1 |
dpavlin |
536 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
|
5 |
dpavlin |
819 |
use Test::More tests => 341; |
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 |
dpavlin |
819 |
cmp_ok( join('-', prefix('', 'x', 'y') ), 'eq', 'x-y', 'prefix empty'); |
195 |
|
|
cmp_ok( join('-', prefix(0, 'x', 'y') ), 'eq', '0x-0y', 'prefix 0'); |
196 |
|
|
|
197 |
dpavlin |
536 |
cmp_ok( join('', suffix('xy', 'cd') ), 'eq', 'cdxy', 'suffix'); |
198 |
dpavlin |
819 |
cmp_ok( join('-', suffix('', 'x', 'y' ) ), 'eq', 'x-y', 'suffix empty'); |
199 |
|
|
cmp_ok( join('-', suffix(0, 'x', 'y' ) ), 'eq', 'x0-y0', 'suffix 0'); |
200 |
|
|
|
201 |
dpavlin |
536 |
cmp_ok( join('', surround('->', '<-', 'a','b','c') ), 'eq', '->a<-->b<-->c<-', 'surround'); |
202 |
dpavlin |
819 |
cmp_ok( join('-', surround('', '', 'x','y','z') ), 'eq', 'x-y-z', 'surround empty'); |
203 |
|
|
cmp_ok( join('-', surround(0, 0, 'x','y','z') ), 'eq', '0x0-0y0-0z0', 'surround 0 0'); |
204 |
dpavlin |
536 |
|
205 |
dpavlin |
791 |
# count |
206 |
|
|
my @el; |
207 |
|
|
for my $i ( 0 .. 10 ) { |
208 |
|
|
cmp_ok( count( @el ), '==', $i, "count($i)"); |
209 |
|
|
push @el, "element $i"; |
210 |
|
|
} |
211 |
|
|
|
212 |
dpavlin |
725 |
# lookups |
213 |
dpavlin |
536 |
|
214 |
dpavlin |
736 |
throws_ok { _set_load_row() } qr/CODE/, 'empty _set_load_row()'; |
215 |
dpavlin |
725 |
|
216 |
dpavlin |
736 |
ok(_set_load_row(sub { |
217 |
dpavlin |
725 |
my ($database,$input,$mfn) = @_; |
218 |
dpavlin |
811 |
diag "load_row( $database, $input, $mfn )" if ($debug); |
219 |
dpavlin |
725 |
cmp_ok( $#_, '==', 2, 'have 3 arguments'); |
220 |
dpavlin |
736 |
ok($database, '_load_row database'); |
221 |
|
|
ok($input, '_load_row input'); |
222 |
|
|
ok($mfn, '_load_row mfn'); |
223 |
dpavlin |
725 |
return { |
224 |
|
|
'900' => [{ x => '900x-' . $mfn , y => '900y-' . $mfn }], |
225 |
|
|
} |
226 |
|
|
|
227 |
dpavlin |
736 |
}), '_set_load_row'); |
228 |
dpavlin |
725 |
|
229 |
|
|
my @v = qw/foo bar baz aaa bbb ccc ddd/; |
230 |
|
|
|
231 |
|
|
my @accumulated; |
232 |
|
|
|
233 |
|
|
for my $i ( 0 .. $#v ) { |
234 |
|
|
|
235 |
|
|
my $mfn = 1000 + $i; |
236 |
|
|
|
237 |
|
|
ok(WebPAC::Normalize::_set_config({ '_mfn' => $mfn }), "_set_config _mfn=$mfn"); |
238 |
|
|
|
239 |
|
|
my $size = $#v + 1; |
240 |
|
|
|
241 |
|
|
cmp_ok( |
242 |
|
|
save_into_lookup('db','input','key', sub { @v }), |
243 |
|
|
'==', $size, "save_into_lookup $size values" |
244 |
|
|
); |
245 |
|
|
|
246 |
|
|
ok(my $l = WebPAC::Normalize::_get_lookup(), '_get_lookup'); |
247 |
dpavlin |
811 |
diag "_get_lookup = ", dump($l) if ($debug); |
248 |
dpavlin |
725 |
|
249 |
|
|
my @lookup; |
250 |
|
|
|
251 |
dpavlin |
736 |
ok(@lookup = lookup( |
252 |
dpavlin |
725 |
sub { |
253 |
dpavlin |
811 |
diag "in show" if ($debug); |
254 |
dpavlin |
725 |
rec('900','x'); |
255 |
|
|
}, |
256 |
|
|
'db','input','key', |
257 |
|
|
sub { |
258 |
|
|
return @v; |
259 |
|
|
} |
260 |
|
|
), |
261 |
|
|
