--- loncom/publisher/loncleanup.pm	2005/05/28 18:53:50	1.3
+++ loncom/publisher/loncleanup.pm	2012/12/14 00:27:08	1.19
@@ -1,7 +1,7 @@
 # The LearningOnline Network with CAPA
 # Handler to cleanup XML files
 #
-# $Id: loncleanup.pm,v 1.3 2005/05/28 18:53:50 albertel Exp $
+# $Id: loncleanup.pm,v 1.19 2012/12/14 00:27:08 raeburn Exp $
 #
 # Copyright Michigan State University Board of Trustees
 #
@@ -34,10 +34,14 @@ use strict;
 use Apache::File;
 use File::Copy;
 use Apache::Constants qw(:common :http :methods);
-use Apache::loncacc;
 use Apache::loncommon();
+use Apache::lonhtmlcommon();
 use Apache::lonlocal;
 use Apache::lonnet;
+use lib '/home/httpd/lib/perl/';
+use LONCAPA;
+use HTML::Entities();
+ 
 
 sub latextrans {
     my $symbolfont=shift;
@@ -218,8 +222,9 @@ sub htmlclean {
     }
 # Generate empty tags, remove wrong end tags
     unless ($blockemptytags) {
-	$raw=~s/\<(br|hr|img|meta|allow|basefont)([^\>\/]*?)\>/\<$1$2 \/\>/gis;
-	$raw=~s/\<\/(br|hr|img|meta|allow|basefont)\>//gis;
+	$raw=~s/\<(br|hr|img|meta|embed|allow|basefont)([^\>]*?)\>/\<$1$2 \/\>/gis;
+	$raw=~s/\<\/(br|hr|img|meta|embed|allow|basefont)\>//gis;
+	$raw=~s/\/ \/\>/\/\>/gs;
 	unless ($full) {
 	    $raw=~s/\<[\/]*(body|head|html)\>//gis;
 	}
@@ -244,21 +249,168 @@ sub htmlclean {
 
 sub phaseone {
     my ($r,$fn,$uname,$udom)=@_;
+    $r->print(
+        &Apache::lonhtmlcommon::start_pick_box()
+       .&Apache::lonhtmlcommon::row_title(&mt('Select actions to attempt'))
+       .'<label>'
+       .'<input type="checkbox" name="linefeed" checked="checked" /> '
+       .&mt('Linefeeds, formfeeds, and carriage returns')
+       .'</label><br />'
+       .'<label>'
+       .'<input type="checkbox" name="empty" checked="checked" /> '
+       .&mt('Empty tags')
+       .'</label><br />'
+       .'<label>'
+       .'<input type="checkbox" name="lower" checked="checked" /> '
+       .&mt('Lower casing')
+       .'</label><br />'
+       .'<label>'
+       .'<input type="checkbox" name="symbol" checked="checked" /> '
+       .&mt('Symbol font')
+       .'</label>'
+       .&Apache::lonhtmlcommon::row_closure(1)
+       .&Apache::lonhtmlcommon::end_pick_box()
+    );
+
+    $r->print(
+        '<input type="hidden" name="phase" value="two" />'
+       .'<p>'
+       .'<input type="submit" value="'.&mt('Next').'" />'
+       .'</p>'
+    );
 }
 
