You are here: Foswiki>Tasks Web>Item626 (30 May 2014, GeorgeClark)Edit Attach

Item626: Diff improvement: Support for diff by word

pencil
Priority: Enhancement
Current State: Confirmed
Released In: n/a
Target Release:
Applies To: Engine
Component: FoswikiUIRDiff
Branches:
Reported By: TimotheLitt
Waiting For:
Last Change By: GeorgeClark
It would be nice if the rdiff script used to compare topic revisions highlighted changes within each line.

With each paragraph in a topic being a very long line, and users who add a comma, delete a space, or fix a single word, it can be difficult to see what happened.

I'm not sure that I understand all the subtleties of the renderer, but I do have a prototype that seems functional.

This patch:
  • Is against TWiki 4.2.3
  • implements renderstyle={sequentialbyword, sequentialbyline}. byword is the default and what you get with renderstyle=sequential. byline is for anyone who relies on the old style. One might argue that the the default should be backwards compatibility - but then I'd have had to patch the differences GUI. (Exercise for the reader who cares.)
  • Modifies UI/RDiff.pm
  • Adds UI/WDiff.pm
  • Requires CPAN's Text:::WordDiff
  • Is a lot simpler than it appears - there are only 13 new/changed executable statements in RDiff and 7 in WDiff - the rest is indentation & comments.
  • Could probably use some more testing.

You're welcome to take it as is, or send it through your new feature process. It seems a simple enough patch that the latter would be overkill.

Enjoy.

--- lib/TWiki/UI/RDiff.pm.4.2.3 2008-09-11 23:41:58.000000000 -0400
+++ lib/TWiki/UI/RDiff.pm       2009-01-01 10:17:56.000000000 -0500
@@ -244,13 +244,13 @@
 #| Parameter: =$left= | the text blob before the opteration |
 #| Parameter: =$right= | the text after the operation |
 #| Return: =$result= | Formatted html text |
 #| TODO: | this should move to Render.pm |
 sub _renderSequential
 {
-    my ( $session, $web, $topic, $diffType, $left, $right ) = @_;
+    my ( $session, $web, $topic, $diffType, $left, $right, $seqByLine ) = @_;
     my $result = '';
     ASSERT($topic) if DEBUG;

     #note: I have made the colspan 9 to make sure that it spans all columns (thought there are only 2 now)
     if ( $diffType eq '-') {
         $result .=
@@ -274,26 +274,40 @@
                           'Unchanged',
                           'Unchanged',
                           _renderCellData( $session, $right, $web, $topic ),
                           'u', '',
                           $session );
     } elsif ( $diffType eq 'c') {
-        $result .=
-          _sequentialRow( '#D0FFD0',
-                          'Changed',
-                          'Deleted',
-                          _renderCellData( $session, $left, $web, $topic ),
-                          '-', '<',
-                          $session );
-        $result .=
-          _sequentialRow( undef,
-                          'Changed',
-                          'Added',
-                          _renderCellData( $session, $right, $web, $topic ),
-                          '+', '>',
-                          $session );
+       if( $seqByLine ) {
+           $result .=
+               _sequentialRow( '#D0FFD0',
+                               'Changed',
+                               'Deleted',
+                               _renderCellData( $session, $left, $web, $topic ),
+                               '-', '<',
+                               $session );
+           $result .=
+               _sequentialRow( undef,
+                               'Changed',
+                               'Added',
+                               _renderCellData( $session, $right, $web, $topic ),
+                               '+', '>',
+                               $session );
+       } else {
+           my $diff = word_diff( \$left, \$right, { STYLE=>'TWiki::UI::WDiff',
+                                                } );
+           $diff = _renderCellData( $session, $diff, $web, $topic );
+
+           $result .= CGI::Tr(CGI::td({bgcolor=>'#D0FFD0',
+                                       class=>"twikiDiffChangedHeader",
+                                       colspan=>9},
+                                      CGI::b( $session->i18n->maketext('Changed').': ')));
+           $result .=CGI::Tr(CGI::td( " " ),
+                             CGI::td({class=>"twikiDiffChangedText", colspan=>9}, $diff));
+
+       }
     } elsif ( $diffType eq 'l' && $left ne '' && $right ne '' ) {
         $result .= CGI::Tr({bgcolor=>$format{l}[0],
                             class=>'twikiDiffLineNumberHeader'},
                            CGI::th({align=>'left',
                                     colspan=>9},
                                     ($session->i18n->maketext('Line: [_1] to [_2]',$left,$right))
@@ -311,13 +325,13 @@
 #| Parameter: =$diffArray= | array generated by parseRevisionDiff |
 #| Parameter: =$renderStyle= | style of rendering { debug, sequential, sidebyside} |
 #| Return: =$text= | output html for one renderes revision diff |
 #| TODO: | move into Render.pm |
 sub _renderRevisionDiff
 {
-    my( $session, $web, $topic, $sdiffArray_ref, $renderStyle ) = @_;
+    my( $session, $web, $topic, $sdiffArray_ref, $renderStyle, $seqByLine ) = @_;