"lookup db/input/key"); |
262 |
|
|
|
263 |
|
|
push @accumulated, '900x-' . $mfn; |
264 |
|
|
|
265 |
|
|
is_deeply(\@lookup, \@accumulated, "lookup db/input/key"); |
266 |
|
|
|
267 |
|
|
shift @v; |
268 |
|
|
|
269 |
|
|
} |
270 |
|
|
|
271 |
|
|
ok(my $l = WebPAC::Normalize::_get_lookup(), '_get_lookup'); |
272 |
dpavlin |
811 |
diag "_get_lookup = ", dump($l) if ($debug); |
273 |
dpavlin |
725 |
|
274 |
|
|
is_deeply( $l, { |
275 |
|
|
db => { |
276 |
|
|
input => { |
277 |
|
|
key => { |
278 |
|
|
foo => { 1000 => 1 }, |
279 |
|
|
bar => { 1000 => 1, 1001 => 1 }, |
280 |
|
|
baz => { 1000 => 1, 1001 => 1, 1002 => 1 }, |
281 |
|
|
aaa => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1 }, |
282 |
|
|
bbb => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1 }, |
283 |
|
|
ccc => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1, 1005 => 1 }, |
284 |
|
|
ddd => { 1000 => 1, 1001 => 1, 1002 => 1, 1003 => 1, 1004 => 1, 1005 => 1, 1006 => 1 }, |
285 |
|
|
}, |
286 |
|
|
}, |
287 |
|
|
}, |
288 |
|
|
}, 'lookup data'); |
289 |
|
|
|
290 |
|
|
####### |
291 |
|
|
|
292 |
dpavlin |
811 |
diag "lookup_hash1 = ", dump($lookup_hash1) if ($debug); |
293 |
dpavlin |
725 |
ok(_set_lookup( $lookup_hash1 ), '_set_lookup $lookup_hash1'); |
294 |
|
|
|
295 |
dpavlin |
736 |
throws_ok { _set_load_row() } qr/CODE/, 'empty _set_load_row()'; |
296 |
dpavlin |
725 |
|
297 |
dpavlin |
736 |
ok(_set_load_row(sub { |
298 |
dpavlin |
725 |
my ($database,$input,$mfn) = @_; |
299 |
dpavlin |
736 |
diag "load_row( $database, $input, $mfn )"; |
300 |
dpavlin |
725 |
cmp_ok( $#_, '==', 2, 'have 3 arguments'); |
301 |
|
|
ok($database, 'database'); |
302 |
|
|
ok($input, 'input'); |
303 |
|
|
ok($mfn, 'mfn'); |
304 |
|
|
|
305 |
dpavlin |
736 |
}), '_set_load_row'); |
306 |
dpavlin |
725 |
|
307 |
|
|
|
308 |
|
|
# cmp_ok(lookup( |
309 |
|
|
# sub { |
310 |
|
|
# 'found' |
311 |
|
|
# }, |
312 |
|
|
# 'db1','input1','key1', |
313 |
|
|
# sub { |
314 |
|
|
# rec('200','a') |
315 |
|
|
# } |
316 |
|
|
# ), 'eq', 'found', 'lookup db1/input1/key1'); |
317 |
|
|
|
318 |
|
|
|
319 |
dpavlin |
536 |
|
320 |
dpavlin |
725 |
# cmp_ok( |
321 |
|
|
# lookup( |
322 |
|
|
# ), |
323 |
|
|
# 'eq', 'lookup 1 i lookup 2', 'join lookup regex rec2'); |
324 |
dpavlin |
536 |
|
325 |
|
|
# check join_with operations |
326 |
|
|
|
327 |
|
|
sub test_join_with_2 { |
328 |
|
|
my ($a,$b,$e) = @_; |
329 |
|
|
|
330 |
|
|
cmp_ok( |
331 |
|
|
join_with(" <1> ", |
332 |
|
|
rec('201',$a), |
333 |
|
|
rec('201',$b), |
334 |
|
|
), |
335 |
|
|
'eq', $e, "join_with $a <1> $b = $e"); |
336 |
|
|
} |
337 |
|
|
|
338 |
|
|
test_join_with_2('_','_',''); |
339 |
|
|
test_join_with_2('x','_','201x'); |
340 |
|
|
test_join_with_2('_','x','201x'); |
341 |
|
|
test_join_with_2('x','y','201x <1> 201y'); |
342 |
|
|
|
343 |
|
|
sub test_join_with_3 { |
344 |
|
|
my ($a,$b,$c,$e) = @_; |
345 |
|
|
|
346 |
|
|
cmp_ok( |
347 |
|
|
join_with(" <1> ", rec('201',$a), |
348 |
|
|
join_with(" <2> ", rec('201',$b), |
