--- trunk/t/002_isis.t 2004/12/31 01:06:21 22 +++ trunk/t/2_isis.t 2006/07/09 21:36:33 61 @@ -3,28 +3,43 @@ use strict; use blib; -use Data::Dumper; +use Test::More tests => 136; +use File::Spec; -use Test::More tests => 104; +BEGIN { + use_ok( 'Biblio::Isis' ); + eval "use Data::Dump"; + + if (! $@) { + *Dumper = *Data::Dump::dump; + } else { + use Data::Dumper; + } +} -BEGIN { use_ok( 'IsisDB' ); } +my $debug = length( shift(@ARGV) || '' ); my $isis; +my $path_winisis = File::Spec->catfile('data', 'winisis', 'BIBL'); +my $path_isismarc = File::Spec->catfile('data', 'isismarc', 'BIBL'); + sub test_data { my $args = {@_}; - isa_ok ($isis, 'IsisDB'); + isa_ok ($isis, 'Biblio::Isis'); - cmp_ok($isis->{maxmfn}, '==', 5, "maxmfn set to 5"); + cmp_ok($isis->count, '==', 5, "count is 5"); # test .CNT data SKIP: { skip "no CNT file for this database", 5 unless $isis->{cnt_file}; - $isis->read_cnt; + ok(my $isis_cnt = $isis->read_cnt, "read_cnt"); + + cmp_ok(scalar keys %{$isis_cnt}, '==', 2, "returns 2 elements"); my $cnt = { '1' => { @@ -53,7 +68,7 @@ foreach my $c (keys %{$cnt}) { foreach my $kn (keys %{$cnt->{$c}}) { - cmp_ok($isis->{cnt}->{$c}->{$kn}, '==', $cnt->{$c}->{$kn}, "cnt $c $kn same"); + cmp_ok($isis_cnt->{$c}->{$kn}, '==', $cnt->{$c}->{$kn}, "cnt $c $kn same"); } } } @@ -86,22 +101,30 @@ '10' => [ '^a0-340-59691-0' ], }, { # identifier test - '225' => [ '1#^aMcGraw-Hill series in Psychology' ], '200' => [ '1#^aPsychology^fCamille B. Wortman, Elizabeth F. Loftus, Mary E. Marshal' ], + 225 => ["1#^aMcGraw-Hill series in Psychology"], + 205 => ["^a4th ed"], } ]; - foreach my $mfn (1 .. $isis->{'maxmfn'}) { + foreach my $mfn (1 .. $isis->count) { + my $rec; ok($rec = $isis->fetch($mfn), "fetch $mfn"); + diag "<<<<< rec = ",Dumper( $rec ), "\n>>>>> data = ", Dumper( $data->[$mfn-1] ) if ($debug); + foreach my $f (keys %{$data->[$mfn-1]}) { my $i = 0; foreach my $v (@{$data->[$mfn-1]->{$f}}) { $v =~ s/^[01# ][01# ]// if ($args->{no_ident}); - cmp_ok($rec->{$f}->[$i], '==', $v, "MFN $mfn $f:$i $v"); + diag "compare '", $rec->{$f}->[$i], "' eq '$v'" if ($debug); + cmp_ok($rec->{$f}->[$i], 'eq', $v, "MFN $mfn field: $f offset: $i"); $i++; } } + + cmp_ok($isis->mfn, '==', $mfn, 'mfn'); + } # test to_ascii @@ -111,33 +134,36 @@ skip "no Digest::MD5 module", 5 if ($@); - foreach my $mfn (1 .. $isis->{'maxmfn'}) { + foreach my $mfn (1 .. $isis->count) { my $md5 = md5_hex($isis->to_ascii($mfn)); - cmp_ok($args->{md5_ascii}[$mfn - 1], 'eq', $md5, "md5 $mfn"); + cmp_ok($md5, 'eq', $args->{md5_ascii}[$mfn - 1], "md5 $mfn"); } } } -$isis = IsisDB->new ( - isisdb => './data/winisis/BIBL', +$isis = Biblio::Isis->new ( + isisdb => $path_winisis, + include_deleted => 1, + debug => $debug > 1 ? ($debug - 1) : 0, ); -print Dumper($isis); +diag "new Biblio::Isis = ", Dumper($isis) if ($debug); test_data( no_ident => 1, md5_ascii => [ qw( a369eff702307ba12eb81656ee0587fe 4fb38537a94f3f5954e40d9536b942b0 - 498cc16c9e7ab0fdc29182533cc35d11 + 579a7c6901c654bdeac10547a98e5b71 7d2adf1675c83283aa9b82bf343e3d85 - daf2cf86ca7e188e8360a185f3b43423 + 4cc1f798bbcf36862f7aa78c3410801a ) ], ); -$isis = IsisDB->new ( - isisdb => './data/isismarc/BIBL', +$isis = Biblio::Isis->new ( + isisdb => $path_isismarc, + include_deleted => 1, ); test_data( @@ -149,3 +175,155 @@ 843b9ebccf16a498fba623c78f21b6c0 ) ], ); + +# check logically deleted + +$isis = Biblio::Isis->new ( + isisdb => $path_winisis, + include_deleted => 1, +); + +ok($isis->fetch(3), "deleted found"); +cmp_ok($isis->{deleted}, '==', 3, "MFN 3 is deleted"); +ok($isis->{record}, "record exists"); + +diag "record = ",Dumper($isis->{record}) if ($debug); + +$isis = Biblio::Isis->new ( + isisdb => $path_winisis, + debug => $debug, +); + +ok(! $isis->fetch(3), "deleted not found"); +cmp_ok($isis->{deleted}, '==', 