|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="ee5c6497-b984-411d-9890-69bed2ae38c5_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
|
|
4
|
<RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="TransformationRuleDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2020-09-19T18:24:53+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<CONFIGURATION>
|
|
11
|
<IMPORTED/>
|
|
12
|
<SCRIPT>
|
|
13
|
<TITLE>xslt_cleaning_datarepo_datacite_NAKALA</TITLE>
|
|
14
|
<CODE>
|
|
15
|
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:oaf="http://namespace.openaire.eu/oaf"
|
|
16
|
xmlns:oai="http://www.openarchives.org/OAI/2.0/"
|
|
17
|
xmlns:TransformationFunction="eu.dnetlib.data.collective.transformation.core.xsl.ext.TransformationFunctionProxy"
|
|
18
|
xmlns:dr="http://www.driver-repository.eu/namespace/dr" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
|
19
|
exclude-result-prefixes="TransformationFunction" extension-element-prefixes="TransformationFunction" version="1.1">
|
|
20
|
<xsl:param name="varOfficialName"/>
|
|
21
|
<xsl:param name="varDsType"/>
|
|
22
|
<xsl:param name="varDataSourceId"/>
|
|
23
|
<xsl:param name="index" select="0"/>
|
|
24
|
<xsl:param name="transDate" select="current-dateTime()"/>
|
|
25
|
<xsl:variable name="tf" select="TransformationFunction:getInstance()"/>
|
|
26
|
<xsl:template match="/">
|
|
27
|
<xsl:variable name="datasourcePrefix" select="normalize-space(//oaf:datasourceprefix)"/>
|
|
28
|
<xsl:call-template name="validRecord"/>
|
|
29
|
</xsl:template>
|
|
30
|
<xsl:template name="terminate">
|
|
31
|
<xsl:message terminate="yes">
|
|
32
|
record is not compliant, transformation is interrupted.
|
|
33
|
</xsl:message>
|
|
34
|
</xsl:template>
|
|
35
|
<xsl:template name="validRecord">
|
|
36
|
<record>
|
|
37
|
<xsl:apply-templates select="//*[local-name() = 'header']"/>
|
|
38
|
<metadata>
|
|
39
|
<xsl:apply-templates select="//*[local-name() = 'metadata']//*[local-name() = 'resource']"/><!-- for CoCoON many deleted records appeared among the transformed records -->
|
|
40
|
<xsl:if test="//oai:header/@status='deleted'">
|
|
41
|
<xsl:call-template name="terminate"/>
|
|
42
|
</xsl:if>
|
|
43
|
<xsl:for-each select="//*[local-name() = 'resource']/*[local-name()='identifier'][@identifierType='Handle'][not(. = '123456789')]">
|
|
44
|
<oaf:identifier>
|
|
45
|
<xsl:attribute name="identifierType">
|
|
46
|
<xsl:value-of select="'handle'"/>
|
|
47
|
</xsl:attribute>
|
|
48
|
<xsl:if test="contains(., '://hdl.handle.net/')">
|
|
49
|
<xsl:value-of select="substring-after(., '://hdl.handle.net/')"/>
|
|
50
|
</xsl:if>
|
|
51
|
<xsl:if test="not(contains(., '://hdl.handle.net/'))">
|
|
52
|
<xsl:value-of select="."/>
|
|
53
|
</xsl:if>
|
|
54
|
</oaf:identifier>
|
|
55
|
</xsl:for-each>
|
|
56
|
<xsl:for-each select="//*[local-name() = 'resource']/*[local-name()='identifier'][@identifierType='DOI']">
|
|
57
|
<oaf:identifier>
|
|
58
|
<xsl:attribute name="identifierType">
|
|
59
|
<xsl:value-of select="'doi'"/>
|
|
60
|
