--- loncom/build/lpml_parse.pl	2001/06/24 23:00:32	1.3
+++ loncom/build/lpml_parse.pl	2001/09/17 20:44:38	1.11
@@ -1,12 +1,30 @@
 #!/usr/bin/perl
 
 # Scott Harrison
+# YEAR=2001
 # May 2001
 # 06/19/2001,06/20,06/24 - Scott Harrison
+# 9/5/2001,9/6,9/7,9/8 - Scott Harrison
 
+###############################################################################
+##                                                                           ##
+## ORGANIZATION OF THIS PERL SCRIPT                                          ##
+## 1. Notes                                                                  ##
+## 2. Get command line arguments                                             ##
+## 3. First pass through (grab distribution-specific information)            ##
+## 4. Second pass through (parse out what is not necessary)                  ##
+## 5. Third pass through (translate markup according to specified mode)      ##
+##                                                                           ##
+###############################################################################
+
+# ----------------------------------------------------------------------- Notes
+#
 # I am using a multiple pass-through approach to parsing
 # the lpml file.  This saves memory and makes sure the server
 # will never be overloaded.
+#
+# This is meant to parse files meeting the lpml document type.
+# See lpml.dtd.  LPML=Linux Packaging Markup Language.
 
 use HTML::TokeParser;
 
@@ -14,9 +32,10 @@ my $usage=<<END;
 **** ERROR ERROR ERROR ERROR ****
 Usage is for lpml file to come in through standard input.
 1st argument is the mode of parsing.
-2nd argument is the distribution (default,redhat6.2,debian2.2,redhat7.1,etc).
-3rd argument is to manually specify a sourceroot.
-4th argument is to manually specify a targetroot.
+2nd argument is the category permissions to use (runtime or development)
+3rd argument is the distribution (default,redhat6.2,debian2.2,redhat7.1,etc).
+4th argument is to manually specify a sourceroot.
+5th argument is to manually specify a targetroot.
 
 Only the 1st argument is mandatory for the program to run.
 
@@ -29,15 +48,21 @@ END
 # ------------------------------------------------- Grab command line arguments
 
 my $mode;
-if (@ARGV) {
+if (@ARGV==5) {
     $mode = shift @ARGV;
 }
 else {
+    @ARGV=();shift @ARGV;
     while(<>){} # throw away the input to avoid broken pipes
     print $usage;
     exit -1; # exit with error status
 }
 
+my $categorytype;
+if (@ARGV) {
+    $categorytype = shift @ARGV;
+}
+
 my $dist;
 if (@ARGV) {
     $dist = shift @ARGV;
@@ -46,10 +71,25 @@ if (@ARGV) {
 my $targetroot;
 my $sourceroot;
 if (@ARGV) {
-    $targetroot = shift @ARGV;
+    $sourceroot = shift @ARGV;
 }
 if (@ARGV) {
-    $sourceroot = shift @ARGV;
+    $targetroot = shift @ARGV;
+}
+$sourceroot=~s/\/$//;
+$targetroot=~s/\/$//;
+
+my $invocation;
+# --------------------------------------------------- Record program invocation
+if ($mode eq 'install') {
+    $invocation=(<<END);
+# Invocation: STDINPUT | lpml_parse.pl
+#             1st argument (mode) is: $mode
+#             2nd argument (category type) is: $categorytype
+#             3rd argument (distribution) is: $dist
+#             4th argument (targetroot) is: described below
+#             5th argument (sourceroot) is: described below
+END
 }
 
 # ---------------------------------------------------- Start first pass through
@@ -104,6 +144,9 @@ while ($token = $parser->get_token()) {
 	$hierarchy[$hloc]++;
 	$key=$token->[1].join(',',@hierarchy[0..($hloc-1)]);
 	my $thisdist=' '.$token->[2]{'dist'}.' ';
+	# This conditional clause is set up to ignore two sets
+	# of invalid conditions before accepting entry into
+	# the cleanstring.
 	if ($hash{$key}==2 and
 	    !($thisdist eq '  ' or $thisdist =~/\s$dist\s/)) {
 	    if ($token->[4]!~/\/>$/) {
@@ -134,6 +177,7 @@ while ($token = $parser->get_token()) {
     }
 }
 $cleanstring=&trim($cleanstring);
+$cleanstring=~s/\>\s*\n\s*\</\>\</g;
 
 # ---------------------------------------------------- Start final pass through
 
@@ -181,6 +225,10 @@ my $command;
 my $status;
 my $dependencies;
 my $dependency;
+my @links;
+my %categoryhash;
+
+my @buildall;
 
