From: Aaron Schulz <aschulz@wikimedia.org>
Date: Fri, 10 Aug 2012 05:23:44 +0000 (-0700)
Subject: [FileBackend] Added option for copy script to skip non-UTF8 filenames.
X-Git-Tag: 1.31.0-rc.0~22776^2
X-Git-Url: http://git.cyclocoop.org/%7B%24www_url%7Dadmin/compta/exercices/modifier.php?a=commitdiff_plain;h=eb65b839f8c8329198c9a1bf0b089b6c3a973ebd;p=lhc%2Fweb%2Fwiklou.git

[FileBackend] Added option for copy script to skip non-UTF8 filenames.

Change-Id: I9b48012d890c2c4589984576a31d84dde535ad5b
---

diff --git a/maintenance/copyFileBackend.php b/maintenance/copyFileBackend.php
index 70a923262e..2363d40ae3 100644
--- a/maintenance/copyFileBackend.php
+++ b/maintenance/copyFileBackend.php
@@ -45,6 +45,7 @@ class CopyFileBackend extends Maintenance {
 		$this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
 		$this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
 		$this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
+		$this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
 		$this->setBatchSize( 50 );
 	}
 
@@ -56,6 +57,10 @@ class CopyFileBackend extends Maintenance {
 
 		$rateFile = $this->getOption( 'ratefile' );
 
+		if ( $this->hasOption( 'utf8only' ) && !extension_loaded( 'mbstring' ) ) {
+			$this->error( "Cannot check for UTF-8, mbstring extension missing.", 1 ); // die
+		}
+
 		$count = 0;
 		foreach ( $containers as $container ) {
 			if ( $subDir != '' ) {
@@ -132,7 +137,10 @@ class CopyFileBackend extends Maintenance {
 		foreach ( $srcPathsRel as $srcPathRel ) {
 			$srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 			$dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
-			if ( $this->filesAreSame( $src, $dst, $srcPath, $dstPath ) ) {
+			if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
+				$this->error( "Detected illegal (non-UTF8) path for $srcPath." );
+				continue;
+			} elseif ( $this->filesAreSame( $src, $dst, $srcPath, $dstPath ) ) {
 				$this->output( "Already have $srcPathRel.\n" );
 				continue; // assume already copied...
 			}