Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER
4
                value="5c777768-c164-41aa-90f1-58d17b4b272d_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
5
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
6
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
7
        <RESOURCE_URI value=""/>
8
        <DATE_OF_CREATION value="2023-04-20T10:32:19+00:00"/>
9
    </HEADER>
10
    <BODY>
11
        <CONFIGURATION>
12
            <SOURCE_METADATA_FORMAT interpretation="cleaned" layout="store" name="odf"/>
13
            <SINK_METADATA_FORMAT name="odf_hbase"/>
14
            <IMPORTED/>
15
            <SCRIPT>
16
                <TITLE>xslt_cleaning_oaire2datacite_hadoop</TITLE>
17
                <CODE><!-- adaptation 2021-08-23 ; ACz/UNIBI -->
18
                    <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
19
                                    xmlns:oaire="http://namespace.openaire.eu/schema/oaire/"
20
                                    xmlns:vocabulary="http://eu/dnetlib/transform/clean"
21
                                    xmlns:dateCleaner="http://eu/dnetlib/transform/dateISO"
22
                                    xmlns:oaf="http://namespace.openaire.eu/oaf"
23
                                    xmlns:oai="http://www.openarchives.org/OAI/2.0/"
24
                                    xmlns:datacite="http://datacite.org/schema/kernel-4"
25
                                    xmlns:dri="http://www.driver-repository.eu/namespace/dri"
26
                                    xmlns:xs="http://www.w3.org/2001/XMLSchema"
27
                                    xmlns:dr="http://www.driver-repository.eu/namespace/dr"
28
                                    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
29
                                    xmlns:dc="http://purl.org/dc/elements/1.1/"
30
                                    exclude-result-prefixes="xsl vocabulary dateCleaner" version="2.0">
31
                        <xsl:param name="varOfficialName"/>
32
                        <xsl:param name="varDataSourceId"/>
33
                        <xsl:param name="varAKA" select="'aka_________::'"/>
34
                        <xsl:param name="varARC" select="'arc_________::'"/>
35
                        <xsl:param name="varANR" select="'anr_________::'"/>
36
                        <xsl:param name="varCHISTERA" select="'chistera____::'"/>
37
                        <xsl:param name="varCONICYT" select="'conicytf____::'"/>
38
                        <xsl:param name="varDFG" select="'dfgf________::'"/>
39
                        <xsl:param name="varEUENVAGENCY" select="'euenvagency_::'"/>
40
                        <xsl:param name="varFCT" select="'fct_________::'"/>
41
                        <xsl:param name="varFP7" select="'corda_______::'"/>
42
                        <xsl:param name="varFWF" select="'fwf_________::'"/>
43
                        <xsl:param name="varGSRI" select="'gsri________::'"/>
44
                        <xsl:param name="varGSRT" select="'gsrt________::'"/>
45
                        <xsl:param name="varH2020" select="'corda__h2020::'"/>
46
                        <xsl:param name="varHEUROPE" select="'corda_____he::'"/>
47
                        <xsl:param name="varHRZZ" select="'irb_hr______::'"/><!-- HRZZ not found -->
48
                        <xsl:param name="varINNOVIRIS" select="'innoviris___::'"/>
49
                        <xsl:param name="varMESTD" select="'mestd_______::'"/>
50
                        <xsl:param name="varMIUR" select="'miur________::'"/>
51
                        <xsl:param name="varMZOS" select="'irb_hr______::'"/>
52
                        <xsl:param name="varNHMRC" select="'nhmrc_______::'"/>
53
                        <xsl:param name="varNIH" select="'nih_________::'"/>
54
                        <xsl:param name="varNSF" select="'nsf_________::'"/>
55
                        <xsl:param name="varNWO" select="'nwo_________::'"/>
56
                        <xsl:param name="varRIF" select="'rif_________::'"/>
57
                        <xsl:param name="varRSF" select="'rsf_________::'"/>
58
                        <xsl:param name="varSFI" select="'sfi_________::'"/>
59
                        <xsl:param name="varSFRS" select="'sfrs________::'"/>
60
                        <xsl:param name="varSGOV"
61
                                   select="'sgov________::'"/><!-- SGOV to be added, awaiting DOI from Pilar, found project ids not in CSV list? -->
62
                        <xsl:param name="varSNSF" select="'snsf________::'"/>
63
                        <xsl:param name="varTARA"
64
                                   select="'taraexp_____::'"/><!-- TARA to be added, awaiting DOI from André -->
65
                        <xsl:param name="varTUBITAK" select="'tubitakf____::'"/>
66
                        <xsl:param name="varUKRI"
67
                                   select="'ukri________::'"/><!-- UKRI replace RCUK namespace:  'rcuk________::'  -->
68
                        <xsl:param name="varWT" select="'wt__________::'"/>
69
                        <xsl:param name="index" select="0"/>
70
                        <xsl:param name="transDate" select="current-dateTime()"/>
71
                        <xsl:template match="/">
72
                            <xsl:variable name="datasourcePrefix"
73
                                          select="normalize-space(//oaf:datasourceprefix)"/><!-- TODO: termination handling at Spark level must be refined! -->
74
                            <xsl:if test="//oai:header/@status='deleted'">
75
                                <xsl:call-template name="terminate"/>
76
                            </xsl:if>
77
                            <xsl:for-each select="//*[local-name() = 'date'][@dateType='Issued']">
78
                                <xsl:if test="string-length(normalize-space(.)) &lt; 4"><!-- xsl:call-template name="terminate"/ --></xsl:if>
79
                            </xsl:for-each>
80
                            <xsl:call-template name="checkIdentifierType"/>
81
                            <xsl:call-template name="validRecord"/>
82
                        </xsl:template>
83
                        <xsl:template name="terminate">
84
                            <xsl:message terminate="yes">
85
                                record is not i+t v4 compliant, transformation is interrupted.