</xsl:attribute>
|
|
61
|
<xsl:if test="contains(., '://dx.doi.org/')">
|
|
62
|
<xsl:value-of select="substring-after(., '://dx.doi.org/')"/>
|
|
63
|
</xsl:if>
|
|
64
|
<xsl:if test="not(contains(., '://dx.doi.org/'))">
|
|
65
|
<xsl:value-of select="."/>
|
|
66
|
</xsl:if>
|
|
67
|
</oaf:identifier>
|
|
68
|
</xsl:for-each>
|
|
69
|
<!-- This is the only difference with the generic datacite repo rule: since all datasets from Nakala are Images -->
|
|
70
|
<xsl:if
|
|
71
|
test="lower-case(//*[local-name()='resourceType']/@*[local-name()='resourceTypeGeneral']) = 'image'">
|
|
72
|
<oaf:concept>
|
|
73
|
<xsl:attribute name="id">
|
|
74
|
<xsl:value-of select="'dariah'"/>
|
|
75
|
</xsl:attribute>
|
|
76
|
</oaf:concept>
|
|
77
|
</xsl:if>
|
|
78
|
<xsl:if test="//*[local-name()='date']/@dateType='Available' and //*[local-name()='datasourceprefix']!='r33ffb097cef'">
|
|
79
|
<xsl:variable name="varEmbargoEndDate" select="TransformationFunction:convertString($tf, normalize-space(//*[local-name()='date'][@dateType='Available']), 'DateISO8601')"/>
|
|
80
|
<xsl:choose>
|
|
81
|
<xsl:when test="string-length($varEmbargoEndDate) > 0">
|
|
82
|
<oaf:embargoenddate>
|
|
83
|
<xsl:value-of select="$varEmbargoEndDate"/>
|
|
84
|
</oaf:embargoenddate>
|
|
85
|
</xsl:when>
|
|
86
|
<xsl:otherwise>
|
|
87
|
<xsl:call-template name="terminate"/>
|
|
88
|
</xsl:otherwise>
|
|
89
|
</xsl:choose>
|
|
90
|
</xsl:if>
|
|
91
|
<xsl:variable name="varTypLst" select="distinct-values((//*[local-name()='resourceType']/(., @resourceTypeGeneral)))"/>
|
|
92
|
<xsl:variable name="varCobjCatLst" select="distinct-values((for $i in $varTypLst return TransformationFunction:convertString($tf, normalize-space($i), 'TextTypologies')))"/>
|
|
93
|
<xsl:variable name="varCobjSupLst" select="for $i in $varCobjCatLst return concat($i, '###', TransformationFunction:convertString($tf, normalize-space($i), 'SuperTypes'))"/>
|
|
94
|
<dr:CobjCategory>
|
|
95
|
<xsl:choose>
|
|
96
|
<xsl:when test="count($varCobjSupLst[not(substring-after(., '###') = 'other') and not(substring-before(., '###') = ('0038', '0039', '0040'))]) > 0">
|
|
97
|
<xsl:variable name="varCobjSup" select="$varCobjSupLst[not(substring-after(., '###') = 'other') and not(substring-before(., '###') = ('0038', '0039', '0040'))][1]"/>
|
|
98
|
<xsl:attribute name="type" select="substring-after($varCobjSup, '###')"/>
|
|
99
|
<xsl:value-of select="substring-before($varCobjSup, '###')"/>
|
|
100
|
</xsl:when>
|
|
101
|
<xsl:when test="count($varCobjSupLst[not(substring-after(., '###') = 'other')]) > 0">
|
|
102
|
<xsl:variable name="varCobjSup" select="$varCobjSupLst[not(substring-after(., '###') = 'other')][1]"/>
|
|
103
|
<xsl:attribute name="type" select="substring-after($varCobjSup, '###')"/>
|
|
104
|
<xsl:value-of select="substring-before($varCobjSup, '###')"/>
|
|
105
|
</xsl:when>
|
|
106
|
<xsl:when test="count($varCobjSupLst[not(substring-before(., '###') = ('0020', '0000'))]) > 0">
|
|
107
|
<xsl:variable name="varCobjSup" select="$varCobjSupLst[not(substring-before(., '###') = ('0020', '0000'))][1]"/>
