|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace EWW\Dpf\Services\ImportExternalMetadata; |
|
4
|
|
|
|
|
5
|
|
|
use Symfony\Component\Serializer\Encoder\XmlEncoder; |
|
6
|
|
|
use EWW\Dpf\Domain\Model\BibTexMetadata; |
|
7
|
|
|
use EWW\Dpf\Domain\Model\ExternalMetadata; |
|
8
|
|
|
|
|
9
|
|
|
use RenanBr\BibTexParser\Listener; |
|
10
|
|
|
use RenanBr\BibTexParser\Parser; |
|
11
|
|
|
use RenanBr\BibTexParser\Processor; |
|
12
|
|
|
|
|
13
|
|
|
class BibTexFileImporter extends AbstractImporter implements FileImporter |
|
14
|
|
|
{ |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @var array |
|
18
|
|
|
*/ |
|
19
|
|
|
protected $mandatoryErrors = []; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Returns the list of all publication types |
|
23
|
|
|
* |
|
24
|
|
|
* @return array |
|
25
|
|
|
*/ |
|
26
|
|
|
public static function types() |
|
27
|
|
|
{ |
|
28
|
|
|
return [ |
|
29
|
|
|
'article', |
|
30
|
|
|
'book', |
|
31
|
|
|
'booklet', |
|
32
|
|
|
'inbook', |
|
33
|
|
|
'incollection', |
|
34
|
|
|
'inproceedings', |
|
35
|
|
|
'manual', |
|
36
|
|
|
'mastersthesis', |
|
37
|
|
|
'misc', |
|
38
|
|
|
'phdthesis', |
|
39
|
|
|
'proceedings', |
|
40
|
|
|
'techreport', |
|
41
|
|
|
'unpublished' |
|
42
|
|
|
]; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @param string $file Can be a path to a file or a string with the file content |
|
47
|
|
|
* @param bool $contentOnly Determines if $file is a path or content as a string |
|
48
|
|
|
* @return array |
|
49
|
|
|
* @throws \ErrorException |
|
50
|
|
|
* @throws \RenanBr\BibTexParser\Exception\ParserException |
|
51
|
|
|
*/ |
|
52
|
|
|
protected function parseFile($file, $contentOnly = false) |
|
53
|
|
|
{ |
|
54
|
|
|
//$data = json_decode($response->__toString(),true); |
|
55
|
|
|
//$encoder = new XmlEncoder(); |
|
56
|
|
|
$listener = new Listener(); |
|
57
|
|
|
$listener->addProcessor(new Processor\TagNameCaseProcessor(CASE_LOWER)); |
|
58
|
|
|
$parser = new Parser(); |
|
59
|
|
|
$parser->addListener($listener); |
|
60
|
|
|
|
|
61
|
|
|
if ($contentOnly) { |
|
62
|
|
|
$parser->parseString($file); |
|
63
|
|
|
} else { |
|
64
|
|
|
$parser->parseFile($file); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
$entries = $listener->export(); |
|
68
|
|
|
|
|
69
|
|
|
foreach ($entries as $index => $fields) { |
|
70
|
|
|
foreach ($fields as $key => $field) { |
|
71
|
|
|
$field = str_replace("{\\\"o}", "ö", $field); |
|
72
|
|
|
$field = str_replace("{\\\"O}", "Ö", $field); |
|
73
|
|
|
$field = str_replace("{\\\"a}", "ä", $field); |
|
74
|
|
|
$field = str_replace("{\\\"A}", "Ä", $field); |
|
75
|
|
|
$field = str_replace("{\\\"u}", "ü", $field); |
|
76
|
|
|
$field = str_replace("{\\\"U}", "Ü", $field); |
|
77
|
|
|
$field = str_replace("\\textendash", "-", $field); |
|
78
|
|
|
$field = str_replace("\\textemdash", "—", $field); |
|
79
|
|
|
$field = str_replace("\\textquoteright", "'", $field); |
|
80
|
|
|
$field = str_replace("\\textquoteleft", "'", $field); |
|
81
|
|
|
$field = str_replace("{\\ss}", "ß", $field); |
|
82
|
|
|
$field = preg_replace("/{\s*[\\\]textunderscore\s*}/", '_', $field); |
|
83
|
|
|
$field = preg_replace("/{\s*[\\\]textquotedbl\s*}/", '"', $field); |
|
84
|
|
|
$entries[$index][$key] = preg_replace("/{\s*[\\\]&\s*}/", '&', $field); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
return $entries; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @param string $filePath |
|
93
|
|
|
* @param string $mandatoryFieldSettings |
|
94
|
|
|
* @param bool $contentOnly Determines if $file is a path or content as a string |
|
95
|
|
|
* @return array |
|
96
|
|
|
*/ |
|
97
|
|
|
public function loadFile($filePath, $mandatoryFieldSettings, $contentOnly = false) |
|
98
|
|
|
{ |
|
99
|
|
|
$results = []; |
|
100
|
|
|
$mandatoryErrors = []; |
|
101
|
|
|
$mandatoryFieldErrors = []; |
|
|
|
|
|
|
102
|
|
|
|
|
103
|
|
|
$mandatoryFields = array_map( |
|
104
|
|
|
'trim', |
|
105
|
|
|
explode(',', $mandatoryFieldSettings) |
|
106
|
|
|
); |
|
107
|
|
|
|
|
108
|
|
|
foreach ($mandatoryFields as $key => $value) { |
|
109
|
|
|
$orFields = array_map( |
|
110
|
|
|
'trim', |
|
111
|
|
|
explode('|', $value) |
|
112
|
|
|
); |
|
113
|
|
|
$mandatoryFields[$key] = $orFields; |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
if ($contentOnly) { |
|
117
|
|
|