 # Make new parser with distribution specific input
 undef $parser;
@@ -217,6 +265,7 @@ $parser->{textify}={
     files => \&format_files,
     file => \&format_file,
     fileglob => \&format_fileglob,
+    links => \&format_links,
     link => \&format_link,
     linkto => \&format_linkto,
     source => \&format_source,
@@ -242,7 +291,7 @@ while ($token = $parser->get_tag('lpml')
     $token = $parser->get_tag('/lpml');
     print $lpml; 
     print "\n";
-    $text=~s/\s*\n\s*\n\s*/\n/g;
+#    $text=~s/\s*\n\s*\n\s*/\n/g;
     print $text;
     print "\n";
     print &end();
@@ -251,7 +300,10 @@ exit;
 
 sub end {
     if ($mode eq 'html') {
-	return "THE END\n";
+	return "<br />THE END\n";
+    }
+    if ($mode eq 'install') {
+	return '';
     }
 }
 
@@ -267,7 +319,22 @@ sub format_lpml {
     my (@tokeninfo)=@_;
     my $date=`date`; chop $date;
     if ($mode eq 'html') {
-	$lpml = "LPML BEGINNING: $date";
+	$lpml = "<br />LPML BEGINNING: $date";
+    }
+    elsif ($mode eq 'install') {
+	print '# LPML install targets. Linux Packaging Markup Language,';
+	print ' by Scott Harrison 2001'."\n";
+	print '# This file was automatically generated on '.`date`;
+	print "\n".$invocation;
+    }
+    elsif ($mode eq 'build') {
+	$lpml = "# Dynamic Makefile generated by LON-CAPA build process\n";
+	$lpml .= '# This file was automatically generated on '.`date`;
+	$lpml .= "\n";
+	$lpml .= "SHELL=\"/bin/sh\"\n\n";
+    }
+    else {
+	return '';
     }
 }
 # --------------------------------------------------- Format targetroot section
@@ -276,7 +343,13 @@ sub format_targetroot {
     $text=$targetroot if $targetroot;
     $parser->get_tag('/targetroot');
     if ($mode eq 'html') {
-	return $targetroot="\nTARGETROOT: $text";
+	return $targetroot="\n<br />TARGETROOT: $text";
+    }
+    elsif ($mode eq 'install') {
+	return '# TARGET INSTALL LOCATION is "'.$targetroot."\"\n";
+    }
+    elsif ($mode eq 'build') {
+	return '# TARGET INSTALL LOCATION is "'.$targetroot."\"\n";
     }
     else {
 	return '';
@@ -288,7 +361,13 @@ sub format_sourceroot {
     $text=$sourceroot if $sourceroot;
     $parser->get_tag('/sourceroot');
     if ($mode eq 'html') {
-	return $sourceroot="\nSOURCEROOT: $text";
+	return $sourceroot="\n<br />SOURCEROOT: $text";
+    }
+    elsif ($mode eq 'install') {
+	return '# SOURCE CODE LOCATION IS "'.$sourceroot."\"\n";;
+    }
+    elsif ($mode eq 'build') {
+	return '# SOURCE CODE LOCATION IS "'.$sourceroot."\"\n";;
     }
     else {
 	return '';
@@ -299,7 +378,8 @@ sub format_categories {
     my $text=&trim($parser->get_text('/categories'));
     $parser->get_tag('/categories');
     if ($mode eq 'html') {
-	return $categories="\nBEGIN CATEGORIES\n$text\nEND CATEGORIES\n";
+	return $categories="\n<br />BEGIN CATEGORIES\n$text\n".
+	    "<br />END CATEGORIES\n";
     }
     else {
 	return '';
@@ -314,10 +394,15 @@ sub format_category {
     $parser->get_text('/category');
     $parser->get_tag('/category');
     if ($mode eq 'html') {
-	return $category="\nCATEGORY $category_att_name $category_att_type ".
-	    "$chmod $chown";
+	return $category="\n<br />CATEGORY $category_att_name ".
+	    "$category_att_type $chmod $chown";
     }
     else {
+	if ($category_att_type eq $categorytype) {
+	    my ($user,$group)=split(/\:/,$chown);
+	    $categoryhash{$category_att_name}='-o '.$user.' -g '.$group.