86
                            </xsl:message>
87
                        </xsl:template><!-- ## check datacite:identifier identifierType vocabulary - https://openaire-guidelines-for-literature-repository-managers.readthedocs.io/en/latest/field_resourceidentifier.html#attribute-identifiertype-m ## -->
88
                        <xsl:template name="checkIdentifierType">
89
                            <xsl:if test="lower-case(//*[local-name() = 'identifier']/@identifierType) != 'ark' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'doi' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'handle' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'igsn' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'purl' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'url' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'urn' and lower-case(//*[local-name() = 'identifier']/@identifierType) != 'w3id'">
90
                                <xsl:call-template name="terminate"/>
91
                            </xsl:if>
92
                        </xsl:template>
93
                        <xsl:template name="validRecord">
94
                            <record>
95
                                <xsl:apply-templates select="//*[local-name() = 'header']"/>
96
                                <metadata><!-- ~~~~~ pre-detect resourceType, superType and store in variable ~~~~~ --><!-- optimize resourceType, superType in case of several statements (e.g. FU Berlin unfortunately contains record with several) --><!--
97
                     <xsl:variable name='varCobjCategory' select="TransformationFunction:convertString($tf, distinct-values(//*[local-name()='resourceType'][1]/@uri), 'TextTypologies')" />
98
                     <xsl:variable name="varSuperType" select="TransformationFunction:convertString($tf, $varCobjCategory, 'SuperTypes')"/>
99
                -->
100
                                    <xsl:variable name="varTypLst"
101
                                                  select="distinct-values((//*[local-name()='resourceType']/(., @*)))"/>
102
                                    <xsl:variable name="varCobjCatLst"
103
                                                  select="distinct-values((for $i in $varTypLst      return vocabulary:clean( normalize-space($i), 'dnet:publication_resource')))"/>
104
                                    <xsl:variable name="varCobjSupLst"
105
                                                  select="for $i in $varCobjCatLst      return concat($i, '###', vocabulary:clean( normalize-space($i), 'dnet:result_typologies'))"/>
106
                                    <xsl:variable name="varCobjSup"
107
                                                  select="(                         $varCobjSupLst[not(substring-after(., '###') = 'other') and not(substring-before(., '###') = ('0038', '0039', '0040'))][1],                          $varCobjSupLst[not(substring-after(., '###') = 'other')][1],                         $varCobjSupLst[not(substring-before(., '###') = ('0020', '0000'))][1],                         $varCobjSupLst[not(substring-before(., '###') = ('0000'))][1],                         $varCobjSupLst[not(substring-before(., '###') = ('0000'))][1],                         '0000')[1]"/>
108
                                    <xsl:variable name="varSuperType" select="substring-after($varCobjSup, '###')"/>
109
                                    <xsl:variable name="varCobjCategory"
110
                                                  select="substring-before($varCobjSup, '###')"/><!-- ~~~~~ pre-detect, -clean, -deduplicat PIDs and store in variable ~~~~~ -->
111
                                    <xsl:variable name="varKnownFileEndings"
112
                                                  select="('.bmp', '.doc', '.docx', '.epub', '.flv', '.jpeg', '.jpg', '.m4v', '.mp4', '.mpg', '.odp', '.pdf', '.png', '.ppt', '.tiv', '.txt', '.xls', '.xlsx', '.zip')"/><!-- landingPage URL --><!-- either generate landing pages (as often not given), or compare URL with baseURL --><!--
113
                     // covered in comparison: URL encoded baseURLs, item URL und baseURL differing only in prefix www. or postfix port
114
                     // not covered in comparison: URL encoded item URL, URLs being a substring of URLs (like item URL http://def.br/... and baseURL http://abcdef.br/...), ports when only in baseURL or only in item URL
115
                     // tries to not consider higher level URLs as item URLs (e.g. journal URLs) by enforcing something after /
116
                -->
117
                                    <xsl:variable name="varOrigBaseUrl"
118
                                                  select="//*[local-name() = 'about']/*[local-name() = 'provenance']//*[local-name() = 'originDescription' and not(./*[local-name() = 'originDescription'])]/*[local-name() = 'baseURL']"/>
119
                                    <xsl:variable name="varAggrBaseUrl"
120
                                                  select="//*[local-name() = 'about']/*[local-name() = 'provenance']//*[local-name() = 'originDescription' and (./*[local-name() = 'originDescription'])]/*[local-name() = 'baseURL']"/>
121
                                    <xsl:variable name="varLandingPage"
122
                                                  select="(                         //datacite:identifier[(contains(substring-after(., '://'), '/') and contains($varOrigBaseUrl, substring-before(substring-after(., '://'), '/'))) or (contains(substring-after(., '://'), ':') and contains($varOrigBaseUrl, substring-before(substring-after(., '://'), ':')))][not(replace(lower-case(.), '.*(\.[a-z]*)$', '$1') = $varKnownFileEndings)],                         //datacite:identifier[(contains(substring-after(., '://'), '/') and contains($varAggrBaseUrl, substring-before(substring-after(., '://'), '/'))) or (contains(substring-after(., '://'), ':') and contains($varAggrBaseUrl, substring-before(substring-after(., '://'), ':')))][not(replace(lower-case(.), '.*(\.[a-z]*)$', '$1') = $varKnownFileEndings)],                         //oaf:datasourceprefix[. = 'od_______268']/concat('https://qspace.library.queensu.ca/handle/', substring-after(//dri:recordIdentifier, 'oai:qspace.library.queensu.ca:')),                         //oaf:datasourceprefix[. = 'od_______307']/concat('http://repositorium.sdum.uminho.pt/handle/', substring-after(//dri:recordIdentifier, 'oai:repositorium.sdum.uminho.pt:')),                         //oaf:datasourceprefix[. = 'od_______317']/concat('https://papyrus.bib.umontreal.ca/xmlui/handle/', substring-after(//dri:recordIdentifier, 'oai:papyrus.bib.umontreal.ca:')),                         //oaf:datasourceprefix[. = 'od______1318']/concat('https://orbi.uliege.be/handle/', substring-after(//dri:recordIdentifier, 'oai:orbi.ulg.ac.be:')),                         //oaf:datasourceprefix[. = 'od______1388']/concat('http://rabida.uhu.es/dspace/handle/', substring-after(//dri:recordIdentifier, 'oai:rabida.uhu.es:')),                         //oaf:datasourceprefix[. = 'od______1472']/concat('https://gredos.usal.es/handle/', substring-after(//dri:recordIdentifier, 'oai:gredos.usal.es:')),                         //oaf:datasourceprefix[. = 'od______1560']/concat('https://riunet.upv.es/handle/', substring-after(//dri:recordIdentifier, 'oai:riunet.upv.es:')),                         //oaf:datasourceprefix[. = 'od______3158']/concat('https://open.uct.ac.za/handle/', substring-after(//dri:recordIdentifier, 'oai:localhost:')),                         //oaf:datasourceprefix[. = 'od______4225']/concat('https://repository.rothamsted.ac.uk/item/', substring-after(//dri:recordIdentifier, 'oai:repository.rothamsted.ac.uk:')),                         //oaf:datasourceprefix[. = 'od______9626']/(//datacite:identifier)[@identifierType='Handle'][not(contains(., '/handle/123456789/'))],                         //oaf:datasourceprefix[not(. = ('od_______268', 'od_______307', 'od______1318', 'od______1388', 'od______1472', 'od______4225'))]/(//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*) = ('url', 'uri')][not(contains(., 'doi.org/')) and not(contains(., '/doi/')) and not(contains(., '/pmc/'))][starts-with(., 'http')]                     )[1]"/><!-- IDs (identifier and alternateIdentifier, not relatedIdentifier) --><!-- container PIDs (ISSNs, ISBNs) often erroneously stuffed in alternateIdentifier for contained items (journal articles, book chapters, ...), are to be shifted into relatedIdentifier instead -->
123
                                    <xsl:variable name="varPidDoi"
124
                                                  select="concat(' DOI:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[contains(., 'doi.org/10.')]/lower-case(substring-after(., 'doi.org/')),                                     (//datacite:identifier, //datacite:alternateIdentifier)[contains(., 'info:doi:')]/substring-after(., 'info:doi:'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='doi'][not(contains(., 'doi.org/10.')) and not(contains(., 'info:doi:')) and not(contains(., '/doi/'))]/lower-case(.),                                     (//datacite:identifier, //datacite:alternateIdentifier)[matches(., '.*/doi/.*/10\..*')]/replace(., '.*/doi/.*/(10\..*)$', '$1')                                 )), ' DOI:::'))"/>
125
                                    <xsl:variable name="varPidHandle"
126
                                                  select="concat(' Handle:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[contains(., 'hdl.handle.net/')][not(contains(., '123456789'))]/substring-after(., 'hdl.handle.net/'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[contains(., 'info:hdl:')][not(contains(., '123456789'))]/substring-after(., 'info:hdl:'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='handle'][not(starts-with(., 'http')) and not(contains(., 'hdl.handle.net/')) and not(contains(., 'info:hdl:')) and not(contains(., '/handle/123456789/'))]                                 )), ' Handle:::'))"/>
127
                                    <xsl:variable name="varPidUrn"
128
                                                  select="concat(' URN:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'urn:nbn:')]/concat('urn:nbn:', substring-after(lower-case(.), 'urn:nbn:')),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'http') and (contains(lower-case(.), '://nbn-resolving.org/urn:nbn:') or contains(lower-case(.), '://nbn-resolving.de/urn/resolver.pl?urn:nbn:') or contains(lower-case(.), '://nbn-resolving.de/urn:nbn:') or contains(lower-case(.), '://resolver.obvsg.at/urn:nbn:') or contains(lower-case(.), '://urn.fi/urn:nbn:') or contains(lower-case(.), '://urn.kb.se/resolve?urn=urn:nbn:'))]/concat('urn:nbn:', substring-after(lower-case(.), 'urn:nbn:'))                                 )), ' URN:::'))"/>
