--- loncom/xml/lonxml.pm 2000/06/19 15:52:29 1.1 +++ loncom/xml/lonxml.pm 2001/03/23 22:08:05 1.60 @@ -1,1305 +1,494 @@ -package Apache::lonxml; - +# The LearningOnline Network with CAPA +# XML Parser Module +# +# last modified 06/26/00 by Alexander Sakharuk +# 11/6 Gerd Kortemeyer +# 6/1/1 Gerd Kortemeyer +# 2/21,3/13 Guy + +package Apache::lonxml; +use vars +qw(@pwd @outputstack $redirection $import @extlinks $metamode); use strict; use HTML::TokeParser; use Safe; +use Safe::Hole; +use Opcode; use Apache::Constants qw(:common); -use Apache::lontexconvert; +sub register { + my $space; + my @taglist; + my $temptag; + ($space,@taglist) = @_; + foreach $temptag (@taglist) { + $Apache::lonxml::alltags{$temptag}=$space; + } +} -#======================================================= Main subroutine: xmlparse +sub printalltags { + my $temp; + foreach $temp (sort keys %Apache::lonxml::alltags) { + &Apache::lonxml::debug("$temp -- $Apache::lonxml::alltags{$temp}"); + } +} +use Apache::style; +use Apache::lontexconvert; +use Apache::run; +use Apache::londefdef; +use Apache::scripttag; +use Apache::edit; +#================================================== Main subroutine: xmlparse +@pwd=(); +@outputstack = (); +$redirection = 0; +$import = 1; +@extlinks=(); +$metamode = 0; sub xmlparse { - my ($target,$content_file_string,%style_for_target) = @_; - my $pars = HTML::TokeParser->new(\$content_file_string); - my $currentstring = ''; - my $finaloutput = ''; - my $newarg = ''; - my $tempostring = ''; - my $tempocont = ''; - my $safeeval = new Safe; - -#------------------------- Redefinition of the target in the case of compound target - ($target, my @tenta) = split('&&',$target); -#------------------------------ Stack definition (in stack we have all current tags) - - my @stack = (); - my @parstack = (); - -#------------------------------------------ Parse input string (content_file_string) - - my $token; + my ($target,$content_file_string,$safeinit,%style_for_target) = @_; + if ($target eq 'meta') { + # meta mode is a bit weird only some output is to be turned off + # tag turns metamode off (defined in londefdef.pm) + $Apache::lonxml::redirection = 0; + $Apache::lonxml::metamode = 1; + $Apache::lonxml::import = 0; + } elsif ($target eq 'grade') { + &startredirection; + $Apache::lonxml::metamode = 0; + $Apache::lonxml::import = 1; + } else { + $Apache::lonxml::metamode = 0; + $Apache::lonxml::redirection = 0; + $Apache::lonxml::import = 1; + } + #&printalltags(); + my @pars = (); + @Apache::lonxml::pwd=(); + my $pwd=$ENV{'request.filename'}; + $pwd =~ s:/[^/]*$::; + &newparser(\@pars,\$content_file_string,$pwd); + my $currentstring = ''; + my $finaloutput = ''; + my $newarg = ''; + my $result; + + my $safeeval = new Safe; + my $safehole = new Safe::Hole; + $safeeval->permit("entereval"); + $safeeval->permit(":base_math"); + $safeeval->deny(":base_io"); + $safehole->wrap(\&Apache::lonnet::EXT,$safeeval,'&EXT'); +#need to inspect this class of ops +# $safeeval->deny(":base_orig"); + $safeinit .= ';$external::target='.$target.';'; + $safeinit .= ';$external::randomseed='.&Apache::lonnet::rndseed().';'; + &Apache::run::run($safeinit,$safeeval); +#-------------------- Redefinition of the target in the case of compound target + + ($target, my @tenta) = split('&&',$target); + + my @stack = (); + my @parstack = (); + &initdepth; + my $token; + while ( $#pars > -1 ) { + while ($token = $pars[$#pars]->get_token) { + if (($token->[0] eq 'T') || ($token->[0] eq 'C') || ($token->[0] eq 'D') ) { + if (!