|
|
108
|
<xsl:attribute name="type" select="substring-after($varCobjSup, '###')"/>
|
|
109
|
<xsl:value-of select="substring-before($varCobjSup, '###')"/>
|
|
110
|
</xsl:when>
|
|
111
|
<xsl:when test="count($varCobjSupLst[not(substring-before(., '###') = ('0000'))]) > 0">
|
|
112
|
<xsl:variable name="varCobjSup" select="$varCobjSupLst[not(substring-before(., '###') = ('0000'))][1]"/>
|
|
113
|
<xsl:attribute name="type" select="substring-after($varCobjSup, '###')"/>
|
|
114
|
<xsl:value-of select="substring-before($varCobjSup, '###')"/>
|
|
115
|
</xsl:when>
|
|
116
|
<xsl:otherwise>
|
|
117
|
<xsl:attribute name="type" select="'other'"/>
|
|
118
|
<xsl:value-of select="'0000'"/>
|
|
119
|
</xsl:otherwise>
|
|
120
|
</xsl:choose>
|
|
121
|
</dr:CobjCategory><!-- review status --><!-- no review hints found in resource type declarations, no version declarations found -->
|
|
122
|
<xsl:variable name="varRefereedConvt" select="for $i in ( //*[local-name()='resourceType']/(., @resourceTypeGeneral), //oai:setSpec, //*[local-name()='description']) return TransformationFunction:convertString($tf, normalize-space($i), 'ReviewLevels')"/>
|
|
123
|
<xsl:variable name="varRefereedIdntf" select="( //*[local-name()=('identifier', 'alternateIdentifier')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier')]) = 1][matches(lower-case(.), '(^|.*[\.\-_\\/\s\(\)%\d#:])pre[\.\-_\\/\s\(\)%\d#:]?prints?([\.\-_\\/\s\(\)%\d#:].*)?$')]/'0002', //*[local-name()=('identifier', 'alternateIdentifier')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier')]) = 1][matches(lower-case(.), '(^|.*[\.\-_\\/\s\(\)%\d#:])refereed([\.\-_\\/\s\(\)%\d#:].*)?$')]/'0001', //*[local-name()=('identifier', 'alternateIdentifier')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier')]) = 1][matches(lower-case(.), '.*-peer-reviewed-(fulltext-)?article-.*')]/'0001')"/>
|
|
124
|
<xsl:variable name="varRefereedVersn" select="(//*[local-name()='version'][matches(lower-case(.), '.*peer[\s\-\.\\_/:%]?reviewed.*')]/'0001', //*[local-name()='version'][matches(normalize-space(lower-case(.)), '^(v|vs|version|rel|release)?[\s\.\-_]*0$')]/'0002', //*[local-name()='version'][matches(lower-case(.), '(^|[\s\-\.\\_/:%].*)(beta|draft|trial|test)([\s\-\.\\_/:%].*|$)')]/'0002', //*[local-name()='version'][matches(lower-case(.), '.*submi(tted|ssion|ttal).*')]/'0002') "/>
|
|
125
|
<xsl:variable name="varRefereedOther" select="(//*[local-name()='publisher'][matches(lower-case(.), '.*[\s\-\.\\_/:%]pre[\s\-\.\\_/:%]?prints?([\s\-\.\\_/:%].*|$)')]/'0002', //*[local-name()='description'][matches(lower-case(.), '^peer[\s\-\.\\_/:%]?reviewed$')]/'0001', //*[local-name()='description'][matches(lower-case(.), '^pre[\s\-\.\\_/:%]?prints?$')]/'0002') "/>
|
|
126
|
<xsl:variable name="varRefereedReltn" select="//*[local-name() = 'relatedIdentifier'][./@relationType/lower-case(.)='isreviewedby']/'0001'"/>
|
|
127
|
