1 |
#!/usr/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
|
5 |
use Test::More tests => 107; |
6 |
use Test::Exception; |
7 |
use Cwd qw/abs_path/; |
8 |
use blib; |
9 |
use File::Slurp; |
10 |
use Getopt::Long; |
11 |
|
12 |
BEGIN { |
13 |
use_ok( 'WebPAC::Normalize' ); |
14 |
} |
15 |
|
16 |
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 |
ok(my $abs_path = abs_path($0), "abs_path"); |
32 |
$abs_path =~ s#/[^/]*$#/#; |
33 |
diag "abs_path: $abs_path" if ($debug); |
34 |
|
35 |
#throws_ok { new WebPAC::Normalize::XML( lookup_regex => 'foo' ) } qr/pair/, "lookup_regex without lookup"; |
36 |
|
37 |
my $rec1 = { |
38 |
'200' => [{ |
39 |
'a' => '200a', |
40 |
'b' => '200b', |
41 |
},{ |
42 |
'c' => '200c', |
43 |
'd' => '200d', |
44 |
},{ |
45 |
'a' => '200a*2', |
46 |
'd' => '200d*2', |
47 |
}], |
48 |
'201' => [{ |
49 |
'x' => '201x', |
50 |
'y' => '201y', |
51 |
}], |
52 |
'900' => [ |
53 |
'900-no_subfield' |
54 |
], |
55 |
'901' => [{ |
56 |
'a' => '900a', |
57 |
}], |
58 |
'902' => [{ |
59 |
'z' => '900', |
60 |
}], |
61 |
}; |
62 |
|
63 |
my $rec2 = { |
64 |
'675' => [ { |
65 |
'a' => '159.9' |
66 |
} ], |
67 |
'210' => [ { |
68 |
'c' => 'New York University press', |
69 |
'a' => 'New York', |
70 |
'd' => 'cop. 1988' |
71 |
} ], |
72 |
'700' => [ { |
73 |
'a' => 'Haynal', |
74 |
'b' => 'André' |
75 |
} ], |
76 |
'801' => [ 'FFZG' ], |
77 |
'991' => [ '8302' ], |
78 |
'000' => [ 1 ], |
79 |
'702' => [ { |
80 |
'a' => 'Holder', |
81 |
'b' => 'Elizabeth' |
82 |
} ], |
83 |
'215' => [ { |
84 |
'c' => 'ilustr', |
85 |
'a' => 'xix, 202 str', |
86 |
'd' => '23cm' |
87 |
} ], |
88 |
'990' => [ |
89 |
'2140', |
90 |
'88', |
91 |
'HAY' |
92 |
], |
93 |
'200' => [ { |
94 |
'e' => 'from Freud and Ferenczi to Michael balint', |
95 |
'a' => 'Controversies in psychoanalytic method', |
96 |
'g' => 'translated by Elizabeth Holder on the basisi of a first draft by Archie Hooton ; with a preface by Daniel N. Stern', |
97 |
'f' => 'by André E. Haynal' |
98 |
} ], |
99 |
'610' => [ 'povijest psihoanalize' ], |
100 |
'994' => [ { |
101 |
'c' => '', |
102 |
'a' => 'PS', |
103 |
'b' => 'MG' |
104 |
} ], |
105 |
'320' => [ 'Kazalo' ], |
106 |
'101' => [ 'ENG' ], |
107 |
'686' => [ '2140' ], |
108 |
'300' => [ 'Prijevod djela: ' ], |
109 |
}; |
110 |
|
111 |
|
112 |
my $lookup1 = { |
113 |
'00900' => [ |
114 |
'lookup 1', |
115 |
'lookup 2', |
116 |
], |
117 |
}; |
118 |
|
119 |
my $lookup2 = { |
120 |
'00900' => 'lookup', |
121 |
}; |
122 |
|
123 |
|
124 |
sub test { |
125 |
print dump( @_ ), ("-" x 78), "\n"; |
126 |
ok( defined(@_) ); |
127 |
} |
128 |
|
129 |
# how much of string evaled to display? |
130 |
my $max_eval_output = 170; |
131 |
|
132 |
sub dump_error { |
133 |
my ($msg,$code) = @_; |
134 |
|
135 |
my @l = split(/[\n\r]/, $code); |
136 |
my $out = "$msg\n"; |
137 |
|
138 |
foreach my $i ( 0 .. $#l ) { |
139 |
$out .= sprintf("%2d: %s\n", $i, $l[$i]); |
