From: Tim Starling <tstarling@wikimedia.org>
Date: Thu, 23 Mar 2017 01:51:43 +0000 (+1100)
Subject: Add benchmarkTidy.php, to benchmark tidy drivers
X-Git-Tag: 1.31.0-rc.0~3439^2
X-Git-Url: http://git.cyclocoop.org/%22%20.%20generer_url_ecrire%28%22auteur_infos%22%2C%22id_auteur=%24connect_id_auteur%22%29%20.%20%22?a=commitdiff_plain;h=448be2ed3e3d705696f1bade68c55a4fbc1b56bf;p=lhc%2Fweb%2Fwiklou.git

Add benchmarkTidy.php, to benchmark tidy drivers

Plus representative input file

Change-Id: I254793fc55c57a98c07ae1e4c27e6005965c9a20
---

diff --git a/autoload.php b/autoload.php
index ba5b77478e..27c27c5692 100644
--- a/autoload.php
+++ b/autoload.php
@@ -192,6 +192,7 @@ $wgAutoloadLocalClasses = [
 	'BenchmarkHooks' => __DIR__ . '/maintenance/benchmarks/benchmarkHooks.php',
 	'BenchmarkParse' => __DIR__ . '/maintenance/benchmarks/benchmarkParse.php',
 	'BenchmarkPurge' => __DIR__ . '/maintenance/benchmarks/benchmarkPurge.php',
+	'BenchmarkTidy' => __DIR__ . '/maintenance/benchmarks/benchmarkTidy.php',
 	'Benchmarker' => __DIR__ . '/maintenance/benchmarks/Benchmarker.php',
 	'BitmapHandler' => __DIR__ . '/includes/media/Bitmap.php',
 	'BitmapHandler_ClientOnly' => __DIR__ . '/includes/media/Bitmap_ClientOnly.php',
diff --git a/includes/parser/MWTidy.php b/includes/parser/MWTidy.php
index 01bf2d0d65..ffc884eb2d 100644
--- a/includes/parser/MWTidy.php
+++ b/includes/parser/MWTidy.php
@@ -83,7 +83,7 @@ class MWTidy {
 	/**
 	 * @return bool|\MediaWiki\Tidy\TidyDriverBase
 	 */
-	protected static function singleton() {
+	public static function singleton() {
 		global $wgUseTidy, $wgTidyInternal, $wgTidyConf, $wgDebugTidy, $wgTidyConfig,
 			$wgTidyBin, $wgTidyOpts;
 
diff --git a/maintenance/benchmarks/README b/maintenance/benchmarks/README
index c021abd293..27da9defd1 100644
--- a/maintenance/benchmarks/README
+++ b/maintenance/benchmarks/README
@@ -5,3 +5,6 @@ To get somehow accurate result, you might want to bound the PHP process
 to a specific CPU with `taskset` and raise its priority with `nice`. Example:
 
  $ taskset 1 nice -n-10 php bench_wfIsWindows.php
+
+australia-untidy.html.gz contains representative input text for
+benchmarkTidy.php. It needs to be decompressed before use.
diff --git a/maintenance/benchmarks/australia-untidy.html.gz b/maintenance/benchmarks/australia-untidy.html.gz
new file mode 100644
index 0000000000..148481da6f
Binary files /dev/null and b/maintenance/benchmarks/australia-untidy.html.gz differ
diff --git a/maintenance/benchmarks/benchmarkTidy.php b/maintenance/benchmarks/benchmarkTidy.php
new file mode 100644
index 0000000000..14791745fe
--- /dev/null
+++ b/maintenance/benchmarks/benchmarkTidy.php
@@ -0,0 +1,78 @@
+<?php
+
+require __DIR__ . '/../Maintenance.php';
+
+class BenchmarkTidy extends Maintenance {
+	public function __construct() {
+		parent::__construct();
+		$this->addOption( 'file', 'A filename which contains the input text', true, true );
+		$this->addOption( 'driver', 'The Tidy driver name, or false to use the configured instance',
+			false,  true );
+		$this->addOption( 'tidy-config', 'JSON encoded value for the tidy configuration array',
+			false, true );
+	}
+
+	public function execute() {
+		$html = file_get_contents( $this->getOption( 'file' ) );
+		if ( $html === false ) {
+			$this->error( "Unable to open input file", 1 );
+		}
+		if ( $this->hasOption( 'driver' ) || $this->hasOption( 'tidy-config' ) ) {
+			$config = json_decode( $this->getOption( 'tidy-config', '{}' ), true );
+			if ( !is_array( $config ) ) {
+				$this->error( "Invalid JSON tidy config", 1 );
+			}
+			$config += [ 'driver' => $this->getOption( 'driver', 'RemexHtml' ) ];
+			$driver = MWTidy::factory( $config );
+		} else {
+			$driver = MWTidy::singleton();
+			if ( !$driver ) {
+				$this->error( "Tidy disabled or not installed", 1 );
+			}
+		}
+
+		$this->benchmark( $driver, $html );
+	}
+
+	private function benchmark( $driver, $html ) {
+		global $wgContLang;
+
+		$times = [];
+		$innerCount = 10;
+		$outerCount = 10;
+		for ( $j = 1; $j <= $outerCount; $j++ ) {
+			$t = microtime( true );
+			for ( $i = 0; $i < $innerCount; $i++ ) {
+				$driver->tidy( $html );
+				print $wgContLang->formatSize( memory_get_usage( true ) ) . "\n";
+			}
+			$t = ( ( microtime( true ) - $t ) / $innerCount ) * 1000;
+			$times[] = $t;
+			print "Run $j: $t\n";
+		}
+		print "\n";
+
+		sort( $times, SORT_NUMERIC );
+		$n = $outerCount;
+		$min = $times[0];
+		$max = end( $times );
+		if ( $n % 2 ) {
+			$median = $times[ ( $n - 1 ) / 2 ];
+		} else {
+			$median = ( $times[$n / 2] + $times[$n / 2 - 1] ) / 2;
+		}
+		$mean = array_sum( $times ) / $n;
+
+		print "Minimum: $min ms\n";
+		print "Median: $median ms\n";
+		print "Mean: $mean ms\n";
+		print "Maximum: $max ms\n";
+		print "Memory usage: " .
+			$wgContLang->formatSize( memory_get_usage( true ) ) . "\n";
+		print "Peak memory usage: " .
+			$wgContLang->formatSize( memory_get_peak_usage( true ) ) . "\n";
+	}
+}
+
+$maintClass = 'BenchmarkTidy';
+require RUN_MAINTENANCE_IF_MAIN;