Secure testing commits - Jan 2008 - r7925 - bin

Author: stef-guest
Date: 2008-01-14 23:08:05 +0000 (Mon, 14 Jan 2008)
New Revision: 7925

Modified:
   bin/check-new-issues
Log:
- preliminary support for embedded-code-copies
- minor improvements at guessing the product name


Modified: bin/check-new-issues
==================================================================---
bin/check-new-issues	2008-01-14 23:05:37 UTC (rev 7924)
+++ bin/check-new-issues	2008-01-14 23:08:05 UTC (rev 7925)
@@ -7,7 +7,7 @@
 use Term::ReadLine;
 
 my %opts;
-getopts(''ln:fhi:t:T'', \%opts);
+getopts(''ln:fhi:t:Tc'', \%opts);
 
 if ($opts{h}) {
     print <<''EOF'';
@@ -20,6 +20,7 @@
   * blank line to skip to next issue
   * .fname to do "apt-file search name"
   * .cname to do "apt-cache search name"
+  * .mpackage to search data/embedded-code-copies for "package"
   * v or e to launch an editor with the current item
   * q to save and quit
   * CTRL-C to quit without saving
@@ -34,6 +35,7 @@
 -i regexp : use regexp to select issues (default:
''CVE-200[3-9]'' )
 -t regexp : use regexp to select todos (default: ''^\s+TODO:
check$'' )
 -T	  : same as -t ''^\s+TODO: check'' (note the missing $)
+-c        : only do syntax check of embedded-code-copies
 
 EOF
 
@@ -56,6 +58,18 @@
 	$basedir="..";
 }
 
+
+my $embed_code = {};
+my $embed_pkg = {};
+my $embed_errors;
+
+read_embedded_copies();
+
+if ($opts{c}) {
+	exit($embed_errors);
+}
+
+
 my $datafile="$basedir/data/CVE/list";
 my $allitemsfile="gunzip -c $basedir/../allitems.txt.gz|";
 my
$allitemsurl="http://cve.mitre.org/data/downloads/allitems.txt.gz";
@@ -100,7 +114,9 @@
 	}
 }
 
-print scalar(@{$CVEs}), "/", scalar(@{$entries}),  "/",
scalar(@todos), "\n";
+print scalar(@{$CVEs}), " CVEs, ",
+      scalar(@{$entries}) - scalar(@{$CVEs}),  " temp issues, ",
+      scalar(@todos),  " todos matching /$todo_regexp/\n";
 
 if ($opts{l}) {
 	#list only
@@ -159,6 +175,14 @@
 			print "===\n";
 			next READ;
 		}
+		elsif ($r=~ /^\.m(.*)$/ ) {
+			my $s = $1;
+			$s =~ s/^\s+//;
+			$s =~ s/\s+$//;
+			print "references to $s in embedded-code-copies:\n";
+			search_embed($s) or print "none\n";
+			next READ;
+		}
 		elsif ($r=~ /^q$/i ) {
 			last TODO;
 		}
@@ -247,7 +271,10 @@
 	
 	my $file;
 	my $prog;
-	if ( $desc =~ / in (\S+\.\S+) in (\S+) / ) {
+	if ( $desc =~ /^(\S+(?: [A-Z]\w*)*) \d/ ) {
+		$prog = $1;
+	}
+	elsif ( $desc =~ / in (\S+\.\S+) in (?:the )?(\S+) / ) {
 		$file = $1;
 		$prog = $2;
 	}
@@ -259,11 +286,15 @@
 		my $ac=`apt-cache search ''$prog'' |wc -l`;
 		chomp $ac;
 		print "\r$ac results from apt-cache search $prog\n";
+
+		foreach my $p (split /\s+/, $prog) {
+			search_embed($p);
+		}
 	}
-	if ( $file eq ''index.php'' ) {
+	if ( $file =~ /^(?:index|default|login|search|admin)\.(?:php3?|asp|cgi|pl)$/i
) {
 		return;
 	}
-	if ( $file =~ /(php3?|asp|cgi)$/ ) {
+	if ( $file =~ /(php3?|asp|cgi|pl)$/ ) {
 		if (! exists $afcache{$file}) {
 			print "doing apt-file search...";
 			$afcache{$file}=`apt-file -i search ''$file'' |wc -l`;
@@ -272,3 +303,71 @@
 		print "\r$afcache{$file} results from apt-file -i search $file\n";
 	}
 }
+
+sub read_embedded_copies {
+	open(my $fh, "$basedir/data/embedded-code-copies");
+
+	# skip comments
+	while (<$fh>) {
+		last if /^---BEGIN/; 
+	}
+	
+	my ($code, $pkg);
+	while (my $line = <$fh>) {
+		if ($line =~ /^([-\w]+)/) {
+			$code = lc($1);
+			$pkg  = undef;
+			if (exists $embed_code->{$code}) {
+				syntax_error("Duplicate embedded code $code")
+			}
+		}
+		elsif ($line =~ /^\s*$/) {
+			$code = undef;
+			$pkg = undef;
+		}
+		elsif ($line =~ /^\s+(?:\[\w+\]\s+)?-\s+(\w[\w.-]+)/) {
+			$pkg = $1;
+			$line =~ s/^\s+//;
+			if ($embed_code->{$code}->{$pkg}) {
+				$embed_code->{$code}->{$pkg} .= $line;
+			}
+			else {
+				$embed_code->{$code}->{$pkg} = $line;
+				push @{$embed_pkg->{$pkg}}, $code;
+			}
+		}
+		elsif ($line =~ /^\s+(?:NOTE|TODO)/) {
+			$line =~ s/^\s+//;
+			if ($pkg) {
+				$embed_code->{$code}->{$pkg} .= $line;
+			}
+		}
+		else {
+			syntax_error("Cannot parse $line");
+		}
+	}
+}
+
+sub syntax_error {
+	$embed_errors=1;
+	print STDERR "embedded-code-copies:$.: @_\n";
+}
+
+sub search_embed {
+	my $text = shift;
+	my $found = 0;
+	$text = lc($text);
+	if (exists $embed_code->{$text}) {
+		print "$text is embedded by: ",
+		      join(" ", sort keys %{$embed_code->{$text}}),
+		      "\n";
+		$found = 1;
+	}
+	if (exists $embed_pkg->{$text}) {
+		print "$text embeds: ",
+		      join(" ", sort @{$embed_pkg->{$text}}),
+		      "\n";
+		$found = 1;
+	}
+	return $found;
+}