140 |
} |
141 |
|
142 |
return $out; |
143 |
} |
144 |
|
145 |
sub test_s { |
146 |
my $t = shift || die; |
147 |
|
148 |
my $eval_t = $t; |
149 |
$eval_t =~ s/[\n\r\s]+/ /gs; |
150 |
$eval_t = substr($eval_t,0,$max_eval_output) . '...' if (length($eval_t) > $max_eval_output); |
151 |
$eval_t =~ s/\\/\\\\/gs; |
152 |
|
153 |
my $v = eval "$t"; |
154 |
ok(! $@, $@ ? dump_error($@, $t) : "eval: $eval_t"); |
155 |
$v; |
156 |
} |
157 |
|
158 |
{ |
159 |
no strict 'subs'; |
160 |
use WebPAC::Normalize; |
161 |
|
162 |
ok(! _set_lookup( undef ), "set_lookup(undef)"); |
163 |
|
164 |
_set_rec( $rec1 ); |
165 |
|
166 |
cmp_ok( join(",", rec2('200','a') ), 'eq', '200a,200a*2', 'join rec2' ); |
167 |
cmp_ok( join(",", rec2('200','a'), rec2('200','b') ), 'eq', '200a,200a*2,200b', 'join rec2 rec2' ); |
168 |
cmp_ok( join(" * ", sort(rec1('200'), rec1('201') )), 'eq', '200a * 200a*2 * 200b * 200c * 200d * 200d*2 * 201x * 201y', 'join sort rec1 rec1' ); |
169 |
diag "is_deeply checks\n"; |
170 |
is_deeply( \[ rec1('200') ], \[ qw/200a 200b 200c 200d 200a*2 200d*2/ ] ); |
171 |
is_deeply( \[ regex( 's/0/o/g', rec1('200') ) ], \[ qw/2ooa 2oob 2ooc 2ood 2ooa*2 2ood*2/ ]); |
172 |
is_deeply( \[ grep { /\*/ } regex( 's/0/o/g', rec1('200') ) ], \[ qw/2ooa*2 2ood*2/ ]); |
173 |
is_deeply( \[ rec('902') ], \[ '900' ] ); |
174 |
|
175 |
cmp_ok( rec('902'), 'eq', rec('902','z'), 'rec sf' ); |
176 |
|
177 |
# simple list manipulatons |
178 |
cmp_ok( join('', prefix('ab', 'cd') ), 'eq', 'abcd', 'prefix'); |
179 |
cmp_ok( join('', suffix('xy', 'cd') ), 'eq', 'cdxy', 'suffix'); |
180 |
cmp_ok( join('', surround('->', '<-', 'a','b','c') ), 'eq', '->a<-->b<-->c<-', 'surround'); |
181 |
|
182 |
|
183 |
_set_lookup( $lookup1 ); |
184 |
|
185 |
cmp_ok( |
186 |
join_with(" i ", |
187 |
lookup( |
188 |
regex( 's/^/00/', |
189 |
rec2('902','z') |
190 |
) |
191 |
) |
192 |
), |
193 |
'eq', 'lookup 1 i lookup 2', 'join lookup regex rec2'); |
194 |
|
195 |
# check join_with operations |
196 |
|
197 |
sub test_join_with_2 { |
198 |
my ($a,$b,$e) = @_; |
199 |
|
200 |
cmp_ok( |
201 |
join_with(" <1> ", |
202 |
rec('201',$a), |
203 |
rec('201',$b), |
204 |
), |
205 |
'eq', $e, "join_with $a <1> $b = $e"); |
206 |
} |
207 |
|
208 |
test_join_with_2('_','_',''); |
209 |
test_join_with_2('x','_','201x'); |
210 |
test_join_with_2('_','x','201x'); |
211 |
test_join_with_2('x','y','201x <1> 201y'); |
212 |
|
213 |
sub test_join_with_3 { |
214 |
my ($a,$b,$c,$e) = @_; |
215 |
|
216 |
cmp_ok( |
217 |
join_with(" <1> ", rec('201',$a), |
218 |
join_with(" <2> ", rec('201',$b), |
219 |
rec('201',$c), |
220 |
) |
221 |
), |
222 |
'eq', $e, "join_with $a <1> $b <2> $c = $e"); |
223 |
}; |
224 |
|
225 |
test_join_with_3('_','_','_',''); |
226 |
test_join_with_3('x','_','_','201x'); |
227 |
test_join_with_3('_','x','_','201x'); |
228 |
test_join_with_3('_','_','x','201x'); |
229 |
test_join_with_3('x','y','_','201x <1> 201y'); |
