| 1 |
dpavlin |
536 |
#!/usr/bin/perl -w |
| 2 |
|
|
|
| 3 |
|
|
use strict; |
| 4 |
|
|
|
| 5 |
dpavlin |
791 |
use Test::More tests => 322; |
| 6 |
dpavlin |
536 |
use Test::Exception; |
| 7 |
|
|
use Cwd qw/abs_path/; |
| 8 |
|
|
use blib; |
| 9 |
|
|
use File::Slurp; |
| 10 |
dpavlin |
555 |
use Getopt::Long; |
| 11 |
dpavlin |
536 |
|
| 12 |
|
|
BEGIN { |
| 13 |
|
|
use_ok( 'WebPAC::Normalize' ); |
| 14 |
|
|
} |
| 15 |
|
|
|
| 16 |
dpavlin |
555 |
use Data::Dump qw/dump/; |
| 17 |
|
|
|
| 18 |
|
|
my $debug = 0; |
| 19 |
|
|
GetOptions( |
| 20 |
|
|
"debug+", \$debug |
| 21 |
|
|
); |
| 22 |
|
|
|
| 23 |
|
|
cmp_ok(_debug(1), '==', 1, '_debug level'); |
| 24 |
|
|
cmp_ok(_debug(0), '==', 0, '_debug level'); |
| 25 |
|
|
|
| 26 |
|
|
diag "debug level for $0 is $debug" if ($debug); |
| 27 |
|
|
if ($debug > 2) { |
| 28 |
|
|
diag "debug level for WebPAC::Normalize is ", _debug( $debug - 2 ); |
| 29 |
|
|
} |
| 30 |
|
|
|
| 31 |
dpavlin |
536 |
ok(my $abs_path = abs_path($0), "abs_path"); |
| 32 |
|
|
$abs_path =~ s#/[^/]*$#/#; |
| 33 |
|
|
diag "abs_path: $abs_path" if ($debug); |
| 34 |
|
|
|
| 35 |
|
|
my $rec1 = { |
| 36 |
|
|
'200' => [{ |
| 37 |
|
|
'a' => '200a', |
| 38 |
|
|
'b' => '200b', |
| 39 |
|
|
},{ |
| 40 |
|
|
'c' => '200c', |
| 41 |
|
|
'd' => '200d', |
| 42 |
|
|
},{ |
| 43 |
|
|
'a' => '200a*2', |
| 44 |
|
|
'd' => '200d*2', |
| 45 |
|
|
}], |
| 46 |
|
|
'201' => [{ |
| 47 |
|
|
'x' => '201x', |
| 48 |
|
|
'y' => '201y', |
| 49 |
|
|
}], |
| 50 |
|
|
'900' => [ |
| 51 |
|
|
'900-no_subfield' |
| 52 |
|
|
], |
| 53 |
|
|
'901' => [{ |
| 54 |
|
|
'a' => '900a', |
| 55 |
|
|
}], |
| 56 |
|
|
'902' => [{ |
| 57 |
|
|
'z' => '900', |
| 58 |
|
|
}], |
| 59 |
|
|
}; |
| 60 |
|
|
|
| 61 |
|
|
my $rec2 = { |
| 62 |
|
|
'675' => [ { |
| 63 |
|
|
'a' => '159.9' |
| 64 |
|
|
} ], |
| 65 |
|
|
'210' => [ { |
| 66 |
|
|
'c' => 'New York University press', |
| 67 |
|
|
'a' => 'New York', |
| 68 |
|
|
'd' => 'cop. 1988' |
| 69 |
|
|
} ], |
| 70 |
|
|
'700' => [ { |
| 71 |
|
|
'a' => 'Haynal', |
| 72 |
|
|
'b' => 'André' |
| 73 |
|
|
} ], |
| 74 |
|
|
'801' => [ 'FFZG' ], |
| 75 |
|
|
'991' => [ '8302' ], |
| 76 |
|
|
'000' => [ 1 ], |
| 77 |
|
|
'702' => [ { |
| 78 |
|
|
'a' => 'Holder', |
| 79 |
|
|
'b' => 'Elizabeth' |
| 80 |
|
|
} ], |
| 81 |
|
|
'215' => [ { |
| 82 |
|
|
'c' => 'ilustr', |
| 83 |
|
|
'a' => 'xix, 202 str', |
| 84 |
|
|
'd' => '23cm' |
| 85 |
|
|
} ], |
| 86 |
|
|
'990' => [ |
| 87 |
|
|
'2140', |
| 88 |
|
|
'88', |
| 89 |
|
|
'HAY' |
| 90 |
|
|
], |
| 91 |
|
|
'200' => [ { |
| 92 |
|
|
'e' => 'from Freud and Ferenczi to Michael balint', |
| 93 |
|
|
'a' => 'Controversies in psychoanalytic method', |
| 94 |
|
|
'g' => 'translated by Elizabeth Holder on the basisi of a first draft by Archie Hooton ; with a preface by Daniel N. Stern', |
| 95 |
|
|
'f' => 'by André E. Haynal' |
| 96 |
|
|
} ], |
| 97 |
|
|
'610' => [ 'povijest psihoanalize' ], |
| 98 |
|
|
'994' => [ { |
| 99 |
|
|
'c' => '', |
| 100 |
|
|
'a' => 'PS', |
| 101 |
|
|
'b' => 'MG' |
| 102 |
|
|
} ], |
| 103 |
|
|
'320' => [ 'Kazalo' ], |
| 104 |
|
|
'101' => [ 'ENG' ], |
| 105 |
|
|
'686' => [ '2140' ], |
| 106 |
|
|
'300' => [ 'Prijevod djela: ' ], |
| 107 |
|
|
}; |
| 108 |
|
|
|
| 109 |
|
|
|
| 110 |
dpavlin |
725 |
my $lookup_hash1 = { |
| 111 |
|
|
'db1' => { |
| 112 |
|
|
'input1' => { |
| 113 |
|
|
'key1' => { 1 => 1 }, |
| 114 |
|
|
'key2' => { 2 => 1 }, |
| 115 |
|
|
}, |
| 116 |
|
|
'input2' => { |
| 117 |
|
|
'key3' => { 3 => 1 }, |
| 118 |
|
|
'key4' => { 4 => 1 }, |
| 119 |
|
|
}, |
| 120 |
|
|
}, |
| 121 |
|
|
'db2' => { |
| 122 |
|
|
'input3' => { |
| 123 |
|
|
'key5' => { 5 => 1 }, |
| 124 |
|
|
'key6' => { 6 => 1 }, |
| 125 |