<xsl:variable name="varRefereedDesct" select="(//*[local-name() = 'description'] [matches(lower-case(.), '.*(this\s*book|this\s*volume|it)\s*constitutes\s*the\s*(thoroughly\s*)?refereed') or matches(lower-case(.), '.*peer[\.\-_/\s\(\)]?review\s*under\s*responsibility\s*of.*') or matches(lower-case(.), '(this|a)\s*(article|preprint)\s*(has\s*been\s*)?(peer[\-\s]*)?reviewed\s*and\s*recommended\s*by\s*peer[\-\s]*community')]/'0001')"/>
|
|
128
|
<xsl:variable name="varRefereed" select="($varRefereedConvt, $varRefereedIdntf, $varRefereedReltn, $varRefereedVersn, $varRefereedOther, $varRefereedReltn, $varRefereedDesct)"/>
|
|
129
|
<xsl:choose>
|
|
130
|
<xsl:when test="count($varRefereed[. = '0001']) > 0">
|
|
131
|
<oaf:refereed>
|
|
132
|
<xsl:value-of select="'0001'"/>
|
|
133
|
</oaf:refereed>
|
|
134
|
</xsl:when>
|
|
135
|
<xsl:when test="count($varRefereed[. = '0002']) > 0">
|
|
136
|
<oaf:refereed>
|
|
137
|
<xsl:value-of select="'0002'"/>
|
|
138
|
</oaf:refereed>
|
|
139
|
</xsl:when>
|
|
140
|
</xsl:choose>
|
|
141
|
<oaf:dateAccepted>
|
|
142
|
<xsl:value-of select="TransformationFunction:convertString($tf, normalize-space(//*[local-name()='publicationYear']), 'DateISO8601')"/>
|
|
143
|
</oaf:dateAccepted>
|
|
144
|
<xsl:choose>
|
|
145
|
<xsl:when test="//*[local-name() = 'rights']/@rightsURI[starts-with(normalize-space(.), 'info:eu-repo/semantics')]">
|
|
146
|
<oaf:accessrights>
|
|
147
|
<xsl:value-of select="TransformationFunction:convertString($tf, //*[local-name() = 'rights']/@rightsURI[starts-with(normalize-space(.), 'info:eu-repo/semantics')], 'AccessRights')"/>
|
|
148
|
</oaf:accessrights>
|
|
149
|
</xsl:when>
|
|
150
|
<xsl:otherwise>
|
|
151
|
<xsl:choose>
|
|
152
|
<xsl:when test="//*[local-name() = 'rights'][starts-with(normalize-space(.), 'http://creativecommons.org') or starts-with(normalize-space(.), 'Creative Commons') or starts-with(normalize-space(.), 'GNU LESSER GENERAL PUBLIC LICENSE')]">
|
|
153
|
<oaf:accessrights>
|
|
154
|
<xsl:text>OPEN</xsl:text>
|
|
155
|
</oaf:accessrights>
|
|
156
|
</xsl:when>
|
|
157
|
<xsl:when test="//*[local-name() = 'rights']/@rightsURI[starts-with(normalize-space(.), 'http://creativecommons.org') or starts-with(normalize-space(.), 'http://opendatacommons.org')]">
|
|
158
|
<oaf:accessrights>
|
|
159
|
<xsl:text>OPEN</xsl:text>
|
|
160
|
</oaf:accessrights>
|
|
161
|
</xsl:when>
|
|
162
|
<xsl:when test="//*[local-name() = 'rights'][starts-with(normalize-space(.), 'Open access data at least for academic use')]">
|
|
163
|
<oaf:accessrights>
|
|
164
|
<xsl:text>RESTRICTED</xsl:text>
|
|
165
|
</oaf:accessrights>
|
|
166
|
</xsl:when>
|
|
167
|
<xsl:otherwise>
|
|
168
|
<oaf:accessrights>
|
|
169
|
<xsl:text>UNKNOWN</xsl:text>
|
|
170
|
</oaf:accessrights>
|
|
171
|
</xsl:otherwise>
|
|
172
|
</xsl:choose>
|
|
173
|
</xsl:otherwise>
|
|
174
|
</xsl:choose>
|
|
175
|
<xsl:for-each select="//*[local-name()='rights']/@rightsURI[starts-with(normalize-space(.), 'http') and matches(., '.*(/licenses|/publicdomain|unlicense.org/|/legal-and-data-protection-notices|/download/license|/open-government-licence).*')]">
|
|
176