$bibTexEntries = $this->parseFile($filePath, $contentOnly); |
|
118
|
|
|
} else { |
|
119
|
|
|
$bibTexEntries = $this->parseFile($filePath); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
$encoder = new XmlEncoder(); |
|
123
|
|
|
|
|
124
|
|
|
foreach ($bibTexEntries as $index => $bibTexItem) { |
|
125
|
|
|
|
|
126
|
|
|
$mandatoryFieldErrors = []; |
|
127
|
|
|
foreach ($mandatoryFields as $combinedMandatoryField) { |
|
128
|
|
|
|
|
129
|
|
|
$mandatoryOk = false; |
|
130
|
|
|
foreach ($combinedMandatoryField as $key => $value) { |
|
131
|
|
|
$mandatoryOk = $mandatoryOk || ( |
|
132
|
|
|
array_key_exists($value, $bibTexItem) |
|
133
|
|
|
&& $bibTexItem[$value] |
|
134
|
|
|
); |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
if (!$mandatoryOk) { |
|
138
|
|
|
$mandatoryFieldErrors[implode('|', $combinedMandatoryField)] = implode( |
|
139
|
|
|
'|', |
|
140
|
|
|
$combinedMandatoryField |
|
141
|
|
|
); |
|
142
|
|
|
$mandatoryErrors[$index] = [ |
|
143
|
|
|
'index' => $index + 1, |
|
144
|
|
|
'title' => $bibTexItem['title'], |
|
145
|
|
|
'fields' => $mandatoryFieldErrors |
|
146
|
|
|
]; |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
if (!$mandatoryErrors[$index]) { |
|
151
|
|
|
|
|
152
|
|
|
$bibTexData = $bibTexItem; |
|
153
|
|
|
|
|
154
|
|
|
if (array_key_exists('author', $bibTexItem)) { |
|
155
|
|
|
$bibTexData['author'] = $this->splitPersons($bibTexItem['author']); |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
if (array_key_exists('editor', $bibTexItem)) { |
|
159
|
|
|
$bibTexData['editor'] = $this->splitPersons($bibTexItem['editor']); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** @var BibTexMetadata $bibTexMetadata */ |
|
163
|
|
|
$bibTexMetadata = $this->objectManager->get(BibTexMetadata::class); |
|
164
|
|
|
|
|
165
|
|
|
$bibTexMetadata->setSource(get_class($this)); |
|
166
|
|
|
$bibTexMetadata->setFeUser($this->security->getUser()->getUid()); |
|
167
|
|
|
$bibTexMetadata->setData($encoder->encode($bibTexData, 'xml')); |
|
168
|
|
|
|
|
169
|
|
|
$results[] = $bibTexMetadata; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
$this->mandatoryErrors = $mandatoryErrors; |
|
175
|
|
|
|
|
176
|
|
|
return $results; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
/** |
|
180
|
|
|
* @return array |
|
181
|
|
|
*/ |
|
182
|
|
|
public function getMandatoryErrors() |
|
183
|
|
|
{ |
|
184
|
|
|
return $this->mandatoryErrors; |
|
185
|
|
|
} |
|
186
|
|
|
|
|
187
|
|
|
/** |
|
188
|
|
|
* @return bool |
|
189
|
|
|
*/ |
|
190
|
|
|
public function hasMandatoryErrors() |
|
191
|
|
|
{ |
|
192
|
|
|
return !empty($this->mandatoryErrors); |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
/** |
|
196
|
|
|
* @return \EWW\Dpf\Domain\Model\TransformationFile|void |
|
197
|
|
|
*/ |
|
198
|
|
|
protected function getDefaultXsltTransformation() |
|
199
|
|
|
{ |
|
200
|
|
|
/** @var \EWW\Dpf\Domain\Model\Client $client */ |
|
201
|
|
|
$client = $this->clientRepository->findAll()->current(); |
|
202
|
|
|
|
|
203
|
|
|
/** @var \EWW\Dpf\Domain\Model\TransformationFile $xsltTransformationFile */ |
|
204
|
|
|
return $client->getBibTexTransformation()->current(); |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* @return string|void |
|
209
|
|
|
*/ |
|
210
|
|
|
protected function getDefaultXsltFilePath() |
|
211
|
|
|
{ |
|
212
|
|
|
return \TYPO3\CMS\Core\Utility\GeneralUtility::getFileAbsFileName( |
|
213
|
|
|
'EXT:dpf/Resources/Private/Xslt/bibtex-default.xsl' |
|
214
|
|
|
); |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
/** |
|
218
|
|
|
* @return string|void |
|
219
|
|
|
*/ |
|
220
|
|
|
protected function getImporterName() |
|
221
|
|
|
{ |
|
222
|
|
|
return 'bibtex'; |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
/** |
|
226
|
|
|
* @param string $persons |
|
227
|
|
|
*/ |
|
228
|
|
|
protected function splitPersons($persons) |
|
229
|
|
|
{ |
|
230
|
|
|
$results = []; |
|
231
|
|
|
|
|
232
|
|
|
$persons = array_map('trim', explode(' and ', $persons)); |
|
233
|
|
|
|
|
234
|
|
|
foreach ($persons as $person) { |
|
235
|
|
|
list($family, $given) = array_map('trim', explode(',', $person)); |
|
236
|
|
|
$results[] = [ |
|
237
|
|
|
'family' => $family, |
|
238
|
|
|
'given' => $given |
|
239
|
|
|
]; |
|
240
|
|
|
} |
|
241
|
|
|
|
|
242
|
|
|
return $results; |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
} |