|
|
}, |
| 126 |
|
|
} |
| 127 |
dpavlin |
536 |
}; |
| 128 |
|
|
|
| 129 |
dpavlin |
725 |
my $lookup_hash2 = { |
| 130 |
|
|
'db3' => { |
| 131 |
|
|
'input4' => { |
| 132 |
|
|
'key7' => { 7 => 1 }, |
| 133 |
|
|
'key8' => { 8 => 1 }, |
| 134 |
|
|
}, |
| 135 |
|
|
} |
| 136 |
dpavlin |
536 |
}; |
| 137 |
|
|
|
| 138 |
|
|
sub test { |
| 139 |
dpavlin |
550 |
print dump( @_ ), ("-" x 78), "\n"; |
| 140 |
dpavlin |
536 |
ok( defined(@_) ); |
| 141 |
|
|
} |
| 142 |
|
|
|
| 143 |
|
|
# how much of string evaled to display? |
| 144 |
|
|
my $max_eval_output = 170; |
| 145 |
|
|
|
| 146 |
|
|
sub dump_error { |
| 147 |
|
|
my ($msg,$code) = @_; |
| 148 |
|
|
|
| 149 |
|
|
my @l = split(/[\n\r]/, $code); |
| 150 |
|
|
my $out = "$msg\n"; |
| 151 |
|
|
|
| 152 |
|
|
foreach my $i ( 0 .. $#l ) { |
| 153 |
|
|
$out .= sprintf("%2d: %s\n", $i, $l[$i]); |
| 154 |
|
|
} |
| 155 |
|
|
|
| 156 |
|
|
return $out; |
| 157 |
|
|
} |
| 158 |
|
|
|
| 159 |
|
|
sub test_s { |
| 160 |
|
|
my $t = shift || die; |
| 161 |
|
|
|
| 162 |
|
|
my $eval_t = $t; |
| 163 |
|
|
$eval_t =~ s/[\n\r\s]+/ /gs; |
| 164 |
|
|
$eval_t = substr($eval_t,0,$max_eval_output) . '...' if (length($eval_t) > $max_eval_output); |
| 165 |
dpavlin |
562 |
$eval_t =~ s/\\/\\\\/gs; |
| 166 |
dpavlin |
536 |
|
| 167 |
dpavlin |
571 |
my @__ret; |
| 168 |
|
|
eval "\@__ret = $t"; |
| 169 |
|
|
ok(! $@, $@ ? dump_error($@, $t) : "eval: $eval_t = " . dump(@__ret)); |
| 170 |
|
|
return \@__ret; |
| 171 |
dpavlin |
536 |
} |
| 172 |
|
|
|
| 173 |
|
|
{ |
| 174 |
|
|
no strict 'subs'; |
| 175 |
|
|
use WebPAC::Normalize; |
| 176 |
|
|
|
| 177 |
dpavlin |
538 |
ok(! _set_lookup( undef ), "set_lookup(undef)"); |
| 178 |
dpavlin |
536 |
|
| 179 |
dpavlin |
538 |
_set_rec( $rec1 ); |
| 180 |
dpavlin |
536 |
|
| 181 |
|
|
cmp_ok( join(",", rec2('200','a') ), 'eq', '200a,200a*2', 'join rec2' ); |
| 182 |
|
|
cmp_ok( join(",", rec2('200','a'), rec2('200','b') ), 'eq', '200a,200a*2,200b', 'join rec2 rec2' ); |
| 183 |
|
|
cmp_ok( join(" * ", sort(rec1('200'), rec1('201') )), 'eq', '200a * 200a*2 * 200b * 200c * 200d * 200d*2 * 201x * 201y', 'join sort rec1 rec1' ); |
| 184 |
|
|
diag "is_deeply checks\n"; |
| 185 |
|
|
is_deeply( \[ rec1('200') ], \[ qw/200a 200b 200c 200d 200a*2 200d*2/ ] ); |
| 186 |
|
|
is_deeply( \[ regex( 's/0/o/g', rec1('200') ) ], \[ qw/2ooa 2oob 2ooc 2ood 2ooa*2 2ood*2/ ]); |
| 187 |
|
|
is_deeply( \[ grep { /\*/ } regex( 's/0/o/g', rec1('200') ) ], \[ qw/2ooa*2 2ood*2/ ]); |
| 188 |
|
|
is_deeply( \[ rec('902') ], \[ '900' ] ); |
| 189 |
|
|
|
| 190 |
|
|
cmp_ok( rec('902'), 'eq', rec('902','z'), 'rec sf' ); |
| 191 |
|
|
|
| 192 |
|
|
# simple list manipulatons |
| 193 |
|
|
cmp_ok( join('', prefix('ab', 'cd') ), 'eq', 'abcd', 'prefix'); |
| 194 |
|
|
cmp_ok( join('', suffix('xy', 'cd') ), 'eq', 'cdxy', 'suffix'); |
| 195 |
|
|
cmp_ok( join('', surround('->', '<-', 'a','b','c') ), 'eq', '->a<-->b<-->c<-', 'surround'); |
| 196 |
|
|
|
| 197 |
dpavlin |
791 |
# count |
| 198 |
|
|
my @el; |
| 199 |
|
|
for my $i ( 0 .. 10 ) { |
| 200 |
|
|
cmp_ok( count( @el ), '==', $i, "count($i)"); |
| 201 |
|
|
push @el, "element $i"; |
| 202 |
|
|
} |
| 203 |
|
|
|
| 204 |
dpavlin |
725 |
# lookups |
| 205 |
dpavlin |
536 |
|
| 206 |
dpavlin |
736 |
throws_ok { _set_load_row() } qr/CODE/, 'empty _set_load_row()'; |
| 207 |
dpavlin |
725 |
|
| 208 |
dpavlin |
736 |
ok(_set_load_row(sub { |
| 209 |
dpavlin |
725 |
my ($database,$input,$mfn) = @_; |
| 210 |
dpavlin |
736 |
diag "load_row( $database, $input, $mfn )"; |
| 211 |
dpavlin |
725 |
cmp_ok( $#_, '==', 2, 'have 3 arguments'); |
| 212 |
dpavlin |
736 |
ok($database, '_load_row database'); |
| 213 |
|
|
ok($input, '_load_row input'); |