129
                                    <xsl:variable name="varPidArk"
130
                                                  select="concat(' ARK:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'http') and contains(., '/ark:/')]/concat('ark:/', substring-after(., '/ark:/'))                                 )), ' ARK:::'))"/>
131
                                    <xsl:variable name="varPidPmid"
132
                                                  select="concat(' PMID:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'http') and contains(., '://www.ncbi.nlm.nih.gov/pmc/articles/')]/substring-after(., '://www.ncbi.nlm.nih.gov/pmc/articles/'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'info:pmid/')]/substring-after(., 'info:pmid/'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'pmid:') or starts-with(lower-case(.), 'pubmed:')]/substring-after(., ':'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='pmc'][not(contains(., 'ncbi.nlm.nih.gov'))][not(contains(., ':'))],                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='pmid'][not(contains(., ':'))]                                 )), ' PMID:::'))"/>
133
                                    <xsl:variable name="varPidPmcid"
134
                                                  select="concat(' PMCID:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'http') and (contains(lower-case(.), '://www.ncbi.nlm.nih.gov/pmc/articles/pmc') or contains(lower-case(.), '://europepmc.org/articles/pmc'))]/substring-after(., '/articles/'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'pmcid:') or starts-with(lower-case(.), 'pmc:')]/substring-after(., ':'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='pmcid']                                 )), ' PMCID:::'))"/>
135
                                    <xsl:variable name="varPidHal"
136
                                                  select="concat(' HAL:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='hal' and (starts-with(., 'hal-') or starts-with(., 'halshs-') or starts-with(., 'halsde-'))],                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'http') and (contains(., '://hal.archives-ouvertes.fr/hal') or contains(., '://hal.inria.fr/hal') or contains(., '://halshs.archives-ouvertes.fr/hal') or contains(., '://halsde.archives-ouvertes.fr/hal'))]/substring-after(., '.fr/')                                 )), ' HAL:::'))"/>
137
                                    <xsl:variable name="varPidBibc"
138
                                                  select="concat(' BIBC:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'bibcode:')]/substring-after(lower-case(.), 'bibcode:'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(., 'http') and contains(lower-case(.), 'bibcode=')]/substring(substring-after(lower-case(.), 'bibcode='), 1, 19)                                 )), ' BIBC:::'))"/>
139
                                    <xsl:variable name="varPidArxiv"
140
                                                  select="concat(' arXiv:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'arxiv:')]/substring-after(lower-case(.), 'arxiv:'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[(starts-with(., 'http') or starts-with(lower-case(.), 'arxiv: http')) and contains(., '://arxiv.org/abs/')]/substring-after(., '://arxiv.org/abs/'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[(starts-with(., 'http') or starts-with(lower-case(.), 'arxiv: http')) and contains(., '://arxiv.org/pdf/')]/substring-before(substring-after(lower-case(.), '://arxiv.org/pdf/'), '.pdf'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='arxiv']                                 )), ' arXiv:::'))"/>
141
                                    <xsl:variable name="varPidWos"
142
                                                  select="concat(' WOS:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='wos'][not(starts-with(lower-case(.), 'wos:'))],                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'wos:') and not(starts-with(lower-case(.), 'wos: wos:'))]/substring-after(lower-case(.), 'wos:'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'wos: wos:')]/substring-after(lower-case(.), 'wos: wos:')                                 )), ' WOS:::'))"/>
143
                                    <xsl:variable name="varPidScp"
144
                                                  select="concat(' SCP:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(lower-case(.), 'scopus_id:') or starts-with(lower-case(.), 'scopus:')]/substring-after(., ':'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(normalize-space(.), 'scopus: eid=')]/substring-after(., 'eid=2-s2.0-'),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(normalize-space(.), 'http') and contains(., '://www.scopus.com/inward/record.ur') and contains(., 'scp=')]/substring-after(., 'scp='),                                     (//datacite:identifier, //datacite:alternateIdentifier)[starts-with(normalize-space(.), 'http') and contains(., '://www.scopus.com/inward/record.ur') and contains(., 'eid=2-s2.0-')]/substring-after(., 'eid=2-s2.0-')                                 )), ' SCP:::'))"/>
145
                                    <xsl:variable name="varPidUrl"
146
                                                  select="concat(' URL:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='url']                                     [not(contains(., 'doi.org/10.')) and not(contains(., 'info:doi:')) and not(matches(., '.*/doi/.*/10\..*'))]                                     [not(contains(., 'hdl.handle.net/')) and not(contains(., 'info:hdl:'))]                                     [not(contains(., '://www.ncbi.nlm.nih.gov/pmc/articles/'))]                                     [not(. = $varLandingPage)]                                 )), ' URL:::'))"/>
147
                                    <xsl:variable name="varPidIsbn"
148
                                                  select="concat(' ISBN:::', string-join(distinct-values((                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='isbn' and not(contains(., ':')) and not($varCobjCategory = '0013')],                                     (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)='isbn' and starts-with(lower-case(.), 'urn:isbn') and not($varCobjCategory = '0013')]/substring-after(lower-case(.), 'urn:isbn:')                                 )), ' ISBN:::'))"/>
149
                                    <xsl:variable name="varPidIssn"
150
                                                  select="string-join(distinct-values((                                 (//datacite:identifier, //datacite:alternateIdentifier)[lower-case(@*)=('issn', 'pissn', 'eissn', 'lissn') and matches(., '.*\d{4}[\s-]?\d{3}[\dXx].*') and not($varCobjCategory = '0001')]/concat(' ', upper-case((@identifierType, @alternateIdentifierType)), ':::', replace(., '^.*(\d{4})[\s-]?(\d{3}[\dXx]).*$', '$1-$2'))                             )), '')"/><!-- heap all IDs, drop landingPage URL -->
151
                                    <xsl:variable name="varPidW3id"
152
                                                  select="concat(' w3id:::', string-join(distinct-values((                                     (//datacite:identifier)[@identifierType='w3id']/text(), ' w3id:::'))))"/><!-- heap all IDs, drop landingPage URL -->
153
                                    <xsl:variable name="varPidPre"
154
                                                  select="concat($varPidDoi[not(. = ' DOI:::')], $varPidHandle[not(. = ' Handle:::')], $varPidUrn[not(. = ' URN:::')], $varPidArk[not(. = ' ARK:::')], $varPidPmid[not(. = ' PMID:::')], $varPidPmcid[not(. = ' PMCID:::')], $varPidHal[not(. = ' HAL:::')], $varPidBibc[not(. = ' BIBC:::')], $varPidArxiv[not(. = ' arXiv:::')], $varPidWos[not(. = ' WOS:::')], $varPidScp[not(. = ' SCP:::')], $varPidUrl[not(. = ' URL:::')], $varPidIsbn[not(. = ' ISBN:::')], $varPidIssn[not(. = ' ISSN:::')], $varPidW3id[not(. = ' w3id:::')])"/><!--
155
                     <xsl:variable name="varPid" select="concat($varPidPre, $varLandingPage[string-length($varPidPre) = 0]/concat('landingPage:::', .))"/>
156
                     <xsl:variable name="varPid" select="concat($varPidPre, concat(' landingPage:::', $varLandingPage[string-length($varPidPre) = 0], .))"/>
157
                -->
158
                                    <xsl:variable name="varPid"
159
                                                  select="concat($varPidPre, //oaf:datasourceprefix[string-length($varPidPre) = 0]/concat(' landingPage:::', $varLandingPage))"/><!-- relatedIdentifier --><!-- container PIDs (ISSNs, ISBNs) often erroneously stuffed in alternateIdentifier for contained items (journal articles, book chapters, ...), are to be shifted into relatedIdentifier instead -->
160
                                    <xsl:variable name="varRelPidDoi"
161
                                                  select="concat(' DOI:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[contains(., 'doi.org/10.')]/concat(lower-case(substring-after(., 'doi.org/')), ':::', @relationType),                                     (//datacite:relatedIdentifier)[contains(., 'info:doi:')]/concat(substring-after(., 'info:doi:'), ':::', @relationType),                                     (//datacite:relatedIdentifier)[@relatedIdentifierType='DOI'][not(contains(., 'doi.org/10.')) and not(contains(., 'info:doi:')) and not(contains(., '/doi/'))]/concat(lower-case(.), ':::', @relationType),                                     (//datacite:relatedIdentifier)[matches(., '.*/doi/.*/10\..*')]/concat(replace(., '.*/doi/.*/(10\..*)$', '$1'), ':::', @relationType)                                 )), ' DOI:::'))"/>
162
                                    <xsl:variable name="varRelPidHandle"
163
                                                  select="concat(' Handle:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[contains(., 'hdl.handle.net/')]/concat(substring-after(., 'hdl.handle.net/'), ':::', @relationType),                                     (//datacite:relatedIdentifier)[contains(., 'info:hdl:')]/concat(substring-after(., 'info:hdl:'), ':::', @relationType),                                     (//datacite:relatedIdentifier)[lower-case(@relatedIdentifierType)='handle'][not(contains(., 'hdl.handle.net/')) and not(contains(., 'info:hdl:'))]/concat(., ':::', @relationType)                                 )), ' Handle:::'))"/>
