|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of dispositif/wikibot application |
|
4
|
|
|
* 2019 : Philippe M. <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, please view the LICENSE file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Domain\Publisher; |
|
11
|
|
|
|
|
12
|
|
|
use App\Domain\Enums\Language; |
|
13
|
|
|
use SimpleXMLElement; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Class BnfMapper |
|
17
|
|
|
* http://catalogue.bnf.fr/api/SRU?version=1.2&operation=searchRetrieve&query=bib.isbn%2Badj%2B%25222844940404%2522. |
|
18
|
|
|
*/ |
|
19
|
|
|
class BnfMapper extends AbstractBookMapper implements MapperInterface |
|
20
|
|
|
{ |
|
21
|
|
|
/** |
|
22
|
|
|
* @var SimpleXMLElement |
|
23
|
|
|
*/ |
|
24
|
|
|
private $xml; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* XML in UniMarc format. |
|
28
|
|
|
* See http://api.bnf.fr/formats-bibliographiques-intermarc-unimarc |
|
29
|
|
|
* https://www.transition-bibliographique.fr/systemes-et-donnees/manuel-unimarc-format-bibliographique/. |
|
30
|
|
|
* |
|
31
|
|
|
* @param $xml |
|
32
|
|
|
* |
|
33
|
|
|
* @return array |
|
34
|
|
|
*/ |
|
35
|
|
|
public function process($xml): array |
|
36
|
|
|
{ |
|
37
|
|
|
if (!$xml instanceof SimpleXMLElement) { |
|
38
|
|
|
return []; |
|
39
|
|
|
} |
|
40
|
|
|
$this->xml = $xml; |
|
41
|
|
|
|
|
42
|
|
|
// skip multi-records |
|
43
|
|
|
$nbResults = (int)$xml->xpath('//srw:numberOfRecords[1]')[0] ?? 0; |
|
44
|
|
|
if (1 !== $nbResults) { |
|
45
|
|
|
echo "BNF : $nbResults records (skip)\n"; |
|
46
|
|
|
|
|
47
|
|
|
return []; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
return [ |
|
51
|
|
|
// Langue |
|
52
|
|
|
'langue' => $this->lang2wiki($this->xpath2string('//mxc:datafield[@tag="101"]/mxc:subfield[@code="a"][1]')), |
|
53
|
|
|
// c : Langue de l’œuvre originale |
|
54
|
|
|
'langue originale' => $this->stripLangFR( |
|
55
|
|
|
$this->lang2wiki( |
|
56
|
|
|
$this->xpath2string('//mxc:datafield[@tag="101"]/mxc:subfield[@code="c"][1]') |
|
57
|
|
|
) |
|
58
|
|
|
), |
|
59
|
|
|
// g : Langue du titre propre (si différent) |
|
60
|
|
|
'langue titre' => $this->stripLangFR( |
|
61
|
|
|
$this->lang2wiki( |
|
62
|
|
|
$this->xpath2string('//mxc:datafield[@tag="101"]/mxc:subfield[@code="g"][1]') |
|
63
|
|
|
) |
|
64
|
|
|
), |
|
65
|
|
|
/* |
|
66
|
|
|
* Bloc 200. |
|
67
|
|
|
* https://www.transition-bibliographique.fr/wp-content/uploads/2019/11/B200-2018.pdf |
|
68
|
|
|
*/ // a : Titre propre |
|
69
|
|
|
'titre' => $this->xpath2string('//mxc:datafield[@tag="200"]/mxc:subfield[@code="a"][1]'), |
|
70
|
|
|
// d : Titre parralèle (autre langue) |
|
71
|
|
|
'titre original' => $this->xpath2string('//mxc:datafield[@tag="200"]/mxc:subfield[@code="d"][1]'), |
|
72
|
|
|
// e : Complément du titre |
|
73
|
|
|
'sous-titre' => $this->xpath2string('//mxc:datafield[@tag="200"]/mxc:subfield[@code="e"][1]', ', '), |
|
74
|
|
|
|
|
75
|
|
|
// Responsabilités : zone 200 trop merdique "Pierre Durand, Paul Dupond" ou "Paul Durand,..." |
|
76
|
|
|
'prénom1' => $this->xpath2string('//mxc:datafield[@tag="700"]/mxc:subfield[@code="b"]'), |
|
77
|
|
|
'nom1' => $this->xpath2string('//mxc:datafield[@tag="700"]/mxc:subfield[@code="a"]'), |
|
78
|
|
|
|
|
79
|
|
|
'prénom2' => $this->xpath2string('//mxc:datafield[@tag="701"]/mxc:subfield[@code="b"]'), |
|
80
|
|
|
'nom2' => $this->xpath2string('//mxc:datafield[@tag="701"]/mxc:subfield[@code="a"]'), |
|
81
|
|
|
|
|
82
|
|
|
// zone 200 |
|
83
|
|
|
// h : Numéro de partie |
|
84
|
|
|
// 'volume' => $this->xpath2string('//mxc:datafield[@tag="200"]/mxc:subfield[@code="h"]'), |
|
85
|
|
|
// i : Titre de partie |
|
86
|
|
|
// v : numéro de volume |
|
87
|
|
|
'volume' => $this->xpath2string('//mxc:datafield[@tag="200"]/mxc:subfield[@code="v"][1]'), |
|
88
|
|
|
|
|
89
|
|
|
// 410 : collection |
|
90
|
|
|