 #combine sequential array elements that are the same diffType
     my @diffArray = ();
        foreach my $ele ( @$sdiffArray_ref ) {
                if( ( @$ele[1] =~ /^\%META\:TOPICINFO/ ) || ( @$ele[2] =~ /^\%META\:TOPICINFO/ ) ) {
                        # do nothing, ignore redundant topic info
@@ -348,26 +362,26 @@
                }
                if (( @$diff_ref[0] eq '-' ) && ( @$next_ref[0] eq '+' )) {
                    $diff_ref = ['c', @$diff_ref[1], @$next_ref[2]];
             $next_ref = undef;
                }
                if ( $renderStyle eq 'sequential' ) {
-                   $result .= _renderSequential( $session, $web, $topic, @$diff_ref );
+                   $result .= _renderSequential( $session, $web, $topic, @$diff_ref, $seqByLine );
                } elsif ( $renderStyle eq 'sidebyside' ) {
             $result .= CGI::Tr(CGI::td({ width=>'50%'}, ''),
                                CGI::td({ width=>'50%'}, ''));
                    $result .= _renderSideBySide( $session, $web, $topic, @$diff_ref );
                } elsif ( $renderStyle eq 'debug' ) {
                    $result .= _renderDebug( @$diff_ref );
                }
                $diff_ref = $next_ref;
        }
     #don't forget the last one ;)
     if ( $diff_ref ) {
         if ( $renderStyle eq 'sequential' ) {
-            $result .= _renderSequential ( $session, $web, $topic, @$diff_ref );
+            $result .= _renderSequential ( $session, $web, $topic, @$diff_ref, $seqByLine );
         } elsif ( $renderStyle eq 'sidebyside' ) {
             $result .= CGI::Tr(CGI::td({ width=>'50%'}, ''),
                                CGI::td({ width=>'50%'}, ''));
             $result .= _renderSideBySide( $session, $web, $topic, @$diff_ref );
         } elsif ( $renderStyle eq 'debug' ) {
             $result .= _renderDebug( @$diff_ref );
@@ -388,13 +402,13 @@
 invoked via the =UI::run= method.

 Renders the differences between version of a TwikiTopic
 | topic | topic that we are showing the differences of |
 | rev1 | the higher revision |
 | rev2 | the lower revision |
-| render | the rendering style {sequential, sidebyside, raw, debug} | (preferences) DIFFRENDERSTYLE, =sequential= |
+| render | the rendering style {sequential, sidebyside, raw, debug} | (preferences) DIFFRENDERSTYLE, =sequential= ; sequentialbyword (default)
 or sequentialbyline |
 | type | {history, diff, last} history diff, version to version, last version to previous | =history= |
 | context | number of lines of context |
 | skin | the skin(s) to use to display the diff |
 TODO:
    * add a {word} render style
    * move the common CGI param handling to one place
@@ -412,12 +426,17 @@
     TWiki::UI::checkWebExists( $session, $webName, $topic, 'diff' );
     TWiki::UI::checkTopicExists( $session, $webName, $topic, 'diff' );

