Changeset 33898 for main/trunk

Timestamp:

2020-02-04T12:59:00+13:00 (4 years ago)

Author:

kjdon

Message:

format_metadata_for_sorting now takes two additional args - casefold and accentfold. Do these to the metadata value if they are set to true

File:

• main/trunk/greenstone2/perllib/sorttools.pm (modified) (7 diffs)

main/trunk/greenstone2/perllib/sorttools.pm

@@ -33 +33 @@
 
 use strict;
+use Unicode::Normalize;
 
 my $has_custom_sort = 0;
@@ -65 +66 @@
 # moved here from BasClas so import can share it
 sub format_metadata_for_sorting {
-    my ($metaname, $metavalue, $doc_obj) = @_;
-
-    if ($has_custom_sort && defined (&customsorttools::format_metadata_for_sorting)) {
-    return &customsorttools::format_metadata_for_sorting($metaname, $metavalue, $doc_obj);
-    }
+    my ($metaname, $metavalue, $doc_obj, $casefold, $accentfold) = @_;
     
     if (!defined $metaname || $metaname !~ /\S/ || ! defined $metavalue || $metavalue !~ /\S/) {
     return "";
+    }
+    
+    if ($has_custom_sort && defined (&customsorttools::format_metadata_for_sorting)) {
+    return &customsorttools::format_metadata_for_sorting($metaname, $metavalue, $doc_obj, $casefold, $accentfold);
     }
 
@@ -85 +86 @@
     }
     $lang = 'en' unless defined $lang;
-    
+
+    if (defined $casefold && $casefold eq "true") {
+    $metavalue = lc($metavalue);
+    }
+    if (defined $accentfold && $accentfold eq "true") {
+    $metavalue = NFKD($metavalue);
+    $metavalue =~ s/\p{NonspacingMark}//g;
+    }
+
     # is this metadata likely to be a name?
     my $function_name="format_string_name_$lang";
@@ -117 +126 @@
     }
     
-    $$stringref = lc($$stringref);
     $$stringref =~ s/&[^\;]+\;//g; # html entities
     $$stringref =~ s/^\s*(the|a|an)\b//; # articles
@@ -133 +141 @@
     }
     
-    $$stringref =~ tr/A-Z/a-z/;
     $$stringref =~ s/&\S+;//g;
 
@@ -163 +170 @@
     }
 
-    $$stringref = lc($$stringref);
     $$stringref =~ s/&[^\;]+\;//g; # html entities
     $$stringref =~ s/^\s*(les?|la|une?)\b//; # articles
@@ -179 +185 @@
     }
 
-    $$stringref = lc($$stringref);
     $$stringref =~ s/&[^\;]+\;//g; # html entities
     $$stringref =~ s/^\s*(la|el)\b//; # articles