Item14293: soffice/libreoffice indexer for ppt

foswiki · Jan 23, 2017 · 72b121e · 72b121e
1 parent c7797fe
commit 72b121e
Show file tree

Hide file tree

Showing 28 changed files with 119 additions and 46 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,8 +1,8 @@
 *.swp
-StringifierContrib.md5
-StringifierContrib.sha1
-StringifierContrib.tgz
-StringifierContrib.txt
-StringifierContrib.zip
-StringifierContrib_installer
-StringifierContrib_installer.pl
+/StringifierContrib.md5
+/StringifierContrib.sha1
+/StringifierContrib.tgz
+/StringifierContrib.txt
+/StringifierContrib.zip
+/StringifierContrib_installer
+/StringifierContrib_installer.pl
diff --git a/data/System/StringifierContrib.txt b/data/System/StringifierContrib.txt
@@ -139,6 +139,7 @@ Foswiki:Development/UnitTests for more information on unit testing.
 %$DEPENDENCIES%
 
 ---++ Change History
+|  23 Jan 2017: | (4.30) added stringifier to index XLS using soffice |
 |  18 Oct 2015: | (4.20) removed dependency on File::MMagic; now using extension-based mime detection |
 |  01 Oct 2015: | (4.10) don't default to pass-through for non-supported document types; fixed unit tests |
 |  29 Sep 2015: | (4.00) added unicode support with Foswiki > 2.0 | 
@@ -155,7 +156,7 @@ Foswiki:Development/UnitTests for more information on unit testing.
 
 %META:FORM{name="PackageForm"}%
 %META:FIELD{name="Author" title="Author" value="Foswiki:Main.MarkusHesse, Foswiki:Main.SvenDowideit, Foswiki:Main.MichaelDaum & Foswiki:Main.AndrewJones"}%
-%META:FIELD{name="Copyright" title="Copyright" value="&copy; 2007, Foswiki:Main.MarkusHesse; &copy; 2009-2015, Foswiki Contributors"}%
+%META:FIELD{name="Copyright" title="Copyright" value="&copy; 2007, Foswiki:Main.MarkusHesse; &copy; 2009-2017, Foswiki Contributors"}%
 %META:FIELD{name="Description" title="Description" value="%25$SHORTDESCRIPTION%25"}%
 %META:FIELD{name="Home" title="Home" value="Foswiki:Extensions/%TOPIC%"}%
 %META:FIELD{name="License" title="License" value="GPL ([[http://www.gnu.org/copyleft/gpl.html][GNU General Public License]])"}%

diff --git a/lib/Foswiki/Contrib/Stringifier.pm b/lib/Foswiki/Contrib/Stringifier.pm
@@ -1,6 +1,6 @@
 # Plugin for Foswiki - The Free and Open Source Wiki, http://foswiki.org/
 #
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -65,7 +65,5 @@ sub stringFor {
   return $plugin->stringForFile($filename);
 }
 
-
-
 1;
 
diff --git a/lib/Foswiki/Contrib/Stringifier/Base.pm b/lib/Foswiki/Contrib/Stringifier/Base.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -54,6 +54,9 @@ __PACKAGE__->plugins;
     sub _programExists {
 	my ($self, $program) = @_;
 
+        # work around a bug in old File::Which that doesn't like absolute paths
+        return $program if -f $program; 
+
 	my $path = File::Which::which($program);
         return defined $path;
     }

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOCX.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOCX.pm
@@ -1,5 +1,5 @@
 # Copyright (C) 2009 TWIKI.NET (http://www.twiki.net)
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_abiword.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_abiword.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_antiword.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_antiword.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_catdoc.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_catdoc.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_soffice.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_soffice.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -28,7 +28,7 @@ my $soffice = $Foswiki::cfg{StringifierContrib}{sofficeCmd} || '/usr/bin/soffice
 
 if (defined($Foswiki::cfg{StringifierContrib}{WordIndexer}) && 
     ($Foswiki::cfg{StringifierContrib}{WordIndexer} eq 'soffice')) {
-    if (-f $soffice){
+    if (__PACKAGE__->_programExists($soffice)) {
         __PACKAGE__->register_handler("application/word", ".doc", "text/docx", ".docx");
     }
 }

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_wv.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/DOC_wv.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/HTML.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/HTML.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/ODT.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/ODT.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2011-2015 Foswiki Contributors
+# Copyright (C) 2011-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/PDF.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/PDF.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/PPT.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/PPT.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/PPTX.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/PPTX.pm
@@ -1,5 +1,5 @@
 # Copyright (C) 2009 TWIKI.NET (http://www.twiki.net)
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/PPT_catdoc.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/PPT_catdoc.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/PPT_soffice.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/PPT_soffice.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -26,18 +26,17 @@ use File::Basename qw(basename);
 
 my $soffice = $Foswiki::cfg{StringifierContrib}{sofficeCmd} || '/usr/bin/soffice';
 
