Fix paginating with multiple initial inputs

Reset paginator state after finishing paginating for one base input,
to enable paginating multiple listings of the same structure.

Also fix a warning when running tests, because of a final new line in
a demo cache file and add memory limit option when running phpstan.

.editorconfig

@@ -15,3 +15,6 @@ trim_trailing_whitespace = false
 
 [*.yml]
 indent_size = 2
+
+[_cachefilecontent]
+insert_final_newline = false

CHANGELOG.md

@@ -6,6 +6,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.3.4] - 2023-12-19
+### Fixed
+* Reset paginator state after finishing paginating for one base input, to enable paginating multiple listings of the same structure.
+
 ## [1.3.3] - 2023-12-01
 ### Fixed
 * Add forgotten getter method to get the DOM query that is attached to an `InvalidDomQueryException` instance.

composer.json

@@ -69,7 +69,7 @@
     "scripts": {
         "test": "pest --exclude-group integration",
         "test-integration": "pest --group integration",
-        "stan": "phpstan analyse",
+        "stan": "@php -d memory_limit=4G vendor/bin/phpstan analyse",
         "cs": "php-cs-fixer fix -v --dry-run",
         "cs-fix": "php-cs-fixer fix -v",
         "add-git-hooks": "@php bin/add-git-hooks"

src/Steps/Loading/Http/AbstractPaginator.php

@@ -44,6 +44,25 @@ public function hasFinished(): bool
         return $this->hasFinished || $this->maxPagesReached();
     }
 
+    /**
+     * When a paginate step is called with multiple inputs, like:
+     *
+     * ['https://www.example.com/listing1', 'https://www.example.com/listing2', ...]
+     *
+     * it always has to start paginating again for each listing base URL.
+     * Therefore, we reset the state after finishing paginating one base input.
+     * Except for $this->found, because if it would be the case that the exact same pages are
+     * discovered whilst paginating, we don't want to load the exact same pages again and again.
+     */
+    public function resetFinished(): void
+    {
+        $this->hasFinished = false;
+
+        $this->loadedCount = 0;
+
+        $this->latestRequest = null;
+    }
+
     public function stopWhen(Closure|StopRule $callback): self
     {
         $this->stopRules[] = $callback;

src/Steps/Loading/Http/Paginate.php

@@ -70,6 +70,10 @@ protected function invoke(mixed $input): Generator
 
         if ($this->logger) {
             $this->paginator->logWhenFinished($this->logger);
+
+            if (method_exists($this->paginator, 'resetFinished')) {
+                $this->paginator->resetFinished();
+            }
         }
     }
 

tests/Cache/_cachefilecontent

@@ -217,4 +217,4 @@ $query = Query::fromString('foo=1&bar=2&baz=3&quz=4')
 <script>hljs.highlightAll();</script>
 </body>
 </html>
-";}
+";}

tests/_Integration/Http/PaginationTest.php

@@ -51,3 +51,15 @@ protected function loader(UserAgentInterface $userAgent, LoggerInterface $logger
 
     expect($this->getActualOutputForAssertion())->toContain('Max pages limit reached');
 });
+
+it('resets the finished paginating state after each processed (/paginated) input', function () {
+    $crawler = new PaginationCrawler();
+
+    $crawler
+        ->inputs(['http://localhost:8000/paginated-listing', 'http://localhost:8000/paginated-listing?foo=bar'])
+        ->addStep(Http::get()->paginate('#pagination', 2)->outputKey('response'));
+
+    $results = helper_generatorToArray($crawler->run());
+
+    expect($results)->toHaveCount(4);
+});

tests/_Integration/Http/QueryParamPaginationTest.php

@@ -89,3 +89,25 @@ protected function loader(UserAgentInterface $userAgent, LoggerInterface $logger
 
     expect($results)->toHaveCount(2);
 });
+
+it('resets the finished paginating state after each processed (/paginated) input', function () {
+    $crawler = new QueryParamPaginationCrawler();
+
+    $crawler
+        ->inputs([
+            'http://localhost:8000/query-param-pagination?page=1',
+            'http://localhost:8000/query-param-pagination?page=1&foo=bar',
+        ])
+        ->addStep(
+            Http::get()
+                ->paginate(
+                    QueryParamsPaginator::paramsInUrl(2)
+                        ->increase('page')
+                        ->stopWhen(PaginatorStopRules::isEmptyInJson('data.items'))
+                )->addToResult(['body'])
+        );
+
+    $results = helper_generatorToArray($crawler->run());
+
+    expect($results)->toHaveCount(4);
+});

tests/_Integration/_Server/PaginatedListing.php

@@ -7,8 +7,13 @@
 <div id="listing">
     <?php
         $page = $_GET['page'] ?? 1;
+        $additionalFooQueryParam = '';
         $itemsPerPage = 3;
 
+        if (!empty($_GET['foo'])) {
+            $additionalFooQueryParam = '&foo=' . $_GET['foo'];
+        }
+
         if ($page < 4) {
             for ($i = 1; $i < 4; $i++) {
                 $itemNumber = (($page - 1) * $itemsPerPage) + $i; ?>
@@ -27,11 +32,11 @@
 
     <div id="pagination">
         <?php if ($page > 1) { ?>
-            <a id="prevPage" href="/paginated-listing?page=<?=($page - 1)?>">&lt;&lt;</a>
+            <a id="prevPage" href="/paginated-listing?page=<?=($page - 1) . $additionalFooQueryParam?>">&lt;&lt;</a>
         <?php } ?>
 
         <?php if ($page < 4) { ?>
-            <a id="nextPage" href="/paginated-listing?page=<?=($page + 1)?>">&gt;&gt;</a>
+            <a id="nextPage" href="/paginated-listing?page=<?=($page + 1) . $additionalFooQueryParam?>">&gt;&gt;</a>
         <?php } ?>
     </div>
 </div>