230 |
test_join_with_3('x','_','y','201x <1> 201y'); |
231 |
test_join_with_3('_','x','y','201x <2> 201y'); |
232 |
test_join_with_3('x','_','y','201x <1> 201y'); |
233 |
test_join_with_3('x','y','x','201x <1> 201y <2> 201x'); |
234 |
|
235 |
# test lookups |
236 |
|
237 |
_set_lookup( $lookup2 ); |
238 |
|
239 |
is_deeply( \[ lookup( prefix( '00', rec('902') ) ) ], \[ 'lookup' ], 'lookup prefix' ); |
240 |
|
241 |
ok(! lookup('non-existent'), 'lookup non-existant' ); |
242 |
|
243 |
_set_rec( $rec2 ); |
244 |
|
245 |
test_s(qq{ |
246 |
tag('Title', |
247 |
rec('200','a') |
248 |
); |
249 |
}); |
250 |
test_s(qq{ |
251 |
tag('Who', |
252 |
join_with(" ", |
253 |
rec('702','a'), |
254 |
rec('702','b') |
255 |
) |
256 |
); |
257 |
}); |
258 |
|
259 |
test_s(qq{ |
260 |
display('Publisher', |
261 |
rec('210','c') |
262 |
) |
263 |
}); |
264 |
|
265 |
test_s(qq{ |
266 |
search('Year', |
267 |
regex( 's/[^\\d]+//', |
268 |
rec('210','d') |
269 |
) |
270 |
) |
271 |
}); |
272 |
|
273 |
ok(my $ds = _get_ds(), "get_ds"); |
274 |
diag "ds = ", dump($ds) if ($debug); |
275 |
|
276 |
|
277 |
sub test_check_ds { |
278 |
|
279 |
my $t = shift; |
280 |
|
281 |
ok($ds = _get_ds(), 'get_ds'); |
282 |
diag dump( $ds ) if ($debug); |
283 |
|
284 |
ok( $ds && $ds->{something}, 'get_ds->something exists' ); |
285 |
ok( $ds && $ds->{something}->{$t}, 'get_ds->something->'.$t.' exists') if ($t); |
286 |
ok( $ds && !$ds->{empty}, 'get_ds->empty doesn\'t' ); |
287 |
|
288 |
return $ds; |
289 |
} |
290 |
|
291 |
_clean_ds(); |
292 |
test_s(qq{ search('something', '42'); }); |
293 |
test_s(qq{ search('empty', ''); }); |
294 |
test_check_ds('search'); |
295 |
|
296 |
_clean_ds(); |
297 |
test_s(qq{ display('something', '42'); }); |
298 |
test_s(qq{ display('empty', ''); }); |
299 |
test_check_ds('display'); |
300 |
|
301 |
_clean_ds(); |
302 |
test_s(qq{ tag('something', '42'); }); |
303 |
test_s(qq{ tag('empty', ''); }); |
304 |
test_check_ds('search'); |
305 |
test_check_ds('display'); |
306 |
|
307 |
_clean_ds(); |
308 |
my $n = read_file( "$abs_path/data/normalize.pl" ); |
309 |
$n .= "\n1;\n"; |
310 |
#diag "normalize code:\n$n\n"; |
311 |
test_s( $n ); |
312 |
|
313 |
ok($ds = _get_ds(), "get_ds"); |
314 |
diag "ds = ", dump($ds) if ($debug); |
315 |
|
316 |
my $rec = { |
317 |
'200' => [{ |
318 |
'a' => '200a', |
319 |
'b' => '200b', |
320 |
}], |
321 |
}; |
322 |
my $rules = qq{ search('mixed', rec('200') ) }; |
323 |
|
324 |
_clean_ds(); |
325 |
_set_rec( $rec ); |
326 |
test_s( $rules ); |
327 |
ok($ds = _get_ds(), "get_ds"); |
328 |
is_deeply( $ds, { |
329 |
'mixed' => { |
330 |
'search' => [ '200a', '200b' ], |
331 |
'tag' => 'mixed' |
332 |
} |
333 |
}, 'correct get_ds'); |
334 |
|
335 |
ok(my $ds2 = WebPAC::Normalize::data_structure( |
336 |
row => $rec, |
337 |
rules => $rules, |
338 |
), 'data_structure'); |
339 |
is_deeply( $ds, $ds2, 'data_structure(s) same'); |