$metamode) { $result=$token->[1]; } + } elsif ($token->[0] eq 'PI') { + if (!$metamode) { $result=$token->[2]; } + } elsif ($token->[0] eq 'S') { + # add tag to stack + push (@stack,$token->[1]); + # add parameters list to another stack + push (@parstack,&parstring($token)); + &increasedepth($token); + if (exists $style_for_target{$token->[1]}) { + if ($metamode) { + $result = $style_for_target{$token->[1]}; + } elsif ($Apache::lonxml::redirection) { + $Apache::lonxml::outputstack['-1'] .= + &recurse($style_for_target{$token->[1]},$target,$safeeval, + \%style_for_target,@parstack); + } else { + $finaloutput .= &recurse($style_for_target{$token->[1]},$target, + $safeeval,\%style_for_target,@parstack); + } + } else { + $result = &callsub("start_$token->[1]", $target, $token,\@parstack, + \@pars, $safeeval, \%style_for_target); + } + } elsif ($token->[0] eq 'E') { + #clear out any tags that didn't end + while ($token->[1] ne $stack[$#stack] && ($#stack > -1)) { + &Apache::lonxml::warning("Unbalanced tags in resource $stack['-1']"); + pop @stack;pop @parstack;&decreasedepth($token); + } + + if (exists $style_for_target{'/'."$token->[1]"}) { + if ($metamode) { + $result = $style_for_target{$token->[1]}; + } elsif ($Apache::lonxml::redirection) { + $Apache::lonxml::outputstack['-1'] .= + &recurse($style_for_target{'/'."$token->[1]"}, + $target,$safeeval,\%style_for_target,@parstack); + } else { + $finaloutput .= &recurse($style_for_target{'/'."$token->[1]"}, + $target,$safeeval,\%style_for_target, + @parstack); + } + + } else { + $result = &callsub("end_$token->[1]", $target, $token, \@parstack, + \@pars,$safeeval, \%style_for_target); + } + } else { + &Apache::lonxml::error("Unknown token event :$token->[0]:$token->[1]:"); + } + #evaluate variable refs in result + if ($result ne "") { + if ( $#parstack > -1 ) { + if ($Apache::lonxml::redirection) { + $Apache::lonxml::outputstack['-1'] .= + &Apache::run::evaluate($result,$safeeval,$parstack[$#parstack]); + } else { + $finaloutput .= &Apache::run::evaluate($result,$safeeval, + $parstack[$#parstack]); + } + } else { + $finaloutput .= &Apache::run::evaluate($result,$safeeval,''); + } + $result = ''; + } + if ($token->[0] eq 'E') { + pop @stack;pop @parstack;&decreasedepth($token); + } + } + pop @pars; + pop @Apache::lonxml::pwd; + } + +# if ($target eq 'meta') { +# $finaloutput.=&endredirection; +# } + return $finaloutput; +} - while ($token = $pars->get_token) { - if ($token->[0] eq 'T') { - $finaloutput .= $token->[1]; - $tempocont .= $token->[1]; - } elsif ($token->[0] eq 'S') { -#------------------------------------------------------------------ add tag to stack - push (@stack,$token->[1]); -#---------------------------------------------- add parameters list to another stack - map {$tempostring .= "$_=$token->[2]->{$_},"} @{$token->[3]}; - push (@parstack,$tempostring); - $tempostring = ''; - $tempocont = ''; - - if (exists $style_for_target{$token->[1]}) { - -#--------------------------------------------------------- use style file definition - - $newarg = $style_for_target{$token->[1]}; - - if (index($newarg,'script') != -1 ) { - my $pat = HTML::TokeParser->new(\$newarg); - my $tokenpat; - my $partstring = ''; - my $oustring = ''; - my $outputstring; - - while ($tokenpat = $pat->get_token) { - if ($tokenpat->[0] eq 'T') { - $oustring .