| 214 |
|
|
ok($mfn, '_load_row mfn'); |
| 215 |
dpavlin |
725 |
return { |
| 216 |
|
|
'900' => [{ x => '900x-' . $mfn , y => '900y-' . $mfn }], |
| 217 |
|
|
} |
| 218 |
|
|
|
| 219 |
dpavlin |
736 |
}), '_set_load_row'); |
| 220 |
dpavlin |
725 |
|
| 221 |
|
|
my @v = qw/foo bar baz aaa bbb ccc ddd/; |
| 222 |
|
|
|
| 223 |
|
|
my @accumulated; |
| 224 |
|
|
|
| 225 |
|
|
for my $i ( 0 .. $#v ) { |
| 226 |
|
|
|
| 227 |
|
|
my $mfn = 1000 + $i; |
| 228 |
|
|
|
| 229 |
|
|
ok(WebPAC::Normalize::_set_config({ '_mfn' => $mfn }), "_set_config _mfn=$mfn"); |
| 230 |
|
|
|
| 231 |
|
|
my $size = $#v + 1; |
| 232 |
|
|
|
| 233 |
|
|
cmp_ok( |
| 234 |
|
|
save_into_lookup('db','input','key', sub { @v }), |
| 235 |
|
|
'==', $size, "save_into_lookup $size values" |
| 236 |
|
|
); |
| 237 |
|
|
|
| 238 |
|
|
ok(my $l = WebPAC::Normalize::_get_lookup(), '_get_lookup'); |
| 239 |
|
|
diag "_get_lookup = ", dump($l); |
| 240 |
|
|
|
| 241 |
|
|
my @lookup; |
| 242 |
|
|
|
| 243 |
dpavlin |
736 |
ok(@lookup = lookup( |
| 244 |
dpavlin |
725 |
sub { |
| 245 |
|
|
diag "in show"; |
| 246 |
|
|
rec('900','x'); |
| 247 |
|
|
}, |
| 248 |
|
|
'db','input','key', |
| 249 |
|
|
sub { |
| 250 |
|
|
return @v; |
| 251 |
|
|
} |
| 252 |
|
|
), |
| 253 |
|
|
"lookup db/input/key"); |
| 254 |
|
|
|
| 255 |
|
|
push @accumulated, '900x-' . $mfn; |
| 256 |
|
|
|
| 257 |
|
|
is_deeply(\@lookup, \@accumulated, "lookup db/input/key"); |
| 258 |
|
|
|
| 259 |
|
|
shift @v; |
| 260 |
|
|
|
| 261 |
|
|
} |
| 262 |
|
|
|
| 263 |
|
|
ok(my $l = WebPAC::Normalize::_get_lookup(), '_get_lookup'); |
| 264 |
|
|
diag "_get_lookup = ", dump($l); |
| 265 |
|
|
|
| 266 |
|
|
is_deeply( $l, { |
| 267 |
|
|
db => { |
| 268 |
|
|
input => { |
| 269 |
|
|
key => { |
| 270 |
|
|
foo => { 1000 => 1 }, |
| 271 |
|
|
bar => { 1000 => 1, 1001 => 1 }, |
| 272 |
|
|
baz => { 1000 => 1, 1001 => 1, 1002 => 1 }, |
| 273 |
|
|
aaa => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1 }, |
| 274 |
|
|
bbb => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1 }, |
| 275 |
|
|
ccc => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1, 1005 => 1 }, |
| 276 |
|
|
ddd => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1, 1005 => 1, 1006 => 1 }, |
| 277 |
|
|
}, |
| 278 |
|
|
}, |
| 279 |
|
|
}, |
| 280 |
|
|
}, 'lookup data'); |
| 281 |
|
|
|
| 282 |
|
|
####### |
| 283 |
|
|
|
| 284 |
|
|
diag "lookup_hash1 = ", dump($lookup_hash1); |
| 285 |
|
|
ok(_set_lookup( $lookup_hash1 ), '_set_lookup $lookup_hash1'); |
| 286 |
|
|
|
| 287 |
dpavlin |
736 |
throws_ok { _set_load_row() } qr/CODE/, 'empty _set_load_row()'; |
| 288 |
dpavlin |
725 |
|
| 289 |
dpavlin |
736 |
ok(_set_load_row(sub { |
| 290 |
dpavlin |
725 |
my ($database,$input,$mfn) = @_; |
| 291 |
dpavlin |
736 |
diag "load_row( $database, $input, $mfn )"; |
| 292 |
dpavlin |
725 |
cmp_ok( $#_, '==', 2, 'have 3 arguments'); |
| 293 |
|
|
ok($database, 'database'); |
| 294 |
|
|
ok($input, 'input'); |
| 295 |
|
|
ok($mfn, 'mfn'); |
| 296 |
|
|
|
| 297 |
dpavlin |
736 |
}), '_set_load_row'); |
| 298 |
dpavlin |
725 |
|
| 299 |
|
|
|
| 300 |
|
|
# cmp_ok(lookup( |
| 301 |
|
|
# sub { |
| 302 |
|
|
# 'found' |
| 303 |
|
|
# }, |
| 304 |
|
|
# 'db1','input1','key1', |
| 305 |
|
|
# sub { |
| 306 |
|
|
# rec('200','a') |
| 307 |
|
|
# } |
| 308 |
|
|
# ), 'eq', 'found', 'lookup db1/input1/key1'); |
| 309 |
|
|
|
| 310 |
|
|
|
| 311 |
dpavlin |
536 |
|
| 312 |
dpavlin |
725 |
# cmp_ok( |
| 313 |
|
|
# lookup( |
| 314 |
|
|
# ), |
| 315 |
|
|
# 'eq', 'lookup 1 i lookup 2', 'join lookup regex rec2'); |
| 316 |
dpavlin |
536 |
|
| 317 |
|
|
# check join_with operations |
| 318 |
|
|
|
| 319 |
|
|
sub test_join_with_2 { |