340 |
|
341 |
# wird and non-valid structure which is supported anyway |
342 |
_clean_ds(); |
343 |
_set_rec({ |
344 |
'200' => [{ |
345 |
'a' => '200a', |
346 |
}, |
347 |
'200-solo' |
348 |
] |
349 |
}); |
350 |
test_s(qq{ search('mixed', rec('200') ) }); |
351 |
ok($ds = _get_ds(), "get_ds"); |
352 |
is_deeply( $ds, { |
353 |
'mixed' => { |
354 |
'search' => [ '200a', '200-solo' ], |
355 |
'tag' => 'mixed' |
356 |
} |
357 |
}, 'correct get_ds'); |
358 |
|
359 |
# |
360 |
# MARC |
361 |
# |
362 |
|
363 |
test_s(qq{ marc_indicators('900',1,2) }); |
364 |
test_s(qq{ marc('900','a', rec('200') ) }); |
365 |
my @marc; |
366 |
ok(@marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
367 |
diag dump( \@marc ) if ($debug); |
368 |
|
369 |
is_deeply( \@marc, [ |
370 |
[ '900', 1, 2, 'a', '200a' ], |
371 |
[ '900', 1, 2, 'a', '200-solo' ] |
372 |
], 'correct marc with indicators'); |
373 |
|
374 |
test_s(qq{ marc_indicators('900',' ',9) }); |
375 |
test_s(qq{ marc_repeatable_subfield('900','a', rec('200') ) }); |
376 |
|
377 |
ok(@marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
378 |
diag dump( \@marc ) if ($debug); |
379 |
|
380 |
is_deeply( \@marc, [ |
381 |
[ '900', 1, 2, 'a', '200a', 'a', '200-solo' ], |
382 |
[ '900', ' ', 9, 'a', '200a', 'a', '200-solo' ] |
383 |
], 'correct marc with repetable subfield'); |
384 |
|
385 |
# |
386 |
# test magic re-ordering of input data |
387 |
# |
388 |
|
389 |
sub test_rec_rules { |
390 |
my ($msg, $rec, $rules, $struct) = @_; |
391 |
|
392 |
_clean_ds(); |
393 |
_set_rec($rec); |
394 |
|
395 |
foreach my $r (split(/;/, $rules)) { |
396 |
$r =~ s/[\s\n\r]+/ /gs; |
397 |
$r =~ s/^\s+//gs; |
398 |
$r =~ s/\s+$//gs; |
399 |
test_s($r) if ($r); |
400 |
} |
401 |
|
402 |
ok(my @marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
403 |
diag dump( \@marc ) if ($debug); |
404 |
diag "expects:\n", dump($struct) if ($debug > 1); |
405 |
is_deeply( \@marc, $struct, $msg ); |
406 |
} |
407 |
|
408 |
test_rec_rules( |
409 |
'correct marc with repetable subfield', |
410 |
{ |
411 |
'200' => [{ |
412 |
'a' => '200a-1', |
413 |
'b' => '200b-1', |
414 |
'c' => '200c-1', |
415 |
}, { |
416 |
'a' => '200a-2', |
417 |
'b' => '200b-2', |
418 |
}, { |
419 |
'a' => '200a-3', |
420 |
}], |
421 |
}, |
422 |
qq{ |
423 |
marc_indicators('900',1 ,0); |
424 |
marc('900','a', rec('200','a') ); |
425 |
marc('900','b', rec('200','b') ); |
426 |
marc('900','c', rec('200','c') ); |
427 |
}, |
428 |
[ |
429 |
[ '900', 1, 0, 'a', '200a-1', 'b', '200b-1', 'c', '200c-1' ], |
430 |
[ '900', 1, 0, 'a', '200a-2', 'b', '200b-2' ], |
431 |
[ '900', 1, 0, 'a', '200a-3' ], |
432 |
], |
433 |
); |
434 |
|
435 |
|
436 |
test_rec_rules( |
437 |
'marc_repeatable_subfield', |
438 |
{ |
439 |
'200' => [{ |
440 |
'a' => '200a-1', |
441 |
'b' => '200b-1', |
442 |
'c' => '200c-1', |
443 |
}, { |
444 |
'a' => '200a-2', |
445 |
'b' => '200b-2', |
446 |
'c' => '200c-2', |
