Navigation

← Previous Changeset
Next Changeset →

Changeset 1970

Timestamp:

12/03/07 14:30:26 (5 months ago)

Author:

joshr

Message:

retabbed files to view prettier from SVN web browser.

Files:

Swishetest/trunk/t/010-C010-trivial-txt.t (modified) (1 diff)
Swishetest/trunk/t/011-C011-trivial-html.t (modified) (1 diff)
Swishetest/trunk/t/012-C012-trivial-xml.t (modified) (1 diff)
Swishetest/trunk/t/020-C020-wordsbasic-txt.t (modified) (1 diff)
Swishetest/trunk/t/030-C030-medsm-xml.t (modified) (3 diffs)
Swishetest/trunk/t/050-C020-largeindex.t (modified) (2 diffs)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

Swishetest/trunk/t/010-C010-trivial-txt.t

r1959	r1970
13	13
14	14	BEGIN {
15		require Carp;
16		$SIG{__WARN__} = sub { Carp::confess $_[0] };
17		use File::Path qw(mkpath);
18		mkpath( ["blib/index"], 0, 0755);
19		my $base = "C010";
20		my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-txt", "blib/index/$base.index" );
21		#print STDERR "out is " . Dumper( \%out );
	15	require Carp;
	16	$SIG{__WARN__} = sub { Carp::confess $_[0] };
	17	use File::Path qw(mkpath);
	18	mkpath( ["blib/index"], 0, 0755);
	19	my $base = "C010";
	20	my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-txt", "blib/index/$base.index" );
	21	#print STDERR "out is " . Dumper( \%out );
22	22
23		cmp_ok( scalar((keys(%out))), '>', 2, "Indexing output: " . scalar(%out) );
24		cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
25		cmp_ok( $out{properties}, '==', 5, 'num properties' );
26		cmp_ok( $out{files}, '==', 3, 'files indexed' );
27		cmp_ok( $out{bytes}, '==', 24, 'bytes indexed' );
28		cmp_ok( $out{words}, '==', 4, 'total words indexed' );
29
	23	cmp_ok( scalar((keys(%out))), '>', 2, "Indexing output: " . scalar(%out) );
	24	cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
	25	cmp_ok( $out{properties}, '==', 5, 'num properties' );
	26	cmp_ok( $out{files}, '==', 3, 'files indexed' );
	27	cmp_ok( $out{bytes}, '==', 24, 'bytes indexed' );
	28	cmp_ok( $out{words}, '==', 4, 'total words indexed' );
	29
30	30	DoSearch::open_index( "blib/index/$base.index" );
31		my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
	31	my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
32	32	DoSearch::close_index( "blib/index/$base.index" );
33		cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
	33	cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
34	34
35	35	};

Swishetest/trunk/t/011-C011-trivial-html.t

r1959	r1970
12	12
13	13	BEGIN {
14		require Carp;
15		$SIG{__WARN__} = sub { Carp::confess $_[0] };
16		use File::Path qw(mkpath);
17		mkpath( ["blib/index"], 0, 0755);
18		my $base = "C011";
19		my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-html", "blib/index/$base.index" );
	14	require Carp;
	15	$SIG{__WARN__} = sub { Carp::confess $_[0] };
	16	use File::Path qw(mkpath);
	17	mkpath( ["blib/index"], 0, 0755);
	18	my $base = "C011";
	19	my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-html", "blib/index/$base.index" );
20	20
21		cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
22		cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
23		cmp_ok( $out{properties}, '==', 5, 'num properties' );
24		cmp_ok( $out{files}, '==', 4, 'files indexed' );
25		cmp_ok( $out{bytes}, '==', 318, 'bytes indexed' );
26		cmp_ok( $out{words}, '==', 8, 'total words indexed' );
27
	21	cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
	22	cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
	23	cmp_ok( $out{properties}, '==', 5, 'num properties' );
	24	cmp_ok( $out{files}, '==', 4, 'files indexed' );
	25	cmp_ok( $out{bytes}, '==', 318, 'bytes indexed' );
	26	cmp_ok( $out{words}, '==', 8, 'total words indexed' );
	27
28	28	DoSearch::open_index( "blib/index/$base.index" );
29		my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
	29	my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
30	30	DoSearch::close_index( "blib/index/$base.index" );
31		cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
	31	cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
32	32
33	33	};