164
                                    <xsl:variable name="varRelPidPmid"
165
                                                  select="concat(' PMID:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[contains(., '://www.ncbi.nlm.nih.gov/pmc/articles/')]/concat(substring-after(., '://www.ncbi.nlm.nih.gov/pmc/articles/'), ':::', @relationType),                                     (//datacite:relatedIdentifier)[lower-case(@*)='pmc'][not(contains(., 'ncbi.nlm.nih.gov'))]/concat(., ':::', @relationType)                                 )), ' PMID:::'))"/>
166
                                    <xsl:variable name="varRelPidPmcid"
167
                                                  select="concat(' PMCID:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[lower-case(@*)='pmcid']                                     /concat(., ':::', @relationType)                                 )), ' PMCID:::'))"/>
168
                                    <xsl:variable name="varRelPidArxiv"
169
                                                  select="concat(' arXiv:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[lower-case(@*)='arxiv']                                     /concat(., ':::', @relationType)                                 )), ' arXiv:::'))"/>
170
                                    <xsl:variable name="varRelPidWos"
171
                                                  select="concat(' WOS:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[lower-case(@*)='wos'][not(starts-with(lower-case(.), 'wos:'))]/concat(., ':::', @relationType),                                     (//datacite:relatedIdentifier)[starts-with(lower-case(.), 'wos:')]/concat(substring-after(lower-case(.), 'wos:'), ':::', @relationType)                                 )), ' WOS:::'))"/>
172
                                    <xsl:variable name="varRelPidUrl"
173
                                                  select="concat(' URL:::', string-join(distinct-values((                                     (//datacite:relatedIdentifier)[lower-case(@*)='url']                                     [not(contains(., 'doi.org/10.')) and not(contains(., 'info:doi:')) and not(matches(., '.*/doi/.*/10\..*'))]                                     [not(contains(., 'hdl.handle.net/')) and not(contains(., 'info:hdl:'))]                                     [not(contains(., '://www.ncbi.nlm.nih.gov/pmc/articles/'))]                                     /concat(., ':::', @relationType)                                 )), ' URL:::'))"/>
174
                                    <xsl:variable name="varRelPidW3id"
175
                                                  select="concat(' w3id:::', string-join(distinct-values(((//datacite:relatedIdentifier)[lower-case(@*) = 'w3id'][not(starts-with(lower-case(.), 'w3id:'))]/concat(., ':::', @relationType), (//datacite:relatedIdentifier)[starts-with(lower-case(.), 'w3id:')]/concat(substring-after(lower-case(.), 'w3id:'), ':::', @relationType))), ' w3id:::'))"/>
176
                                    <xsl:variable name="varRelPid"
177
                                                  select="concat($varRelPidDoi[not(. = ' DOI:::')], $varRelPidHandle[not(. = ' Handle:::')], $varRelPidPmid[not(. = ' PMID:::')], $varRelPidPmcid[not(. = ' PMCID:::')], $varRelPidArxiv[not(. = ' arXiv:::')], $varRelPidWos[not(. = ' WOS:::')], $varRelPidW3id[not(. = ' w3id:::')], $varRelPidUrl[not(. = ' URL:::')])"/><!-- container PIDs --><!-- TODO --><!-- drop oaire resource --><!-- <oaire:resource xmlns="http://namespace.openaire.eu/schema/oaire/"> -->
178
                                    <datacite:resource><!-- ~~~~~ IDs ~~~~~ --><!-- choose 'best' PID among identifiers and alternateIdentifers, and cut off DOI resolver prefix to just get the number part -->
179
                                        <datacite:identifier>
180
                                            <xsl:attribute name="identifierType"
181
                                                           select="substring-before(tokenize(normalize-space($varPid), ' ')[1], ':::')"/>
182
                                            <xsl:value-of
183
                                                    select="substring-after(tokenize(normalize-space($varPid), ' ')[1], ':::')"/>
184
                                        </datacite:identifier>
185
                                        <datacite:alternateIdentifiers>
186
                                            <xsl:for-each
187
                                                    select="tokenize(normalize-space($varPid), ' ')[position() &gt; 1][string-length(substring-after(., ':::')) &gt; 0]">
188
                                                <datacite:alternateIdentifier>
189
                                                    <xsl:attribute name="alternateIdentifierType"
190
                                                                   select="substring-before(., ':::')"/>
191
                                                    <xsl:value-of select="substring-after(., ':::')"/>
192
                                                </datacite:alternateIdentifier>
193
                                            </xsl:for-each>
194
                                            <xsl:for-each
195
                                                    select="(//datacite:alternateIdentifier, //datacite:identifier)                             [not($varCobjCategory = '0001' and ./@*[local-name()=('identifierType', 'alternateIdentifierType')]/lower-case(.) = ('issn', 'pissn', 'eissn', 'lissn'))]                             [not($varCobjCategory = '0013' and ./@*[local-name()=('identifierType', 'alternateIdentifierType')]/lower-case(.) = 'isbn')]                             [not(ends-with(., 'pdf'))]                             [not(@*[local-name()=('identifierType', 'alternateIdentifierType')]/lower-case(.) = ('doi', 'handle', 'urn', 'pmid', 'pmcid', 'arxiv', 'wos', 'url', 'isbn', 'issn', 'eissn', 'lissn', 'pissn'))]                             [not(. = $varLandingPage)]">
196
                                                <datacite:alternateIdentifier>
197
                                                    <xsl:attribute name="alternateIdentifierType"
198
                                                                   select="./@*[local-name()=('identifierType', 'alternateIdentifierType')]"/>
199
                                                    <xsl:value-of select="."/>
200
                                                </datacite:alternateIdentifier>
201
                                            </xsl:for-each>
202
                                            <xsl:if test="not(starts-with(normalize-space(lower-case($varPid)), 'landingpage')) and string-length($varLandingPage) &gt; 0">
203
                                                <datacite:alternateIdentifier>
204
                                                    <xsl:attribute name="alternateIdentifierType"
205
                                                                   select="'landingPage'"/>
206
                                                    <xsl:value-of select="$varLandingPage"/>
207
                                                </datacite:alternateIdentifier>
208
                                            </xsl:if>
209
                                        </datacite:alternateIdentifiers>
210
                                        <datacite:relatedIdentifiers>
211
                                            <xsl:for-each
212
                                                    select="(//datacite:relatedIdentifier)                             [not(@relatedIdentifierType/lower-case(.) = ('doi', 'handle', 'pmid', 'pmcid', 'wos', 'url'))]                             [not(starts-with(lower-case(.), 'wos:') or contains(., 'hdl.handle.net/') or contains(., 'doi.org/10.'))]                             [not(@relatedIdentifierType/lower-case(.) = ('issn', 'pissn', 'eissn', 'lissn', 'isbn') and @relationType/lower-case(.) = 'ispartof')]">
213
                                                <datacite:relatedIdentifier>
214
                                                    <xsl:attribute name="relatedIdentifierType"
215
                                                                   select="./@relatedIdentifierType"/>
216
                                                    <xsl:attribute name="relationType" select="./@relationType"/>
217
                                                    <xsl:value-of
218
                                                            select=".[not(@relatedIdentifierType/lower-case(.) = ('issn', 'pissn', 'eissn', 'lissn'))],                                     .[@relatedIdentifierType/lower-case(.) = ('issn', 'pissn', 'eissn', 'lissn')]/concat(substring(., 1, 4), '-', substring(., string-length(.)-3, 4))"/>
219
                                                </datacite:relatedIdentifier>
220
                                            </xsl:for-each>
221
                                        </datacite:relatedIdentifiers><!-- change namespace/format from oaire to datacite for description, language, rights, ... -->
222
                                        <xsl:for-each select="//oaire:resourceType">
223
                                            <datacite:resourceType>
224
                                                <xsl:attribute name="xs:anyURI" select="@uri"/>
225
                                                <xsl:value-of select="."/>
226
                                            </datacite:resourceType>
227
                                        </xsl:for-each>
228
                                        <xsl:for-each select="distinct-values(//oaire:version)">
229
                                            <datacite:version>
230
                                                <xsl:value-of select="."/>
231
                                            </datacite:version>
232
                                        </xsl:for-each>
233
                                        <xsl:if test="(//datacite:rights, //oaire:licenseCondition)[string-length(.) &gt; 0 or string-length(@rightsURI) &gt; 0 or string-length(@uri) &gt; 0]">
234
                                            <datacite:rightsList>
235
                                                <xsl:for-each