 sub phasetwo {
+    # Check original file
     my ($r,$fn,$uname,$udom)=@_;
+    my $text='';
+    my $londocroot = $r->dir_config('lonDocRoot');
+    if (open(IN,"<$londocroot/priv/$udom/$uname".$fn)) {
+        while (my $line=<IN>) {
+	    $text.=$line;
+        }
+        close(IN);
+    }
+    # Check if any selection was made
+    if ($env{'form.linefeed'} ne 'on' &&
+        $env{'form.empty'} ne 'on' &&
+        $env{'form.lower'} ne 'on' &&
+        $env{'form.symbol'} ne 'on') {
+        $r->print(
+            '<p class="LC_warning">'
+           .&mt('Please select at least one option.')
+           .'</p>'
+           .'<p><a href="javascript:history.back();">'.&mt('Back').'</p>'
+        );
+        return;
+    }
+
+    my $uri="/priv/$udom/$uname".$fn;
+    my $result=&Apache::lonnet::ssi_body($uri,
+					 ('grade_target'=>'web',
+					  'return_only_error_and_warning_counts' => 1));
+    my ($errorcount,$warningcount)=split(':',$result);
+
+    # Display results for original file
+    $r->print(
+        &Apache::lonhtmlcommon::start_pick_box()
+       .&Apache::lonhtmlcommon::row_title(&mt('Original file'))
+       .&Apache::lonhtmlcommon::confirm_success(
+           &mt('[quant,_1,error]',$errorcount), $errorcount)
+       .'<br />'
+       .&Apache::lonhtmlcommon::confirm_success(
+           &mt('[quant,_1,warning]',$warningcount), $warningcount)
+       .&Apache::lonhtmlcommon::row_closure()
+    );
+
+    # Clean up file
+    $text=&htmlclean($text,1,
+               ($env{'form.linefeed'} ne 'on'),
+               ($env{'form.empty'} ne 'on'),
+               ($env{'form.lower'} ne 'on'),
+               ($env{'form.symbol'} ne 'on'));
+    my ($main,$ext)=($fn=~/^(.*)\.(\w+)/);
+    my $newfn=$main.'_Auto_Cleaned_Up.'.$ext;
+    if (open(OUT,">$londocroot/priv/$udom/$uname".$newfn)) {
+        print OUT $text;
+        close(OUT);
+    }
+    my $newuri="/priv/$udom/$uname".$newfn;
+    $result=&Apache::lonnet::ssi_body($newuri,
+                                         ('grade_target'=>'web',
+                                          'return_only_error_and_warning_counts' => 1));
+    ($errorcount,$warningcount)=split(':',$result);
+
+    # Display results for cleaned up file
+    $r->print(
+        &Apache::lonhtmlcommon::row_title(&mt('Cleaned up file'))
+       .&Apache::lonhtmlcommon::confirm_success(
+           &mt('[quant,_1,error]',$errorcount), $errorcount)
+       .'<br />'
+       .&Apache::lonhtmlcommon::confirm_success(
+           &mt('[quant,_1,warning]',$warningcount), $warningcount)
+       .&Apache::lonhtmlcommon::row_closure()
+    );
+
+    # Display actions
+    $r->print(
+        &Apache::lonhtmlcommon::row_title(&mt('Actions'))
+       .'<ul>'
+       .'<li><a href="'.$newuri.'" target="prev">'
+       .&mt('Open (and edit) cleaned up file in new window')
+       .'</a></li>'
+       .'<li><a href="'
+       .&HTML::Entities::encode(
+            '/adm/diff?filename='.&escape($uri)
+           .'&versionone=priv&filetwo='.&escape($newuri))
+       .'" target="prev">'
+       .&mt('Show diffs in new window')
+       .'</a></li>'
+       .'</ul>'
+       .&Apache::lonhtmlcommon::row_closure(1)
+       .&Apache::lonhtmlcommon::end_pick_box()
+       .'<p>'
+       .'<input type="hidden" name="phase" value="three" />'
+       .'<input type="submit" name="accept" value="'
+       .&mt('Clean Up').'" />'
+       .' <input type="submit" name="reject" value="'
+       .&mt('Cancel').'" />'
+       .'</p>'
+    );
 }
 
 sub phasethree {
     my ($r,$fn,$uname,$udom)=@_;
+    my $old=$r->dir_config('lonDocRoot')."/priv/$udom/$uname".$fn;
+    my ($main,$ext)=($fn=~/^(.*)\.(\w+)/);
+    my $newfn=$main.'_Auto_Cleaned_Up.'.$ext;
+    my $new=$r->dir_config('lonDocRoot')."/priv/$udom/$uname".$newfn;
+    if ($env{'form.accept'}) {
+        $r->print(
+        '<p class="LC_info">'
+       .&mt('Accepting changes...')
+       .'</p>'
+    );
+        move($new,$old);
+    } else {
+        $r->print(
+        '<p class="LC_info">'
+       .&mt('Rejecting changes...')
+       .'</p>'
+        );
+        unlink($new);
+    }
+    $r->print(
+        '<p>'
+       .&Apache::lonhtmlcommon::confirm_success(&mt('Done')));
+        '</p>'
 }
 