-if (defined($Foswiki::cfg{StringifierContrib}{PowerpointIndexer})
-  && $Foswiki::cfg{StringifierContrib}{PowerpointIndexer} eq 'soffice')
+if ((!defined($Foswiki::cfg{StringifierContrib}{PowerpointIndexer}) || $Foswiki::cfg{StringifierContrib}{PowerpointIndexer} eq 'soffice')
+  && __PACKAGE__->_programExists($soffice))
 {
-  if (-f $soffice) {
-    __PACKAGE__->register_handler("text/ppt", ".ppt", "text/pptx", ".pptx");
-  }
+  __PACKAGE__->register_handler("text/ppt", ".ppt", "text/pptx", ".pptx");
 }
 
+
 sub stringForFile {
     my ($self, $file) = @_;
     my $tmpDir = File::Temp->newdir();
-    
+
     # first convert to pdf as the thing can't do txt:Text directly as reliably
     my $cmd = $soffice . ' --convert-to pdf --invisible --headless --minimized --outdir %OUTDIR|F% %FILENAME|F%';
 

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/Text.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/Text.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/XLS.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/XLS.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -22,9 +22,12 @@ our @ISA = qw( Foswiki::Contrib::Stringifier::Base );
 
 my $xls2txt = $Foswiki::cfg{StringifierContrib}{xls2txtCmd} || 'xls2txt.pl';
 
-# Only if xls2txt.pl exists, I register myself.
-if (__PACKAGE__->_programExists($xls2txt)){
+if (defined($Foswiki::cfg{StringifierContrib}{ExcelIndexer})
+  && ($Foswiki::cfg{StringifierContrib}{ExcelIndexer} eq 'script'))
+{
+  if (__PACKAGE__->_programExists($xls2txt)) {
     __PACKAGE__->register_handler("application/excel", ".xls");
+  }
 }
 
 sub stringForFile {

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/XLSX.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/XLSX.pm
@@ -1,5 +1,5 @@
 # Copyright (C) 2009 TWIKI.NET (http://www.twiki.net)
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License

diff --git a/lib/Foswiki/Contrib/Stringifier/Plugins/XLS_soffice.pm b/lib/Foswiki/Contrib/Stringifier/Plugins/XLS_soffice.pm
@@ -0,0 +1,57 @@
+# Copyright (C) 2009-2017 Foswiki Contributors
+#
+# For licensing info read LICENSE file in the Foswiki root.
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License
+# as published by the Free Software Foundation; either version 2
+# of the License, or (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details, published at 
+# http://www.gnu.org/copyleft/gpl.html
+
+package Foswiki::Contrib::Stringifier::Plugins::XLS_soffice;
+
+use strict;
+use warnings;
+
+use Foswiki::Contrib::Stringifier::Base ();
+our @ISA = qw( Foswiki::Contrib::Stringifier::Base );
+
+use Foswiki::Func ();
+use File::Temp ();
+use File::Basename qw(basename);
+
+my $soffice = $Foswiki::cfg{StringifierContrib}{sofficeCmd} || '/usr/bin/soffice';
+
+if (defined($Foswiki::cfg{StringifierContrib}{ExcelIndexer}) && 
+    ($Foswiki::cfg{StringifierContrib}{ExcelIndexer} eq 'soffice')) {
+    if (__PACKAGE__->_programExists($soffice)) {
+        __PACKAGE__->register_handler("application/excel", ".xls");
+    }
+}
+
+sub stringForFile {
+    my ($self, $file) = @_;
+    my $tmpDir = File::Temp->newdir();
+
+    my $cmd = $soffice . ' --convert-to html --invisible --headless --minimized --outdir %OUTDIR|F% %FILENAME|F%';
+
+    my ($data, $exit) = Foswiki::Sandbox->sysCommand(
+        $cmd,
+        OUTDIR => $tmpDir->dirname,
+        FILENAME  => $file,
+    );
+
+    return '' unless ($exit == 0);
+
+    my $stringifier = Foswiki::Contrib::Stringifier::Plugins::HTML->new();
+    my $tmpFile = $tmpDir->dirname . '/' . basename($file, ".xls", ".xlsx") . '.html';
+
+    return $stringifier->stringForFile($tmpFile);
+}
+
+1;
+
diff --git a/lib/Foswiki/Contrib/StringifierContrib.pm b/lib/Foswiki/Contrib/StringifierContrib.pm
@@ -1,6 +1,6 @@
 # Plugin for Foswiki - The Free and Open Source Wiki, http://foswiki.org/
 #
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or
@@ -18,8 +18,8 @@ package Foswiki::Contrib::StringifierContrib;
 use strict;
 use warnings;
 