+		' -m '.$chmod;
+	}
 	return '';
     }
 }
@@ -348,7 +433,7 @@ sub format_rpm {
     my $text=&trim($parser->get_text('/rpm'));
     $parser->get_tag('/rpm');
     if ($mode eq 'html') {
-	return $rpm="\nBEGIN RPM\n$text\nEND RPM";
+	return $rpm="\n<br />BEGIN RPM\n$text\n<br />END RPM";
     }
     else {
 	return '';
@@ -359,7 +444,7 @@ sub format_rpmSummary {
     my $text=&trim($parser->get_text('/rpmSummary'));
     $parser->get_tag('/rpmSummary');
     if ($mode eq 'html') {
-	return $rpmSummary="\nRPMSUMMARY $text";
+	return $rpmSummary="\n<br />RPMSUMMARY $text";
     }
     else {
 	return '';
@@ -370,7 +455,7 @@ sub format_rpmName {
     my $text=&trim($parser->get_text('/rpmName'));
     $parser->get_tag('/rpmName');
     if ($mode eq 'html') {
-	return $rpmName="\nRPMNAME $text";
+	return $rpmName="\n<br />RPMNAME $text";
     }
     else {
 	return '';
@@ -381,7 +466,7 @@ sub format_rpmVersion {
     my $text=$parser->get_text('/rpmVersion');
     $parser->get_tag('/rpmVersion');
     if ($mode eq 'html') {
-	return $rpmVersion="\nRPMVERSION $text";
+	return $rpmVersion="\n<br />RPMVERSION $text";
     }
     else {
 	return '';
@@ -392,7 +477,7 @@ sub format_rpmRelease {
     my $text=$parser->get_text('/rpmRelease');
     $parser->get_tag('/rpmRelease');
     if ($mode eq 'html') {
-	return $rpmRelease="\nRPMRELEASE $text";
+	return $rpmRelease="\n<br />RPMRELEASE $text";
     }
     else {
 	return '';
@@ -403,7 +488,7 @@ sub format_rpmVendor {
     my $text=$parser->get_text('/rpmVendor');
     $parser->get_tag('/rpmVendor');
     if ($mode eq 'html') {
-	return $rpmVendor="\nRPMVENDOR $text";
+	return $rpmVendor="\n<br />RPMVENDOR $text";
     }
     else {
 	return '';
@@ -414,7 +499,7 @@ sub format_rpmBuildRoot {
     my $text=$parser->get_text('/rpmBuildRoot');
     $parser->get_tag('/rpmBuildRoot');
     if ($mode eq 'html') {
-	return $rpmBuildRoot="\nRPMBUILDROOT $text";
+	return $rpmBuildRoot="\n<br />RPMBUILDROOT $text";
     }
     else {
 	return '';
@@ -425,7 +510,7 @@ sub format_rpmCopyright {
     my $text=$parser->get_text('/rpmCopyright');
     $parser->get_tag('/rpmCopyright');
     if ($mode eq 'html') {
-	return $rpmCopyright="\nRPMCOPYRIGHT $text";
+	return $rpmCopyright="\n<br />RPMCOPYRIGHT $text";
     }
     else {
 	return '';
@@ -436,7 +521,7 @@ sub format_rpmGroup {
     my $text=$parser->get_text('/rpmGroup');
     $parser->get_tag('/rpmGroup');
     if ($mode eq 'html') {
-	return $rpmGroup="\nRPMGROUP $text";
+	return $rpmGroup="\n<br />RPMGROUP $text";
     }
     else {
 	return '';
@@ -447,7 +532,7 @@ sub format_rpmSource {
     my $text=$parser->get_text('/rpmSource');
     $parser->get_tag('/rpmSource');
     if ($mode eq 'html') {
-	return $rpmSource="\nRPMSOURCE $text";
+	return $rpmSource="\n<br />RPMSOURCE $text";
     }
     else {
 	return '';
@@ -458,7 +543,7 @@ sub format_rpmAutoReqProv {
     my $text=$parser->get_text('/rpmAutoReqProv');
     $parser->get_tag('/rpmAutoReqProv');
     if ($mode eq 'html') {
-	return $rpmAutoReqProv="\nRPMAUTOREQPROV $text";
+	return $rpmAutoReqProv="\n<br />RPMAUTOREQPROV $text";
     }
     else {
 	return '';
@@ -469,7 +554,7 @@ sub format_rpmdescription {
     my $text=$parser->get_text('/rpmdescription');
     $parser->get_tag('/rpmdescription');
