lookbehind penalty

[perl/list-index.git] / lib / List / Index.pm

diff --git a/lib/List/Index.pm b/lib/List/Index.pm
index 3663c2459422638225058204dbc9854f667f315f..06c27273fcbd67193e9eb4f751ab280a93f0a8ac 100644 (file)
--- a/lib/List/Index.pm
+++ b/lib/List/Index.pm
@@ -23,14 +23,15 @@ sub ranges {
        my $pages    = $options->{pages   } || 1 + int $#$self / $pagesize;
 
        $pagesize = @$self / $pages;
-       my $offset = $pagesize;
+       my $offset = $pagesize + .5;
+       my $penalty = 0;
        my @links = ('');
        while ($offset < @$self) {
                my $link = substr $self->[$offset], 0, $length;
                if ($context) {
-                       {
+                       if ($offset > $context - 1 + $penalty) {
                                # take a value slightly before the current offset
-                               my $before = $offset > $context ? $self->[$offset - $context] : '.';
+                               my $before = $self->[$offset - $context - 1 + $penalty];
                                # see how much of it matches the current link
                                my $trim = 1;
                                for my $match (split //, $before) {
@@ -41,6 +42,7 @@ sub ranges {
                                substr($link, $trim) = '' unless $trim > length $link;
                        }
 
+                       $penalty = 0;
                        if ($offset + $context < $#$self) {
                                # take a value after the current offset
                                my $after = $self->[$offset + $context];
@@ -53,6 +55,9 @@ sub ranges {
                                # use this link if it's shorter
                                if ($trim < length $link) {
                                        $link = substr $after, 0, $trim;
+#                                      $offset += $context + 1;
+                                       $self->[$offset + ++$penalty] =~ /^\Q$link/
+                                               while $offset + $penalty < $#$self;
                                }
                        }
                }
@@ -78,6 +83,7 @@ sub rangematch {
        my ($link) = @_;
        my ($s1, $s2) = $link =~ /([^-]*) - ([^-]*)/x
                or return qr/^\Q$link/i;
+       $s1 =~ s/\.$//;
        my @allow;
 
        if (length $s1) {
@@ -88,9 +94,12 @@ sub rangematch {
                my $prefix = '';
                my $char;
                for my $i (0 .. length($s1) - 1) {
+                       my $lasti = $i == length($s1) - 1;
                        $char = substr $s1, $i, 1;
                        my $next = $char;
-                       $next = chr( ord($char) + 1 ) if length $s1 > $i + 1;
+                       # do not include prefix character in final range
+                       $next = chr( ord($char) + 1 ) unless $lasti;
+
                        my $last = 'z';
                        next if $next gt $last;
                        if (length $s2 > $i) {
@@ -101,10 +110,23 @@ sub rangematch {
                                        next if $next gt $last;
                                }
                        }
+
+                       if ($char eq '.') {
+                               if ($last eq 'z') {
+#                                      push @allow, $prefix if $i and $lasti;
+#                                      next;
+                               }
+#                              if ($last eq 'z') {
+#                                      push @allow, $prefix if $i and $lasti;
+#                                      next;
+#                              }
+                               $next = 'a';
+                       }
+
                        push @allow, $prefix."[$next-$last]";
                }
                continue {
-                       $prefix .= $char;
+                       $prefix .= $char eq '.' ? '[^a-z]' : $char;
                }
        }
 
@@ -120,11 +142,16 @@ sub rangematch {
                                        next if $c1 le $char;
                                }
                        }
-                       push @allow, $prefix."(?![$char-$last])"
+
+                       if ($char eq '.') {
+                               next if $i < length($s2) - 1;
+                       }
+
+                       push @allow, $prefix.'(?!['.($char eq '.' ? 'a' : $char)."-$last])"
                                if $i or $s1 eq '';
                }
                continue {
-                       $prefix .= $char;
+                       $prefix .= $char eq '.' ? '[^a-z]' : $char;
                }
 
                push @allow, $prefix
@@ -142,18 +169,18 @@ __END__
 
 =head1 NAME
 
-List::Index - Paginate alphabetic entries by finding minimal prefixes
+List::Index - Find and apply prefix ranges to paginate keywords
 
 =head1 SYNOPSIS
 
        use List::Index;
        my $index = List::Index->new(\@values);
        my @pages = $index->ranges({pagesize => 50});
-       printf '<a href="?q=%s-%s">%1$s</a> ', @$_ for @pages;
+       say "<a href='?q=$_'>$_</a>" for @pages;
 
        use List::Index 'rangematch';
-       my $limit = rangematch('b-bmq');  # matches prefix like 'baa'..'bmq'
-       @results = grep { $limit } @results;
+       my $limit = rangematch('b-bmq');  # ge 'b' && le 'bmq'
+       @request = grep { $limit } @values;
 
 =head1 DESCRIPTION