X-Git-Url: http://git.shiar.nl/perl/list-index.git/blobdiff_plain/a0438ade03450c7141821f7f8aa25d3a8e39d849..9d8572d98bcc078f2946094bacc6a47dfbbd934b:/t/10-ranges.t

diff --git a/t/10-ranges.t b/t/10-ranges.t
index 55df649..01971c8 100644
--- a/t/10-ranges.t
+++ b/t/10-ranges.t
@@ -2,7 +2,7 @@
 use strict;
 use warnings;
 
-use Test::More tests => 7;
+use Test::More tests => 9;
 use Test::NoWarnings;
 use Data::Dump 'pp';
 
@@ -10,7 +10,7 @@ BEGIN { use_ok('List::Index'); }
 ok(eval { List::Index->VERSION(1) }, 'version 1.00 compatibility');
 
 subtest 'single-char alphabet' => sub {
-	plan tests => 4;
+	plan tests => 5;
 	my @uniform = 'a'..'z';
 	my $index = List::Index->new(\@uniform) or return;
 	is_deeply(\@uniform, ['a'..'z'], 'original data unaltered');
@@ -19,6 +19,7 @@ subtest 'single-char alphabet' => sub {
 	is_deeply($index->ranges({pagesize => @uniform / 2.1}), [qw(
 		-i j-q r-
 	)], 'equivalent pagesize');
+	is_deeply($index->ranges({ pages => 500 }), ['-a', 'b'..'y', 'z-'], 'max pages');
 };
 
 subtest 'uniform alphanumeric' => sub {
@@ -32,6 +33,121 @@ subtest 'uniform alphanumeric' => sub {
 	is_deeply($index->ranges({pagesize => 300}), [qw(-c d-n o-)], 'large pagesize');
 };
 
+subtest 'context' => sub {
+	plan tests => 9;
+	my $index = List::Index->new([qw(
+		kkeg kl km kmlu knsy    koxb kpeo kuaa kuab kuac
+		kuapa kuq kur kux kzb   lc lg lgu lgua lguc
+		lguq lgur lgws lgx lka  lkq lks lln llq llx
+	)]) or return;
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>0, length=>5 }),
+		# ranges should match offsets exactly
+		[qw(-kuap. kuapa-lgup lguq-)],
+		'no context'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>0 }),
+		# default length limits to 4 chars
+		[qw(-kuao kuap-lgup lguq-)],
+		'default length'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>1 }),
+		# lookbehinds aren't shorter (kuac<kuap, lguc<lguq)
+		# 'kuap' can advance to 'kuq'
+		[qw(-kup kuq-lgup lguq-)],
+		'lookahead'
+	);
+TODO: {
+	local $TODO = 'backtrack';
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>2 }),
+		# allowed to advance to 'kur', but provides no benefits over 'kuq'
+		[qw(-kup kuq-lgup lguq-)],
+		'minimal lookahead'
+	);
+}
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>3 }),
+		# shorten 'kuap' to 'ku' because lookbehind is 'kp...'
+		# 'lguq' matches 'lg', but may only backtrack to 'lgu'
+		[qw(-kt ku-lgt lgu-)],
+		'lookbehind'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>4 }),
+		[qw(-kt ku-lf lg-)],
+		'maximal lookahead'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>5 }),
+		# after forwarding 'kuap' to 'lc'
+		# disallow backtracking of 'lguq' to 'lc' to prevent qw[-k l-]
+		# so only lookahead (to 'lkq') remains
+		[qw(-k l-lj lk-)],
+		'lookbehind forbidden'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>9 }),
+		# allow a single (10-9) entry (l-lf = lc) to remain
+		[qw(-k l-lf lg-)],
+		'lookbehind penalty'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>10 }),
+		# allow the last page to go back upto 'lc', replacing the 2nd page
+		[qw(-k l-)],
+		'full overlap'
+	);
+};
+
+subtest 'distribution' => sub {
+	plan tests => 2;
+	my $index = List::Index->new([qw(
+		gnihka gniub go gsearnrqns gtdvcxyt gw gwoufolwcvmtueyg gysgphci h habkdgifjfxoh
+		hbbvjf hbqleexnqts hccg hd hdoeqwdmgqwaoya hfbegicieuxz hfm hj hkoysmws hmylu
+		hnvtvpievbdlkrmb hs hvdvcqn hvn hyrybeur iaiaab ib ibavqyar idfniqvxpohbk idh
+	)]) or return;
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>8 }),
+		[qw(-g h i-)],
+		'large context'
+	);
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>7 }),
+		# after 2nd page is enlarged by lookbehind to 'h', limit subsequent lookahead
+		# to prevent the page from getting too large (17 entries if forwarded to 'i')
+		[qw(-g h-hm hn-)],
+		'lookahead penalty'
+	);
+	# page #14 [gn-g] (8): gnihka gniub go gsearnrqns gtdvcxyt gwawkvmueovdjtfj gwoufolwcvmtueyg gysgphci
+	# page #15 [h] (17): h habkdgifjfxoh hbbvjf hbqleexnqts hccgszftbaymfu hdaqzkow hdoeqwdmgqwaoya hfbegicieu hfmlpzzioqjbthz hj hkoysmws hmylu hnvtvpievbdlkrmb hsodfpkatk hvdvcqn hvn hyrybeurqtevjfmi
+	# page #16 [i-ie] (5): i iaab ibiavqyar idfniqvxpohbk idh
+};
+
+subtest 'modulo' => sub {
+	plan tests => 2;
+	my $index = List::Index->new([qw(
+		a b ccb   ccd  cce gf gg   gh  i j
+	)]) or return;
+	# 10 entries at 4 per page requires 3 pages
+	# so actual target page sizes should be 3,4,3 (not 4,4,2)
+
+	is_deeply(
+		$index->ranges({ pagesize=>4, context=>0 }),
+		[qw(-ccc ccd-gg gh-)],
+		'uniform page sizes'
+	);
+{ local $TODO = 'early lookbehind causing [c-gg]';
+	is_deeply(
+		$index->ranges({ pagesize=>4, context=>1 }),
+		[qw(-b c-h i-)],
+		'context at new intervals'
+	);
+}
+};
+
 subtest 'context' => sub {
 	plan tests => 4;
 	my $index = List::Index->new([qw(
@@ -53,27 +169,3 @@ subtest 'context' => sub {
 	#pp($index->ranges({pagesize => 2, context => 2, length => 1}));
 };
 
-subtest 'distribution' => sub {
-	plan tests => 2;
-	my $index = List::Index->new([qw(
-		kkeg kl km kmlu knsy   koxb kpeo kqbt krzu ktyp
-		kuap kuy kvbc kyy kzb  lc lg lgaa lgbv lgbw
-		lgu lij ljr ljs lka    lkq lks lln llq llx
-	)]) or return;
-TODO: {
-	local $TODO = 'under development';
-	is_deeply(
-		$index->ranges({ pagesize=>10, context=>5 }),
-		# after 'kuap' forwards to 'kzb', 'lgu' shouldn't go back to 'lc'
-		# otherwise we get qw[-k l-]
-		[qw(-k l-lg lgu-)],
-		'lookbehind after full lookahead'
-	);
-}
-	is_deeply(
-		$index->ranges({ pagesize=>10, context=>4 }),
-		[qw(-kt ku-lf lg-)],
-		'maximal lookahead'
-	);
-}
-