= $tokenpat->[1]; - } elsif ($tokenpat->[0] eq 'S') { - if ($tokenpat->[1] eq 'script') { - while ($tokenpat = $pat->get_token and $tokenpat->[1] ne 'script') { - if ($tokenpat->[0] eq 'S') { - $partstring .= $tokenpat->[4]; - } elsif ($tokenpat->[0] eq 'T') { - $partstring .= $tokenpat->[1]; - } elsif ($tokenpat->[0] eq 'E') { - $partstring .= $tokenpat->[2]; - } - } - - map {$partstring =~ s/\$$_/$token->[2]->{$_}/g; } @{$token->[3]}; - - &run($partstring,$safeeval); - $partstring = ''; - } elsif ($tokenpat->[1] eq 'evaluate') { - $outputstring = &evaluate($tokenpat->[2]{expression},$safeeval); - $oustring .= $outputstring; - } else { - $oustring .= $tokenpat->[4]; - } - } elsif ($tokenpat->[0] eq 'E' and $tokenpat->[1] ne 'evaluate') { - $oustring .= $tokenpat->[1]; - } - } - $newarg = $oustring; - } else { - map {$newarg =~ s/\$$_/$token->[2]->{$_}/g; } @{$token->[3]}; - } - $finaloutput .= $newarg; - } else { -#----------------------------------------------------- use default definition of tag - my $sub="start_$token->[1]"; - - { - no strict 'refs'; - if (defined (&$sub)) { - $currentstring = &$sub($target,$token,\@parstack); - $finaloutput .= $currentstring; - $currentstring = ''; - } else { - $finaloutput .= $token->[4]; - } - use strict 'refs'; - } - } - } elsif ($token->[0] eq 'E') { - pop @stack; - unless (exists $style_for_target{$token->[1]}) { - my $sub="end_$token->[1]"; - { - no strict 'refs'; - if (defined (&$sub)) { - $currentstring = &$sub($target,$token,\@parstack); - $finaloutput .= $currentstring; - $currentstring = ''; - } else { - $finaloutput .= $token->[4]; - } - use strict 'refs'; - } - } -#------------------------------------------------------- end tag from the style file - if (exists $style_for_target{'/'."$token->[1]"}) { - $newarg = $style_for_target{'/'."$token->[1]"}; - my @very_temp = split(',',@parstack[$#parstack]); - map {my @ret= split('=',$_); $newarg =~ s/\$$ret[0]/$ret[1]/g; } @very_temp; - $finaloutput .= $newarg; - } - pop @parstack; +sub recurse { + + my @innerstack = (); + my @innerparstack = (); + my ($newarg,$target,$safeeval,$style_for_target,@parstack) = @_; + my @pat = (); + &newparser(\@pat,\$newarg); + my $tokenpat; + my $partstring = ''; + my $output=''; + my $decls=''; + while ( $#pat > -1 ) { + while ($tokenpat = $pat[$#pat]->get_token) { + if (($tokenpat->[0] eq 'T') || ($tokenpat->[0] eq 'C') || ($tokenpat->[0] eq 'D') ) { + $partstring = $tokenpat->[1]; + } elsif ($tokenpat->[0] eq 'PI') { + $partstring = $tokenpat->[2]; + } elsif ($tokenpat->[0] eq 'S') { + push (@innerstack,$tokenpat->[1]); + push (@innerparstack,&parstring($tokenpat)); + &increasedepth($tokenpat); + $partstring = &callsub("start_$tokenpat->[1]", + $target, $tokenpat, \@innerparstack, + \@pat, $safeeval, $style_for_target); + } elsif ($tokenpat->[0] eq 'E') { + #clear out any tags that didn't end + while ($tokenpat->[1] ne $innerstack[$#innerstack] + && ($#innerstack > -1)) { + &Apache::lonxml::warning("Unbalanced tags in resource $innerstack['-1']"); + pop @innerstack;pop @innerparstack;&decreasedepth($tokenpat); + } + $partstring = &callsub("end_$tokenpat->[1]", + $target, $tokenpat, \@innerparstack, + \@pat, $safeeval, $style_for_target); + } else { + &Apache::lonxml::error("Unknown token event :$tokenpat->[0]:$tokenpat->[1]:"); + } + #pass both the variable to the style tag, and the tag we + #are processing inside the + if ( $partstring ne "" ) { + if ( $#parstack > -1 ) { + if ( $#innerparstack > -1 ) { + $decls= $parstack[$#parstack].