Swishetest/trunk/t/012-C012-trivial-xml.t

r1959	r1970
12	12
13	13	BEGIN {
14		require Carp;
15		$SIG{__WARN__} = sub { Carp::confess $_[0] };
16		use File::Path qw(mkpath);
17		mkpath( ["blib/index"], 0, 0755);
18		my $base = "C012";
19		my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-xml", "blib/index/$base.index" );
	14	require Carp;
	15	$SIG{__WARN__} = sub { Carp::confess $_[0] };
	16	use File::Path qw(mkpath);
	17	mkpath( ["blib/index"], 0, 0755);
	18	my $base = "C012";
	19	my (%out) = BuildIndex::build_index_from_directory( "data/$base-trivial-xml", "blib/index/$base.index" );
20	20
21		cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
22		cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
23		cmp_ok( $out{properties}, '==', 5, 'num properties' );
24		cmp_ok( $out{files}, '==', 4, 'files indexed' );
25		cmp_ok( $out{bytes}, '==', 301, 'bytes indexed' );
26		cmp_ok( $out{words}, '==', 8, 'total words indexed' );
27
	21	cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
	22	cmp_ok( $out{unique}, '==', 2, 'unique words indexed' );
	23	cmp_ok( $out{properties}, '==', 5, 'num properties' );
	24	cmp_ok( $out{files}, '==', 4, 'files indexed' );
	25	cmp_ok( $out{bytes}, '==', 301, 'bytes indexed' );
	26	cmp_ok( $out{words}, '==', 8, 'total words indexed' );
	27
28	28	DoSearch::open_index( "blib/index/$base.index" );
29		my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
	29	my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
30	30	DoSearch::close_index( "blib/index/$base.index" );
31		cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
	31	cmp_ok(scalar(@rows), '==', 3, "num results from 'swishe OR test'")
32	32
33	33	};

Swishetest/trunk/t/020-C020-wordsbasic-txt.t

r1959	r1970
12	12
13	13	BEGIN {
14		use File::Path qw(mkpath);
15		mkpath( ["blib/index"], 0, 0755);
16		my $base = "C020";
17		my (%out) = BuildIndex::build_index_from_directory( "data/$base-words-txt", "blib/index/$base.index" );
	14	use File::Path qw(mkpath);
	15	mkpath( ["blib/index"], 0, 0755);
	16	my $base = "C020";
	17	my (%out) = BuildIndex::build_index_from_directory( "data/$base-words-txt", "blib/index/$base.index" );
18	18
19		cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
20		cmp_ok( $out{unique}, '==', 252983, 'unique words indexed' );
21		cmp_ok( $out{properties}, '==', 5, 'num properties' );
22		cmp_ok( $out{files}, '==', 2, 'files indexed' );
23		cmp_ok( $out{bytes}, '==', 2896130, 'bytes indexed' );
24		cmp_ok( $out{words}, '==', 280381, 'total words indexed' );
	19	cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
	20	cmp_ok( $out{unique}, '==', 252983, 'unique words indexed' );
	21	cmp_ok( $out{properties}, '==', 5, 'num properties' );
	22	cmp_ok( $out{files}, '==', 2, 'files indexed' );
	23	cmp_ok( $out{bytes}, '==', 2896130, 'bytes indexed' );
	24	cmp_ok( $out{words}, '==', 280381, 'total words indexed' );
25	25
26	26	DoSearch::open_index( "blib/index/$base.index" );
27		my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
	27	my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
28	28	DoSearch::close_index( "blib/index/$base.index" );
29		cmp_ok(scalar(@rows), '==', 2, "num results from 'swishe OR test'")
	29	cmp_ok(scalar(@rows), '==', 2, "num results from 'swishe OR test'")
30	30	};
31	31