236
                                                        select="(//datacite:rights | //oaire:licenseCondition)[string-length(.) &gt; 0 or string-length(@rightsURI) &gt; 0 or string-length(@uri) &gt; 0]">
237
                                                    <datacite:rights>
238
                                                        <xsl:attribute name="rightsURI" select="(@rightsURI, @uri)[1]"/>
239
                                                        <xsl:value-of select="."/>
240
                                                    </datacite:rights>
241
                                                </xsl:for-each>
242
                                            </datacite:rightsList>
243
                                        </xsl:if>
244
                                        <xsl:if test="//datacite:title">
245
                                            <datacite:titles>
246
                                                <xsl:for-each select="//datacite:title">
247
                                                    <datacite:title>
248
                                                        <xsl:for-each select="@*">
249
                                                            <xsl:copy/>
250
                                                        </xsl:for-each>
251
                                                        <xsl:value-of select="."/>
252
                                                    </datacite:title>
253
                                                </xsl:for-each>
254
                                            </datacite:titles>
255
                                        </xsl:if>
256
                                        <xsl:if test="//dc:description">
257
                                            <datacite:descriptions>
258
                                                <xsl:for-each select="//dc:description">
259
                                                    <datacite:description>
260
                                                        <xsl:attribute name="descriptionType" select="'Abstract'"/>
261
                                                        <xsl:for-each select="@*">
262
                                                            <xsl:copy/>
263
                                                        </xsl:for-each>
264
                                                        <xsl:value-of select="."/>
265
                                                    </datacite:description>
266
                                                </xsl:for-each>
267
                                            </datacite:descriptions>
268
                                        </xsl:if>
269
                                        <xsl:for-each
270
                                                select="distinct-values(//dc:language)[string-length(normalize-space(.)) &gt; 0]">
271
                                            <datacite:language>
272
                                                <xsl:value-of select="normalize-space(.)"/>
273
                                            </datacite:language>
274
                                        </xsl:for-each>
275
                                        <xsl:for-each
276
                                                select="distinct-values(//dc:publisher)[string-length(normalize-space(.)) &gt; 0]">
277
                                            <datacite:publisher>
278
                                                <xsl:value-of select="normalize-space(.)"/>
279
                                            </datacite:publisher>
280
                                        </xsl:for-each>
281
                                        <xsl:if test="//dc:format[string-length(normalize-space(.)) &gt; 0]">
282
                                            <datacite:formats>
283
                                                <xsl:for-each
284
                                                        select="distinct-values(//dc:format)[string-length(normalize-space(.)) &gt; 0]">
285
                                                    <datacite:format>
286
                                                        <xsl:value-of select="normalize-space(.)"/>
287
                                                    </datacite:format>
288
                                                </xsl:for-each>
289
                                            </datacite:formats>
290
                                        </xsl:if>
291
                                        <xsl:if test="//oaire:fundingReference[./oaire:awardNumber]">
292
                                            <datacite:fundingReferences>
293
                                                <xsl:for-each select="//oaire:fundingReference[./oaire:awardNumber]">
294
                                                    <datacite:fundingReference>
295
                                                        <datacite:funderName>
296
                                                            <xsl:value-of select="./oaire:funderName"/>
297
                                                        </datacite:funderName>
298
                                                        <datacite:funderIdentifier>
299
                                                            <xsl:attribute name="funderIdentifierType"
300
                                                                           select="'Crossref Funder ID'"/>
301
                                                            <xsl:value-of select="./oaire:funderIdentifier"/>
302
                                                        </datacite:funderIdentifier>
303
                                                        <datacite:awardNumber>
304
                                                            <xsl:value-of select="./oaire:awardNumber"/>
305
                                                        </datacite:awardNumber><!--<xsl:value-of select="."/>-->
306
                                                    </datacite:fundingReference>
307
                                                </xsl:for-each>
308
                                            </datacite:fundingReferences>
309
                                        </xsl:if>
310
                                        <xsl:apply-templates
311
                                                select="(//*[local-name()='resource'], //*[local-name() = 'oai_openaire'])/*[not(local-name() = ('identifier', 'alternateIdentifiers', 'alternateIdentifier', 'relatedIdentifiers', 'relatedIdentifier', 'description', 'titles', 'title', 'language', 'publisher', 'resourceType', 'version', 'fundingReferences', 'fundingReference', 'rights', 'licenseCondition', 'file', 'format', 'audience', 'source', 'coverage'))][not(starts-with(local-name(), 'citation'))]"/><!-- </oaire:resource> -->
312
                                    </datacite:resource><!-- ~~~~~ put oaf elements below datacite:resource ~~~~~ --><!-- oaf:identifier -->
313
                                    <xsl:for-each
314
                                            select="tokenize(normalize-space($varPid), ' ')[string-length(substring-after(normalize-space(.), ':::')) &gt; 0]">
315
                                        <oaf:identifier>
316
                                            <xsl:attribute name="identifierType"
317
                                                           select="substring-before(normalize-space(.), ':::')"/>
318
                                            <xsl:value-of select="substring-after(normalize-space(.), ':::')"/>
319
                                        </oaf:identifier>
320
                                    </xsl:for-each>
321
                                    <xsl:if test="not(starts-with(normalize-space(lower-case($varPid)), 'landingpage')) and string-length($varLandingPage) &gt; 0">
322
                                        <oaf:identifier>
323
                                            <xsl:attribute name="identifierType" select="'landingPage'"/>
324
                                            <xsl:value-of select="$varLandingPage"/>
325
                                        </oaf:identifier>
326
                                    </xsl:if>
327
                                    <xsl:if test="//*[local-name() = 'about']/*[local-name() = 'provenance']/*[local-name() = 'originDescription']/*[local-name() = 'originDescription']/*[local-name() = 'identifier' and string-length(.) &gt; 0]">
328
                                        <oaf:identifier>
329
                                            <xsl:attribute name="identifierType" select="'oai-original'"/>
330
                                            <xsl:value-of
331
                                                    select="//*[local-name() = 'about']/*[local-name() = 'provenance']//*[local-name() = 'originDescription' and not(./*[local-name() = 'originDescription'])]/*[local-name() = 'identifier']"/>
332
                                        </oaf:identifier>
333
                                    </xsl:if>
334
                                    <xsl:variable name="varEmbargoEndDate"
335
                                                  select="dateCleaner:dateISO(normalize-space(//*[local-name()='date'][@dateType='Available']))"/><!-- resourceType, superType -->
336
                                    <xsl:choose>
337
                                        <xsl:when
338
                                                test="lower-case(//*[local-name()='resourceType']/@resourceTypeGeneral) = ('dataset', 'software', 'literature', 'publication', 'other research product') or not(//*[local-name()='resourceType']/@resourceTypeGeneral)">
339
                                            <dr:CobjCategory>
340
                                                <xsl:variable name="varCobjCategory"
341
                                                              select="vocabulary:clean( //*[local-name()='resourceType']/@resourceTypeGeneral, 'dnet:publication_resource')"/>
342
                                                <xsl:variable name="varSuperType"
343
                                                              select="vocabulary:clean( $varCobjCategory, 'dnet:result_typologies')"/>
344
                                                <xsl:attribute name="type">
345
                                                    <xsl:value-of select="$varSuperType"/>
346
                                                </xsl:attribute>