| 320 |
|
|
my ($a,$b,$e) = @_; |
| 321 |
|
|
|
| 322 |
|
|
cmp_ok( |
| 323 |
|
|
join_with(" <1> ", |
| 324 |
|
|
rec('201',$a), |
| 325 |
|
|
rec('201',$b), |
| 326 |
|
|
), |
| 327 |
|
|
'eq', $e, "join_with $a <1> $b = $e"); |
| 328 |
|
|
} |
| 329 |
|
|
|
| 330 |
|
|
test_join_with_2('_','_',''); |
| 331 |
|
|
test_join_with_2('x','_','201x'); |
| 332 |
|
|
test_join_with_2('_','x','201x'); |
| 333 |
|
|
test_join_with_2('x','y','201x <1> 201y'); |
| 334 |
|
|
|
| 335 |
|
|
sub test_join_with_3 { |
| 336 |
|
|
my ($a,$b,$c,$e) = @_; |
| 337 |
|
|
|
| 338 |
|
|
cmp_ok( |
| 339 |
|
|
join_with(" <1> ", rec('201',$a), |
| 340 |
|
|
join_with(" <2> ", rec('201',$b), |
| 341 |
|
|
rec('201',$c), |
| 342 |
|
|
) |
| 343 |
|
|
), |
| 344 |
|
|
'eq', $e, "join_with $a <1> $b <2> $c = $e"); |
| 345 |
|
|
}; |
| 346 |
|
|
|
| 347 |
|
|
test_join_with_3('_','_','_',''); |
| 348 |
|
|
test_join_with_3('x','_','_','201x'); |
| 349 |
|
|
test_join_with_3('_','x','_','201x'); |
| 350 |
|
|
test_join_with_3('_','_','x','201x'); |
| 351 |
|
|
test_join_with_3('x','y','_','201x <1> 201y'); |
| 352 |
|
|
test_join_with_3('x','_','y','201x <1> 201y'); |
| 353 |
|
|
test_join_with_3('_','x','y','201x <2> 201y'); |
| 354 |
|
|
test_join_with_3('x','_','y','201x <1> 201y'); |
| 355 |
|
|
test_join_with_3('x','y','x','201x <1> 201y <2> 201x'); |
| 356 |
|
|
|
| 357 |
|
|
# test lookups |
| 358 |
|
|
|
| 359 |
dpavlin |
725 |
_set_lookup( $lookup_hash2 ); |
| 360 |
dpavlin |
536 |
|
| 361 |
dpavlin |
725 |
throws_ok { lookup() } qr/need/, 'empty lookup'; |
| 362 |
dpavlin |
536 |
|
| 363 |
dpavlin |
725 |
#is_deeply( \[ lookup( prefix( '00', rec('902') ) ) ], \[ 'lookup' ], 'lookup prefix' ); |
| 364 |
dpavlin |
536 |
|
| 365 |
dpavlin |
725 |
#ok(! lookup('non-existent'), 'lookup non-existant' ); |
| 366 |
|
|
|
| 367 |
dpavlin |
538 |
_set_rec( $rec2 ); |
| 368 |
dpavlin |
536 |
|
| 369 |
|
|
test_s(qq{ |
| 370 |
|
|
tag('Title', |
| 371 |
|
|
rec('200','a') |
| 372 |
|
|
); |
| 373 |
|
|
}); |
| 374 |
|
|
test_s(qq{ |
| 375 |
|
|
tag('Who', |
| 376 |
|
|
join_with(" ", |
| 377 |
|
|
rec('702','a'), |
| 378 |
|
|
rec('702','b') |
| 379 |
|
|
) |
| 380 |
|
|
); |
| 381 |
|
|
}); |
| 382 |
|
|
|
| 383 |
|
|
test_s(qq{ |
| 384 |
|
|
display('Publisher', |
| 385 |
|
|
rec('210','c') |
| 386 |
|
|
) |
| 387 |
|
|
}); |
| 388 |
|
|
|
| 389 |
|
|
test_s(qq{ |
| 390 |
|
|
search('Year', |
| 391 |
|
|
regex( 's/[^\\d]+//', |
| 392 |
|
|
rec('210','d') |
| 393 |
|
|
) |
| 394 |
|
|
) |
| 395 |
|
|
}); |
| 396 |
|
|
|
| 397 |
dpavlin |
538 |
ok(my $ds = _get_ds(), "get_ds"); |
| 398 |
dpavlin |
550 |
diag "ds = ", dump($ds) if ($debug); |
| 399 |
dpavlin |
536 |
|
| 400 |
|
|
|
| 401 |
|
|
sub test_check_ds { |
| 402 |
|
|
|
| 403 |
|
|
my $t = shift; |
| 404 |
|
|
|
| 405 |
dpavlin |
538 |
ok($ds = _get_ds(), 'get_ds'); |
| 406 |
dpavlin |
550 |
diag dump( $ds ) if ($debug); |
| 407 |
dpavlin |
536 |
|
| 408 |
|
|
ok( $ds && $ds->{something}, 'get_ds->something exists' ); |
| 409 |
|
|
ok( $ds && $ds->{something}->{$t}, 'get_ds->something->'.$t.' exists') if ($t); |
| 410 |
|
|
ok( $ds && !$ds->{empty}, 'get_ds->empty doesn\'t' ); |
| 411 |
|
|
|
| 412 |
|
|
return $ds; |
| 413 |
|
|
} |
| 414 |
|
|
|
| 415 |
dpavlin |
538 |
_clean_ds(); |
| 416 |
dpavlin |
536 |
test_s(qq{ search('something', '42'); }); |
| 417 |
|
|
test_s(qq{ search('empty', ''); }); |
| 418 |
|
|
test_check_ds('search'); |
| 419 |
|
|
|
| 420 |
dpavlin |
538 |
_clean_ds(); |
| 421 |
dpavlin |
536 |
test_s(qq{ display('something', '42'); }); |
| 422 |
|
|
test_s(qq{ display('empty', ''); }); |
| 423 |
|
|
test_check_ds('display'); |
| 424 |
|
|
|
| 425 |
dpavlin |
538 |
_clean_ds(); |
| 426 |
dpavlin |
536 |
test_s(qq{ tag('something', '42'); }); |
| 427 |
|
|
test_s(qq{ tag('empty', ''); }); |
| 428 |
|
|
test_check_ds('search'); |
| 429 |
|
|
test_check_ds('display'); |
| 430 |
|
|
|
| 431 |
dpavlin |
538 |
_clean_ds(); |
| 432 |
dpavlin |
536 |
my $n = read_file( "$abs_path/data/normalize.pl" ); |
| 433 |
|
|
$n .= "\n1;\n"; |
| 434 |
|
|
#diag "normalize code:\n$n\n"; |
| 435 |
|
|
test_s( $n ); |
| 436 |
|
|
|
| 437 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
| 438 |
dpavlin |
550 |
diag "ds = ", dump($ds) if ($debug); |
| 439 |
dpavlin |
536 |
|
| 440 |
|
|
my $rec = { |
| 441 |
|
|
'200' => [{ |
| 442 |
|
|
'a' => '200a', |
| 443 |
|
|
'b' => '200b', |
| 444 |
|
|
}], |
| 445 |
|
|
}; |
| 446 |
|
|
my $rules = qq{ search('mixed', rec('200') ) }; |
| 447 |
|
|
|
| 448 |
dpavlin |
538 |
_clean_ds(); |
| 449 |
|
|
_set_rec( $rec ); |
| 450 |
dpavlin |
536 |
test_s( $rules ); |
| 451 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
| 452 |
dpavlin |
536 |
is_deeply( $ds, { |
| 453 |
|
|
'mixed' => { |
| 454 |
|
|
'search' => [ '200a', '200b' ], |
| 455 |
|
|
'tag' => 'mixed' |
| 456 |
|
|
} |
| 457 |
|
|
}, 'correct get_ds'); |
| 458 |
|
|
|
| 459 |
|
|
ok(my $ds2 = WebPAC::Normalize::data_structure( |
| 460 |
|
|
row => $rec, |
| 461 |
|
|
rules => $rules, |
| 462 |
|
|
), 'data_structure'); |
| 463 |
|
|
is_deeply( $ds, $ds2, 'data_structure(s) same'); |
| 464 |
|
|
|
| 465 |
|
|
# wird and non-valid structure which is supported anyway |
| 466 |
dpavlin |
538 |
_clean_ds(); |
| 467 |
|
|
_set_rec({ |
| 468 |
dpavlin |
536 |
'200' => [{ |
| 469 |
|
|
'a' => '200a', |
| 470 |
|
|
}, |
| 471 |
|
|
'200-solo' |
| 472 |
|
|
] |
| 473 |
|
|
}); |
| 474 |
|
|
test_s(qq{ search('mixed', rec('200') ) }); |
| 475 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
| 476 |
dpavlin |
536 |
is_deeply( $ds, { |
| 477 |
|
|
'mixed' => { |
| 478 |
|
|
'search' => [ '200a', '200-solo' ], |
| 479 |
|
|
'tag' => 'mixed' |
| 480 |
|
|
} |
| 481 |
|
|
}, 'correct get_ds'); |
| 482 |
|
|
|
| 483 |
dpavlin |
555 |
# |
| 484 |
dpavlin |
540 |
# MARC |
| 485 |
dpavlin |
555 |
# |
| 486 |
dpavlin |
725 |
#_debug( 4 ); |
| 487 |
dpavlin |
555 |
|
| 488 |
dpavlin |
547 |
test_s(qq{ marc_indicators('900',1,2) }); |
| 489 |
|
|
test_s(qq{ marc('900','a', rec('200') ) }); |
| 490 |
dpavlin |
579 |
my $marc; |
| 491 |
|
|
ok($marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
| 492 |
|
|
diag dump( $marc ) if ($debug); |
| 493 |
dpavlin |
547 |
|
| 494 |
dpavlin |
579 |
is_deeply( $marc, [ |
| 495 |
dpavlin |
547 |
[ '900', 1, 2, 'a', '200a' ], |
| 496 |
|
|
[ '900', 1, 2, 'a', '200-solo' ] |
| 497 |
|
|
], 'correct marc with indicators'); |
| 498 |
|
|
|
| 499 |
|
|
test_s(qq{ marc_indicators('900',' ',9) }); |
| 500 |
|
|
test_s(qq{ marc_repeatable_subfield('900','a', rec('200') ) }); |
| 501 |
|
|
|
| 502 |
dpavlin |
579 |
ok($marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
| 503 |
|
|
diag dump( $marc ) if ($debug); |
| 504 |
dpavlin |
547 |
|
| 505 |
dpavlin |
579 |
is_deeply( $marc, [ |
| 506 |
dpavlin |
547 |
[ '900', 1, 2, 'a', '200a', 'a', '200-solo' ], |
| 507 |
|
|
[ '900', ' ', 9, 'a', '200a', 'a', '200-solo' ] |
| 508 |
|
|
], 'correct marc with repetable subfield'); |
| 509 |
dpavlin |
550 |
|
| 510 |
dpavlin |
555 |
# |
| 511 |
|
|
# test magic re-ordering of input data |
| 512 |
|
|
# |
| 513 |
dpavlin |
550 |
|
| 514 |
dpavlin |
555 |
sub test_rec_rules { |
| 515 |
|
|
my ($msg, $rec, $rules, $struct) = @_; |
| 516 |
dpavlin |
550 |
|
| 517 |
dpavlin |
555 |
_clean_ds(); |
| 518 |
|
|
_set_rec($rec); |
| 519 |
|
|
|
| 520 |
|
|
foreach my $r (split(/;/, $rules)) { |
| 521 |
|
|
