sbofixinfo: typo correction

1 files changed, 99 insertions, 2 deletions

diff --git a/sbofixinfo b/sbofixinfo
index eeba16e..7c8028e 100755
--- a/sbofixinfo
+++ b/sbofixinfo
@@ -70,6 +70,18 @@ Any capitalized hex digits in MD5SUM or MD5SUM_x86_64 will be lowercased.
 
 Lowercase key names (e.g. md5sum or download) will be uppercased.
 
+=item -
+
+Misspelled (typo'ed) key names will be corrected, if they're not too
+mangled. When this happens, a warning is given on standard error. The
+correction algorithm isn't perfect, so make sure you compare the
+original and 'corrected' key names using your own eyeballs and brain.
+
+=item -
+
+If there is no PRGNAM, it will be generated from the .info file's name
+(minus the I<.info> extension).
+
 =back
 
 B<sbofixinfo> doesn't attempt to detect any other errors or dubious
@@ -142,6 +154,13 @@ for($arg0) {
 		exec("pod2man --stderr -s1 -csbo-maintainer-tools -r$VERSION $0") ||
 			die "$SELF: can't exec pod2man\n";
 	};
+	/^--keytest/ && do {
+		while(<STDIN>) {
+			chomp;
+			fix_key($_);
+		}
+		exit 0;
+	};
 }
 
 push @ARGV, "." unless @ARGV;
@@ -152,7 +171,75 @@ exit $errcnt;
 
 sub fix_key {
 	my $key = shift;
-	$key =~ tr/-a-wyz/_A-WYZ/; # leave "x" lowercase!
+	my $newkey = fix_key_guts($key);
+	if($key ne $newkey) {
+		warn "$SELF: typo correction: $key => $newkey\n";
+	}
+	return $newkey;
+}
+
+sub fix_key_guts {
+	my $key = shift;
+
+	# fix case if needed
+	$key =~ tr/-a-wXyz/_A-WxYZ/; # leave "x" lowercase!
+
+	# if it's a recognized key, use it
+	if(grep { $_ eq $key } @keyorder) {
+		return $key;
+	}
+
+	# if it's not a recognized key, try to detect typos.
+	# most of the keys look different enough that the matching algo
+	# below does OK with them, except if someone types MAIL in place
+	# of EMAIL (the algo will match MAINTAINER, not EMAIL). so special
+	# case it here:
+	if((length $key < 8) && ($key =~ /^[ME]+[AI]+.?L.?/)) {
+		return "EMAIL";
+	}
+
+	my $x64 = 0;
+	if(/_x|_\d|x\d|86|68|64|46/) {
+		$x64 = 1;
+	}
+
+	my $bestmatch;
+	my $bmcount = 0;
+
+	# typo-correction matching algorithm:
+	for my $candidate (@keyorder) {
+		next if $candidate =~ /x86/;
+
+		#warn "checking $key against $candidate\n";
+
+		# if the first 2 characters match, assume the candidate is OK.
+		if(substr($key, 0, 2) eq substr($candidate, 0, 2)) {
+			$bestmatch = $candidate;
+			last;
+		}
+
+		# get rid of extraneous characters (not in the candidate key).
+		my $trykey = eval "\$key =~ tr/$candidate//cdr";
+
+		# if nothing's left, this is obviously the wrong candidate.
+		next unless length $trykey;
+
+		#warn "  trykey is " . $trykey;
+
+		if(length($trykey) > $bmcount) {
+			$bestmatch = $candidate;
+			$bmcount = length($trykey);
+		}
+	}
+
+	if(defined $bestmatch) {
+		$key = $bestmatch;
+		if(($key eq 'DOWNLOAD') || ($key eq 'MD5SUM')) {
+			$key .= "_x86_64" if $x64;
+		}
+	}
+
+	# if nothing matched, the key gets returned as-is.
 	return $key;
 }
 
@@ -170,6 +257,11 @@ sub fix_info {
 		return;
 	};
 
+	# this only gets used if the .info file is missing the PRGNAM key.
+	my $prgnam_guess = $file;
+	$prgnam_guess =~ s,.*/,,;
+	$prgnam_guess =~ s,\.[^.]*$,,;
+
 	my $key = 'INVALID_STUFF';
 	my %info;
 
@@ -220,6 +312,11 @@ sub fix_info {
 		return;
 	};
 
+	if(not defined $info{'PRGNAM'}) {
+		warn "$SELF: $file missing PRGNAM, guessing \"$prgnam_guess\" from filename.\n";
+		$info{'PRGNAM'} = $prgnam_guess;
+	}
+
 	# Reconstitute info file from the values, with correct indentation and
 	# quoting, backslashes, etc.
 	for $key (@keyorder) {
@@ -241,7 +338,7 @@ sub fix_info {
 
 	my $result = system("diff -u --color $file.bak $file");
 	if($result == 0) {
-		system("rm $file.bak");
+		unlink("$file.bak");
 		warn "$SELF: no changes to $file\n";
 	}
 }