Swishetest/trunk/t/030-C030-medsm-xml.t

r1959	r1970
18	18	use MinMax;
19	19	use File::Path qw(mkpath);
20		use GetDictionaryWords;
21		use Test::More qw(no_plan);
22		my $max_words = MinMax::min(1_000_000, ($ENV{MAX_INDEX_FILES} \|\| 1_000_000));
23		# predict number of tests based on number of files in dictionaries and number of index types
24		my @dicts = qw( data/C020-words-txt/words-linux-fc1.txt data/C020-words-txt/words-osx-10_3.txt);
	20	use GetDictionaryWords;
	21	use Test::More qw(no_plan);
	22	my $max_words = MinMax::min(1_000_000, ($ENV{MAX_INDEX_FILES} \|\| 1_000_000));
	23	# predict number of tests based on number of files in dictionaries and number of index types
	24	my @dicts = qw( data/C020-words-txt/words-linux-fc1.txt data/C020-words-txt/words-osx-10_3.txt);
25	25	my @filetypes = qw(html xml txt);
26	26	my $numdicts = scalar(@dicts);
…	…
28	28	# three tests plus one for each word, for each dictionary and filetype.
29	29	#plan tests => ($numdicts * $numfiletypes * (3 + $max_words ));
30		mkpath( ["blib/index"], 0, 0755);
31		my $base = "C030";
32		for my $dict (@dicts) {
33		for my $filetype ( @filetypes ) {
34		( my $dictname = $dict ) =~ s/^.*-(([^.]\|-)+)\.txt$/$1/;
35		~~#( my $dictname = $dict ) =~ s/\.txt$//;~~ # this doesn't work
36		~~#$dictname =~ s![^/]*/!!;~~ # this doesn't work
	30	mkpath( ["blib/index"], 0, 0755);
	31	my $base = "C030";
	32	for my $dict (@dicts) {
	33	for my $filetype ( @filetypes ) {
	34	( my $dictname = $dict ) =~ s/^.*-(([^.]\|-)+)\.txt$/$1/;
	35	#( my $dictname = $dict ) =~ s/\.txt$//; # this doesn't work
	36	#$dictname =~ s![^/]*/!!; # this doesn't work
37	37
38		my $index = "blib/index/${base}_${dictname}_${filetype}.index";
39		my ($words, $word_count) = GetDictionaryWords::get_dictionary_words( $dict, 0, $max_words);
40		# this filename should come from somewhere factored
41		die "Couldn't get words from $dict" unless @$words;
42
43		# make a collection from dict, one word per document
44		my $cmd = "./make_collection --dict=$dict --norand --noenglishify " .
45		"--filetype=$filetype --min_words=1 --max_words=1 --num_files=" . scalar(@$words);
46		print STDERR "Using $cmd\n" if $ENV{TEST_VERBOSE};
	38	my $index = "blib/index/${base}_${dictname}_${filetype}.index";
	39	my ($words, $word_count) = GetDictionaryWords::get_dictionary_words( $dict, 0, $max_words);
	40	# this filename should come from somewhere factored
	41	die "Couldn't get words from $dict" unless @$words;
	42
	43	# make a collection from dict, one word per document
	44	my $cmd = "./make_collection --dict=$dict --norand --noenglishify " .
	45	"--filetype=$filetype --min_words=1 --max_words=1 --num_files=" . scalar(@$words);
	46	print STDERR "Using $cmd\n" if $ENV{TEST_VERBOSE};
47	47
48		my (%out) = BuildIndex::build_index_from_external_program( $cmd, $index);
	48	my (%out) = BuildIndex::build_index_from_external_program( $cmd, $index);
49	49
50		# first three tests: did the indexing seem to work?
51		~~cmp_ok( scalar(%out), '>', 0,~~ "Indexing output" );
52		cmp_ok( $out{files}, '==', scalar(@$words), 'files indexed' );
53		~~cmp_ok( $out{properties}, '==', 5,~~ 'num properties' );
	50	# first three tests: did the indexing seem to work?
	51	cmp_ok( scalar(%out), '>', 0, "Indexing output" );
	52	cmp_ok( $out{files}, '==', scalar(@$words), 'files indexed' );
	53	cmp_ok( $out{properties}, '==', 5, 'num properties' );
54	54
55	55	DoSearch::open_index($index);
56		~~for my $word (@$words) {~~ # then, one test for each word in the test
57		my @rows = DoSearch::do_search($index, "'$word'"); # quote the word
58		~~my ($num_expected_rows) = (~~ # look up the count unless it's AND, OR, or NOT
59		($word =~ /^\s(and\|or\|not\|near)\s$/i) ? 0 : ($word_count->{lc($word)} \|\| 1));
60		cmp_ok(scalar(@rows), "==", $num_expected_rows, "search '$word' ($filetype index from $dict)");
61		}
	56	for my $word (@$words) { # then, one test for each word in the test
	57	my @rows = DoSearch::do_search($index, "'$word'"); # quote the word
	58	my ($num_expected_rows) = ( # look up the count unless it's AND, OR, or NOT
	59	($word =~ /^\s(and\|or\|not\|near)\s$/i) ? 0 : ($word_count->{lc($word)} \|\| 1));
	60	cmp_ok(scalar(@rows), "==", $num_expected_rows, "search '$word' ($filetype index from $dict)");
	61	}
62	62	DoSearch::close_index($index);
63		$words = undef;
64		$word_count = undef;
65		}
66		}
	63	$words = undef;
	64	$word_count = undef;
	65	}
	66	}
67	67	};
68	68
…	…
70	70
71	71	BEGIN {
72		use File::Path qw(mkpath);
73		mkpath( ["blib/index"], 0, 0755);
74		my $base = "C030";
75		my (%out) = build_index(
76		"data/C030-medsm-xml", "blib/index/$base.index");
	72	use File::Path qw(mkpath);
	73	mkpath( ["blib/index"], 0, 0755);
	74	my $base = "C030";
	75	my (%out) = build_index(
	76	"data/C030-medsm-xml", "blib/index/$base.index");
77	77
78		cmp_ok( scalar(%out), '>', 2, "Indexing output" );
79		cmp_ok( $out{unique}, '==', 117468, 'unique words indexed' );
80		cmp_ok( $out{properties}, '==', 4, 'num properties' );
81		cmp_ok( $out{files}, '==', 1000, 'files indexed' );
82		cmp_ok( $out{bytes}, '==', 16626260, 'bytes indexed' );
83		cmp_ok( $out{words}, '==', 1513714, 'total words indexed' );
	78	cmp_ok( scalar(%out), '>', 2, "Indexing output" );
	79	cmp_ok( $out{unique}, '==', 117468, 'unique words indexed' );
	80	cmp_ok( $out{properties}, '==', 4, 'num properties' );
	81	cmp_ok( $out{files}, '==', 1000, 'files indexed' );
	82	cmp_ok( $out{bytes}, '==', 16626260, 'bytes indexed' );
	83	cmp_ok( $out{words}, '==', 1513714, 'total words indexed' );
84	84
85
86		my @rows = do_search(
87		"blib/index/$base.index", "swishe OR test");
88		cmp_ok(scalar(@rows), '==', 14, "num results from 'swishe OR test'")
	85
	86	my @rows = do_search(
	87	"blib/index/$base.index", "swishe OR test");
	88	cmp_ok(scalar(@rows), '==', 14, "num results from 'swishe OR test'")
89	89	};
90	90

