perf tests in development branch

tests/Spout/Reader/CSV/ReaderPerfTest.php

@@ -0,0 +1,62 @@
+<?php
+
+namespace Box\Spout\Reader\CSV;
+
+use Box\Spout\Common\Type;
+use Box\Spout\Reader\Common\Creator\ReaderEntityFactory;
+use Box\Spout\TestUsingResource;
+use PHPUnit\Framework\TestCase;
+
+/**
+ * Class ReaderPerfTest
+ * Performance tests for CSV Reader
+ */
+class ReaderPerfTest extends TestCase
+{
+    use TestUsingResource;
+
+    /**
+     * 1 million rows (each row containing 3 cells) should be read
+     * in less than 1 minute and the execution should not require
+     * more than 1MB of memory
+     *
+     * @group perf-tests
+     *
+     * @return void
+     */
+    public function testPerfWhenReadingOneMillionRowsCSV()
+    {
+        // getting current memory peak to avoid taking into account the memory used by PHPUnit
+        $beforeMemoryPeakUsage = memory_get_peak_usage(true);
+
+        $expectedMaxExecutionTime = 60; // 1 minute in seconds
+        $expectedMaxMemoryPeakUsage = 1 * 1024 * 1024; // 1MB in bytes
+        $startTime = time();
+
+        $fileName = 'csv_with_one_million_rows.csv';
+        $resourcePath = $this->getResourcePath($fileName);
+
+        $reader = ReaderEntityFactory::createReader(Type::CSV);
+        $reader->open($resourcePath);
+
+        $numReadRows = 0;
+
+        /** @var Sheet $sheet */
+        foreach ($reader->getSheetIterator() as $sheet) {
+            foreach ($sheet->getRowIterator() as $row) {
+                $numReadRows++;
+            }
+        }
+
+        $reader->close();
+
+        $expectedNumRows = 1000000;
+        $this->assertEquals($expectedNumRows, $numReadRows, "$expectedNumRows rows should have been read");
+
+        $executionTime = time() - $startTime;
+        $this->assertTrue($executionTime < $expectedMaxExecutionTime, "Reading 1 million rows should take less than $expectedMaxExecutionTime seconds (took $executionTime seconds)");
+
+        $memoryPeakUsage = memory_get_peak_usage(true) - $beforeMemoryPeakUsage;
+        $this->assertTrue($memoryPeakUsage < $expectedMaxMemoryPeakUsage, 'Reading 1 million rows should require less than ' . ($expectedMaxMemoryPeakUsage / 1024 / 1024) . ' MB of memory (required ' . round($memoryPeakUsage / 1024 / 1024, 2) . ' MB)');
+    }
+}

tests/Spout/Reader/ODS/ReaderPerfTest.php

@@ -0,0 +1,62 @@
+<?php
+
+namespace Box\Spout\Reader\ODS;
+
+use Box\Spout\Common\Type;
+use Box\Spout\Reader\Common\Creator\ReaderEntityFactory;
+use Box\Spout\TestUsingResource;
+use PHPUnit\Framework\TestCase;
+
+/**
+ * Class ReaderPerfTest
+ * Performance tests for ODS Reader
+ */
+class ReaderPerfTest extends Testcase
+{
+    use TestUsingResource;
+
+    /**
+     * 1 million rows (each row containing 3 cells) should be read
+     * in less than 10 minutes and the execution should not require
+     * more than 1MB of memory
+     *
+     * @group perf-tests
+     *
+     * @return void
+     */
+    public function testPerfWhenReadingOneMillionRowsODS()
+    {
+        // getting current memory peak to avoid taking into account the memory used by PHPUnit
+        $beforeMemoryPeakUsage = memory_get_peak_usage(true);
+
+        $expectedMaxExecutionTime = 600; // 10 minutes in seconds
+        $expectedMaxMemoryPeakUsage = 1 * 1024 * 1024; // 1MB in bytes
+        $startTime = time();
+
+        $fileName = 'ods_with_one_million_rows.ods';
+        $resourcePath = $this->getResourcePath($fileName);
+
+        $reader = ReaderEntityFactory::createReader(Type::ODS);
+        $reader->open($resourcePath);
+
+        $numReadRows = 0;
+
+        /** @var Sheet $sheet */
+        foreach ($reader->getSheetIterator() as $sheet) {
+            foreach ($sheet->getRowIterator() as $row) {
+                $numReadRows++;
+            }
+        }
+
+        $reader->close();
+
+        $expectedNumRows = 1000000;
+        $this->assertEquals($expectedNumRows, $numReadRows, "$expectedNumRows rows should have been read");
+
+        $executionTime = time() - $startTime;
+        $this->assertTrue($executionTime < $expectedMaxExecutionTime, "Reading 1 million rows should take less than $expectedMaxExecutionTime seconds (took $executionTime seconds)");
+
+        $memoryPeakUsage = memory_get_peak_usage(true) - $beforeMemoryPeakUsage;
+        $this->assertTrue($memoryPeakUsage < $expectedMaxMemoryPeakUsage, 'Reading 1 million rows should require less than ' . ($expectedMaxMemoryPeakUsage / 1024 / 1024) . ' MB of memory (required ' . round($memoryPeakUsage / 1024 / 1024, 2) . ' MB)');
+    }
+}

tests/Spout/Reader/XLSX/ReaderPerfTest.php

@@ -0,0 +1,76 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX;
+
+use Box\Spout\Common\Type;
+use Box\Spout\Reader\Common\Creator\ReaderEntityFactory;
+use Box\Spout\TestUsingResource;
+use PHPUnit\Framework\TestCase;
+
+/**
+ * Class ReaderPerfTest
+ * Performance tests for XLSX Reader
+ */
+class ReaderPerfTest extends TestCase
+{
+    use TestUsingResource;
+
+    /**
+     * @return array
+     */
+    public function dataProviderForTestPerfWhenReading300kRowsXLSX()
+    {
+        return [
+            [$shouldUseInlineStrings = true, $expectedMaxExecutionTime = 390], // 6.5 minutes in seconds
+            [$shouldUseInlineStrings = false, $expectedMaxExecutionTime = 600], // 10 minutes in seconds
+        ];
+    }
+
+    /**
+     * 300,000 rows (each row containing 3 cells) should be read
+     * in less than 6.5 minutes for inline strings, 10 minutes for
+     * shared strings and the execution should not require
+     * more than 3MB of memory.
+     *
+     * @dataProvider dataProviderForTestPerfWhenReading300kRowsXLSX
+     * @group perf-tests
+     *
+     * @param bool $shouldUseInlineStrings
+     * @param int $expectedMaxExecutionTime
+     * @return void
+     */
+    public function testPerfWhenReading300kRowsXLSX($shouldUseInlineStrings, $expectedMaxExecutionTime)
+    {
+        // getting current memory peak to avoid taking into account the memory used by PHPUnit
+        $beforeMemoryPeakUsage = memory_get_peak_usage(true);
+
+        $expectedMaxMemoryPeakUsage = 3 * 1024 * 1024;
+        $startTime = time();
+
+        $fileName = ($shouldUseInlineStrings) ? 'xlsx_with_300k_rows_and_inline_strings.xlsx' : 'xlsx_with_300k_rows_and_shared_strings.xlsx';
+        $resourcePath = $this->getResourcePath($fileName);
+
+        $reader = ReaderEntityFactory::createReader(Type::XLSX);
+        $reader->open($resourcePath);
+
+        $numReadRows = 0;
+
+        /** @var Sheet $sheet */
+        foreach ($reader->getSheetIterator() as $sheet) {
+            foreach ($sheet->getRowIterator() as $row) {
+                $numReadRows++;
+            }
+        }
+
+        $reader->close();
+
+        $expectedNumRows = 300000;
+        $this->assertEquals($expectedNumRows, $numReadRows, "$expectedNumRows rows should have been read");
+
+        $executionTime = time() - $startTime;
+        $this->assertTrue($executionTime < $expectedMaxExecutionTime, "Reading 300,000 rows should take less than $expectedMaxExecutionTime seconds (took $executionTime seconds)");
+
+        $memoryPeakUsage = memory_get_peak_usage(true) - $beforeMemoryPeakUsage;
+        $this->assertTrue($memoryPeakUsage < $expectedMaxMemoryPeakUsage, 'Reading 300,000 rows should require less than ' . ($expectedMaxMemoryPeakUsage / 1024 / 1024) . ' MB of memory (required ' . ($memoryPeakUsage / 1024 / 1024) . ' MB)');
+    }
+}