Fixes Out-of-memory errors during Scrutinizer run (#443)

* run gc_collect_cycles in testParseFile to avoid out-of-memory error

* Added gc_collect_cycles() to setUp

* Update ParserTest.php

* Update ParserTest.php

* simplified ParserTest::testParseFile

only test a given set of files, not all in samples/bugs

* check if scrutinizer fails only when executing testParseFile

* restored ParserTest::testParseFile; refined testRetainImageContentImpact

* restored old version of testParseFile; marked it and other tests as memory-heavy ...

... exclude all memory-heavy tests from scrutinizer run

* refined memory checks in testRetainImageContentImpact

* increased memory value in testRetainImageContentImpact: Reason is, that it otherwise fails in Github because
more memory is used as in local development environment. Its still not an error!

.scrutinizer.yml

@@ -9,7 +9,7 @@ build:
                         # to not enough memory.
                         # @https://github.com/smalot/pdfparser/issues/410
                         # @https://github.com/smalot/pdfparser/pull/412
-                        command: make prepare-for-scrutinizer && make install-dev-tools && make run-phpunit ARGS="--coverage-clover coverage/clover.xml"
+                        command: make prepare-for-scrutinizer && make install-dev-tools && make run-phpunit ARGS="--exclude-group memory-heavy --coverage-clover coverage/clover.xml"
                         coverage:
                             file: coverage/clover.xml
                             format: clover

tests/Integration/ParserTest.php

@@ -48,6 +48,11 @@ protected function setUp(): void
         $this->fixture = new Parser();
     }
 
+    /**
+     * Notice: it may fail to run in Scrutinizer because of memory limitations.
+     *
+     * @group memory-heavy
+     */
     public function testParseFile()
     {
         $directory = $this->rootDir.'/samples/bugs';
@@ -305,6 +310,8 @@ public function testUsageOfConfigObject()
     /**
      * Tests the impact of the retainImageContent config setting on memory usage
      *
+     * @group memory-heavy
+     *
      * @see https://github.com/smalot/pdfparser/issues/104#issuecomment-883422508
      */
     public function testRetainImageContentImpact()
@@ -314,29 +321,46 @@ public function testRetainImageContentImpact()
         }
 
         $filename = $this->rootDir.'/samples/bugs/Issue104a.pdf';
-        $iterations = 2;
+        $iterations = 1;
 
-        // check default (= true)
+        /*
+         * check default (= true)
+         */
         $this->fixture = new Parser([]);
         $this->assertTrue($this->fixture->getConfig()->getRetainImageContent());
+        $document = null;
 
         for ($i = 0; $i < $iterations; ++$i) {
             $document = $this->fixture->parseFile($filename);
         }
 
-        $this->assertTrue(memory_get_usage() > 200000000);
+        $usedMemory = memory_get_usage(true);
+        $this->assertTrue($usedMemory > 100000000, 'Memory is only '.$usedMemory);
+        $this->assertTrue(null != $document && 0 < \strlen($document->getText()));
 
         // force garbage collection
         $this->fixture = $document = null;
         gc_collect_cycles();
 
-        // check false
+        /*
+         * check false
+         */
         $config = new Config();
         $config->setRetainImageContent(false);
         $this->fixture = new Parser([], $config);
         $this->assertEquals($config, $this->fixture->getConfig());
 
-        $this->assertTrue(memory_get_usage() < 200000000);
+        for ($i = 0; $i < $iterations; ++$i) {
+            $document = $this->fixture->parseFile($filename);
+        }
+
+        $usedMemory = memory_get_usage(true);
+        /*
+         * note: the following memory value is set manually and may differ from system to system.
+         *       it must be high enough to not produce a false negative though.
+         */
+        $this->assertTrue($usedMemory < 106000000, 'Memory is '.$usedMemory);
+        $this->assertTrue(0 < \strlen($document->getText()));
     }
 }