summaryrefslogtreecommitdiff
path: root/maintenance/populateImageSha1.php
diff options
context:
space:
mode:
Diffstat (limited to 'maintenance/populateImageSha1.php')
-rw-r--r--maintenance/populateImageSha1.php74
1 files changed, 60 insertions, 14 deletions
diff --git a/maintenance/populateImageSha1.php b/maintenance/populateImageSha1.php
index 4cd2747c..37429a34 100644
--- a/maintenance/populateImageSha1.php
+++ b/maintenance/populateImageSha1.php
@@ -17,15 +17,22 @@
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
+ * @file
* @ingroup Maintenance
*/
-require_once( dirname( __FILE__ ) . '/Maintenance.php' );
+require_once( __DIR__ . '/Maintenance.php' );
+/**
+ * Maintenance script to populate the img_sha1 field.
+ *
+ * @ingroup Maintenance
+ */
class PopulateImageSha1 extends LoggedUpdateMaintenance {
public function __construct() {
parent::__construct();
$this->mDescription = "Populate the img_sha1 field";
+ $this->addOption( 'force', "Recalculate sha1 for rows that already have a value" );
$this->addOption( 'method', "Use 'pipe' to pipe to mysql command line,\n" .
"\t\tdefault uses Database class", false, true );
$this->addOption( 'file', 'Fix for a specific file, without File: namespace prefixed', false, true );
@@ -49,11 +56,13 @@ class PopulateImageSha1 extends LoggedUpdateMaintenance {
public function doDBUpdates() {
$method = $this->getOption( 'method', 'normal' );
- $file = $this->getOption( 'file' );
+ $file = $this->getOption( 'file', '' );
+ $force = $this->getOption( 'force' );
+ $isRegen = ( $force || $file != '' ); // forced recalculation?
$t = -microtime( true );
$dbw = wfGetDB( DB_MASTER );
- if ( $file ) {
+ if ( $file != '' ) {
$res = $dbw->select(
'image',
array( 'img_name' ),
@@ -66,15 +75,24 @@ class PopulateImageSha1 extends LoggedUpdateMaintenance {
}
$this->output( "Populating img_sha1 field for specified files\n" );
} else {
- $res = $dbw->select( 'image',
- array( 'img_name' ), array( 'img_sha1' => '' ), __METHOD__ );
- $this->output( "Populating img_sha1 field\n" );
+ if ( $force ) {
+ $conds = array();
+ $this->output( "Populating and recalculating img_sha1 field\n" );
+ } else {
+ $conds = array( 'img_sha1' => '' );
+ $this->output( "Populating img_sha1 field\n" );
+ }
+ $res = $dbw->select( 'image', array( 'img_name' ), $conds, __METHOD__ );
}
$imageTable = $dbw->tableName( 'image' );
+ $oldImageTable = $dbw->tableName( 'oldimage' );
if ( $method == 'pipe' ) {
- // @todo FIXME: Kill this and replace with a second unbuffered DB connection.
+ // Opening a pipe allows the SHA-1 operation to be done in parallel
+ // with the database write operation, because the writes are queued
+ // in the pipe buffer. This can improve performance by up to a
+ // factor of 2.
global $wgDBuser, $wgDBserver, $wgDBpassword, $wgDBname;
$cmd = 'mysql -u' . wfEscapeShellArg( $wgDBuser ) .
' -h' . wfEscapeShellArg( $wgDBserver ) .
@@ -87,21 +105,49 @@ class PopulateImageSha1 extends LoggedUpdateMaintenance {
$i = 0;
foreach ( $res as $row ) {
if ( $i % $this->mBatchSize == 0 ) {
- $this->output( sprintf( "Done %d of %d, %5.3f%% \r", $i, $numRows, $i / $numRows * 100 ) );
+ $this->output( sprintf(
+ "Done %d of %d, %5.3f%% \r", $i, $numRows, $i / $numRows * 100 ) );
wfWaitForSlaves();
}
$file = wfLocalFile( $row->img_name );
if ( !$file ) {
continue;
}
+ // Upgrade the current file version...
$sha1 = $file->getRepo()->getFileSha1( $file->getPath() );
- if ( strval( $sha1 ) !== '' ) {
- $sql = "UPDATE $imageTable SET img_sha1=" . $dbw->addQuotes( $sha1 ) .
- " WHERE img_name=" . $dbw->addQuotes( $row->img_name );
- if ( $method == 'pipe' ) {
- fwrite( $pipe, "$sql;\n" );
+ if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
+ if ( $isRegen && $file->getSha1() !== $sha1 ) {
+ // The population was probably done already. If the old SHA1
+ // does not match, then both fix the SHA1 and the metadata.
+ $file->upgradeRow();
} else {
- $dbw->query( $sql, __METHOD__ );
+ $sql = "UPDATE $imageTable SET img_sha1=" . $dbw->addQuotes( $sha1 ) .
+ " WHERE img_name=" . $dbw->addQuotes( $file->getName() );
+ if ( $method == 'pipe' ) {
+ fwrite( $pipe, "$sql;\n" );
+ } else {
+ $dbw->query( $sql, __METHOD__ );
+ }
+ }
+ }
+ // Upgrade the old file versions...
+ foreach ( $file->getHistory() as $oldFile ) {
+ $sha1 = $oldFile->getRepo()->getFileSha1( $oldFile->getPath() );
+ if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
+ if ( $isRegen && $oldFile->getSha1() !== $sha1 ) {
+ // The population was probably done already. If the old SHA1
+ // does not match, then both fix the SHA1 and the metadata.
+ $oldFile->upgradeRow();
+ } else {
+ $sql = "UPDATE $oldImageTable SET oi_sha1=" . $dbw->addQuotes( $sha1 ) .
+ " WHERE (oi_name=" . $dbw->addQuotes( $oldFile->getName() ) . " AND" .
+ " oi_archive_name=" . $dbw->addQuotes( $oldFile->getArchiveName() ) . ")";
+ if ( $method == 'pipe' ) {
+ fwrite( $pipe, "$sql;\n" );
+ } else {
+ $dbw->query( $sql, __METHOD__ );
+ }
+ }
}
}
$i++;