349 |
|
|
rec('201',$c), |
350 |
|
|
) |
351 |
|
|
), |
352 |
|
|
'eq', $e, "join_with $a <1> $b <2> $c = $e"); |
353 |
|
|
}; |
354 |
|
|
|
355 |
|
|
test_join_with_3('_','_','_',''); |
356 |
|
|
test_join_with_3('x','_','_','201x'); |
357 |
|
|
test_join_with_3('_','x','_','201x'); |
358 |
|
|
test_join_with_3('_','_','x','201x'); |
359 |
|
|
test_join_with_3('x','y','_','201x <1> 201y'); |
360 |
|
|
test_join_with_3('x','_','y','201x <1> 201y'); |
361 |
|
|
test_join_with_3('_','x','y','201x <2> 201y'); |
362 |
|
|
test_join_with_3('x','_','y','201x <1> 201y'); |
363 |
|
|
test_join_with_3('x','y','x','201x <1> 201y <2> 201x'); |
364 |
|
|
|
365 |
|
|
# test lookups |
366 |
|
|
|
367 |
dpavlin |
725 |
_set_lookup( $lookup_hash2 ); |
368 |
dpavlin |
536 |
|
369 |
dpavlin |
725 |
throws_ok { lookup() } qr/need/, 'empty lookup'; |
370 |
dpavlin |
536 |
|
371 |
dpavlin |
725 |
#is_deeply( \[ lookup( prefix( '00', rec('902') ) ) ], \[ 'lookup' ], 'lookup prefix' ); |
372 |
dpavlin |
536 |
|
373 |
dpavlin |
725 |
#ok(! lookup('non-existent'), 'lookup non-existant' ); |
374 |
|
|
|
375 |
dpavlin |
538 |
_set_rec( $rec2 ); |
376 |
dpavlin |
536 |
|
377 |
|
|
test_s(qq{ |
378 |
dpavlin |
912 |
search_display('Title', |
379 |
dpavlin |
536 |
rec('200','a') |
380 |
|
|
); |
381 |
|
|
}); |
382 |
|
|
test_s(qq{ |
383 |
dpavlin |
912 |
search_display('Who', |
384 |
dpavlin |
536 |
join_with(" ", |
385 |
|
|
rec('702','a'), |
386 |
|
|
rec('702','b') |
387 |
|
|
) |
388 |
|
|
); |
389 |
|
|
}); |
390 |
|
|
|
391 |
|
|
test_s(qq{ |
392 |
|
|
display('Publisher', |
393 |
|
|
rec('210','c') |
394 |
|
|
) |
395 |
|
|
}); |
396 |
|
|
|
397 |
|
|
test_s(qq{ |
398 |
|
|
search('Year', |
399 |
|
|
regex( 's/[^\\d]+//', |
400 |
|
|
rec('210','d') |
401 |
|
|
) |
402 |
|
|
) |
403 |
|
|
}); |
404 |
|
|
|
405 |
dpavlin |
538 |
ok(my $ds = _get_ds(), "get_ds"); |
406 |
dpavlin |
550 |
diag "ds = ", dump($ds) if ($debug); |
407 |
dpavlin |
536 |
|
408 |
|
|
|
409 |
|
|
sub test_check_ds { |
410 |
|
|
|
411 |
|
|
my $t = shift; |
412 |
|
|
|
413 |
dpavlin |
538 |
ok($ds = _get_ds(), 'get_ds'); |
414 |
dpavlin |
550 |
diag dump( $ds ) if ($debug); |
415 |
dpavlin |
536 |
|
416 |
|
|
ok( $ds && $ds->{something}, 'get_ds->something exists' ); |
417 |
|
|
ok( $ds && $ds->{something}->{$t}, 'get_ds->something->'.$t.' exists') if ($t); |
418 |
|
|
ok( $ds && !$ds->{empty}, 'get_ds->empty doesn\'t' ); |
419 |
|
|
|
420 |
|
|
return $ds; |
421 |
|
|
} |
422 |
|
|
|
423 |
dpavlin |
538 |
_clean_ds(); |
424 |
dpavlin |
536 |
test_s(qq{ search('something', '42'); }); |
425 |
|
|
test_s(qq{ search('empty', ''); }); |
426 |
|
|
test_check_ds('search'); |
427 |
|
|
|
428 |
dpavlin |
538 |
_clean_ds(); |
429 |
dpavlin |
536 |
test_s(qq{ display('something', '42'); }); |
430 |
|
|
test_s(qq{ display('empty', ''); }); |
431 |
|
|
test_check_ds('display'); |
432 |
|
|
|
433 |
dpavlin |
538 |
_clean_ds(); |
434 |
dpavlin |
912 |
test_s(qq{ search_display('something', '42'); }); |