-our $VERSION = '4.20';
-our $RELEASE = '18 Oct 2015';
+our $VERSION = '4.30';
+our $RELEASE = '23 Jan 2017';
 our $SHORTDESCRIPTION = 'Helper library to stringify binary document formats';
 our $NO_PREFS_IN_TOPIC = 1;
 

diff --git a/lib/Foswiki/Contrib/StringifierContrib/Config.spec b/lib/Foswiki/Contrib/StringifierContrib/Config.spec
@@ -1,26 +1,37 @@
 # ---+ Extensions
 # ---++ StringifierContrib
 
-# **SELECT abiword, antiword, catdoc, soffice, wv **
+# **SELECT abiword, antiword, catdoc, soffice, wv, none **
 # Select which MS Word indexer to use (you need to have antiword, abiword or wvText installed)
 # <dl>
 # <dt>abiword</dt><dd></dd>
 # <dt>antiword</dt><dd>chould be used on Linux/Unix but may have problems with doc files generated by OpenOffice</dd>
 # <dt>catdoc</dt><dd></dd>
 # <dt>soffice</dt><dd>is the most capable of all converters but may be slow</dd>
 # <dt>wvText</dt><dd>is the default</dd>
+# <dt>none</dt><dd>don't index word documents</dd>
 # </dl>
 $Foswiki::cfg{StringifierContrib}{WordIndexer} = 'antiword';
 
-# **SELECT catppt,script,soffice**
+# **SELECT catppt,script,soffice, none**
 # Select which indexer to use to extract the text of a Powerpoint file (ppt, pptx) 
 # <dl>
 # <dt>catppt</dt><dd>is the default</dd>
 # <dt>script</dt><dd>uses ppthtml </dd>
 # <dt>soffice</dt><dd>is the most capable of all converters but may be slow</dd>
+# <dt>none</dt><dd>don't index powerpoint documents</dd>
 # </dl>
 $Foswiki::cfg{StringifierContrib}{PowerpointIndexer} = 'script';
 
+# **SELECT script,soffice, none**
+# Select which indexer to use to extract the text of an XLS file
+# <dl>
+# <dt>script</dt><dd>default </dd>
+# <dt>soffice</dt><dd>is the most capable of all converters but may be slow</dd>
+# <dt>none</dt><dd>don't index excel documents</dd>
+# </dl>
+$Foswiki::cfg{StringifierContrib}{ExcelIndexer} = 'script';
+
 # **COMMAND**
 # Path to your abiword command (used to convert MS word documents: .doc)
 $Foswiki::cfg{StringifierContrib}{abiwordCmd} = 'abiword';

diff --git a/lib/Foswiki/Contrib/StringifierContrib/MANIFEST b/lib/Foswiki/Contrib/StringifierContrib/MANIFEST
@@ -18,6 +18,7 @@ lib/Foswiki/Contrib/Stringifier/Plugins/PPT_soffice.pm 0644
 lib/Foswiki/Contrib/Stringifier/Plugins/PPTX.pm 0644
 lib/Foswiki/Contrib/Stringifier/Plugins/Text.pm 0644
 lib/Foswiki/Contrib/Stringifier/Plugins/XLS.pm 0644
+lib/Foswiki/Contrib/Stringifier/Plugins/XLS_soffice.pm 0644
 lib/Foswiki/Contrib/Stringifier/Plugins/XLSX.pm 0644
 lib/Foswiki/Contrib/Stringifier.pm 0644
 tools/docx2txt.pl 0755

diff --git a/lib/Foswiki/Contrib/StringifierContrib/build.pl b/lib/Foswiki/Contrib/StringifierContrib/build.pl
@@ -1,4 +1,4 @@
-#!/usr/bin/perl -w
+#!/usr/bin/env perl
 BEGIN {
     unshift @INC, split( /:/, $ENV{FOSWIKI_LIBS} );
 }

diff --git a/tools/docx2txt.pl b/tools/docx2txt.pl
@@ -2,7 +2,7 @@
 
 # docx2txt, a command-line utility to convert Docx documents to text format.
 # Copyright (C) 2008-2009 Sandeep Kumar
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by

diff --git a/tools/pptx2txt.pl b/tools/pptx2txt.pl
@@ -5,7 +5,7 @@
 # Microsoft Office presentation files.
 # Copyright (C) 2009 - Sopan Shewale - sopan.shewale@gmail.com
 #                      TWIKI.NET     - sales@twiki.net  
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # The development of this tools is completely based on other similar 
 # tool called docx2txt - available at http://docx2txt.sourceforge.net/

diff --git a/tools/stringify b/tools/stringify
@@ -1,6 +1,6 @@
 #!/usr/bin/env perl
 #
-# Copyright (C) 2009-2015 Foswiki Contributors
+# Copyright (C) 2009-2017 Foswiki Contributors
 #
 # For licensing info read LICENSE file in the Foswiki root.
 # This program is free software; you can redistribute it and/or