3, "MFN 3 is deleted"); +ok(! $isis->{record}, 'no record'); + +$isis->{record} = { + 900 => [ '^a900a^b900b^c900c' ], + 901 => [ + '^a901a-1^b901b-1^c901c-1', + '^a901a-2^b901b-2', + '^a901a-3', + ], + 902 => [ + '^aa1^aa2^aa3^bb1^aa4^bb2^cc1^aa5', + ], +}; +$isis->{current_mfn} = 42; + +ok(my $hash = $isis->to_hash( $isis->mfn ), 'to_hash'); +diag "to_hash = ",Dumper( $hash ) if ($debug); +is_deeply( $hash, { + "000" => [42], + 900 => [{ a => "900a", b => "900b", c => "900c" }], + 901 => [ + { a => "901a-1", b => "901b-1", c => "901c-1" }, + { a => "901a-2", b => "901b-2" }, + { a => "901a-3" }, + ], + 902 => [ + { a => ["a1", "a2", "a3", "a4", "a5"], b => ["b1", "b2"], c => "c1" }, + ], +}, 'hash is_deeply'); + +ok(my $ascii = $isis->to_ascii( $isis->mfn ), 'to_ascii'); +diag "to_ascii = \n", $ascii if ($debug); +cmp_ok($ascii, 'eq', <<'__END_OF_ASCII__', 'to_ascii output'); +0 42 +900 ^a900a^b900b^c900c +901 ^a901a-1^b901b-1^c901c-1 +901 ^a901a-2^b901b-2 +901 ^a901a-3 +902 ^aa1^aa2^aa3^bb1^aa4^bb2^cc1^aa5 +__END_OF_ASCII__ + +ok(my $hash2 = $isis->to_hash({ mfn => $isis->mfn }), 'to_hash(mfn)'); +is_deeply( $hash2, $hash, 'same hash' ); + +ok($hash = $isis->to_hash({ mfn => $isis->mfn, include_subfields => 1 }), 'to_hash(mfn,include_subfields)'); +diag "to_hash = ",Dumper( $hash ) if ($debug); +is_deeply( $hash, { + "000" => [42], + 900 => [ + { a => "900a", b => "900b", c => "900c", subfields => ["a", 0, "b", 0, "c", 0] }, + ], + 901 => [ + { a => "901a-1", b => "901b-1", c => "901c-1", subfields => ["a", 0, "b", 0, "c", 0] }, + { a => "901a-2", b => "901b-2", subfields => ["a", 0, "b", 0] }, + { a => "901a-3", subfields => ["a", 0] }, + ], + 902 => [ + { a => ["a1", "a2", "a3", "a4", "a5"], b => ["b1", "b2"], c => "c1", + subfields => ["a", 0, "a", 1, "a", 2, "b", 0, "a", 3, "b", 1, "c", 0, "a", 4], + }, + ], +}, 'hash is_deeply'); + +ok($hash = $isis->to_hash({ mfn => $isis->mfn, join_subfields_with => ' ; ' }), 'to_hash(mfn,join_subfields_with)'); +diag "to_hash = ",Dumper( $hash ) if ($debug); +is_deeply( $hash, { + "000" => [42], + 900 => [{ a => "900a", b => "900b", c => "900c" }], + 901 => [ + { a => "901a-1", b => "901b-1", c => "901c-1" }, + { a => "901a-2", b => "901b-2" }, + { a => "901a-3" }, + ], + 902 => [{ a => "a1 ; a2 ; a3 ; a4 ; a5", b => "b1 ; b2", c => "c1" }], +}, 'hash is_deeply'); + +# split_on_delimiters + +$isis->{record} = { + 900 => [ '^a900a : 900b ; 900c' ], + 901 => [ + '^a901a-1 x 901b-1 y 901c-1', + '^b901b-2 x 901b-2', + '^c901c-3 y 901c-3', + ], +}; + +my $split_on_delimiters = { + 900 => { a => { ' : ' => 'b', ' ; ' => 'c' } }, + 901 => { + a => { ' x ' => 'b', ' y ' => 'c' }, + b => { ' x ' => 'b' }, + c => { ' y ' => 'c' } + }, +}; + +diag "split_on_delimiters = ", Dumper($split_on_delimiters); + +my $regexpes; + +foreach my $f (keys %$split_on_delimiters) { +warn "--- f: $f\n"; + foreach my $sf (keys %{ $split_on_delimiters->{$f} }) { +warn "---- sf: $sf\n"; + foreach my $patt (keys %{ $split_on_delimiters->{$f}->{$sf} }) { + my $new_sf = $split_on_delimiters->{$f}->{$sf}->{$patt} || die "no field?"; +warn "----- patt: $patt -> $new_sf\n"; + + # fool-proof spaces in pattern + $patt =~ s/\s+/\\s\+/g; + + push @{ $regexpes->{$f} }, + 's/\^' . $sf . '([^\^]+)' . $patt . '([^\^]+)/^' . $sf . '$1^' . $new_sf . '$2/g'; + } + } +} + +diag "generated regexpes = ", Dumper( $regexpes ); + + +ok($hash = $isis->to_hash({ mfn => $isis->mfn, regexps => $regexpes, }), 'to_hash(mfn,regexpes)'); +diag "to_hash = ",Dumper( $hash ) if ($debug); +is_deeply( $hash, { + "000" => [42], + 900 => [{ a => "900a", b => "900b", c => "900c" }], + 901 => [ + { a => "901a-1", b => "901b-1", c => "901c-1" }, + { b => ["901b-2", "901b-2"] }, + { c => ["901c-3", "901c-3"] }, + ], +}, 'hash is_deeply');