DB cleanup: Remove some non-manpages & fix wrongly-detected locales

sql/update-2016-10-16.sql

@@ -0,0 +1,19 @@
+-- Various non-manpages
+DELETE FROM man
+ WHERE filename ~ '/Makefile\.(in|am)$'
+    OR filename ~ '/\.cvsignore(\.gz)?$'
+    OR filename !~ '/[^/]*\.[^/]*$'
+    OR filename ~ '/man\.tmp$';
+
+-- Wrong locales, found with:
+--   SELECT DISTINCT Locale FROM man ORDER BY locale;
+UPDATE man SET locale = NULL
+ WHERE locale = '5man'
+    OR locale = 'c'
+    OR locale ~ '^man.?$'
+    OR locale ~ '^Man-Part[12]$';
+
+-- Man page containing only a '$1'. Likely a build failure in earlier FreeBSD releases.
+DELETE FROM man WHERE hash = '\x5ea7b8101325c704551852f70b652e0a2b0d7c12';
+
+DELETE FROM contents c WHERE NOT EXISTS(SELECT 1 FROM man m WHERE m.hash = c.hash);

util/add_dir.pl

@@ -154,9 +154,23 @@ find sub {
   return warn "abs_path($File::Find::name): $!\n" if !$path;
   return warn "$vpath ($path) points outside of the tar directory!\n" if $path !~ s/^\Q$dir\E//;
   # Note: fltk also creates pre-formatted pages in /cat$sectre/, but those are ignored.
-  # TODO: Also ignore html and INDEX sections
+
   return warn "Ignoring $vpath\n" if $vpath !~ m{man(?:/([^/]+))?/man[0-9n]/([^/]+)$};
-  addman $pkgid, $vpath, $2, $1;
+  my($locale, $fn) = ($1, $2);
+  return warn "Ignoring $vpath\n" if
+       $fn =~ /^Makefile\.(in|am)$/
+    || $fn =~ /^\.cvsignore(\.gz)?$/
+    || $fn !~ /\./ # Also excludes INDEX files
+    || $fn eq 'man.tmp';
+
+  $locale = undef if $locale && (
+       $locale eq '5man'
+    || $locale eq 'c'
+    || $locale =~ /^man.?$/
+    || $locale =~ /^Man-Part[12]/
+  );
+
+  addman $pkgid, $vpath, $fn, $locale;
   $found++;
 }, $dir;