X-Git-Url: http://git.shiar.nl/perl/list-index.git/blobdiff_plain/40c02d74069ab5542b5c5c7990dba7e018a3d3fa..fd5749ab518cdf3c68e94fbad168aa2a7725d592:/t/10-ranges.t

diff --git a/t/10-ranges.t b/t/10-ranges.t
index c5cc3e6..cf3b0a4 100644
--- a/t/10-ranges.t
+++ b/t/10-ranges.t
@@ -2,54 +2,149 @@
 use strict;
 use warnings;
 
-use Test::More tests => 16;
+use Test::More tests => 8;
 use Test::NoWarnings;
 use Data::Dump 'pp';
 
 BEGIN { use_ok('List::Index'); }
 ok(eval { List::Index->VERSION(1) }, 'version 1.00 compatibility');
 
-{
-my @uniform = 'a'..'z';
-ok(my $index = List::Index->new(\@uniform), 'object (single-char values)');
-is_deeply(\@uniform, ['a'..'z'], 'original data unaltered');
-is_deeply($index->ranges, [['a','']], 'single page');
-is_deeply($index->ranges({pages => 3}), [map { [split /-/, $_, 2] } qw(
-	a-h i-q r-
-)], 'given pages');
-is_deeply($index->ranges({pagesize => @uniform / 2.1}), [map { [split /-/, $_, 2] } qw(
-	a-h i-q r-
-)], 'equivalent pagesize');
-}
+subtest 'single-char alphabet' => sub {
+	plan tests => 4;
+	my @uniform = 'a'..'z';
+	my $index = List::Index->new(\@uniform) or return;
+	is_deeply(\@uniform, ['a'..'z'], 'original data unaltered');
+	is_deeply($index->ranges, ['-'], 'single page');
+	is_deeply($index->ranges({pages => 3}), [qw(-i j-q r-)], 'given pages');
+	is_deeply($index->ranges({pagesize => @uniform / 2.1}), [qw(
+		-i j-q r-
+	)], 'equivalent pagesize');
+};
+
+subtest 'uniform alphanumeric' => sub {
+	plan tests => 2;
+	my $index = List::Index->new(['aa'..'zz', 1..202]) or return;
+	is_deeply($index->ranges, [qw(
+		-.
+		.-bp bq-dm dn-fi fj-hf hg-i j-k l-m n-os ot-qp qq-sm sn-uj uk-wf wg-x y-
+
+	)], 'default ranges');
+	is_deeply($index->ranges({pagesize => 300}), [qw(-c d-n o-)], 'large pagesize');
+};
 
-{
-ok(my $index = List::Index->new(['aa'..'zz', 1..193]), 'non-alphabetic values (uniform)');
-is_deeply($index->ranges, [map { [split /-/, $_, 2] } qw(
-	.-. ..-. ..-.. ...-.
-	a-bv bw-dr ds-fn fo-hk hl-jg jh-k l-m n-ov ow-qr qs-sn so-uk ul-wg wh-x y-
-)], 'default ranges');
-is_deeply($index->ranges({pagesize => 300}), [map { [split /-/, $_, 2] } qw(
-	.-c d-ov ow-
-)], '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 lgus 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'
+	);
+};
 
-{
-ok(my $index = List::Index->new([qw(
-	baa1 baa2  baa3 baaa  bbc cbc  daaa ea  eaaa zed
-)]), 'variable length values');
-is_deeply($index->ranges({pagesize => 2, context => 0}), [
-	map { [split /-/, $_, 2] } qw(baa.-baa baa.-bbb bbc-daa. daaa-eaa. eaaa-)
-], 'no context');
-is_deeply($index->ranges({pagesize => 2}), [
-	map { [split /-/, $_, 2] } qw(b-baa baa.-ba bb-c d-ea. eaa-)
-], 'default context');  # context should be 1
-is_deeply($index->ranges({pagesize => 2, context => 2}), [
-	map { [split /-/, $_, 2] } qw(b-a b-ba bb-c d-d e-)
-], 'overlap');  # first item equals second due to large context
-is_deeply($index->ranges({pagesize => 2, length => 1}), [
-	map { [split /-/, $_, 2] } qw(b-a b-a b-c d-d e-)
-], 'single char');
-
-#pp($index->ranges({pagesize => 2, context => 2, length => 1}));
+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=>6 }),
+		[qw(-g h i-)],
+		'large context'
+	);
+{ local $TODO = '?';
+	is_deeply(
+		$index->ranges({ pagesize=>10, context=>5 }),
+		# 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 'context' => sub {
+	plan tests => 4;
+	my $index = List::Index->new([qw(
+		baa1 baa2  baa3 baaa  bbc cbc  daaa ea  eaaa zed
+	)]) or return;
+	is_deeply($index->ranges({pagesize => 2, context => 0}), [
+		qw(-baa. baa.-bbb bbc-daa. daaa-eaa. eaaa-)
+	], 'no context');
+	is_deeply($index->ranges({pagesize => 2}), [
+		qw(-a b c d e-)
+	], 'default context');  # context should be 1
+	is_deeply($index->ranges({pagesize => 2, context => 2}), [
+		qw(-a b-c d e-)
+	], 'overlap');  # first item equals second due to large context
+	is_deeply($index->ranges({pagesize => 2, context => 0, length => 1}), [
+		qw(-a b-c d e-)
+	], 'single char');
+
+	#pp($index->ranges({pagesize => 2, context => 2, length => 1}));
+};