$r =~ s/[\s\n\r]+/ /gs; |
| 522 |
|
|
$r =~ s/^\s+//gs; |
| 523 |
|
|
$r =~ s/\s+$//gs; |
| 524 |
|
|
test_s($r) if ($r); |
| 525 |
|
|
} |
| 526 |
|
|
|
| 527 |
dpavlin |
579 |
ok(my $marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
| 528 |
|
|
diag dump( $marc ) if ($debug); |
| 529 |
dpavlin |
562 |
diag "expects:\n", dump($struct) if ($debug > 1); |
| 530 |
dpavlin |
579 |
is_deeply( $marc, $struct, $msg ); |
| 531 |
dpavlin |
555 |
} |
| 532 |
|
|
|
| 533 |
|
|
test_rec_rules( |
| 534 |
|
|
'correct marc with repetable subfield', |
| 535 |
|
|
{ |
| 536 |
|
|
'200' => [{ |
| 537 |
|
|
'a' => '200a-1', |
| 538 |
|
|
'b' => '200b-1', |
| 539 |
|
|
'c' => '200c-1', |
| 540 |
|
|
}, { |
| 541 |
|
|
'a' => '200a-2', |
| 542 |
|
|
'b' => '200b-2', |
| 543 |
|
|
}, { |
| 544 |
|
|
'a' => '200a-3', |
| 545 |
|
|
}], |
| 546 |
|
|
}, |
| 547 |
|
|
qq{ |
| 548 |
|
|
marc_indicators('900',1 ,0); |
| 549 |
|
|
marc('900','a', rec('200','a') ); |
| 550 |
|
|
marc('900','b', rec('200','b') ); |
| 551 |
|
|
marc('900','c', rec('200','c') ); |
| 552 |
|
|
}, |
| 553 |
|
|
[ |
| 554 |
|
|
[ '900', 1, 0, 'a', '200a-1', 'b', '200b-1', 'c', '200c-1' ], |
| 555 |
|
|
[ '900', 1, 0, 'a', '200a-2', 'b', '200b-2' ], |
| 556 |
|
|
[ '900', 1, 0, 'a', '200a-3' ], |
| 557 |
|
|
], |
| 558 |
|
|
); |
| 559 |
|
|
|
| 560 |
|
|
|
| 561 |
|
|
test_rec_rules( |
| 562 |
|
|
'marc_repeatable_subfield', |
| 563 |
|
|
{ |
| 564 |
|
|
'200' => [{ |
| 565 |
|
|
'a' => '200a-1', |
| 566 |
|
|
'b' => '200b-1', |
| 567 |
|
|
'c' => '200c-1', |
| 568 |
|
|
}, { |
| 569 |
|
|
'a' => '200a-2', |
| 570 |
|
|
'b' => '200b-2', |
| 571 |
|
|
'c' => '200c-2', |
| 572 |
|
|
}, { |
| 573 |
|
|
'a' => '200a-3', |
| 574 |
|
|
'c' => '200c-3', |
| 575 |
|
|
}], |
| 576 |
|
|
}, |
| 577 |
|
|
qq{ |
| 578 |
|
|
marc_indicators('900',1 ,0); |
| 579 |
|
|
marc_repeatable_subfield('900','a', rec('200','a') ); |
| 580 |
|
|
marc('900','b', rec('200','b') ); |
| 581 |
|
|
marc('900','c', rec('200','c') ); |
| 582 |
|
|
}, |
| 583 |
|
|
[ |
| 584 |
|
|
[ '900', 1, 0, 'a', '200a-1', 'a', '200a-2', 'a', '200a-3', 'b', '200b-1', 'c', '200c-1' ], |
| 585 |
|
|
[ '900', 1, 0, 'b', '200b-2', 'c', '200c-2' ], |
| 586 |
|
|
[ '900', 1, 0, 'c', '200c-3' ], |
| 587 |
|
|
], |
| 588 |
|
|
); |
| 589 |
dpavlin |
562 |
|
| 590 |
|
|
test_rec_rules( |
| 591 |
|
|
'marc_compose', |
| 592 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
| 593 |
|
|
qq{ |
| 594 |
|
|
marc_compose('900', |
| 595 |
|
|
'c', rec(200,'b'), |
| 596 |
|
|
'b', rec(200,'a'), |
| 597 |
|
|
'a', rec(200,'c'), |
| 598 |
|
|
); |
| 599 |
|
|
}, |
| 600 |
|
|
[ |
| 601 |
|
|
[ '900', ' ', ' ', 'c', 42, 'b', 'foo ; bar', 'a', 'baz' ] |
| 602 |
|
|
], |
| 603 |
|
|
); |
| 604 |
|
|
|
| 605 |
dpavlin |
603 |
test_rec_rules( |
| 606 |
|
|
'marc_compose with + subfields', |
| 607 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
| 608 |
|
|
qq{ |
| 609 |
|
|
marc_compose('900', |
| 610 |
|
|
'a', rec(200,'a'), |
| 611 |
|
|
'+', prefix(" * ", rec(200,'c')), |
| 612 |
|
|
'b', rec(200,'b'), |
| 613 |
|
|
'+', prefix(" : ", rec(200,'c')), |
| 614 |
|
|
); |
| 615 |
|
|
}, |
| 616 |
|
|
[ |
| 617 |
|
|
[ '900', ' ', ' ', 'a', 'foo ; bar * baz', 'b', '42 : baz' ] |
| 618 |
|
|
], |
| 619 |
|
|
); |
| 620 |
|
|
|
| 621 |
dpavlin |
562 |
# |
| 622 |
|
|
# test rules |
| 623 |
|
|
# |
| 624 |
|
|
sub test_rule { |
| 625 |
|
|
my ($msg, $rec, $rule, $struct) = @_; |
| 626 |
|
|
_clean_ds(); |
| 627 |
|
|
_set_rec( $rec ); |
| 628 |
|
|
$rule =~ s/\\/\\/gs; |
| 629 |
|
|
my $r = test_s( $rule ); |
| 630 |
|
|
diag "for ", dump($rec), " got:\n", dump($r), "\nexpect:\n" if ($debug > 1); |
| 631 |
|
|
diag dump($struct) if ($debug); |
| 632 |
