Changeset 9581

Timestamp:

2005-04-07T14:02:48+12:00 (19 years ago)

Author:

jrm21

Message:

couple of extra languages for formatting

File:

• trunk/gsdl/perllib/sorttools.pm (modified) (4 diffs)

trunk/gsdl/perllib/sorttools.pm

@@ -61 +61 @@
 }
 
-# format an english string for sorting
-# i.e. convert to lowercase, remove the, a or an
-# from beginning of string etc.
+### language-specific sorting functions (called by format_metadata_for_sorting)
+
+## format_string_$lang() converts to lowercase (where appropriate), and
+# removes punctuation, articles from the start of string, etc
+## format_string_name_$lang() converts to lowercase, puts the surname first,
+# removes punctuation, etc
+
 sub format_string_en {
     my $stringref = shift;
 
-    $$stringref =~ tr/A-Z/a-z/;
-    $$stringref =~ s/&[^\;]+\;//g;
-    $$stringref =~ s/[^a-z0-9 ]//g;
-    $$stringref =~ s/^\s*(the|a|an)\b//;
+    $$stringref = lc($$stringref);
+    $$stringref =~ s/&[^\;]+\;//g; # html entities
+    $$stringref =~ s/[^[:alnum:]]//g;
+    $$stringref =~ s/^\s*(the|a|an)\b//; # articles
     $$stringref =~ s/\s+/ /g;
     $$stringref =~ s/^\s+//;
@@ -76 +80 @@
 }
 
-# format an english name for sorting
-# i.e. convert to lowercase, put surname before
-# first names etc.    
 sub format_string_name_en {
     my ($stringref) = @_;
@@ -86 +87 @@
 
     my $comma_format = ($$stringref =~ m/^.+,.+$/);
-
-    $$stringref =~ s/[^a-z0-9 ]//g;
+    $$stringref =~ s/[[:punct:]]//g;
     $$stringref =~ s/\s+/ /g;
     $$stringref =~ s/^\s+//;
@@ -103 +103 @@
     $$stringref = $surname . " " . $$stringref;
     }
+    print "name: $$stringref\n";
 }
+
+
+sub format_string_fr {
+    my $stringref = shift;
+
+    $$stringref = lc($$stringref);
+    $$stringref =~ s/&[^\;]+\;//g; # html entities
+    $$stringref =~ s/[^[:alpha:]]//g;
+    $$stringref =~ s/^\s*(les?|la|une?)\b//; # articles
+    $$stringref =~ s/\s+/ /g;
+    $$stringref =~ s/^\s+//;
+    $$stringref =~ s/\s+$//;
+}
+
+sub format_string_es {
+    my $stringref = shift;
+
+    $$stringref = lc($$stringref);
+    $$stringref =~ s/&[^\;]+\;//g; # html entities
+    $$stringref =~ s/[^[:alpha:]]//g;
+    $$stringref =~ s/^\s*(la|el)\b//; # articles
+    $$stringref =~ s/\s+/ /g;
+    $$stringref =~ s/^\s+//;
+    $$stringref =~ s/\s+$//;
+}
+
+### end of language-specific functions
 
 # takes arguments of day, month, year and converts to