X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=maintenance%2FcopyFileBackend.php;h=b39ff55eda1b68697dfc8677b3b2b1d0cbbf12c4;hb=78b6bde01f1ff2af9c20bf624908f65666347639;hp=09b9295e1a16e57bbe2f3d8c167f654663c81e71;hpb=93131fcd385cdae84ecb7ccb6faa1c7340214211;p=lhc%2Fweb%2Fwiklou.git

diff --git a/maintenance/copyFileBackend.php b/maintenance/copyFileBackend.php
index 09b9295e1a..b39ff55eda 100644
--- a/maintenance/copyFileBackend.php
+++ b/maintenance/copyFileBackend.php
@@ -35,7 +35,7 @@ require_once __DIR__ . '/Maintenance.php';
  * @ingroup Maintenance
  */
 class CopyFileBackend extends Maintenance {
-	/** @var Array|null (path sha1 => stat) Pre-computed dst stat entries from listings */
+	/** @var array|null (path sha1 => stat) Pre-computed dst stat entries from listings */
 	protected $statCache = null;
 
 	public function __construct() {
@@ -226,8 +226,8 @@ class CopyFileBackend extends Maintenance {
 			}
 			$t_start = microtime( true );
 			$fsFiles = $src->getLocalReferenceMulti( array( 'srcs' => $srcPaths, 'latest' => 1 ) );
-			$ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
-			$this->output( "\n\tDownloaded these file(s) [{$ellapsed_ms}ms]:\n\t" .
+			$elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
+			$this->output( "\n\tDownloaded these file(s) [{$elapsed_ms}ms]:\n\t" .
 				implode( "\n\t", $srcPaths ) . "\n\n" );
 		}
 
@@ -281,12 +281,12 @@ class CopyFileBackend extends Maintenance {
 			sleep( 10 ); // wait and retry copy again
 			$status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 		}
-		$ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
+		$elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 		if ( !$status->isOK() ) {
 			$this->error( print_r( $status->getErrorsArray(), true ) );
 			$this->error( "$wikiId: Could not copy file batch.", 1 ); // die
 		} elseif ( count( $copiedRel ) ) {
-			$this->output( "\n\tCopied these file(s) [{$ellapsed_ms}ms]:\n\t" .
+			$this->output( "\n\tCopied these file(s) [{$elapsed_ms}ms]:\n\t" .
 				implode( "\n\t", $copiedRel ) . "\n\n" );
 		}
 	}
@@ -318,12 +318,12 @@ class CopyFileBackend extends Maintenance {
 			sleep( 10 ); // wait and retry copy again
 			$status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 		}
-		$ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
+		$elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 		if ( !$status->isOK() ) {
 			$this->error( print_r( $status->getErrorsArray(), true ) );
 			$this->error( "$wikiId: Could not delete file batch.", 1 ); // die
 		} elseif ( count( $deletedRel ) ) {
-			$this->output( "\n\tDeleted these file(s) [{$ellapsed_ms}ms]:\n\t" .
+			$this->output( "\n\tDeleted these file(s) [{$elapsed_ms}ms]:\n\t" .
 				implode( "\n\t", $deletedRel ) . "\n\n" );
 		}
 	}
@@ -347,15 +347,34 @@ class CopyFileBackend extends Maintenance {
 		} else {
 			$dstStat = $dst->getFileStat( array( 'src' => $dPath ) );
 		}
-		return (
+		// Initial fast checks to see if files are obviously different
+		$sameFast = (
 			is_array( $srcStat ) // sanity check that source exists
 			&& is_array( $dstStat ) // dest exists
 			&& $srcStat['size'] === $dstStat['size']
-			&& ( !$skipHash || $srcStat['mtime'] <= $dstStat['mtime'] )
-			&& ( $skipHash || $src->getFileSha1Base36( array( 'src' => $sPath, 'latest' => 1 ) )
-				=== $dst->getFileSha1Base36( array( 'src' => $dPath, 'latest' => 1 ) )
-			)
 		);
+		// More thorough checks against files
+		if ( !$sameFast ) {
+			$same = false; // no need to look farther
+		} elseif ( isset( $srcStat['md5'] ) && isset( $dstStat['md5'] ) ) {
+			// If MD5 was already in the stat info, just use it.
+			// This is useful as many objects stores can return this in object listing,
+			// so we can use it to avoid slow per-file HEADs.
+			$same = ( $srcStat['md5'] === $dstStat['md5'] );
+		} elseif ( $skipHash ) {
+			// This mode is good for copying to a backup location or resyncing clone
+			// backends in FileBackendMultiWrite (since they get writes second, they have
+			// higher timestamps). However, when copying the other way, this hits loads of
+			// false positives (possibly 100%) and wastes a bunch of time on GETs/PUTs.
+			$same = ( $srcStat['mtime'] <= $dstStat['mtime'] );
+		} else {
+			// This is the slowest method which does many per-file HEADs (unless an object
+			// store tracks SHA-1 in listings).
+			$same = ( $src->getFileSha1Base36( array( 'src' => $sPath, 'latest' => 1 ) )
+				=== $dst->getFileSha1Base36( array( 'src' => $dPath, 'latest' => 1 ) ) );
+		}
+
+		return $same;
 	}
 }