347
                                                <xsl:value-of select="$varCobjCategory"/>
348
                                            </dr:CobjCategory><!--
349
                                            <dr:CobjCategory>
350
                                                <xsl:attribute name="type" select="//oaf:datasourceprefix[. = '_______qeios' and contains(//dri:recordIdentifier, '/definition/')]/'other', //oaf:datasourceprefix[not(. = '_______qeios' and contains(//dri:recordIdentifier, '/definition/'))]/$varSuperType"/>
351
                                                <xsl:value-of select="$varCobjCategory"/>
352
                                            </dr:CobjCategory>
353
-->
354
                                        </xsl:when><!-- drop journals --><!--  TODO: termination handling must refine -->
355
                                        <xsl:when
356
                                                test="lower-case(//*[local-name()='resourceType']/@uri) = 'http://purl.org/coar/resource_type/c_0640'">
357
                                            <xsl:call-template name="terminate"/>
358
                                        </xsl:when><!--
359
                                        <xsl:otherwise>
360
                                            <xsl:call-template name="terminate"/>
361
                                        </xsl:otherwise>
362
-->
363
                                    </xsl:choose><!-- review status -->
364
                                    <xsl:variable name="varRefereedConvt"
365
                                                  select="for $i in (                             //*[local-name()='resourceType']/(., @uri), //oai:setSpec, //*[local-name()='description'])                      return vocabulary:clean( normalize-space($i), 'dnet:review_levels')"/>
366
                                    <xsl:variable name="varRefereedIdntf"
367
                                                  select="(                         //*[local-name()=('identifier', 'alternateIdentifier', 'file')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier', 'file')]) = 1][matches(lower-case(.), '(^|.*[\.\-_\\/\s\(\)%\d#:])pre[\.\-_\\/\s\(\)%\d#:]?prints?([\.\-_\\/\s\(\)%\d#:].*)?$')]/'0002',                          //*[local-name()=('identifier', 'alternateIdentifier', 'file')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier', 'file')]) = 1][matches(lower-case(.), '(^|.*[\.\-_\\/\s\(\)%\d#:])refereed([\.\-_\\/\s\(\)%\d#:].*)?$')]/'0001',                         //*[local-name()=('identifier', 'alternateIdentifier', 'file')][count(//*[local-name()=('metadata', 'resource')]//*[local-name()=('identifier', 'alternateIdentifier', 'file')]) = 1][matches(lower-case(.), '.*-peer-reviewed-(fulltext-)?article-.*')]/'0001')"/>
368
                                    <xsl:variable name="varRefereedSourc"
369
                                                  select="(//*[local-name()=('publisher', 'source', 'citationTitle')][matches(lower-case(.), '.*[\s\-\.\\_/:%]pre[\s\-\.\\_/:%]?prints?([\s\-\.\\_/:%].*|$)')]/'0002')"/>
370
                                    <xsl:variable name="varRefereedReltn"
371
                                                  select="//*[local-name() = 'relatedIdentifier'][./@relationType/lower-case(.)='isreviewedby']/'0001'"/>
372
                                    <xsl:variable name="varRefereedDesct"
373
                                                  select="(//*[local-name() = 'description']                         [matches(lower-case(.), '.*(this\s*book|this\s*volume|it)\s*(constitutes|presents)\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')"/>
374
                                    <xsl:variable name="varRefereedTitle"
375
                                                  select="(//*[local-name()=('title')][matches(lower-case(.), '.*\[.*peer[\s\-\._]*review\s*:.*\]\s*$')]/'0001')"/>
376
                                    <xsl:variable name="varRefereedVersn"
377
                                                  select="vocabulary:clean( normalize-space( //*[local-name()='version']), 'dnet:review_levels')"/>
378
                                    <xsl:variable name="varRefereed"
379
                                                  select="($varRefereedConvt, $varRefereedIdntf, $varRefereedSourc, $varRefereedReltn, $varRefereedDesct, $varRefereedTitle, $varRefereedVersn)"/>
380
                                    <xsl:choose>
381
                                        <xsl:when test="count($varRefereed[. = '0001']) &gt; 0">
382
                                            <oaf:refereed>
383
                                                <xsl:value-of select="'0001'"/>
384
                                            </oaf:refereed>
385
                                        </xsl:when>
386
                                        <xsl:when test="count($varRefereed[. = '0002']) &gt; 0">
387
                                            <oaf:refereed>
388
                                                <xsl:value-of select="'0002'"/>
389
                                            </oaf:refereed>
390
                                        </xsl:when>
391
                                    </xsl:choose>
392
                                    <oaf:dateAccepted>
393
                                        <xsl:value-of
394
                                                select="dateCleaner:dateISO( normalize-space(//datacite:date[@dateType = 'Issued']))"/>
395
                                    </oaf:dateAccepted><!--
396
                     <oaf:accessrights>
397
                     <xsl:variable name='varAccessRights' select="TransformationFunction:convertString($tf, (//*[local-name() = 'rights']/(@uri, @rightsURI))[1], 'AccessRights')" />
398
                     </oaf:accessrights>
399
                     <xsl:choose>
400
                     <xsl:when test="not($varAccessRights = 'EMBARGO' and not((xs:date( max( ($varEmbargoEndDate, '0001-01-01') ) ) gt current-date())))">
401
                     <xsl:value-of select="TransformationFunction:convertString($tf, (//*[local-name() = 'rights']/(@uri, @rightsURI))[1], 'AccessRights')"/>
402
                     </xsl:when>
403
                     <xsl:when test="$varAccessRights = 'EMBARGO' and not((xs:date( max( ($varEmbargoEndDate, '0001-01-01') ) ) gt current-date()))">
404
                     <xsl:value-of select="'OPEN'"/>
405
                     </xsl:when>
406
                     </xsl:choose>
407
                -->
408
                                    <xsl:variable name="varAccessRights"
409
                                                  select="string-join((for $i in ((//*[local-name() = 'rights'], //*[lower-case(local-name())='licensecondition'])/(@*, .)) return  vocabulary:clean( $i, 'dnet:access_modes')), ' ')"/>
410
                                    <oaf:accessrights>
411
                                        <xsl:choose>
412
                                            <xsl:when test="contains($varAccessRights, 'OPEN SOURCE')">
413
                                                <xsl:value-of select="'OPEN SOURCE'"/>
414
                                            </xsl:when>
415
                                            <xsl:when test="contains($varAccessRights, 'OPEN')">
416
                                                <xsl:value-of select="'OPEN'"/>
417
                                            </xsl:when>
418
                                            <xsl:when
419
                                                    test="contains($varAccessRights, 'EMBARGO') and not((xs:date( max( ($varEmbargoEndDate, '0001-01-01') ) ) gt current-date()))">
420
                                                <xsl:value-of select="'OPEN'"/>
421
                                            </xsl:when>
422
                                            <xsl:when
423
                                                    test="contains($varAccessRights, 'EMBARGO') and (xs:date( max( ($varEmbargoEndDate, '0001-01-01') ) ) gt current-date())">
424
                                                <xsl:value-of select="'EMBARGO'"/>
425
                                            </xsl:when>
426
                                            <xsl:when test="contains($varAccessRights, 'RESTRICTED')">
427
                                                <xsl:value-of select="'RESTRICTED'"/>
428
                                            </xsl:when>
429
                                            <xsl:when test="contains($varAccessRights, 'CLOSED')">
430
                                                <xsl:value-of select="'CLOSED'"/>
431
                                            </xsl:when>
432
                                        </xsl:choose>
433
                                    </oaf:accessrights>
434
                                    <xsl:for-each
435
                                            select="//*[local-name()='licenseCondition']                     [string-length(.) &gt; 0 or string-length(@uri) &gt; 0]                     [(starts-with(@uri, 'http') and (contains(@uri, '://creativecommons.org/licenses/') or contains(@uri, '://creativecommons.org/publicdomain/') or contains(@uri, '://opensource.org/licenses/') or contains(@uri, '://opendatacommons.org/licenses/') or contains(@uri, '://rightsstatements.org/page/') or contains(@uri, '://rightsstatements.org/vocab/') or contains(@uri, '://www.opendefinition.org/licenses/') or contains(@uri, '://www.gnu.org/licenses/') or contains(@uri, '://artlibre.org/licence/') or contains(@uri, '://repositorio.uca.edu.sv/jspui/licencias/') or contains(@uri, '://bibliotecavirtual.unl.edu.ar/licencia/'))) or matches(., '^CC[- ]BY([- ](NC([- ](ND|SA))?|ND|SA))([- ]\d(\.\d)?)?$', 'i')]">
436
                                        <oaf:license>
437
                                            <xsl:value-of select=".[not(./@uri)], .[./@uri]/@uri"/>
438
                                        </oaf:license>
439
                                    </xsl:for-each>
440
                                    <oaf:language>
441
                                        <xsl:value-of
442
                                                select="vocabulary:clean( //*[local-name()='language'], 'dnet:languages')"/>
443
                                    </oaf:language>
444
                                    <xsl:call-template name="funding"/>
445
                                    <oaf:hostedBy>
446
