X-Git-Url: http://git.shiar.nl/perl/list-index.git/blobdiff_plain/5d1c609f28d64e60c9595c9a207b8d19e823b122..9d8572d98bcc078f2946094bacc6a47dfbbd934b:/lib/List/Index.pm

diff --git a/lib/List/Index.pm b/lib/List/Index.pm
index 23ab261..e67e5ae 100644
--- a/lib/List/Index.pm
+++ b/lib/List/Index.pm
@@ -22,32 +22,71 @@ sub ranges {
 	my $length   = $options->{length  } || 4;
 	my $pages    = $options->{pages   } || 1 + int $#$self / $pagesize;
 
-	$pagesize = @$self / $pages;
-	my $offset = 0;
-	my @links;
+	$pagesize = $pages >= $#$self ? 1 : @$self / $pages;
+	my $offset = $pagesize + .5;
+	my $lookbehind = -$context;
+	my $lookahead  =  $context;
+	my @links = ('');
 	while ($offset < @$self) {
 		my $link = substr $self->[$offset], 0, $length;
 		if ($context) {
-			my $trim = 1;
-			my $before = $offset > $context ? $self->[$offset - $context] : '';
-			for my $match (split //, $before) {
-				scalar $link =~ /\G\Q$match/g or last;
-				$trim++;
+			my $penalty = 0;
+			# take a value slightly before the current offset
+			if ((my $before = $offset + $lookbehind) > 0) {
+				# see how much of it matches the current link
+				my $trim = 1;
+				for my $match (split //, $self->[$before - 1]) {
+					scalar $link =~ /\G\Q$match/g or last;
+					$trim++;
+				}
+				# truncate link upto where the earlier value starts to differ
+				if ($trim < length $link) {
+					substr($link, $trim) = '';
+					for (reverse $before .. $offset) {
+						$self->[$offset - $penalty] =~ /^\Q$link/ or last;
+						$penalty++;
+					}
+				}
 			}
-			substr($link, $trim) = '' unless $trim > length $link;
+
+			$lookbehind = -$context;
+
+			# take a value after the current offset
+			if ((my $after = $offset + $lookahead) < $#$self) {
+				# see how much of it matches the current link
+				my $trim = 1;
+				for my $match (split //, $self->[$after]) {
+					scalar $link =~ /\G\Q$match/g or last;
+					$trim++;
+				}
+				# use this link if it's shorter
+				if ($trim < length $link) {
+					$link = substr $self->[$after], 0, $trim;
+					# advance lookbehind offset on the next page
+					$penalty = 0;
+					for ($offset .. $after) {
+						last if $self->[$_] =~ /^\Q$link/;
+						$lookbehind++;
+					}
+				}
+			}
+
+			$lookahead = $context - $penalty;
 		}
 
-		push @links, [$link];
+		push @links, $link;
 		$offset += $pagesize;
 	}
 
+	use List::MoreUtils 'uniq';
+	@links = uniq @links;
 	for my $i (0 .. $#links - 1) {
-		my ($link, $lastchar) = $links[$i + 1]->[0] =~ /(.*)(.)/;
-		$link .= $lastchar eq '.' ? 'a' : chr( ord($lastchar) - 1 )
-			unless $lastchar eq 'a';
-		$links[$i]->[1] = $link;
+		my ($link, $lastchar) = $links[$i + 1] =~ /(.*)(.)/;
+		$link .= $lastchar le 'a' ? '.' : chr( ord($lastchar) - 1 );
+		next if $link eq $links[$i] and $i;
+		$links[$i] .= '-'.$link;
 	}
-	$links[-1]->[1] = '';
+	$links[-1] .= '-';
 
 	return \@links;
 }
@@ -56,16 +95,25 @@ sub rangematch {
 	my ($link) = @_;
 	my ($s1, $s2) = $link =~ /([^-]*) - ([^-]*)/x
 		or return qr/^\Q$link/i;
+	$s1 =~ s/\.$//;
 	my @allow;
 
 	if (length $s1) {
+		if (length $s2) {
+			$s1 le $s2 or $s1 =~ /^\Q$s2/ or return undef;
+		}
+
 		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 > $i) {
 				if ($s2 =~ /^\Q$prefix/) {
 					$last = substr $s2, $i, 1;
@@ -74,10 +122,23 @@ 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;
 		}
 	}
 
@@ -93,11 +154,16 @@ 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
@@ -115,18 +181,18 @@ __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