$innerparstack[$#innerparstack]; + } else { + $decls= $parstack[$#parstack]; + } + } else { + if ( $#innerparstack > -1 ) { + $decls=$innerparstack[$#innerparstack]; + } else { + $decls=''; + } } + $output .= &Apache::run::evaluate($partstring,$safeeval,$decls); + $partstring = ''; + } + if ($tokenpat->[0] eq 'E') { pop @innerstack;pop @innerparstack; + &decreasedepth($tokenpat);} } - return $finaloutput; + pop @pat; + pop @Apache::lonxml::pwd; + } + return $output; } +sub callsub { + my ($sub,$target,$token,$parstack,$parser,$safeeval,$style)=@_; + my $currentstring=''; + { + my $sub1; + no strict 'refs'; + if ($target eq 'edit' && $token->[0] eq 'S') { + $currentstring = &Apache::edit::tag_start($target,$token,$parstack,$parser, + $safeeval,$style); + } + if (my $space=$Apache::lonxml::alltags{$token->[1]}) { + #&Apache::lonxml::debug("Calling sub $sub in $space
\n"); + $sub1="$space\:\:$sub"; + $Apache::lonxml::curdepth=join('_',@Apache::lonxml::depthcounter); + $currentstring .= &$sub1($target,$token,$parstack,$parser, + $safeeval,$style); + } else { + #&Apache::lonxml::debug("NOT Calling sub $sub in $space
\n"); + if (defined($token->[4]) && !$metamode) { + $currentstring .= $token->[4]; + } else { + $currentstring .= $token->[2]; + } + } + if ($target eq 'edit' && $token->[0] eq 'E') { + $currentstring = &Apache::edit::tag_end($target,$token,$parstack,$parser, + $safeeval,$style); + } + use strict 'refs'; + } + return $currentstring; +} -#================================================================== style subroutine +sub startredirection { + $Apache::lonxml::redirection++; + push (@Apache::lonxml::outputstack, ''); +} -sub styleparser { +sub endredirection { + if (!$Apache::lonxml::redirection) { + &Apache::lonxml::error("Endredirection was called, before a startredirection, perhaps you have unbalanced tags. Some debuggin information:".join ":",caller); + return ''; + } + $Apache::lonxml::redirection--; + pop @Apache::lonxml::outputstack; +} - my ($target,$content_style_string) = @_; +sub initdepth { + @Apache::lonxml::depthcounter=(); + $Apache::lonxml::depth=-1; + $Apache::lonxml::olddepth=-1; +} -#------------------------------------------------ target redefinition (if necessary) - - my @target_string = ''; - my $element; - - ($element,@target_string) = split ('&&',$target); - - map {$content_style_string =~ s/\<(.*)$_\>/\<$1$element\>/g; } @target_string; - - $target = $element; - -#------------------------------------------------- create a table for defined target -#---------------------------------------------- from the information from Style File - - my @value_style = (); - my $current_key = ''; - my $current_value = ''; - - my $pstyle = HTML::TokeParser->new(\$content_style_string); - - my $stoken; - - while ($stoken = $pstyle->get_token) { -#---------------------------------------------------------- start for tag definition - if ($stoken->[0] eq 'S' and $stoken->[1] eq 'definetag') { -#------------------------------------------------------------------- new key in hash - $current_key = $stoken->[2]{name}; - if ($target eq 'meta') { -#-------------------------------------------------- reserved for the metadate output - - - } else { -#-------------------------------------------------------------------- outtext output - while ($stoken = $pstyle->get_token and $stoken->[1] ne 'outtext') { - } - while ($stoken = $pstyle->get_token and $stoken->[0] ne 'S') { - $current_value .