435 |
|
|
test_s(qq{ search_display('empty', ''); }); |
436 |
dpavlin |
536 |
test_check_ds('search'); |
437 |
|
|
test_check_ds('display'); |
438 |
|
|
|
439 |
dpavlin |
538 |
_clean_ds(); |
440 |
dpavlin |
536 |
my $n = read_file( "$abs_path/data/normalize.pl" ); |
441 |
|
|
$n .= "\n1;\n"; |
442 |
|
|
#diag "normalize code:\n$n\n"; |
443 |
|
|
test_s( $n ); |
444 |
|
|
|
445 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
446 |
dpavlin |
550 |
diag "ds = ", dump($ds) if ($debug); |
447 |
dpavlin |
536 |
|
448 |
|
|
my $rec = { |
449 |
|
|
'200' => [{ |
450 |
|
|
'a' => '200a', |
451 |
|
|
'b' => '200b', |
452 |
|
|
}], |
453 |
|
|
}; |
454 |
|
|
my $rules = qq{ search('mixed', rec('200') ) }; |
455 |
|
|
|
456 |
dpavlin |
538 |
_clean_ds(); |
457 |
|
|
_set_rec( $rec ); |
458 |
dpavlin |
536 |
test_s( $rules ); |
459 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
460 |
dpavlin |
536 |
is_deeply( $ds, { |
461 |
|
|
'mixed' => { |
462 |
|
|
'search' => [ '200a', '200b' ], |
463 |
|
|
} |
464 |
|
|
}, 'correct get_ds'); |
465 |
|
|
|
466 |
|
|
ok(my $ds2 = WebPAC::Normalize::data_structure( |
467 |
|
|
row => $rec, |
468 |
|
|
rules => $rules, |
469 |
|
|
), 'data_structure'); |
470 |
|
|
is_deeply( $ds, $ds2, 'data_structure(s) same'); |
471 |
|
|
|
472 |
|
|
# wird and non-valid structure which is supported anyway |
473 |
dpavlin |
538 |
_clean_ds(); |
474 |
|
|
_set_rec({ |
475 |
dpavlin |
536 |
'200' => [{ |
476 |
|
|
'a' => '200a', |
477 |
|
|
}, |
478 |
|
|
'200-solo' |
479 |
|
|
] |
480 |
|
|
}); |
481 |
|
|
test_s(qq{ search('mixed', rec('200') ) }); |
482 |
dpavlin |
538 |
ok($ds = _get_ds(), "get_ds"); |
483 |
dpavlin |
536 |
is_deeply( $ds, { |
484 |
|
|
'mixed' => { |
485 |
|
|
'search' => [ '200a', '200-solo' ], |
486 |
|
|
} |
487 |
|
|
}, 'correct get_ds'); |
488 |
|
|
|
489 |
dpavlin |
555 |
# |
490 |
dpavlin |
540 |
# MARC |
491 |
dpavlin |
555 |
# |
492 |
dpavlin |
725 |
#_debug( 4 ); |
493 |
dpavlin |
555 |
|
494 |
dpavlin |
547 |
test_s(qq{ marc_indicators('900',1,2) }); |
495 |
|
|
test_s(qq{ marc('900','a', rec('200') ) }); |
496 |
dpavlin |
579 |
my $marc; |
497 |
|
|
ok($marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
498 |
|
|
diag dump( $marc ) if ($debug); |
499 |
dpavlin |
547 |
|
500 |
dpavlin |
579 |
is_deeply( $marc, [ |
501 |
dpavlin |
547 |
[ '900', 1, 2, 'a', '200a' ], |
502 |
|
|
[ '900', 1, 2, 'a', '200-solo' ] |
503 |
|
|
], 'correct marc with indicators'); |
504 |
|
|
|
505 |
|
|
test_s(qq{ marc_indicators('900',' ',9) }); |
506 |
|
|
test_s(qq{ marc_repeatable_subfield('900','a', rec('200') ) }); |
507 |
|
|
|
508 |
dpavlin |
579 |
ok($marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
509 |
|
|
diag dump( $marc ) if ($debug); |
510 |
dpavlin |
547 |
|
511 |
dpavlin |
579 |
is_deeply( $marc, [ |
512 |
dpavlin |
547 |
[ '900', 1, 2, 'a', '200a', 'a', '200-solo' ], |
513 |
|
|
[ '900', ' ', 9, 'a', '200a', 'a', '200-solo' ] |
514 |
|
|