'collection' => $this->xpath2string('//mxc:datafield[@tag="410"]/mxc:subfield[@code="a"][1]'), |
|
91
|
|
|
|
|
92
|
|
|
// Auteur : voir plutôt 7XX |
|
93
|
|
|
// https://www.transition-bibliographique.fr/wp-content/uploads/2018/07/B7XX-6-2011.pdf |
|
94
|
|
|
|
|
95
|
|
|
// multi-zones |
|
96
|
|
|
'lieu' => $this->getLocation(), |
|
97
|
|
|
'éditeur' => $this->getPublisher(), |
|
98
|
|
|
'date' => $this->getPublishDate(), |
|
99
|
|
|
// 215 |
|
100
|
|
|
'pages totales' => $this->convertPages(), |
|
101
|
|
|
|
|
102
|
|
|
// 'bnf' => $this->convertBnfIdent(), // pertinent si isbn ? |
|
103
|
|
|
'isbn2' => $this->xpath2string('//mxc:datafield[@tag="010"]/mxc:subfield[@code="a"][2]'), |
|
104
|
|
|
'isbn' => $this->extractISBN(), |
|
105
|
|
|
|
|
106
|
|
|
// hidden data |
|
107
|
|
|
'infos' => [ |
|
108
|
|
|
'source' => 'BnF', |
|
109
|
|
|
'sourceTag' => $this->sourceTag(), |
|
110
|
|
|
'bnfAuteur1' => $this->xpath2string('//mxc:datafield[@tag="700"][1]/mxc:subfield[@code="3"][1]'), |
|
111
|
|
|
'ISNIAuteur1' => $this->formatISNI( |
|
112
|
|
|
$this->xpath2string('//mxc:datafield[@tag="700"][1]/mxc:subfield[@code="o"][1]') |
|
113
|
|
|
), |
|
114
|
|
|
'yearsAuteur1' => $this->xpath2string('//mxc:datafield[@tag="700"][1]/mxc:subfield[@code="f"][1]'), |
|
115
|
|
|
], |
|
116
|
|
|
]; |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
private function xpath2string(string $path, ?string $glue = ', '): ?string |
|
120
|
|
|
{ |
|
121
|
|
|
$elements = $this->xml->xpath($path); |
|
122
|
|
|
|
|
123
|
|
|
$res = []; |
|
124
|
|
|
foreach ($elements as $element) { |
|
125
|
|
|
if (isset($element) && $element instanceof SimpleXMLElement) { |
|
126
|
|
|
$res[] = (string)$element; |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
if (!empty($res)) { |
|
131
|
|
|
return implode($glue, $res); |
|
|
|
|
|
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
return null; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
private function extractISBN(): ?string |
|
138
|
|
|
{ |
|
139
|
|
|
$isbn = $this->xpath2string('//mxc:datafield[@tag="010"]/mxc:subfield[@code="a"][1]'); |
|
140
|
|
|
|
|
141
|
|
|
// data pourrie fréquente : "9789004232891, 9004232893" |
|
142
|
|
|
if (preg_match('#([0-9]{13})#', $isbn, $matches)) { |
|
143
|
|
|
return $matches[1]; |
|
144
|
|
|
} |
|
145
|
|
|
if (preg_match('#([0-9]{10})#', $isbn, $matches)) { |
|
146
|
|
|
return $matches[1]; |
|
147
|
|
|
} |
|
148
|
|
|
// ISBN avec tiret |
|
149
|
|
|
if (preg_match('#([0-9\-]{10,17})#', $isbn, $matches)) { |
|
150
|
|
|
return $matches[1]; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
return null; |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
private function formatISNI(?string $raw = null): ?string |
|
157
|
|
|
{ |
|
158
|
|
|
if (!$raw) { |
|
159
|
|
|
return null; |
|
160
|
|
|
} |
|
161
|
|
|
if (preg_match('#^0000(000[0-4])([0-9]{4})([0-9]{3}[0-9X])$#', $raw, $matches) > 0) { |
|
162
|
|
|
return $raw; |
|
163
|
|
|
} |
|
164
|
|
|
// BnF curious format of ISNI |
|
165
|
|
|
if (preg_match('#^ISNI0000(000[0-4])([0-9]{4})([0-9]{3}[0-9X])$#', $raw, $matches) > 0) { |
|
166
|
|
|
return sprintf('0000 %s %s %s', $matches[1], $matches[2], $matches[3]); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
return null; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
/** |
|
173
|
|
|
* Strip FR lang |
|
174
|
|
|
* |
|
175
|
|
|
* @param string|null $lang |
|
176
|
|
|
* |
|
177
|
|
|
* @return string|null |
|
178
|
|
|
*/ |
|
179
|
|
|
private function stripLangFR(?string $lang = null): ?string |
|
180
|
|
|
{ |
|
181
|
|
|
return ('fr' !== $lang) ? $lang : null; |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
/** |
|
185
|
|
|
* Convert number of pages. |
|
186
|
|
|
* "1 vol. (126 p.)". |
|
187
|
|
|
* |
|
188
|
|
|
* @return string|null |
|
189
|
|
|
*/ |
|
190
|
|
|