= $stoken->[1]; - } - while ($stoken->[1] ne 'definetag') { - if ($stoken->[0] eq 'S' and $stoken->[1] eq $target) { - while ($stoken = $pstyle->get_token) { - if ($stoken->[1] ne $target) { - if ($stoken->[0] eq 'S') { - $current_value .= $stoken->[4]; - } - if ($stoken->[0] eq 'E') { - $current_value .= $stoken->[2]; - } - if ($stoken->[0] eq 'T') { - $current_value .= $stoken->[1]; - } - } else { - last; - } - } - } elsif ($stoken->[0] eq 'S' and $stoken->[1] ne $target) { - while ($stoken = $pstyle->get_token and $stoken->[0] ne 'E') { - } - } - - while ($stoken = $pstyle->get_token) { - if ($stoken->[0] eq 'T') { - $current_value .= $stoken->[1]; - } - if ($stoken->[0] eq 'E') { - last; - } - if ($stoken->[0] eq 'S') { - last; - } - } - - } - } - - } - push (@value_style,lc $current_key,$current_value); - $current_key = ''; - $current_value = ''; +sub increasedepth { + my ($token) = @_; + $Apache::lonxml::depth++; + $Apache::lonxml::depthcounter[$Apache::lonxml::depth]++; + if ($Apache::lonxml::depthcounter[$Apache::lonxml::depth]==1) { + $Apache::lonxml::olddepth=$Apache::lonxml::depth; + } + my $curdepth=join('_',@Apache::lonxml::depthcounter); + &Apache::lonxml::debug("s $Apache::lonxml::depth : $Apache::lonxml::olddepth : $curdepth : $token->[1]\n"); +#print "
s $Apache::lonxml::depth : $Apache::lonxml::olddepth : $curdepth : $token->[1]\n"; +} - } - - my %style_for_target = @value_style; - -#-------------------------------------------------------------------- check printing -# while (($current_key,$current_value) = each %style_for_target) { -# print "$current_key => $current_value\n"; -# } +sub decreasedepth { + my ($token) = @_; + $Apache::lonxml::depth--; + if ($Apache::lonxml::depth<$Apache::lonxml::olddepth-1) { + $#Apache::lonxml::depthcounter--; + $Apache::lonxml::olddepth=$Apache::lonxml::depth+1; + } + if ( $Apache::lonxml::depth < -1) { + &Apache::lonxml::warning("Unbalanced tags in resource"); + $Apache::lonxml::depth='-1'; + } + my $curdepth=join('_',@Apache::lonxml::depthcounter); + &Apache::lonxml::debug("e $Apache::lonxml::depth : $Apache::lonxml::olddepth : $token->[1] : $curdepth\n"); +#print "
e $Apache::lonxml::depth : $Apache::lonxml::olddepth : $token->[1] : $curdepth\n"; +} - return %style_for_target; - +sub get_all_text { + + my($tag,$pars)= @_; + my $depth=0; + my $token; + my $result=''; + if ( $tag =~ m:^/: ) { + my $tag=substr($tag,1); +# &Apache::lonxml::debug("have:$tag:"); + while (($depth >=0) && ($token = $pars->get_token)) { +# &Apache::lonxml::debug("e token:$token->[0]:$depth:$token->[1]"); + if (($token->[0] eq 'T')||($token->[0] eq 'C')||($token->[0] eq 'D')) { + $result.=$token->[1]; + } elsif ($token->[0] eq 'PI') { + $result.=$token->[2]; + } elsif ($token->[0] eq 'S') { + if ($token->[1] eq $tag) { $depth++; } + $result.=$token->[4]; + } elsif ($token->[0] eq 'E') { + if ( $token->[1] eq $tag) { $depth--; } + #skip sending back the last end tag + if ($depth > -1) { $result.=$token->[2]; } else { + $pars->unget_token($token); + } + } + } + } else { + while ($token = $pars->get_token) { +# &Apache::lonxml::debug("s token:$token->[0]:$depth:$token->[1]"); + if (($token->[0] eq 'T')||($token->[0] eq 'C')||($token->[0] eq 'D')) { + $result.=$token->[1]; + } elsif ($token->[0] eq 'PI') { + $result.