|
|
is_deeply( $r, $struct, $msg ); |
| 633 |
|
|
} |
| 634 |
|
|
|
| 635 |
|
|
# test split_rec_on |
| 636 |
|
|
test_rule( |
| 637 |
|
|
'split_rec_on', |
| 638 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
| 639 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 1) }, |
| 640 |
dpavlin |
571 |
[ 'foo' ], |
| 641 |
dpavlin |
562 |
); |
| 642 |
|
|
test_rule( |
| 643 |
|
|
'split_rec_on', |
| 644 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
| 645 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 2) }, |
| 646 |
dpavlin |
571 |
[ 'bar' ], |
| 647 |
dpavlin |
562 |
); |
| 648 |
dpavlin |
566 |
test_rule( |
| 649 |
|
|
'split_rec_on no part', |
| 650 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
| 651 |
dpavlin |
571 |
qq{ split_rec_on('200','a', qr/\\s*;\\s*/) }, |
| 652 |
dpavlin |
566 |
[ 'foo', 'bar' ], |
| 653 |
|
|
); |
| 654 |
|
|
test_rule( |
| 655 |
|
|
'split_rec_on no record', |
| 656 |
|
|
{}, |
| 657 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/) }, |
| 658 |
dpavlin |
571 |
[ '' ], |
| 659 |
dpavlin |
566 |
); |
| 660 |
dpavlin |
562 |
|
| 661 |
|
|
test_rec_rules( |
| 662 |
|
|
'marc_compose+split_rec_on', |
| 663 |
|
|
{ '200' => [{ a => 'foo ! bar', b => 42, c => 'baz' }] }, |
| 664 |
|
|
qq{ |
| 665 |
|
|
marc_compose('900', |
| 666 |
|
|
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 1), |
| 667 |
|
|
'c', rec(200,'c'), |
| 668 |
|
|
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 2), |
| 669 |
|
|
'b', rec(200,'b'), |
| 670 |
|
|
); |
| 671 |
|
|
}, |
| 672 |
|
|
[ |
| 673 |
|
|
[ '900', ' ', ' ', |
| 674 |
|
|
'a', 'foo', |
| 675 |
|
|
'c', 'baz', |
| 676 |
|
|
'a', 'bar', |
| 677 |
|
|
'b', 42, |
| 678 |
|
|
] |
| 679 |
|
|
], |
| 680 |
|
|
); |
| 681 |
dpavlin |
564 |
|
| 682 |
|
|
cmp_ok(marc_leader('06',42), '==', 42, 'marc_leader'); |
| 683 |
|
|
cmp_ok(marc_leader('11',5), '==', 5, 'marc_leader'); |
| 684 |
|
|
ok(marc_leader(), 'marc_leader get'); |
| 685 |
|
|
diag "leader: ", dump(marc_leader()) if ($debug); |
| 686 |
|
|
is_deeply(marc_leader(), { '06' => 42, 11 => 5 }, "marc_leader full"); |
| 687 |
dpavlin |
571 |
|
| 688 |
|
|
test_rule( |
| 689 |
|
|
'rec1(000)', |
| 690 |
|
|
{ '000' => [ 42 ]}, |
| 691 |
|
|
qq{ rec('000') }, |
| 692 |
|
|
[ 42 ], |
| 693 |
|
|
); |
| 694 |
|
|
|
| 695 |
|
|
test_rec_rules( |
| 696 |
dpavlin |
574 |
'marc(001,rec(000))', |
| 697 |
dpavlin |
571 |
{ '000' => [ 42 ]}, |
| 698 |
|
|
qq{ |
| 699 |
|
|
marc('001', rec('000') ); |
| 700 |
|
|
}, |
| 701 |
|
|
[ |
| 702 |
dpavlin |
574 |
[ '001', 42, ] |
| 703 |
dpavlin |
571 |
], |
| 704 |
|
|
); |
| 705 |
dpavlin |
574 |
|
| 706 |
|
|
test_rec_rules( |
| 707 |
dpavlin |
579 |
'marc_remove subfield', |
| 708 |
dpavlin |
574 |
{ '200' => [{ a => 42, b => 'bar', c => 'baz' }] }, |
| 709 |
|
|
qq{ |
| 710 |
|
|
marc('900', 'a', rec('200','a') ); |
| 711 |
|
|
marc('900', 'b', rec('200','b') ); |
| 712 |
|
|
marc_remove('900','b'); |
| 713 |
|
|
marc('900', 'b', rec('200','c') ); |
| 714 |
|
|
marc_remove('900','a'); |
| 715 |
|
|
}, |
| 716 |
|
|
[ |
| 717 |
|
|
[ '900', ' ', ' ', 'b', 'baz' ], |
| 718 |
|
|
], |
| 719 |
|
|
); |
| 720 |
|
|
|
| 721 |
|
|
test_rec_rules( |
| 722 |
dpavlin |
579 |
'marc_remove field', |
| 723 |
|
|
{ '200' => [{ a => 42, b => 'bar', c => 'baz' }] }, |
| 724 |
|
|
qq{ |
| 725 |
|
|
marc('900', 'a', rec('200','a') ); |
| 726 |
|
|
marc('900', 'b', rec('200','b') ); |
| 727 |
|
|
marc('901', 'b', rec('200','b') ); |
| 728 |
|
|
marc('901', 'c', rec('200','c') ); |
| 729 |
|
|
marc_remove('900'); |
| 730 |
|
|
}, |
| 731 |
|
|
[ |
| 732 |
|
|
[ '901', ' ', ' ', 'b', 'bar', 'c', 'baz' ], |
| 733 |
|
|
], |
| 734 |
|
|
); |
| 735 |
dpavlin |
