media: Port DjVuImage::retrieveMetaData() to use BoxedCommand

The shellouts to djvudump and djvutxt are combined with a wrapping shell
script, scripts/retrieveMetaData.sh so the djvu file only needs to be
transmitted to a Shellbox service once.

Windows users can change the new $wgDjvuShell setting to point to a
different shell.

Bug: T289228
Change-Id: Ic4e3dff246792692a952da0e8f9b4a83d1b6b963

RELEASE-NOTES-1.38

@@ -32,6 +32,9 @@ For notes on 1.36.x and older releases, see HISTORY.
 * $wgMaxImageArea - This setting may now be set to false to disable size
   checking before scaling. Extensions can still override its value by
   using the BitmapHandlerCheckImageArea hook.
+* $wgDjvuDump, $wgDjvuRenderer, $wgDjvuTxt now only accept explicit path.
+  'djvudump' will not work and must be changed to '/usr/bin/djvudump'
+  for example.
 * …
 
 ==== Removed configuration ====

includes/DefaultSettings.php

@@ -1790,21 +1790,21 @@ $wgImagePreconnect = false;
 /**
  * Path of the djvudump executable
  * Enable this and $wgDjvuRenderer to enable djvu rendering
- * example: $wgDjvuDump = 'djvudump';
+ * example: $wgDjvuDump = '/usr/bin/djvudump';
  */
 $wgDjvuDump = null;
 
 /**
  * Path of the ddjvu DJVU renderer
  * Enable this and $wgDjvuDump to enable djvu rendering
- * example: $wgDjvuRenderer = 'ddjvu';
+ * example: $wgDjvuRenderer = '/usr/bin/ddjvu';
  */
 $wgDjvuRenderer = null;
 
 /**
  * Path of the djvutxt DJVU text extraction utility
  * Enable this and $wgDjvuDump to enable text layer extraction from djvu files
- * example: $wgDjvuTxt = 'djvutxt';
+ * example: $wgDjvuTxt = '/usr/bin/djvutxt';
  */
 $wgDjvuTxt = null;
 
@@ -1820,6 +1820,14 @@ $wgDjvuPostProcessor = 'pnmtojpeg';
  */
 $wgDjvuOutputExtension = 'jpg';
 
+/**
+ * Shell to use when getting djvu metadata
+ *
+ * @var string
+ * @since 1.37
+ */
+$wgDjvuShell = '/bin/sh';
+
 // endregion -- end of DJvu
 
 // endregion -- end of file uploads

includes/media/DjVuImage.php

@@ -24,7 +24,7 @@
  * @ingroup Media
  */
 
-use MediaWiki\Shell\Shell;
+use MediaWiki\MediaWikiServices;
 
 /**
  * Support for detecting/validating DjVu image files and getting
@@ -249,28 +249,51 @@ class DjVuImage {
 	 * @return string|null|false
 	 */
 	public function retrieveMetaData() {
-		global $wgDjvuDump, $wgDjvuTxt;
+		global $wgDjvuDump, $wgDjvuTxt, $wgDjvuShell;
 
 		if ( !$this->isValid() ) {
 			return false;
 		}
 
-		if ( isset( $wgDjvuDump ) ) {
-			# djvudump is faster than djvutoxml (now abandoned) as of version 3.5
-			# https://sourceforge.net/p/djvu/bugs/71/
-			$cmd = Shell::escape( $wgDjvuDump ) . ' ' . Shell::escape( $this->mFilename );
-			$dump = wfShellExec( $cmd );
+		$command = MediaWikiServices::getInstance()->getShellCommandFactory()
+			->createBoxed( 'media' )
+			->disableNetwork()
+			->firejailDefaultSeccomp()
+			->routeName( 'djvu-metadata' );
+		$command
+			->params( $wgDjvuShell, 'scripts/retrieveMetaData.sh' )
+			->inputFileFromFile(
+				'scripts/retrieveMetaData.sh',
+				__DIR__ . '/scripts/retrieveMetaData.sh' )
+			->inputFileFromFile( 'file.djvu', $this->mFilename )
+			->memoryLimit( self::DJVUTXT_MEMORY_LIMIT );
+
+		$env = [];
+		if ( $wgDjvuDump !== null ) {
+			$env['DJVU_DUMP'] = $wgDjvuDump;
+			$command->outputFileToString( 'dump' );
+		}
+		if ( $wgDjvuTxt !== null ) {
+			$env['DJVU_TXT'] = $wgDjvuTxt;
+			$command->outputFileToString( 'txt' );
+			$command->outputFileToString( 'txt_exit_code' );
+		}
+
+		$result = $command
+			->environment( $env )
+			->execute();
+		if ( $wgDjvuDump !== null ) {
+			$dump = $result->getFileContents( 'dump' );
 			$xml = $this->convertDumpToXML( $dump );
 		} else {
 			$xml = null;
 		}
+
 		# Text layer
-		if ( isset( $wgDjvuTxt ) ) {
-			$cmd = Shell::escape( $wgDjvuTxt ) . ' --detail=page ' . Shell::escape( $this->mFilename );
-			wfDebug( __METHOD__ . ": $cmd" );
-			$retval = '';
-			$txt = wfShellExec( $cmd, $retval, [], [ 'memory' => self::DJVUTXT_MEMORY_LIMIT ] );
+		if ( $wgDjvuTxt !== null ) {
+			$retval = (int)trim( $result->getFileContents( 'txt_exit_code' ) );
 			if ( $retval == 0 ) {
+				$txt = $result->getFileContents( 'txt' );
 				# Strip some control characters
 				# Ignore carriage returns
 				$txt = preg_replace( "/\\\\013/", "", $txt );

includes/media/scripts/retrieveMetaData.sh

@@ -0,0 +1,29 @@
+#!/bin/sh
+
+# Get parameters from environment
+
+export DJVU_DUMP="${DJVU_DUMP:-djvudump}"
+export DJVU_TXT="${DJVU_TXT:-djvutxt}"
+
+runDump() {
+	# djvudump is faster than djvutoxml (now abandoned) as of version 3.5
+	# https://sourceforge.net/p/djvu/bugs/71/
+	"$DJVU_DUMP" file.djvu > dump
+}
+
+runTxt() {
+	# Text layer
+	"$DJVU_TXT" \
+		--detail=page \
+		file.djvu > txt
+	# Store exit code so we can use it later
+	echo $? > txt_exit_code
+}
+
+if [ -x "$DJVU_DUMP" ]; then
+	runDump
+fi
+
+if [ -x "$DJVU_TXT" ]; then
+	runTxt
+fi