=$token->[2]; + } elsif ($token->[0] eq 'S') { + if ( $token->[1] eq $tag) { + $pars->unget_token($token); last; + } else { + $result.=$token->[4]; + } + } elsif ($token->[0] eq 'E') { + $result.=$token->[2]; + } + } + } +# &Apache::lonxml::debug("Exit:$result:"); + return $result } +sub newparser { + my ($parser,$contentref,$dir) = @_; + push (@$parser,HTML::TokeParser->new($contentref)); + $$parser['-1']->xml_mode('1'); + if ( $dir eq '' ) { + push (@Apache::lonxml::pwd, $Apache::lonxml::pwd[$#Apache::lonxml::pwd]); + } else { + push (@Apache::lonxml::pwd, $dir); + } +# &Apache::lonxml::debug("pwd:$#Apache::lonxml::pwd"); +# &Apache::lonxml::debug("pwd:$Apache::lonxml::pwd[$#Apache::lonxml::pwd]"); +} +sub parstring { + my ($token) = @_; + my $temp=''; + map { + unless ($_=~/\W/) { + my $val=$token->[2]->{$_}; + $val =~ s/([\%\@\\])/\\$1/g; + #if ($val =~ m/^[\%\@]/) { $val="\\".$val; } + $temp .= "my \$$_=\"$val\";" + } + } @{$token->[3]}; + return $temp; +} -#=============================================================== Subroutine definition -#--------------------------------------------------------------------------------- Run - sub evaluate { - my ($expression,$safeeval) = @_; - return $safeeval->reval($expression); - } - - sub run { - my ($code,$safeeval) = @_; - $safeeval->reval($code); - } - -#===================================================================== TAG SUBROUTINES -#----------------------------------------------------------------------------- tag - sub start_m { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = "\$out = lontexconvert::converted(\$in = '\$'.\""; - } elsif ($target eq 'tex') { - $currentstring = "\$"; - } - return $currentstring; - } - sub end_m { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = "\".'\$') "; - } elsif ($target eq 'tex') { - $currentstring = "\$"; - } - return $currentstring; - } -#-------------------------------------------------------------------------- tag - sub start_html { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[4]; - } - return $currentstring; - } - sub end_html { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[2]; - } - return $currentstring; - } -#-------------------------------------------------------------------------- tag - sub start_head { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[4]; - } - return $currentstring; - } - sub end_head { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[2]; - } - return $currentstring; - } -#--------------------------------------------------------------------------- tag - sub start_map { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[4]; - } - return $currentstring; - } - sub end_map { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[2]; - } - return $currentstring; - } -#------------------------------------------------------------------------ tag - sub start_applet { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[4]; - } - return $currentstring; - } - sub end_applet { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[2]; - } - return $currentstring; - } -#------------------------------------------------------------------------ tag - sub start_input { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[4]; - } - return $currentstring; - } - sub end_input { - my ($target,$token) = @_; - my $currentstring = ''; - if ($target eq 'web') { - $currentstring = $token->[2]; - } - return $currentstring; - } -#----------------------------------------------------------------------