], 'correct marc with repetable subfield'); |
515 |
dpavlin |
550 |
|
516 |
dpavlin |
555 |
# |
517 |
|
|
# test magic re-ordering of input data |
518 |
|
|
# |
519 |
dpavlin |
550 |
|
520 |
dpavlin |
555 |
sub test_rec_rules { |
521 |
|
|
my ($msg, $rec, $rules, $struct) = @_; |
522 |
dpavlin |
550 |
|
523 |
dpavlin |
555 |
_clean_ds(); |
524 |
|
|
_set_rec($rec); |
525 |
|
|
|
526 |
|
|
foreach my $r (split(/;/, $rules)) { |
527 |
|
|
$r =~ s/[\s\n\r]+/ /gs; |
528 |
|
|
$r =~ s/^\s+//gs; |
529 |
|
|
$r =~ s/\s+$//gs; |
530 |
|
|
test_s($r) if ($r); |
531 |
|
|
} |
532 |
|
|
|
533 |
dpavlin |
579 |
ok(my $marc = WebPAC::Normalize::_get_marc_fields(), "_get_marc_fields"); |
534 |
|
|
diag dump( $marc ) if ($debug); |
535 |
dpavlin |
562 |
diag "expects:\n", dump($struct) if ($debug > 1); |
536 |
dpavlin |
579 |
is_deeply( $marc, $struct, $msg ); |
537 |
dpavlin |
555 |
} |
538 |
|
|
|
539 |
|
|
test_rec_rules( |
540 |
|
|
'correct marc with repetable subfield', |
541 |
|
|
{ |
542 |
|
|
'200' => [{ |
543 |
|
|
'a' => '200a-1', |
544 |
|
|
'b' => '200b-1', |
545 |
|
|
'c' => '200c-1', |
546 |
|
|
}, { |
547 |
|
|
'a' => '200a-2', |
548 |
|
|
'b' => '200b-2', |
549 |
|
|
}, { |
550 |
|
|
'a' => '200a-3', |
551 |
|
|
}], |
552 |
|
|
}, |
553 |
|
|
qq{ |
554 |
|
|
marc_indicators('900',1 ,0); |
555 |
|
|
marc('900','a', rec('200','a') ); |
556 |
|
|
marc('900','b', rec('200','b') ); |
557 |
|
|
marc('900','c', rec('200','c') ); |
558 |
|
|
}, |
559 |
|
|
[ |
560 |
|
|
[ '900', 1, 0, 'a', '200a-1', 'b', '200b-1', 'c', '200c-1' ], |
561 |
|
|
[ '900', 1, 0, 'a', '200a-2', 'b', '200b-2' ], |
562 |
|
|
[ '900', 1, 0, 'a', '200a-3' ], |
563 |
|
|
], |
564 |
|
|
); |
565 |
|
|
|
566 |
|
|
|
567 |
|
|
test_rec_rules( |
568 |
|
|
'marc_repeatable_subfield', |
569 |
|
|
{ |
570 |
|
|
'200' => [{ |
571 |
|
|
'a' => '200a-1', |
572 |
|
|
'b' => '200b-1', |
573 |
|
|
'c' => '200c-1', |
574 |
|
|
}, { |
575 |
|
|
'a' => '200a-2', |
576 |
|
|
'b' => '200b-2', |
577 |
|
|
'c' => '200c-2', |
578 |
|
|
}, { |
579 |
|
|
'a' => '200a-3', |
580 |
|
|
'c' => '200c-3', |
581 |
|
|
}], |
582 |
|
|
}, |
583 |
|
|
qq{ |
584 |
|
|
marc_indicators('900',1 ,0); |
585 |
|
|
marc_repeatable_subfield('900','a', rec('200','a') ); |
586 |
|
|
marc('900','b', rec('200','b') ); |
587 |
|
|
marc('900','c', rec('200','c') ); |
588 |
|
|
}, |
589 |
|
|
[ |
590 |
|
|
[ '900', 1, 0, 'a', '200a-1', 'a', '200a-2', 'a', '200a-3', 'b', '200b-1', 'c', '200c-1' ], |
591 |
|
|
[ '900', 1, 0, 'b', '200b-2', 'c', '200c-2' ], |
592 |
|
|
[ '900', 1, 0, 'c', '200c-3' ], |
593 |
|
|
], |
594 |
|
|
); |
595 |
dpavlin |
562 |
|
596 |
|
|
test_rec_rules( |
597 |
|
|
'marc_compose', |
598 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
599 |
|
|
qq{ |
600 |
|
|
marc_compose('900', |
601 |
|
|
'c', rec(200,'b'), |
602 |
|
|
'b', rec(200,'a'), |
603 |
|
|
'a', rec(200,'c'), |
604 |
|
|
); |
605 |
|
|
}, |
606 |
|
|
[ |
607 |
|
|
[ '900', ' ', ' ', 'c', 42, 'b', 'foo ; bar', 'a', 'baz' ] |
608 |
|
|
], |
609 |