|
<oaf:license>
|
|
177
|
<xsl:value-of select="."/>
|
|
178
|
</oaf:license>
|
|
179
|
</xsl:for-each>
|
|
180
|
<oaf:language>
|
|
181
|
<xsl:value-of select="TransformationFunction:convert($tf, //*[local-name()='language'], 'Languages')"/>
|
|
182
|
</oaf:language>
|
|
183
|
<oaf:hostedBy>
|
|
184
|
<xsl:attribute name="name">
|
|
185
|
<xsl:value-of select="$varOfficialName"/>
|
|
186
|
</xsl:attribute>
|
|
187
|
<xsl:attribute name="id">
|
|
188
|
<xsl:value-of select="$varDataSourceId"/>
|
|
189
|
</xsl:attribute>
|
|
190
|
</oaf:hostedBy>
|
|
191
|
<oaf:collectedFrom>
|
|
192
|
<xsl:attribute name="name">
|
|
193
|
<xsl:value-of select="$varOfficialName"/>
|
|
194
|
</xsl:attribute>
|
|
195
|
<xsl:attribute name="id">
|
|
196
|
<xsl:value-of select="$varDataSourceId"/>
|
|
197
|
</xsl:attribute>
|
|
198
|
</oaf:collectedFrom>
|
|
199
|
</metadata>
|
|
200
|
<xsl:copy-of select="//*[local-name() = 'about']"/>
|
|
201
|
</record>
|
|
202
|
</xsl:template>
|
|
203
|
|
|
204
|
<xsl:template match="node()|@*">
|
|
205
|
<xsl:copy>
|
|
206
|
<xsl:apply-templates select="node()|@*"/>
|
|
207
|
</xsl:copy>
|
|
208
|
</xsl:template>
|
|
209
|
|
|
210
|
<xsl:template match="//*[local-name() = 'metadata']//*[local-name() = 'resource']">
|
|
211
|
<xsl:copy>
|
|
212
|
<xsl:apply-templates select="node()|@*"/>
|
|
213
|
</xsl:copy>
|
|
214
|
</xsl:template>
|
|
215
|
|
|
216
|
<xsl:template match="//*[local-name() = 'resource']/*[local-name()='identifier']">
|
|
217
|
<xsl:if test=".[@identifierType='Handle'][contains(., '://hdl.handle.net/')]">
|
|
218
|
<xsl:element name="identifier" namespace="http://datacite.org/schema/kernel-4">
|
|
219
|
<xsl:attribute name="identifierType">
|
|
220
|
<xsl:value-of select="'Handle'"/>
|
|
221
|
</xsl:attribute>
|
|
222
|
<xsl:value-of select="."/>
|
|
223
|
</xsl:element>
|
|
224
|
</xsl:if>
|
|
225
|
</xsl:template>
|
|
226
|
|
|
227
|
<xsl:template match="//*[local-name() = 'resource']/*[local-name()='alternateIdentifier']">
|
|
228
|
<xsl:choose>
|
|
229
|
<xsl:when test="@alternateIdentifierType">
|
|
230
|
<xsl:copy-of select="."/>
|
|
231
|
</xsl:when>
|
|
232
|
<xsl:otherwise>
|
|
233
|
<xsl:element name="alternateIdentifier" namespace="http://datacite.org/schema/kernel-4">
|
|
234
|
<xsl:attribute name="alternateIdentifierType"><xsl:value-of select="./@identifierType"/></xsl:attribute>
|
|
235
|
<xsl:value-of select="."/>
|
|
236
|
</xsl:element>
|
|
237
|
</xsl:otherwise>
|
|
238
|
</xsl:choose>
|
|
239
|
</xsl:template>
|
|
240
|
|
|
241
|
|
|
242
|
<xsl:template match="//*[local-name() = 'header']">
|
|
243
|
<xsl:copy>
|
|
244
|
<xsl:apply-templates select="node()|@*"/>
|
|
245
|
<xsl:element name="dr:dateOfTransformation">
|
|
246
|
<xsl:value-of select="$transDate"/>
|
|
247
|
</xsl:element>
|
|
248
|
</xsl:copy>
|
|
249
|
</xsl:template>
|
|
250
|
</xsl:stylesheet>
|
|
251
|
|
|
252
|
</CODE>
|
|
253
|
</SCRIPT>
|
|
254
|
</CONFIGURATION>
|
|
255
|
<STATUS/>
|
|
256
|
<SECURITY_PARAMETERS/>
|
|
257
|
</BODY>
|
|
258
|
</RESOURCE_PROFILE>
|