787 |
|
| 736 |
|
|
test_s(qq{ marc_remove('*'); }); |
| 737 |
|
|
ok(! WebPAC::Normalize::_get_marc_fields(), 'marc_remove(*)'); |
| 738 |
|
|
|
| 739 |
dpavlin |
579 |
test_rec_rules( |
| 740 |
dpavlin |
574 |
'marc_duplicate', |
| 741 |
|
|
{ '200' => [{ a => 42, b => 'bar', c => 'baz', d => 'bing', e => 'bong' }] }, |
| 742 |
|
|
qq{ |
| 743 |
|
|
marc('900', 'a', rec('200','a') ); |
| 744 |
|
|
marc('900', 'b', rec('200','b') ); |
| 745 |
|
|
marc_duplicate; |
| 746 |
|
|
marc_remove('900','b'); |
| 747 |
|
|
marc('900', 'b', rec('200','c') ); |
| 748 |
|
|
marc_duplicate; |
| 749 |
|
|
marc_remove('900','b'); |
| 750 |
|
|
marc('900', 'b', rec('200','d') ); |
| 751 |
|
|
marc_duplicate; |
| 752 |
|
|
marc_remove('900','b'); |
| 753 |
|
|
marc('900', 'b', rec('200','e') ); |
| 754 |
|
|
}, |
| 755 |
|
|
[ |
| 756 |
|
|
# this will return FIRST record |
| 757 |
|
|
[ '900', ' ', ' ', 'a', 42, 'b', 'bar' ], |
| 758 |
|
|
], |
| 759 |
|
|
); |
| 760 |
|
|
|
| 761 |
|
|
my $i = 0; |
| 762 |
|
|
foreach my $v ( qw/bar baz bing bong/ ) { |
| 763 |
|
|
|
| 764 |
dpavlin |
579 |
ok($marc = WebPAC::Normalize::_get_marc_fields( offset => $i ), |
| 765 |
dpavlin |
574 |
"_get_marc_fields( offset => $i )" |
| 766 |
|
|
); |
| 767 |
dpavlin |
579 |
diag "marc $i = ", dump( $marc ) if ($debug); |
| 768 |
|
|
is_deeply( $marc, |
| 769 |
dpavlin |
574 |
[ [ '900', ' ', ' ', 'a', 42, 'b', $v ] ], |
| 770 |
|
|
"MARC copy $i has $v", |
| 771 |
|
|
); |
| 772 |
|
|
$i++; |
| 773 |
|
|
} |
| 774 |
dpavlin |
604 |
|
| 775 |
|
|
test_rec_rules( |
| 776 |
|
|
'marc_original_order', |
| 777 |
|
|
{ |
| 778 |
|
|
'200' => [ { |
| 779 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
| 780 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
| 781 |
|
|
}, { |
| 782 |
|
|
a => [ 'a3', 'a4', 'a5' ], b => 'b3', c => 'c3', |
| 783 |
|
|
subfields => [ qw/a 0 a 1 b 0 c 0 a 2/ ], |
| 784 |
|
|
} ], |
| 785 |
|
|
}, |
| 786 |
|
|
qq{ |
| 787 |
dpavlin |
616 |
marc_original_order(900,200); |
| 788 |
dpavlin |
604 |
}, |
| 789 |
|
|
[ |
| 790 |
|
|
[ '900', ' ', ' ', 'a', 'a1', 'b', 'b1', 'a', 'a2', 'b', 'b2', 'c', 'c1', 'c', 'c2', ], |
| 791 |
|
|
[ '900', ' ', ' ', 'a', 'a3', 'a', 'a4', 'b', 'b3', 'c', 'c3', 'a', 'a5', ], |
| 792 |
|
|
], |
| 793 |
|
|
); |
| 794 |
dpavlin |
631 |
|
| 795 |
|
|
test_rule( |
| 796 |
|
|
'rec1 skips subfields', |
| 797 |
|
|
{ |
| 798 |
|
|
'200' => [ { |
| 799 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
| 800 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
| 801 |
|
|
}, { |
| 802 |
|
|
a => [ 'a3', 'a4', 'a5' ], b => 'b3', c => 'c3', |
| 803 |
|
|
subfields => [ qw/a 0 a 1 b 0 c 0 a 2/ ], |
| 804 |
|
|
} ], |
| 805 |
|
|
}, |
| 806 |
|
|
qq{ |
| 807 |
|
|
rec1(200); |
| 808 |
|
|
}, |
| 809 |
dpavlin |
641 |
['a1', 'b1', 'a2', 'b2', 'c1', 'c2', 'a3', 'a4', 'b3', 'c3', 'a5' ], |
| 810 |
dpavlin |
631 |
); |
| 811 |
dpavlin |
641 |
|
| 812 |
|
|
is_deeply( |
| 813 |
|
|
[ _pack_subfields_hash({ |
| 814 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
| 815 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
| 816 |
|
|
}) ], |
| 817 |
|
|
['a1', 'b1', 'a2', 'b2', 'c1', 'c2'], |
| 818 |
|
|
'_pack_subfields_hash( $h )' |
| 819 |
|
|
); |
| 820 |
|
|
|
| 821 |
dpavlin |
669 |
cmp_ok( |
| 822 |
|
|
_pack_subfields_hash({ |
| 823 |
dpavlin |
641 |
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
| 824 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
| 825 |
dpavlin |
669 |
}, 1), |
| 826 |
|
|
'eq', |
| 827 |
|
|
'^aa1^bb1^aa2^bb2^cc1^cc2', |
| 828 |
dpavlin |
641 |
'_pack_subfields_hash( $h, 1 )' |
| 829 |
|
|
); |
| 830 |
dpavlin |
536 |
} |
| 831 |
|
|
|
Parent Directory
| 
Revision Log