|
|
); |
610 |
|
|
|
611 |
dpavlin |
603 |
test_rec_rules( |
612 |
|
|
'marc_compose with + subfields', |
613 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
614 |
|
|
qq{ |
615 |
|
|
marc_compose('900', |
616 |
|
|
'a', rec(200,'a'), |
617 |
|
|
'+', prefix(" * ", rec(200,'c')), |
618 |
|
|
'b', rec(200,'b'), |
619 |
|
|
'+', prefix(" : ", rec(200,'c')), |
620 |
|
|
); |
621 |
|
|
}, |
622 |
|
|
[ |
623 |
|
|
[ '900', ' ', ' ', 'a', 'foo ; bar * baz', 'b', '42 : baz' ] |
624 |
|
|
], |
625 |
|
|
); |
626 |
|
|
|
627 |
dpavlin |
562 |
# |
628 |
|
|
# test rules |
629 |
|
|
# |
630 |
|
|
sub test_rule { |
631 |
|
|
my ($msg, $rec, $rule, $struct) = @_; |
632 |
|
|
_clean_ds(); |
633 |
|
|
_set_rec( $rec ); |
634 |
|
|
$rule =~ s/\\/\\/gs; |
635 |
|
|
my $r = test_s( $rule ); |
636 |
|
|
diag "for ", dump($rec), " got:\n", dump($r), "\nexpect:\n" if ($debug > 1); |
637 |
|
|
diag dump($struct) if ($debug); |
638 |
|
|
is_deeply( $r, $struct, $msg ); |
639 |
|
|
} |
640 |
|
|
|
641 |
|
|
# test split_rec_on |
642 |
|
|
test_rule( |
643 |
|
|
'split_rec_on', |
644 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
645 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 1) }, |
646 |
dpavlin |
571 |
[ 'foo' ], |
647 |
dpavlin |
562 |
); |
648 |
|
|
test_rule( |
649 |
|
|
'split_rec_on', |
650 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
651 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/, 2) }, |
652 |
dpavlin |
571 |
[ 'bar' ], |
653 |
dpavlin |
562 |
); |
654 |
dpavlin |
566 |
test_rule( |
655 |
|
|
'split_rec_on no part', |
656 |
|
|
{ '200' => [{ a => 'foo ; bar', b => 42, c => 'baz' }] }, |
657 |
dpavlin |
571 |
qq{ split_rec_on('200','a', qr/\\s*;\\s*/) }, |
658 |
dpavlin |
566 |
[ 'foo', 'bar' ], |
659 |
|
|
); |
660 |
|
|
test_rule( |
661 |
|
|
'split_rec_on no record', |
662 |
|
|
{}, |
663 |
|
|
qq{ split_rec_on('200','a', qr/\\s*;\\s*/) }, |
664 |
dpavlin |
571 |
[ '' ], |
665 |
dpavlin |
566 |
); |
666 |
dpavlin |
562 |
|
667 |
|
|
test_rec_rules( |
668 |
|
|
'marc_compose+split_rec_on', |
669 |
|
|
{ '200' => [{ a => 'foo ! bar', b => 42, c => 'baz' }] }, |
670 |
|
|
qq{ |
671 |
|
|
marc_compose('900', |
672 |
|
|
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 1), |
673 |
|
|
'c', rec(200,'c'), |
674 |
|
|
'a', split_rec_on(200,'a', qr/\\s*!\\s*/, 2), |
675 |
|
|
'b', rec(200,'b'), |
676 |
|
|
); |
677 |
|
|
}, |
678 |
|
|
[ |
679 |
|
|
[ '900', ' ', ' ', |
680 |
|
|
'a', 'foo', |
681 |
|
|
'c', 'baz', |
682 |
|
|
'a', 'bar', |
683 |
|
|
'b', 42, |
684 |
|
|
] |
685 |
|
|
], |
686 |
|
|
); |
687 |
dpavlin |
564 |
|
688 |
|
|
cmp_ok(marc_leader('06',42), '==', 42, 'marc_leader'); |
689 |
|
|
cmp_ok(marc_leader('11',5), '==', 5, 'marc_leader'); |
690 |
|
|
ok(marc_leader(), 'marc_leader get'); |
691 |
|
|
diag "leader: ", dump(marc_leader()) if ($debug); |
692 |
|
|
is_deeply(marc_leader(), { '06' => 42, 11 => 5 }, "marc_leader full"); |
693 |
dpavlin |
571 |
|
694 |
|
|
test_rule( |
695 |
|
