Open-ILS/src/sql/Pg/upgrade/0441.schema.naco-normalized-crossed-d.sql

   1 BEGIN;
   2
   3 INSERT INTO config.upgrade_log (version) VALUES ('0441'); -- gmc
   4
   5 CREATE OR REPLACE FUNCTION public.naco_normalize( TEXT, TEXT ) RETURNS TEXT AS $func$
   6         use Unicode::Normalize;
   7         use Encode;
   8
   9         # When working with Unicode data, the first step is to decode it to
  10         # a byte string; after that, lowercasing is safe
  11         my $txt = lc(decode_utf8(shift));
  12         my $sf = shift;
  13
  14         $txt = NFD($txt);
  15         $txt =~ s/\pM+//go;     # Remove diacritics
  16
  17         $txt =~ s/\xE6/AE/go;   # Convert ae digraph
  18         $txt =~ s/\x{153}/OE/go;# Convert oe digraph
  19         $txt =~ s/\xFE/TH/go;   # Convert Icelandic thorn
  20
  21         $txt =~ tr/\x{2070}\x{2071}\x{2072}\x{2073}\x{2074}\x{2075}\x{2076}\x{2077}\x{2078}\x{2079}\x{207A}\x{207B}/0123456789+-/;# Convert superscript numbers
  22         $txt =~ tr/\x{2080}\x{2081}\x{2082}\x{2083}\x{2084}\x{2085}\x{2086}\x{2087}\x{2088}\x{2089}\x{208A}\x{208B}/0123456889+-/;# Convert subscript numbers
  23
  24         $txt =~ tr/\x{0251}\x{03B1}\x{03B2}\x{0262}\x{03B3}/AABGG/;             # Convert Latin and Greek
  25         $txt =~ tr/\x{2113}\xF0\x{111}\!\"\(\)\-\{\}\<\>\;\:\.\?\xA1\xBF\/\\\@\*\%\=\xB1\+\xAE\xA9\x{2117}\$\xA3\x{FFE1}\xB0\^\_\~\`/LDD /;     # Convert Misc
  26         $txt =~ tr/\'\[\]\|//d;                                                 # Remove Misc
  27
  28         if ($sf && $sf =~ /^a/o) {
  29                 my $commapos = index($txt,',');
  30                 if ($commapos > -1) {
  31                         if ($commapos != length($txt) - 1) {
  32                                 my @list = split /,/, $txt;
  33                                 my $first = shift @list;
  34                                 $txt = $first . ',' . join(' ', @list);
  35                         } else {
  36                                 $txt =~ s/,/ /go;
  37                         }
  38                 }
  39         } else {
  40                 $txt =~ s/,/ /go;
  41         }
  42
  43         $txt =~ s/\s+/ /go;     # Compress multiple spaces
  44         $txt =~ s/^\s+//o;      # Remove leading space
  45         $txt =~ s/\s+$//o;      # Remove trailing space
  46
  47         # Encoding the outgoing string is good practice, but not strictly
  48         # necessary in this case because we've stripped everything from it
  49         return encode_utf8($txt);
  50 $func$ LANGUAGE 'plperlu' STRICT IMMUTABLE;
  51
  52 COMMIT;