Most indexers now index UTF-8 on Linux

git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@5525 c91229c3-7414-0410-bfa2-8a42b809f60b

Most indexers now index UTF-8 on Linux
git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@5525 c91229c3-7414-0410-bfa2-8a42b809f60b
Bryn Divey
1 parent 3b597345
Showing 5 changed files with 11 additions and 7 deletions
plugins/ktstandard/contents/BaseIndexer.php
plugins/ktstandard/contents/ExcelIndexer.php
plugins/ktstandard/contents/PdfIndexer.php
plugins/ktstandard/contents/RtfIndexer.php
plugins/ktstandard/contents/WordIndexer.php
@@ -99,8 +99,8 @@ class KTBaseIndexerTrigger {
         $contents = $this->extract_contents($intermediate, $myfilename);
-        unlink($myfilename);
-        if (OS_WINDOWS) { unlink($intermediate); }
+        @unlink($myfilename);
+        if (OS_WINDOWS) { @unlink($intermediate); }
         if (empty($contents)) {
             return;
         }
@@ -136,7 +136,7 @@ class KTBaseIndexerTrigger {
         }
         KTUtil::pexec($cmdline, $aOptions);
         $contents = file_get_contents($sTempFilename);
-        
+
         return $contents;
     }
 }
@@ -36,13 +36,14 @@ class KTExcelIndexerTrigger extends KTBaseIndexerTrigger {
     );
     var $command = 'xls2csv';          // could be any application.
     var $commandconfig = 'indexer/xls2csv';          // could be any application.
-    var $args = array("-q", "0", "-c", " ");
+    var $args = array("-d", "UTF-8", "-q", "0", "-c", " ");
     var $use_pipes = true;
     // see BaseIndexer for how the extraction works.
     //
     function extract_contents($sFilename, $sTempFilename) {
         if (!OS_WINDOWS) {
+	    putenv('LANG=en_US.UTF-8');
             $res = parent::extract_contents($sFilename, $sTempFilename);
             if (!empty($res)) {
                 return $res;
@@ -34,11 +34,12 @@ class KTPdfIndexerTrigger extends KTBaseIndexerTrigger {
     );
     var $command = 'pdftotext';          // could be any application.
     var $commandconfig = 'indexer/pdftotext';          // could be any application.
-    var $args = array("-nopgbrk");
+    var $args = array("-nopgbrk", "-enc", "UTF-8");
     var $use_pipes = false;
     // see BaseIndexer for how the extraction works.
     function findLocalCommand() {   
+	putenv('LANG=en_US.UTF-8');
         $sCommand = KTUtil::findCommand($this->commandconfig, $this->command);
         return $sCommand;
     }    
@@ -34,11 +34,12 @@ class KTRtfIndexerTrigger extends KTBaseIndexerTrigger {
     );
     var $command = 'catdoc';          // could be any application.
     var $commandconfig = 'indexer/catdoc';          // could be any application.
-    var $args = array("-w", "-d", "utf-8");
+    var $args = array("-w", "-d", "UTF-8");
     var $use_pipes = true;
     function findLocalCommand() {
         $sCommand = KTUtil::findCommand($this->commandconfig, $this->command);
+	putenv('LANG=en_US.UTF-8');
         return $sCommand;
     }
@@ -34,7 +34,7 @@ class KTWordIndexerTrigger extends KTBaseIndexerTrigger {
     );
     var $command = 'catdoc';          // could be any application.
     var $commandconfig = 'indexer/catdoc';          // could be any application.
-    var $args = array("-w");
+    var $args = array("-w", "-d", "UTF-8");
     var $use_pipes = true;
     function extract_contents($sFilename, $sTempFilename) {
@@ -47,6 +47,7 @@ class KTWordIndexerTrigger extends KTBaseIndexerTrigger {
 	    $sDir = dirname(dirname($sCommand));
 	    putenv('HOME=' . $sDir);
         }
+	putenv('LANG=en_US.UTF-8');
         return parent::extract_contents($sFilename, $sTempFilename);
     }