private function convertPages(): ?string |
|
191
|
|
|
{ |
|
192
|
|
|
$raw = $this->xpath2string('//mxc:datafield[@tag="215"]/mxc:subfield[@code="a"]'); |
|
193
|
|
|
if (!empty($raw) && preg_match('#([0-9]{2,}) p\.#', $raw, $matches) > 0) { |
|
194
|
|
|
return (string)$matches[1]; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
|
|
return null; |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
/** |
|
201
|
|
|
* todo gestion bilingue fr+en |
|
202
|
|
|
* ISO 639-1 http://www.loc.gov/standards/iso639-2/php/French_list.php. |
|
203
|
|
|
* |
|
204
|
|
|
* @param string|null $lang |
|
205
|
|
|
* |
|
206
|
|
|
* @return string|null |
|
207
|
|
|
*/ |
|
208
|
|
|
private function lang2wiki(?string $lang = null): ?string |
|
209
|
|
|
{ |
|
210
|
|
|
if (!empty($lang)) { |
|
211
|
|
|
return Language::iso2b2wiki($lang); |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
return null; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
private function getPublisher(): ?string |
|
218
|
|
|
{ |
|
219
|
|
|
// zone 210 |
|
220
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="210"]/mxc:subfield[@code="c"]', ' / ')) { |
|
221
|
|
|
return $tac; |
|
222
|
|
|
} |
|
223
|
|
|
// 214 : nouvelle zone 2019 |
|
224
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="214"]/mxc:subfield[@code="c"]', ' / ')) { |
|
225
|
|
|
return $tac; |
|
226
|
|
|
} |
|
227
|
|
|
|
|
228
|
|
|
// 219 ancienne zone ? |
|
229
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="219"]/mxc:subfield[@code="c"]', ' / ')) { |
|
230
|
|
|
return $tac; |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
return null; |
|
234
|
|
|
} |
|
235
|
|
|
|
|
236
|
|
|
private function getLocation(): ?string |
|
237
|
|
|
{ |
|
238
|
|
|
// zone 210 |
|
239
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="210"]/mxc:subfield[@code="a"]', '/')) { |
|
240
|
|
|
return $tac; |
|
241
|
|
|
} |
|
242
|
|
|
// 214 : nouvelle zone 2019 |
|
243
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="214"]/mxc:subfield[@code="a"]', '/')) { |
|
244
|
|
|
return $tac; |
|
245
|
|
|
} |
|
246
|
|
|
// ancienne zone ? |
|
247
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="219"]/mxc:subfield[@code="a"]', '/')) { |
|
248
|
|
|
return $tac; |
|
249
|
|
|
} |
|
250
|
|
|
|
|
251
|
|
|
return null; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
private function getPublishDate(): ?string |
|
255
|
|
|
{ |
|
256
|
|
|
// zone 210 d : Date de publication, de diffusion, etc. |
|
257
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="210"]/mxc:subfield[@code="d"][1]')) { |
|
258
|
|
|
return $tac; |
|
259
|
|
|
} |
|
260
|
|
|
// 214 : nouvelle zone 2019 |
|
261
|
|
|
if ($tac = $this->xpath2string('//mxc:datafield[@tag="214"]/mxc:subfield[@code="d"][1]')) { |
|
262
|
|
|
return $tac; |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
return null; |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
// private function convertBnfIdent(): ?string |
|
269
|
|
|
// { |
|
270
|
|
|
// // ark:/12148/cb453986124 |
|
271
|
|
|
// $raw = $this->xpath2string('//srw:recordIdentifier[1]/text()'); |
|
272
|
|
|
// |
|
273
|
|
|
// if ($raw && preg_match('#ark:/[0-9]+/cb([0-9]+)#', $raw, $matches) > 0) { |
|
274
|
|
|
// return (string)$matches[1]; |
|
275
|
|
|
// } |
|
276
|
|
|
// |
|
277
|
|
|
// return null; |
|
278
|
|
|
// } |
|
279
|
|
|
|
|
280
|
|
|
private function sourceTag(): ?string |
|
281
|
|
|
{ |
|
282
|
|
|
$raw = $this->xpath2string('//srw:extraRecordData[1]/ixm:attr[@name="LastModificationDate"][1]'); |
|
283
|
|
|
// 20190922 |
|
284
|
|
|
if ($raw && preg_match('#^([0-9]{4})[0-9]{4}$#', $raw, $matches) > 0) { |
|
285
|
|
|
return sprintf('BnF:%s', $matches[1]); |
|
286
|
|
|
} |
|
287
|
|
|
|
|
288
|
|
|
return null; |
|
289
|
|
|
} |
|
290
|
|
|
} |
|
291
|
|
|
|
Dispositif /
Wikibot
Loading history...