|
'rec1(000)', |
696 |
|
|
{ '000' => [ 42 ]}, |
697 |
|
|
qq{ rec('000') }, |
698 |
|
|
[ 42 ], |
699 |
|
|
); |
700 |
|
|
|
701 |
|
|
test_rec_rules( |
702 |
dpavlin |
574 |
'marc(001,rec(000))', |
703 |
dpavlin |
571 |
{ '000' => [ 42 ]}, |
704 |
|
|
qq{ |
705 |
|
|
marc('001', rec('000') ); |
706 |
|
|
}, |
707 |
|
|
[ |
708 |
dpavlin |
574 |
[ '001', 42, ] |
709 |
dpavlin |
571 |
], |
710 |
|
|
); |
711 |
dpavlin |
574 |
|
712 |
|
|
test_rec_rules( |
713 |
dpavlin |
579 |
'marc_remove subfield', |
714 |
dpavlin |
574 |
{ '200' => [{ a => 42, b => 'bar', c => 'baz' }] }, |
715 |
|
|
qq{ |
716 |
|
|
marc('900', 'a', rec('200','a') ); |
717 |
|
|
marc('900', 'b', rec('200','b') ); |
718 |
|
|
marc_remove('900','b'); |
719 |
|
|
marc('900', 'b', rec('200','c') ); |
720 |
|
|
marc_remove('900','a'); |
721 |
|
|
}, |
722 |
|
|
[ |
723 |
|
|
[ '900', ' ', ' ', 'b', 'baz' ], |
724 |
|
|
], |
725 |
|
|
); |
726 |
|
|
|
727 |
|
|
test_rec_rules( |
728 |
dpavlin |
579 |
'marc_remove field', |
729 |
|
|
{ '200' => [{ a => 42, b => 'bar', c => 'baz' }] }, |
730 |
|
|
qq{ |
731 |
|
|
marc('900', 'a', rec('200','a') ); |
732 |
|
|
marc('900', 'b', rec('200','b') ); |
733 |
|
|
marc('901', 'b', rec('200','b') ); |
734 |
|
|
marc('901', 'c', rec('200','c') ); |
735 |
|
|
marc_remove('900'); |
736 |
|
|
}, |
737 |
|
|
[ |
738 |
|
|
[ '901', ' ', ' ', 'b', 'bar', 'c', 'baz' ], |
739 |
|
|
], |
740 |
|
|
); |
741 |
dpavlin |
787 |
|
742 |
|
|
test_s(qq{ marc_remove('*'); }); |
743 |
|
|
ok(! WebPAC::Normalize::_get_marc_fields(), 'marc_remove(*)'); |
744 |
|
|
|
745 |
dpavlin |
579 |
test_rec_rules( |
746 |
dpavlin |
574 |
'marc_duplicate', |
747 |
|
|
{ '200' => [{ a => 42, b => 'bar', c => 'baz', d => 'bing', e => 'bong' }] }, |
748 |
|
|
qq{ |
749 |
dpavlin |
813 |
marc_leader('06',42); |
750 |
|
|
marc_leader('11',0); |
751 |
dpavlin |
574 |
marc('900', 'a', rec('200','a') ); |
752 |
|
|
marc('900', 'b', rec('200','b') ); |
753 |
|
|
marc_duplicate; |
754 |
dpavlin |
813 |
marc_leader('11',1); |
755 |
dpavlin |
574 |
marc_remove('900','b'); |
756 |
|
|
marc('900', 'b', rec('200','c') ); |
757 |
|
|
marc_duplicate; |
758 |
dpavlin |
813 |
marc_leader('11',2); |
759 |
dpavlin |
574 |
marc_remove('900','b'); |
760 |
|
|
marc('900', 'b', rec('200','d') ); |
761 |
|
|
marc_duplicate; |
762 |
dpavlin |
813 |
marc_leader('11',3); |
763 |
dpavlin |
574 |
marc_remove('900','b'); |
764 |
|
|
marc('900', 'b', rec('200','e') ); |
765 |
|
|
}, |
766 |
|
|
[ |
767 |
|
|
# this will return FIRST record |
768 |
|
|
[ '900', ' ', ' ', 'a', 42, 'b', 'bar' ], |
769 |
|
|
], |
770 |
|
|
); |
771 |
|
|
|
772 |
dpavlin |
813 |
cmp_ok( marc_count(), '==', 3, 'marc_count' ); |
773 |
|
|
|
774 |
dpavlin |
574 |
my $i = 0; |
775 |
|
|
foreach my $v ( qw/bar baz bing bong/ ) { |
776 |
|
|
|
777 |
dpavlin |
579 |
ok($marc = WebPAC::Normalize::_get_marc_fields( offset => $i ), |
778 |
dpavlin |
574 |
"_get_marc_fields( offset => $i )" |
779 |
|
|
); |
780 |
dpavlin |
579 |
