
+    /**
+     * Advances the internal SearchResponseIterator and resets the currentHitIndex to 0
+     *
+     * @internal
+     */
+    private function readPageData(): void
+    {
+        if ($this->searchResponses->valid()) {
+            $currentPage = $this->searchResponses->current();
+            $this->currentHitIndex = 0;
+            $this->currentHitData = $currentPage['hits']['hits'][$this->currentHitIndex];
+        } else {
+            $this->currentHitData = null;
+        }
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    public function count(): int
+    {
+        return $this->count;
+    }
+}

src/Helper/Iterators/SearchResponseIterator.php

@@ -0,0 +1,192 @@
+<?php
+/**
+ * Elasticsearch PHP Client
+ *
+ * @link      https://github.com/elastic/elasticsearch-php
+ * @copyright Copyright (c) Elasticsearch B.V (https://www.elastic.co)
+ * @license   https://opensource.org/licenses/MIT MIT License
+ *
+ * Licensed to Elasticsearch B.V under one or more agreements.
+ * Elasticsearch B.V licenses this file to you under the MIT License.
+ * See the LICENSE file in the project root for more information.
+ */
+declare(strict_types = 1);
+
+namespace Elastic\Elasticsearch\Helper\Iterators;
+
+use Elastic\Elasticsearch\ClientInterface;
+use Elastic\Elasticsearch\Exception\ClientResponseException;
+use Elastic\Elasticsearch\Exception\ServerResponseException;
+use Iterator;
+
+class SearchResponseIterator implements Iterator
+{
+
+    /**
+     * @var ClientInterface
+     */
+    private ClientInterface $client;
+
+    /**
+     * @var array
+     */
+    private array $params;
+
+    /**
+     * @var int
+     */
+    private int $currentKey = 0;
+
+    /**
+     * @var array
+     */
+    private array $currentScrolledResponse;
+
+    /**
+     * @var string|null
+     */
+    private ?string $scrollId;
+
+    /**
+     * @var string duration
+     */
+    private $scrollTtl;
+
+    /**
+     * Constructor
+     *
+     * @param ClientInterface $client
+     * @param array  $searchParams Associative array of parameters
+     * @see   ClientInterface::search()
+     */
+    public function __construct(ClientInterface $client, array $searchParams)
+    {
+        $this->client = $client;
+        $this->params = $searchParams;
+
+        if (isset($searchParams['scroll'])) {
+            $this->scrollTtl = $searchParams['scroll'];
+        }
+    }
+
+    /**
+     * Destructor
+     *
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     */
+    public function __destruct()
+    {
+        $this->clearScroll();
+    }
+
+    /**
+     * Sets the time to live duration of a scroll window
+     *
+     * @param  string $timeToLive
+     * @return $this
+     */
+    public function setScrollTimeout(string $timeToLive): SearchResponseIterator
+    {
+        $this->scrollTtl = $timeToLive;
+        return $this;
+    }
+
+    /**
+     * Clears the current scroll window if there is a scroll_id stored
+     *
+     * @return void
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     */
+    private function clearScroll(): void
+    {
+        if (!empty($this->scrollId)) {
+            /* @phpstan-ignore-next-line */
+            $this->client->clearScroll(
+                [
+                    'body' => [
+                        'scroll_id' => $this->scrollId
+                    ],
+                    'client' => [
+                        'ignore' => 404
+                    ]
+                ]
+            );
+            $this->scrollId = null;
+        }
+    }
+
+    /**
+     * Rewinds the iterator by performing the initial search.
+     *
+     * @return void
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     * @see    Iterator::rewind()
+     */
+    public function rewind(): void
+    {
+        $this->clearScroll();
+        $this->currentKey = 0;
+        /* @phpstan-ignore-next-line */
+        $this->currentScrolledResponse = $this->client->search($this->params)->asArray();
+        $this->scrollId = $this->currentScrolledResponse['_scroll_id'];
+    }
+
+    /**
+     * Fetches every "page" after the first one using the lastest "scroll_id"
+     *
+     * @return void
+     * @throws ClientResponseException
+     * @throws ServerResponseException
+     * @see    Iterator::next()
+     */
+    public function next(): void
+    {
+        /* @phpstan-ignore-next-line */
+        $this->currentScrolledResponse = $this->client->scroll(
+            [
+                'body' => [
+                    'scroll_id' => $this->scrollId,
+                    'scroll'    => $this->scrollTtl
+                ]
+            ]
+        )->asArray();
+        $this->scrollId = $this->currentScrolledResponse['_scroll_id'];
+        $this->currentKey++;
+    }
+
+    /**
+     * Returns a boolean value indicating if the current page is valid or not
+     *
+     * @return bool
+     * @see    Iterator::valid()
+     */
+    public function valid(): bool
+    {
+        return isset($this->currentScrolledResponse['hits']['hits'][0]);
+    }
+
+    /**
+     * Returns the current "page"
+     *
+     * @return array
+     * @see    Iterator::current()
+     */
+    public function current(): array
+    {
+        return $this->currentScrolledResponse;
+    }
+
+    /**
+     * Returns the current "page number" of the current "page"
+     *
+     * @return int
+     * @see    Iterator::key()
+     */
+    public function key(): int
+    {
+        return $this->currentKey;
+    }
+}