     if ($mode eq 'html') {
-	return $rpmdescription="\nRPMDESCRIPTION $text";
+	return $rpmdescription="\n<br />RPMDESCRIPTION $text";
     }
     else {
 	return '';
@@ -480,7 +565,7 @@ sub format_rpmpre {
     my $text=$parser->get_text('/rpmpre');
     $parser->get_tag('/rpmpre');
     if ($mode eq 'html') {
-	return $rpmpre="\nRPMPRE $text";
+	return $rpmpre="\n<br />RPMPRE $text";
     }
     else {
 	return '';
@@ -488,11 +573,15 @@ sub format_rpmpre {
 }
 # -------------------------------------------------- Format directories section
 sub format_directories {
-    my $text=&trim($parser->get_text('/directories'));
+    my $text=$parser->get_text('/directories');
     $parser->get_tag('/directories');
     if ($mode eq 'html') {
-	return $directories="\nBEGIN DIRECTORIES\n$text\nEND DIRECTORIES\n";
+	return $directories="\n<br />BEGIN DIRECTORIES\n$text\n<br />".
+	    "END DIRECTORIES\n";
     }
+    elsif ($mode eq 'install') {
+	return "\n".'directories:'."\n".$text;
+   }
     else {
 	return '';
     }
@@ -504,7 +593,12 @@ sub format_directory {
     $parser->get_text('/directory');
     $parser->get_tag('/directory');
     if ($mode eq 'html') {
-	return $directory="\nDIRECTORY $targetdir $categoryname $description";
+	return $directory="\n<br />DIRECTORY $targetdir $categoryname ".
+	    "$description";
+    }
+    elsif ($mode eq 'install') {
+	return "\t".'install '.$categoryhash{$categoryname}.' -d '.
+	    $targetroot.'/'.$targetdir."\n";
     }
     else {
 	return '';
@@ -536,7 +630,7 @@ sub format_categoryname {
 sub format_description {
     my @tokeninfo=@_;
     $description='';
-    my $text=&trim($parser->get_text('/description'));
+    my $text=&htmlsafe(&trim($parser->get_text('/description')));
     if ($text) {
 	$parser->get_tag('/description');
 	$description=$text;
@@ -545,10 +639,43 @@ sub format_description {
 }
 # -------------------------------------------------------- Format files section
 sub format_files {
-    my $text=&trim($parser->get_text('/files'));
+    my $text=$parser->get_text('/files');
     $parser->get_tag('/files');
     if ($mode eq 'html') {
-	return $directories="\nBEGIN FILES\n$text\nEND FILES\n";
+	return $directories="\n<br />BEGIN FILES\n$text\n<br />END FILES\n";
+    }
+    elsif ($mode eq 'install') {
+	return "\n".'files:'."\n".$text.
+	    "\n".'links:'."\n".join('',@links);
+    }
+    elsif ($mode eq 'build') {
+	my $binfo;
+	my $tword;
+	my $command2;
+	my @deps;
+	foreach my $bi (@buildinfo) {
+	    my ($source,$command,$trigger,@deps)=split(/\;/,$bi);
+	    $tword=''; $tword=' alwaysrun' if $trigger eq 'always run'; 
+	    $command=~s/\/([^\/]*)$//;
+	    $command2="cd $command; sh ./$1;\\";
+	    my $depstring;
+	    foreach my $dep (@deps) {
+		$depstring.="\telif !(test -r $command/$dep);\\\n";
+		$depstring.="\t\tthen echo ".
+		"\"**** LON-CAPA WARNING **** missing the file: ".
+ 	        "$command/$dep\";\\\n";
+	    }
+	    $binfo.="$source: $tword\n".
+		"\t\@if !(echo \"\");\\\n\t\tthen echo ".
+		"\"**** LON-CAPA WARNING **** Strange shell. ".
+ 	        "Check your path settings.\";\\\n".
+		$depstring.
+		"\telse \\\n\t\t$command2\n\tfi\n\n";
+	}
+	return 'all: '.join(' ',@buildall)."\n\n".
+  	        $text.
+		$binfo."\n".
+		"alwaysrun:\n\n";
     }
     else {
 	return '';
@@ -559,8 +686,19 @@ sub format_fileglobs {
 