                                        <xsl:attribute name="name">
447
                                            <xsl:value-of select="$varOfficialName"/>
448
                                        </xsl:attribute>
449
                                        <xsl:attribute name="id">
450
                                            <xsl:value-of select="$varDataSourceId"/>
451
                                        </xsl:attribute>
452
                                    </oaf:hostedBy>
453
                                    <oaf:collectedFrom>
454
                                        <xsl:attribute name="name">
455
                                            <xsl:value-of select="$varOfficialName"/>
456
                                        </xsl:attribute>
457
                                        <xsl:attribute name="id">
458
                                            <xsl:value-of select="$varDataSourceId"/>
459
                                        </xsl:attribute>
460
                                    </oaf:collectedFrom>
461
                                    <xsl:for-each select="//*[local-name()='subject']">
462
                                        <xsl:if test="starts-with(./text(), 'EOSC::')">
463
                                            <oaf:eoscifguidelines>
464
                                                <xsl:attribute name="code" select="./text()"/>
465
                                                <xsl:attribute name="label" select="./text()"/>
466
                                                <xsl:attribute name="url"/>
467
                                                <xsl:attribute name="semanticrelation" select="string('compliesWith')"/>
468
                                            </oaf:eoscifguidelines>
469
                                        </xsl:if>
470
                                    </xsl:for-each>
471
                                    <!-- oaf:container --><!-- TODO: set @typ --><!--
472
                <xsl:variable name="varCitation" select="//*[starts-with(local-name(), 'citation')]" />
473
                <xsl:variable name="varSource" select="//dc:source[//oaf:datasourceprefix[.=('od______1514', 'od______3158')]]" />
474
--><!-- test
475
                     <xsl:if test="contains($varContainerPid, ':::')">
476
                     <oaf:container>
477
                     <xsl:for-each select="tokenize(normalize-space($varContainerPid), ' ')">
478
                     <xsl:attribute name="{lower-case(substring-before(., ':::'))}" select="substring-before(substring-after(., ':::'), ':::')"/>
479
                     <xsl:attribute name="typ" select="substring-after(substring-after(., ':::'), ':::')"/>
480
                     </xsl:for-each>
481
                     <xsl:choose>
482
                     <xsl:when test="lower-case(substring-before(., ':::')) = ('issn', 'pissn', 'eissn', 'lissn')">
483
                     <xsl:attribute name="vol" select="$varCitation[local-name(.) = 'citationVolume']"/>
484
                     <xsl:attribute name="iss" select="$varCitation[local-name(.) = 'citationIssue']"/>
485
                     <xsl:attribute name="sp" select="$varCitation[local-name(.) = 'citationStartPage']"/>
486
                     <xsl:attribute name="ep" select="$varCitation[local-name(.) = 'citationEndPage']"/>
487
                     </xsl:when>
488
                     <xsl:when test="lower-case(substring-before(., ':::')) = 'isbn'">
489
                     <xsl:attribute name="edt" select="$varCitation[local-name(.) = 'citationEdition']"/>
490
                     <xsl:attribute name="sp" select="$varCitation[local-name(.) = 'citationStartPage']"/>
491
                     <xsl:attribute name="ep" select="$varCitation[local-name(.) = 'citationEndPage']"/>
492
                     </xsl:when>
493
                     </xsl:choose>
494
                     <xsl:value-of select="$varCitation[local-name(.) = 'citationTitle'], $varSource" />
495
                     </oaf:container>
496
                     </xsl:if>
497
                     test --><!-- Huelva marks L, E ISSNs as ISSNs, with mark within field in spaces or after blanc --><!-- Qeios declares many records as text, although many seem to be definitions which are also related to 'journal' volumes/issues --><!-- oaf:fulltext--><!-- toDo: clarify if fulltext should be filled when URL given and rights oa --><!--
498
                     <xsl:if test="//*[local-name() = 'file']">
499
                     <oaf:fulltext>
500
                     <xsl:value-of select="//*[local-name() = 'file']"/>
501
                     </oaf:fulltext>
502
                     </xsl:if>
503
                -->
504
                                </metadata>
505
                                <xsl:copy-of select="//*[local-name() = 'about']"/>
506
                            </record>
507
                        </xsl:template>
508
                        <!-- Remove EOSC:: subjects, they need to go to eoscifguidelines  -->
509
                        <xsl:template match="//*[local-name()='subject' and starts-with(./text(), 'EOSC::')]"/>
510
                        <xsl:template match="node()|@*">
511
                            <xsl:copy copy-namespaces="no">
512
                                <xsl:apply-templates select="node()|@*"/>
513
                            </xsl:copy>
514
                        </xsl:template>
515
                        <xsl:template match="//*[local-name() = 'metadata']//*[local-name() = 'resource']">
516
                            <xsl:copy copy-namespaces="no">
517
                                <xsl:apply-templates select="node()|@*"/>
518
                            </xsl:copy>
519
                        </xsl:template>
520
                        <xsl:template name="funding"><!-- funding -->
521
                            <xsl:for-each
522
                                    select="//*[local-name()='fundingReference'][./*[local-name()='awardNumber']]">
523
                                <xsl:choose><!-- FP7 -->
524
                                    <xsl:when
525
                                            test="(./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100004963|10.13039/100011199|10.13039/100011102)\s*$')]                          or (./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000780)$')] and ./*[local-name()='fundingStream'] = ('Framework Programme Seven', 'FP7'))                         or (./*[local-name()='funderName'] = 'European Commission' and ./*[local-name()='fundingStream'] = ('Framework Programme Seven', 'FP7')))                     and ./*[local-name()='awardNumber'][matches(., '.*(^|[^\d])\d\d\d\d\d\d($|[^\d]).*')]">
526
                                        <oaf:projectid>
527
                                            <xsl:value-of
528
                                                    select="concat($varFP7, replace(./*[local-name()='awardNumber'], '.*(^|[^\d])(\d\d\d\d\d\d)($|[^\d]).*', '$2'))"/>
529
                                        </oaf:projectid>
530
                                    </xsl:when><!-- H2020 (Horizon 2020 Framework Programme) --><!-- -->
531
                                    <xsl:when
532
                                            test="(./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100010661)\s*')]                         or (./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000780)$')] and ./*[local-name()='fundingStream'] = ('Horizon 2020 Framework Programme', 'H2020'))                         or (./*[local-name()='funderName'] = 'European Commission' and ./*[local-name()='fundingStream'] = ('Horizon 2020 Framework Programme', 'H2020')))                     and ./*[local-name()='awardNumber'][matches(., '.*(^|[^\d])\d\d\d\d\d\d($|[^\d]).*')]">
533
                                        <oaf:projectid>
534
                                            <xsl:value-of
535
                                                    select="concat($varH2020, replace(./*[local-name()='awardNumber'], '.*(^|[^\d])(\d\d\d\d\d\d)($|[^\d]).*', '$2'))"/>
536
                                        </oaf:projectid>
537
                                    </xsl:when><!-- HE (Horizon Europe Framework Programme) --><!-- -->
538
                                    <xsl:when
539
                                            test="(./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100018693)\s*')]                         or (./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000780)$')] and ./*[local-name()='fundingStream'] = ('Horizon Europe Framework Programme', 'HE'))                         or (./*[local-name()='funderName'] = 'European Commission' and ./*[local-name()='fundingStream'] = ('Horizon Europe Framework Programme', 'HE')))                     and ./*[local-name()='awardNumber'][matches(., '.*(^|[^\d])\d\d\d\d\d\d\d\d($|[^\d]).*')]">
540
                                        <oaf:projectid>
541
                                            <xsl:value-of
542
                                                    select="concat($varHEUROPE, replace(./*[local-name()='awardNumber'], '.*(^|[^\d])(\d\d\d\d\d\d\d\d)($|[^\d]).*', '$2'))"/>
543
                                        </oaf:projectid>
544
                                    </xsl:when><!-- AKA -->
545
                                    <xsl:when
546
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100002341)\s*$') ]                     or ./*[local-name()='funderName'][contains(., 'Suomen Akatemia') or contains(., 'Academy of Finland')]">
547
                                        <oaf:projectid>
548
                                            <xsl:value-of select="concat($varAKA, ./*[local-name()='awardNumber'])"/>
549
                                        </oaf:projectid>
550
                                    </xsl:when><!-- ARC -->
551
                                    <xsl:when
552
                                            test="(./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000923)\s*$')]                          or ./*[local-name()='funderName'][contains(., 'Australian Research Council')])                      and ./*[local-name()='awardNumber'][matches(., '^\d{6}$')]">
553
                                        <oaf:projectid>
554
                                            <xsl:value-of
555
                                                    select="concat($varAKA, replace(./*[local-name()='awardNumber'], '.*(^\d{6}$).*', '$2'))"/>
556