447 |
}, { |
448 |
'a' => '200a-3', |
449 |
'c' => '200c-3', |
450 |
}], |
451 |
}, |
452 |
qq{ |
453 |
marc_indicators('900',1 ,0); |
454 |
marc_repeatable_subfield('900','a', rec('200','a') ); |
455 |
marc('900','b', rec('200','b') ); |
456 |
marc('900','c', rec('200','c') ); |
457 |
}, |
458 |
[ |
459 |
[ '900', 1, 0, 'a', '200a-1', 'a', '200a-2', 'a', '200a-3', 'b', '200b-1', 'c', '200c-1' ], |
460 |
[ '900', 1, 0, 'b', '200b-2', 'c', '200c-2' ], |
461 |
[ '900', 1, 0, 'c', '200c-3' ], |
462 |
], |
463 |
); |
464 |
|
465 |
test_rec_rules( |
466 |
'marc_compose', |
467 |
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
468 |
qq{ |
469 |
marc_compose('900', |
470 |
'c', rec(200,'b'), |
471 |
'b', rec(200,'a'), |
472 |
'a', rec(200,'c'), |
473 |
); |
474 |
}, |
475 |
[ |
476 |
[ '900', ' ', ' ', 'c', 42, 'b', 'foo ; bar', 'a', 'baz' ] |
477 |
], |
478 |
); |
479 |
|
480 |
# |
481 |
# test rules |
482 |
# |
483 |
sub test_rule { |
484 |
my ($msg, $rec, $rule, $struct) = @_; |
485 |
_clean_ds(); |
486 |
_set_rec( $rec ); |
487 |
$rule =~ s/\\/\\/gs; |
488 |
my $r = test_s( $rule ); |
489 |
diag "for ", dump($rec), " got:\n", dump($r), "\nexpect:\n" if ($debug > 1); |
490 |
diag dump($struct) if ($debug); |
491 |
is_deeply( $r, $struct, $msg ); |
492 |
} |
493 |
|
494 |
# test split_rec_on |
495 |
test_rule( |
496 |
'split_rec_on', |
497 |
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
498 |
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 1) }, |
499 |
'foo', |
500 |
); |
501 |
test_rule( |
502 |
'split_rec_on', |
503 |
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
504 |
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 2) }, |
505 |
'bar', |
506 |
); |
507 |
test_rule( |
508 |
'split_rec_on no part', |
509 |
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
510 |
q! split_rec_on('200','a', qr/\\s*;\\s*/) !, |
511 |
[ 'foo', 'bar' ], |
512 |
); |
513 |
test_rule( |
514 |
'split_rec_on no record', |
515 |
{}, |
516 |
qq{ split_rec_on('200','a', qr/\\s*;\\s*/) }, |
517 |
'', |
518 |
); |
519 |
|
520 |
test_rec_rules( |
521 |
'marc_compose+split_rec_on', |
522 |
{ '200' => [{ a => 'foo ! bar', b => 42, c => 'baz' }] }, |
523 |
qq{ |
524 |
marc_compose('900', |
525 |
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 1), |
526 |
'c', rec(200,'c'), |
527 |
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 2), |
528 |
'b', rec(200,'b'), |
529 |
); |
530 |
}, |
531 |
[ |
532 |
[ '900', ' ', ' ', |
533 |
'a', 'foo', |
534 |
'c', 'baz', |
535 |
'a', 'bar', |
536 |
'b', 42, |
537 |
] |
538 |
], |
539 |
); |
540 |
|
541 |
cmp_ok(marc_leader('06',42), '==', 42, 'marc_leader'); |
542 |
cmp_ok(marc_leader('11',5), '==', 5, 'marc_leader'); |
543 |
ok(marc_leader(), 'marc_leader get'); |
544 |
diag "leader: ", dump(marc_leader()) if ($debug); |
545 |
is_deeply(marc_leader(), { '06' => 42, 11 => 5 }, "marc_leader full"); |
546 |
} |
547 |
|