 }
 # -------------------------------------------------------- Format links section
+# deprecated.. currently <link></link>'s are included in <files></files>
 sub format_links {
-
+    my $text=$parser->get_text('/links');
+    $parser->get_tag('/links');
+    if ($mode eq 'html') {
+	return $links="\n<br />BEGIN LINKS\n$text\n<br />END LINKS\n";
+    }
+    elsif ($mode eq 'install') {
+	return "\n".'links:'."\n\t".$text;
+    }
+    else {
+	return '';
+    }
 }
 # --------------------------------------------------------- Format file section
 sub format_file {
@@ -571,11 +709,27 @@ sub format_file {
     if ($source) {
 	$parser->get_tag('/file');
 	if ($mode eq 'html') {
-	    return ($file="\nBEGIN FILE\n".
+	    return ($file="\n<br />BEGIN FILE\n".
 		"$source $target $categoryname $description $note " .
 		"$build $status $dependencies" .
 		"\nEND FILE");
 	}
+	elsif ($mode eq 'install' && $categoryname ne 'conf') {
+	    return "\t".'@test -e '.$sourceroot.'/'.$source.
+		' && install '.
+		$categoryhash{$categoryname}.' '.
+		$sourceroot.'/'.$source.' '.
+		$targetroot.'/'.$target.
+		' || echo "**** LON-CAPA WARNING '.
+		'**** CVS source file does not exist: '.$sourceroot.'/'.
+		$source.'"'."\n";
+	}
+	elsif ($mode eq 'build' && $build) {
+	    push @buildall,$sourceroot.'/'.$source;
+	    push @buildinfo,$sourceroot.'/'.$source.';'.$build.';'.
+		$dependencies;
+#	    return '# need to build '.$source.";
+	}
 	else {
 	    return '';
 	}
@@ -591,11 +745,19 @@ sub format_link {
     if ($linkto) {
 	$parser->get_tag('/link');
 	if ($mode eq 'html') {
-	    return $link="\nBEGIN LINK\n".
+	    return $link="\n<br />BEGIN LINK\n".
 		"$linkto $target $categoryname $description $note " .
 		"$build $status $dependencies" .
 		    "\nEND LINK";
 	}
+	elsif ($mode eq 'install') {
+	    my @targets=map {s/^\s*//;s/\s$//;$_} split(/\;/,$target);
+	    foreach my $tgt (@targets) {
+		push @links,"\t".'ln -fs /'.$linkto.' /'.$targetroot.$tgt.
+		    "\n";
+	    }
+	    return '';
+	}
 	else {
 	    return '';
 	}
@@ -613,11 +775,17 @@ sub format_fileglob {
     if ($sourcedir) {
 	$parser->get_tag('/fileglob');
 	if ($mode eq 'html') {
-	    return $fileglob="\nBEGIN FILEGLOB\n".
+	    return $fileglob="\n<br />BEGIN FILEGLOB\n".
 		"$glob sourcedir $targetdir $categoryname $description $note ".
 		"$build $status $dependencies $filenames" .
 		    "\nEND FILEGLOB";
 	}
+	elsif ($mode eq 'install') {
+	    return "\t".'install '.
+		$categoryhash{$categoryname}.' '.
+		$sourceroot.'/'.$sourcedir.'[^CVS]'.$glob.' '.
+		$targetroot.'/'.$targetdir.'.'."\n";
+	}
 	else {
 	    return '';
 	}
@@ -676,7 +844,7 @@ sub format_build {
     my $text=&trim($parser->get_text('/build'));
     if ($text) {
 	$parser->get_tag('/build');
-	$build=$text;
+	$build=$sourceroot.'/'.$text.';'.$tokeninfo[2]{'trigger'};
     }
     return '';
 }
@@ -698,7 +866,8 @@ sub format_dependencies {
     my $text=&trim($parser->get_text('/dependencies'));
     if ($text) {
 	$parser->get_tag('/dependencies');
-	$dependencies=$text;
+	$dependencies=join(';',
+			      (map {s/^\s*//;s/\s$//;$_} split(/\;/,$text)));
     }
     return '';
 }
@@ -716,7 +885,6 @@ sub format_glob {
 # ---------------------------------------------------- Format filenames section
 sub format_filenames {
     my @tokeninfo=@_;
-    $glob='';
     my $text=&trim($parser->get_text('/filenames'));
     if ($text) {
 	$parser->get_tag('/filenames');
@@ -727,7 +895,6 @@ sub format_filenames {
 # ------------------------------------------------------- Format linkto section
 sub format_linkto {
     my @tokeninfo=@_;
-    $glob='';
     my $text=&trim($parser->get_text('/linkto'));
     if ($text) {
 	$parser->get_tag('/linkto');
@@ -735,6 +902,13 @@ sub format_linkto {
     }
     return '';
 }
+# ------------------------------------- Render less-than and greater-than signs
+sub htmlsafe {
+    my $text=@_[0];
+    $text =~ s/</&lt;/g;
+    $text =~ s/>/&gt;/g;
+    return $text;
+}
 # --------------------------------------- remove starting and ending whitespace
 sub trim {
     my ($s)=@_; $s=~s/^\s*//; $s=~s/\s*$//; return $s;