1 package OpenILS::Utils::Normalize;
4 use Unicode::Normalize;
8 use MARC::File::XML ( BinaryEncoding => 'UTF-8' );
9 use OpenILS::Application::AppUtils;
11 use Exporter 'import';
12 our @EXPORT_OK = qw( clean_marc naco_normalize search_normalize );
18 # Apply NACO normalization to input string; based on
19 # https://www.loc.gov/aba/pcc/naco/documents/SCA_PccNormalization_Final_revised.pdf
21 # Note that unlike a strict reading of the NACO normalization rules,
22 # output is returned as lowercase instead of uppercase for compatibility
23 # with previous versions of the Evergreen naco_normalize routine.
24 $str = _normalize_substitutions($str, $sf);
26 # Remove apostrophes, per NACO specs
29 $str = _normalize_codes($str, $sf);
34 sub search_normalize {
38 $str = _normalize_substitutions($str, $sf);
39 $str = _normalize_codes($str, $sf);
44 sub _normalize_substitutions {
48 # Convert to upper-case first; even though final output will be lowercase, doing this will
49 # ensure that the German eszett (ß) and certain ligatures (ff, fi, ffl, etc.) will be handled correctly.
50 # If there are any bugs in Perl's implementation of upcasing, they will be passed through here.
54 # remove non-filing strings
55 $str =~ s/\x{0098}.*?\x{009C}//g;
57 # Replace curly single and double quote-like characters with straight single and double quotes
58 $str =~ s/[\x{2018}\x{2019}\x{201B}\x{FF07}\x{201A}]/\x{0027}/g;
59 $str =~ s/[\x{201C}\x{201D}\x{201F}\x{FF0C}\x{201E}\x{2E42}]/\x{0022}/g;
63 # additional substitutions - 3.6.
64 $str =~ s/\x{00C6}/AE/g;
65 $str =~ s/\x{00DE}/TH/g;
66 $str =~ s/\x{0152}/OE/g;
67 $str =~ tr/\x{0110}\x{00D0}\x{00D8}\x{0141}\x{2113}\x{02BB}\x{02BC}][/DDOLl/d;
72 sub _normalize_codes {
76 # transformations based on Unicode category codes
77 $str =~ s/[\p{Cc}\p{Cf}\p{Co}\p{Cs}\p{Lm}\p{Mc}\p{Me}\p{Mn}]//g;
79 if ($sf && $sf =~ /^a/o) {
80 my $commapos = index($str, ',');
82 if ($commapos != length($str) - 1) {
83 $str =~ s/,/\x07/; # preserve first comma
88 # since we've stripped out the control characters, we can now
89 # use a few as placeholders temporarily
90 $str =~ tr/+&@\x{266D}\x{266F}#/\x01\x02\x03\x04\x05\x06/;
91 $str =~ s/[\p{Pc}\p{Pd}\p{Pe}\p{Pf}\p{Pi}\p{Po}\p{Ps}\p{Sk}\p{Sm}\p{So}\p{Zl}\p{Zp}\p{Zs}]/ /g;
92 $str =~ tr/\x01\x02\x03\x04\x05\x06\x07/+&@\x{266D}\x{266F}#,/;
95 $str =~ tr/\x{0660}-\x{0669}\x{06F0}-\x{06F9}\x{07C0}-\x{07C9}\x{0966}-\x{096F}\x{09E6}-\x{09EF}\x{0A66}-\x{0A6F}\x{0AE6}-\x{0AEF}\x{0B66}-\x{0B6F}\x{0BE6}-\x{0BEF}\x{0C66}-\x{0C6F}\x{0CE6}-\x{0CEF}\x{0D66}-\x{0D6F}\x{0E50}-\x{0E59}\x{0ED0}-\x{0ED9}\x{0F20}-\x{0F29}\x{1040}-\x{1049}\x{1090}-\x{1099}\x{17E0}-\x{17E9}\x{1810}-\x{1819}\x{1946}-\x{194F}\x{19D0}-\x{19D9}\x{1A80}-\x{1A89}\x{1A90}-\x{1A99}\x{1B50}-\x{1B59}\x{1BB0}-\x{1BB9}\x{1C40}-\x{1C49}\x{1C50}-\x{1C59}\x{A620}-\x{A629}\x{A8D0}-\x{A8D9}\x{A900}-\x{A909}\x{A9D0}-\x{A9D9}\x{AA50}-\x{AA59}\x{ABF0}-\x{ABF9}\x{FF10}-\x{FF19}/0-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-90-9/;
97 # intentionally skipping step 8 of the NACO algorithm; if the string
98 # gets normalized away, that's fine.
100 # leading and trailing spaces
108 # Cleans up a MARC::Record or MARCXML string for storage in the
111 # Takes either a MARC::Record or a string of MARCXML.
113 # Returns a string of MARCXML as Open-ILS likes to store it.
115 # Assumes input is already in UTF-8.
118 my $xml = (UNIVERSAL::isa($input, 'MARC::Record')) ? $input->as_xml_record() : $input;
120 $xml =~ s/^<\?xml.+\?\s*>//go;
121 $xml =~ s/>\s+</></go;
122 $xml =~ s/\p{Cc}//go;
123 $xml = OpenILS::Application::AppUtils->entityize($xml);
124 $xml =~ s/[\x00-\x1f]//go;