 # ---------------------------------------------------------------- Main Handler
 sub handler {
 
     my $r=shift;
-
+    my $fn='';
 
 # Get query string for limited number of parameters
 
@@ -267,7 +419,7 @@ sub handler {
 
     if ($env{'form.filename'}) {
 	$fn=$env{'form.filename'};
-	$fn=~s/^http\:\/\/[^\/]+//;
+	$fn=~s{^https?\://[^/]+}{};
     } else {
 	$r->log_reason($env{'user.name'}.' at '.$env{'user.domain'}.
 		       ' unspecified filename for cleanup', $r->filename); 
@@ -284,8 +436,7 @@ sub handler {
     my $uname;
     my $udom;
 
-    ($uname,$udom)=
-	&Apache::loncacc::constructaccess($fn,$r->dir_config('lonDefDomain'));
+    ($uname,$udom)=&Apache::lonnet::constructaccess($fn);
     unless (($uname) && ($udom)) {
 	$r->log_reason($uname.' at '.$udom.
 		       ' trying to cleanup file '.$env{'form.filename'}.
@@ -294,24 +445,46 @@ sub handler {
 	return HTTP_NOT_ACCEPTABLE;
     }
 
-    $fn=~s/\/\~(\w+)//;
-
     &Apache::loncommon::content_type($r,'text/html');
     $r->send_http_header;
 
-    $r->print('<html><head><title>LON-CAPA Construction Space</title></head>');
-
-    $r->print(&Apache::loncommon::bodytag('Cleanup XML Document'));
-
-    if ($env{'form.phase'} eq 'three') {
-	&phasethree($r,$fn,$uname,$udom);
-    } elsif ($env{'form.phase'} eq 'two') {
-	&phasetwo($r,$fn,$uname,$udom);
+    # Breadcrumbs
+    my $brcrum = [{'href' => &Apache::loncommon::authorspace($fn),
+                   'text' => 'Construction Space'},
+                  {'href' => '',
+                   'text' => 'Cleanup XML Document'}];
+
+    $fn=~s{^/priv/$LONCAPA::domain_re/$LONCAPA::username_re}{};
+
+    $r->print(&Apache::loncommon::start_page('Cleanup XML Document',
+                                             undef,
+                                             {'bread_crumbs' => $brcrum,}));
+    $r->print('<h2>'.$fn.'</h2>'.
+              '<form action="/adm/cleanup" method="post">'.
+              '<input type="hidden" name="filename" value="'.$env{'form.filename'}.'" />');
+    unless ($fn=~/\.(problem|exam|quiz|assess|survey|form|library|xml|html|htm|xhtml|xhtm|sty)$/) {
+	$r->print('<p class="LC_warning">'.&mt('Cannot cleanup this filetype').'</p>');
     } else {
-	&phaseone($r,$fn,$uname,$udom);
+	if ($env{'form.phase'} eq 'three') {
+	    &phasethree($r,$fn,$uname,$udom);
+	} elsif ($env{'form.phase'} eq 'two') {
+	    &phasetwo($r,$fn,$uname,$udom);
+	} else {
+	    &phaseone($r,$fn,$uname,$udom);
+	}
     }
+    my $dir=$fn;
+    $dir=~s{[^/]+$}{};
+    $r->print(
+        '</form>'
+       .&Apache::lonhtmlcommon::actionbox(
+            ['<a href="/priv/'.$udom.'/'.$uname.$fn.'">'.
+                 &mt('Back to Source File').'</a>',
+            '<a href="/priv/'.$udom.'/'.$uname.$dir.'">'.
+                &mt('Back to Source Directory').'</a>'])
+       .&Apache::loncommon::end_page()
+    );
 
-    $r->print('</body></html>');
     return OK;  
 }