diag "marc $i = ", dump( $marc ) if ($debug); |
781 |
|
|
is_deeply( $marc, |
782 |
dpavlin |
574 |
[ [ '900', ' ', ' ', 'a', 42, 'b', $v ] ], |
783 |
|
|
"MARC copy $i has $v", |
784 |
|
|
); |
785 |
dpavlin |
813 |
is_deeply(WebPAC::Normalize::_get_marc_leader(), { '06' => 42, 11 => $i }, "_get_marc_leader copy $i"); |
786 |
dpavlin |
574 |
$i++; |
787 |
|
|
} |
788 |
dpavlin |
604 |
|
789 |
|
|
test_rec_rules( |
790 |
|
|
'marc_original_order', |
791 |
|
|
{ |
792 |
|
|
'200' => [ { |
793 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
794 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
795 |
|
|
}, { |
796 |
|
|
a => [ 'a3', 'a4', 'a5' ], b => 'b3', c => 'c3', |
797 |
|
|
subfields => [ qw/a 0 a 1 b 0 c 0 a 2/ ], |
798 |
|
|
} ], |
799 |
|
|
}, |
800 |
|
|
qq{ |
801 |
dpavlin |
616 |
marc_original_order(900,200); |
802 |
dpavlin |
604 |
}, |
803 |
|
|
[ |
804 |
|
|
[ '900', ' ', ' ', 'a', 'a1', 'b', 'b1', 'a', 'a2', 'b', 'b2', 'c', 'c1', 'c', 'c2', ], |
805 |
|
|
[ '900', ' ', ' ', 'a', 'a3', 'a', 'a4', 'b', 'b3', 'c', 'c3', 'a', 'a5', ], |
806 |
|
|
], |
807 |
|
|
); |
808 |
dpavlin |
631 |
|
809 |
|
|
test_rule( |
810 |
|
|
'rec1 skips subfields', |
811 |
|
|
{ |
812 |
|
|
'200' => [ { |
813 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
814 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
815 |
|
|
}, { |
816 |
|
|
a => [ 'a3', 'a4', 'a5' ], b => 'b3', c => 'c3', |
817 |
|
|
subfields => [ qw/a 0 a 1 b 0 c 0 a 2/ ], |
818 |
|
|
} ], |
819 |
|
|
}, |
820 |
|
|
qq{ |
821 |
|
|
rec1(200); |
822 |
|
|
}, |
823 |
dpavlin |
641 |
['a1', 'b1', 'a2', 'b2', 'c1', 'c2', 'a3', 'a4', 'b3', 'c3', 'a5' ], |
824 |
dpavlin |
631 |
); |
825 |
dpavlin |
641 |
|
826 |
|
|
is_deeply( |
827 |
|
|
[ _pack_subfields_hash({ |
828 |
|
|
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
829 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
830 |
|
|
}) ], |
831 |
|
|
['a1', 'b1', 'a2', 'b2', 'c1', 'c2'], |
832 |
|
|
'_pack_subfields_hash( $h )' |
833 |
|
|
); |
834 |
|
|
|
835 |
dpavlin |
669 |
cmp_ok( |
836 |
|
|
_pack_subfields_hash({ |
837 |
dpavlin |
641 |
a => [ 'a1', 'a2' ], b => [ 'b1', 'b2' ], c => [ 'c1', 'c2' ], |
838 |
|
|
subfields => [ qw/a 0 b 0 a 1 b 1 c 0 c 1/ ], |
839 |
dpavlin |
669 |
}, 1), |
840 |
|
|
'eq', |
841 |
|
|
'^aa1^bb1^aa2^bb2^cc1^cc2', |
842 |
dpavlin |
641 |
'_pack_subfields_hash( $h, 1 )' |
843 |
|
|
); |
844 |
dpavlin |
815 |
|
845 |
|
|
_clean_ds(); |
846 |
|
|
test_s(qq{ |
847 |
|
|
marc_fixed('008', 0, 'abcdef'); |
848 |
|
|
marc_fixed('000', 5, '5'); |
849 |
|
|
marc_fixed('000', 10, 'A'); |
850 |
|
|
marc_fixed('000', 0, '0'); |
851 |
|
|
}); |
852 |
|
|
ok( my $m = WebPAC::Normalize::_get_marc_fields(), '_get_marc_fields'); |
853 |
|
|
diag dump( $m ); |
854 |
|
|
is_deeply( WebPAC::Normalize::_get_marc_fields(), |
855 |
|
|
[ |
856 |
|
|
["008", "abcdef"], |
857 |
|
|
# 0....5....10 |
858 |
|
|
["000", "0 5 A"] |
859 |
|
|
] |
860 |
|
|
); |
861 |
dpavlin |
536 |
} |
862 |
|
|
|