     my $renderStyle = $query->param('render') ||
       $session->{prefs}->getPreferencesValue( 'DIFFRENDERSTYLE' ) ||
         'sequential';
+    my $seqByLine = ($renderStyle eq 'sequentialbyline');
+    $renderStyle =~ s/^sequential.*$/sequential/;
+    if( $renderStyle eq 'sequential' && !$seqByLine ) {
+       use Text::WordDiff;
+    }
     my $diffType = $query->param('type') || 'history';
     my $contextLines = $query->param('context');
     unless( defined $contextLines ) {
         $session->{prefs}->getPreferencesValue( 'DIFFCONTEXTLINES' );
         $contextLines = 3 unless defined $contextLines;
     }
@@ -491,13 +510,13 @@
         # eliminate white space to prevent wrap around in HR table:
         $rInfo =~ s/\s+/ /g;
         $rInfo2 =~ s/\s+/ /g;
         my $diffArrayRef = $session->{store}->getRevisionDiff(
             $session->{user}, $webName, $topic, $r2, $r1, $contextLines );
         $text = _renderRevisionDiff( $session, $webName, $topic,
-                                     $diffArrayRef, $renderStyle );
+                                     $diffArrayRef, $renderStyle, $seqByLine );
         $diff =~ s/%REVINFO1%/$rInfo/go;
         $diff =~ s/%REVINFO2%/$rInfo2/go;
         $diff =~ s/%TEXT%/$text/go;
         $page .= $diff;
         $r1 = $r1 - 1;
         $r2 = $r2 - 1;
--- /dev/null   2008-04-26 06:27:25.709194980 -0400
+++ lib/TWiki/UI/WDiff.pm       2009-01-01 10:22:56.000000000 -0500
@@ -0,0 +1,43 @@
+package TWiki::UI::WDiff;
+
+use strict;
+use HTML::Entities qw(encode_entities);
+use vars qw($VERSION @ISA);
+
+$VERSION = '0.01';
+@ISA = qw(Text::WordDiff::Base);
+
+sub file_header {
+    return '';
+}
+
+sub same_items {
+    shift;
+    return encode_entities( join '', @_ );
+}
+
+sub delete_items {
+    shift;
+    return '<del style="background:#ff9999;" class="twikiDiffDeletedMarker">' . encode_entities( join'', @_ ) . '</del>';
+}
+
+sub insert_items {
+    shift;
+    return '<ins style="background:#ccccff;" class="twikiDiffAddedMarker">' . encode_entities( join'', @_ ) . '</ins>';
+}
+
+1;
+__END__
+
+
+=begin comment
+
+=head1 Copyright and License
+
+Copyright (c) 2005-2008 David Wheeler. Some Rights Reserved.
+Copyright (c) 2009 Timothe Litt. Some Rights Reserved.
+
+This module is free software; you can redistribute it and/or modify it under the
+same terms as Perl itself.
+
+=cut

-- TimotheLitt

Without having tried anything above - no time these days because of 1.0.0 work - have you looked at CompareRevisionsAddOn?

This does word by word and integrates well with TWiki and with Foswiki (I have released also a Foswiki version).

And used with HistoryPlugin and maybe also RevCommentPlugin you get a pretty cool enhancement of TWiki/Foswiki.

It has been my wish to get the 3 plugins either includes with the releases or integrated into core.

The CompareRevisionsAddOn / HistoryPlugins also makes a pretty good UI for the whole thing. I can highly recommend the suite. I have used them in production for years and they are rock solid. On Foswiki I have taken over the maintenance of them.

I missed those in my search, thanks for the pointer. I installed the TWiki version. Noticed that CompareRevisionsAddonPlugin is still using commonTagsHandler (instead of registerTagHandler ; it also isn't returning 1 on load, except by accident. You might want to fix these if they're still present in the foswiki version.

I agree that CompareRevisions and History should be integrated - they look much better than rdiff.

However, they do fall back to rdiff in several cases - the "sequential" link from History for one - and the "more topic actions" diffs for another.

My simple patch improves those considerably, so you still may want to consider applying it to rdiff.

RevCommentPlugin - my opinion - a good thing to include, but does need to be optional. In a development environment, change comments are a good thing. In an environment where it's a struggle to get users to edit at all, it's one more think to persuade them to do (or ignore.)

I'd also make CompareRevisions be the first choice under more topic actions -- you really want the first impression to be the best you can do. And word diffs are much more approachable than paragraph diffs.

ItemTemplate edit

Summary Diff improvement: Support for diff by word
ReportedBy TimotheLitt
Codebase
SVN Range TWiki-4.2.3, Wed, 06 Aug 2008, build 17396
AppliesTo Engine
Component FoswikiUIRDiff
Priority Enhancement
CurrentState Confirmed
WaitingFor
Checkins
ReleasedIn n/a
CheckinsOnBranches
trunkCheckins
Release01x01Checkins
Topic revision: r5 - 30 May 2014, GeorgeClark - This page was cached on 18 Sep 2021 - 12:15.

The copyright of the content on this website is held by the contributing authors, except where stated elsewhere. See Copyright Statement. Creative Commons License    Legal Imprint    Privacy Policy