Swishetest/trunk/t/050-C020-largeindex.t

r1968	r1970
13	13	BEGIN {
14	14	exit(0) unless $ENV{TEST_HUGE_INDEX};
15		use File::Path qw(mkpath);
16		mkpath( ["blib/index"], 0, 0755);
17		my $base = "T050-$$"; # test 050
	15	use File::Path qw(mkpath);
	16	mkpath( ["blib/index"], 0, 0755);
	17	my $base = "T050-$$"; # test 050
18	18	warn "base is $base\n";
19		my (%out) = BuildIndex::build_index_from_external_program(
	19	my (%out) = BuildIndex::build_index_from_external_program(
20	20	#"./make_collection -min_words=1000 -max_words=1000 -num_files=100", # this makes 920K of data, 2.33MB index, 476k propfile
21	21	#"./make_collection -min_words=10000 -max_words=10000 -num_files=1000", # this makes 38M index, 40MB prop
…	…
27	27	);
28	28
29		cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
30		cmp_ok( $out{unique}, '==', 252983, 'unique words indexed' );
31		cmp_ok( $out{properties}, '==', 5, 'num properties' );
32		cmp_ok( $out{files}, '==', 2, 'files indexed' );
33		cmp_ok( $out{bytes}, '==', 2896130, 'bytes indexed' );
34		cmp_ok( $out{words}, '==', 280381, 'total words indexed' );
	29	cmp_ok( scalar(keys(%out)), '>', 2, "Indexing output" );
	30	cmp_ok( $out{unique}, '==', 252983, 'unique words indexed' );
	31	cmp_ok( $out{properties}, '==', 5, 'num properties' );
	32	cmp_ok( $out{files}, '==', 2, 'files indexed' );
	33	cmp_ok( $out{bytes}, '==', 2896130, 'bytes indexed' );
	34	cmp_ok( $out{words}, '==', 280381, 'total words indexed' );
35	35
36	36	DoSearch::open_index( "blib/index/$base.index" );
37		my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
	37	my @rows = DoSearch::do_search( "blib/index/$base.index", "swishe OR test");
38	38	DoSearch::close_index( "blib/index/$base.index" );
39		cmp_ok(scalar(@rows), '==', 2, "num results from 'swishe OR test'")
	39	cmp_ok(scalar(@rows), '==', 2, "num results from 'swishe OR test'")
40	40	};
41	41

Download in other formats: