Improve logging for Pdf's retrieveMetadata.sh

- Don't check for file execution, but for command execution. This way
  pdfinfo and pdftext work without specifying the path
- Only pipe the stdout content of the commands to the outputfiles
- Exit as failure when the pdfinfo command is available, but it's
  execution failed
- Check and log the error output of retrieveMetadata.sh

Bug: T299521
Change-Id: Ia072469f4df6cce51793ab48823c7f4e4e13997b

includes/PdfImage.php

@@ -132,6 +132,13 @@ class PdfImage {
 		MediaWikiServices::getInstance()->getStatsdDataFactory()
 			->increment( 'pdfhandler.shell.retrieve_meta_data' );
 
+		// Metadata retrieval is allowed to fail, but we'd like to know why
+		if ( $result->getExitCode() != 0 ) {
+			wfDebug( __METHOD__ . ': retrieveMetaData.sh' .
+			"\n\nExitcode: " . $result->getExitCode() . "\n\n"
+			. $result->getStderr() );
+		}
+
 		$resultMeta = $result->getFileContents( 'meta' );
 		$resultPages = $result->getFileContents( 'pages' );
 		if ( $resultMeta !== null || $resultPages !== null ) {

scripts/retrieveMetaData.sh

@@ -13,14 +13,23 @@ runInfo() {
 	"$PDFHANDLER_INFO" \
 		-enc 'UTF-8' \
 		-meta \
-		file.pdf > meta
+		file.pdf 1> meta
+
+	# Check for errors and forward them
+	if [ $? -ne 0 ]; then
+		exit 1;
+	fi
 
 	# Report metadata as UTF-8 text...and report page sizes for all pages
 	"$PDFHANDLER_INFO" \
 		-enc 'UTF-8' \
 		-l 9999999 \
-		file.pdf > pages
+		file.pdf 1> pages
 
+	# Check for errors and forward them
+	if [ $? -ne 0 ]; then
+		exit 1;
+	fi
 }
 
 runToText() {
@@ -34,10 +43,10 @@ runToText() {
 	echo $? > text_exit_code
 }
 
-if [ -x "$PDFHANDLER_INFO" ]; then
+if [ -x "$(command -v $PDFHANDLER_INFO)" ]; then
 	runInfo
 fi
 
-if [ -x "$PDFHANDLER_TOTEXT" ]; then
+if [ -x "$(command -v $PDFHANDLER_TOTEXT)" ]; then
 	runToText
 fi