                                        </oaf:projectid>
557
                                    </xsl:when><!-- CONICYT -->
558
                                    <xsl:when
559
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100002848)\s*$')]                     or ./*[local-name()='funderName'][contains(., 'Comisión Nacional de Investigación Científica y Tecnológica') or contains(., 'CONICYT')]">
560
                                        <oaf:projectid>
561
                                            <xsl:value-of
562
                                                    select="concat($varCONICYT, ./*[local-name()='awardNumber'])"/>
563
                                        </oaf:projectid>
564
                                    </xsl:when><!-- DFG -->
565
                                    <xsl:when
566
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100001659)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Deutsche Forschungsgemeinschaft') or contains(., 'DFG') or contains(., 'Deutsche Forschungsgemeinschaft (DFG)')]">
567
                                        <oaf:projectid>
568
                                            <xsl:value-of select="concat($varDFG, ./*[local-name()='awardNumber'])"/>
569
                                        </oaf:projectid>
570
                                    </xsl:when><!----><!-- FCT -->
571
                                    <xsl:when
572
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100001871)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Fundação para a Ciência e a Tecnologia')]">
573
                                        <oaf:projectid>
574
                                            <xsl:value-of select="concat($varFCT, ./*[local-name()='awardNumber'])"/>
575
                                        </oaf:projectid>
576
                                    </xsl:when><!-- FWF -->
577
                                    <xsl:when
578
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100002428)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Fonds zur Förderung der Wissenschaftlichen Forschung') or contains(., 'Austrian Science Fund')]">
579
                                        <oaf:projectid>
580
                                            <xsl:value-of select="concat($varFCT, ./*[local-name()='awardNumber'])"/>
581
                                        </oaf:projectid>
582
                                    </xsl:when><!-- MESTD -->
583
                                    <xsl:when
584
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100001871)\s*$')]                      or ./*[local-name()='funderName'][(contains(., 'Ministarstvo Prosvete, Nauke i Tehnolo') and contains(., 'kog Razvoja')) or contains(., 'MESTD')]">
585
                                        <oaf:projectid>
586
                                            <xsl:value-of select="concat($varMESTD, ./*[local-name()='awardNumber'])"/>
587
                                        </oaf:projectid>
588
                                    </xsl:when><!-- MZOS -->
589
                                    <xsl:when
590
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100006588)\s*$')]                     or ./*[local-name()='funderName'][contains(., 'Ministarstvo Znanosti, Obrazovanja i Sporta') or contains(., 'Ministry of Science, Education and Sports')]">
591
                                        <oaf:projectid>
592
                                            <xsl:value-of select="concat($varMZOS, ./*[local-name()='awardNumber'])"/>
593
                                        </oaf:projectid>
594
                                    </xsl:when><!-- NHMRC -->
595
                                    <xsl:when
596
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000925)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'National Health and Medical Research Council') or contains(., 'NHMRC')]">
597
                                        <oaf:projectid>
598
                                            <xsl:value-of select="concat($varNHMRC, ./*[local-name()='awardNumber'])"/>
599
                                        </oaf:projectid>
600
                                    </xsl:when><!-- NIH -->
601
                                    <xsl:when
602
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100000002)\s*$')]                     or ./*[local-name()='funderName'][contains(., 'National Institutes of Health')]">
603
                                        <oaf:projectid>
604
                                            <xsl:value-of select="concat($varNIH, ./*[local-name()='awardNumber'])"/>
605
                                        </oaf:projectid>
606
                                    </xsl:when><!-- NSF -->
607
                                    <xsl:when
608
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100000001)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'National Science Foundation')]">
609
                                        <oaf:projectid>
610
                                            <xsl:value-of select="concat($varNSF, ./*[local-name()='awardNumber'])"/>
611
                                        </oaf:projectid>
612
                                    </xsl:when><!-- NWO -->
613
                                    <xsl:when
614
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100003246)\s*$')]                     or ./*[local-name()='funderName'][contains(., 'Netherlands Organisation for Scientific Research') or contains(., 'Nederlandse Organisatie voor Wetenschappelijk Onderzoek')]">
615
                                        <oaf:projectid>
616
                                            <xsl:value-of select="concat($varNWO, ./*[local-name()='awardNumber'])"/>
617
                                        </oaf:projectid>
618
                                    </xsl:when><!-- RCUK -->
619
                                    <xsl:when
620
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100000690)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Research Councils UK') or contains(., 'RCUK')]">
621
                                        <oaf:projectid>
622
                                            <xsl:value-of select="concat($varUKRI, ./*[local-name()='awardNumber'])"/>
623
                                        </oaf:projectid>
624
                                    </xsl:when><!-- SFI -->
625
                                    <xsl:when
626
                                            test="(./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100001602)\s*$')]                          or ./*[local-name()='funderName'][contains(., 'Science Foundation Ireland')])                      and ./*[local-name()='awardNumber'][matches(., '.*([\dA-Za-z\.\-]+/)+[\dA-Za-z\.\-]+.*')]">
627
                                        <oaf:projectid>
628
                                            <xsl:value-of
629
                                                    select="concat($varSFI, replace(./*[local-name()='awardNumber'], '.*(^|\s)(([\dA-Za-z\.\-]+/)+[\dA-Za-z\.\-]+)($|\s).*', '$2'))"/>
630
                                        </oaf:projectid>
631
                                    </xsl:when><!-- SNSF -->
632
                                    <xsl:when
633
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100001711)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Swiss National Science Foundation') or contains(., 'Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung')]">
634
                                        <oaf:projectid>
635
                                            <xsl:value-of select="concat($varSNSF, ./*[local-name()='awardNumber'])"/>
636
                                        </oaf:projectid>
637
                                    </xsl:when><!-- TUBITAK -->
638
                                    <xsl:when
639
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/501100004410)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Turkish National Science and Research Council') or (contains(., 'Türkiye Bilimsel ve Teknolojik Ara') and contains(., 'rma Kurumu'))]">
640
                                        <oaf:projectid>
641
                                            <xsl:value-of
642
                                                    select="concat($varTUBITAK, ./*[local-name()='awardNumber'])"/>
643
                                        </oaf:projectid>
644
                                    </xsl:when><!-- UKRI -->
645
                                    <xsl:when
646
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100014013)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'UK Research and Innovation') or contains(., 'UKRI')]">
647
                                        <oaf:projectid>
648
                                            <xsl:value-of select="concat($varUKRI, ./*[local-name()='awardNumber'])"/>
649
                                        </oaf:projectid>
650
                                    </xsl:when><!-- WT -->
651
                                    <xsl:when
652
                                            test="./*[local-name()='funderIdentifier'][matches(., '.*(doi.org/)?(10.13039/100004440)\s*$')]                      or ./*[local-name()='funderName'][contains(., 'Wellcome Trust') or . = 'WT']">
653
                                        <oaf:projectid>
654
                                            <xsl:value-of select="concat($varWT, ./*[local-name()='awardNumber'])"/>
655
                                        </oaf:projectid>
656
                                    </xsl:when>
657
                                </xsl:choose>
658
                            </xsl:for-each>
659
                        </xsl:template><!-- toDo: improve by selecting all attributes -->
660
                        <xsl:template match="//*[local-name()='nameIdentifier']">
661
                            <datacite:nameIdentifier>
662
                                <xsl:attribute name="nameIdentifierScheme"
663
                                               select="./@*[local-name()=('nameIdentifierScheme')]"/>
664
                                <xsl:attribute name="schemeURI" select="./@*[local-name()=('schemeURI')]"/>
665
                                <xsl:choose>
666
                                    <xsl:when test=".[contains(., 'orcid.org/')]">
667
                                        <xsl:value-of select="substring-after(., 'orcid.org/')"/>
668
                                    </xsl:when>
669
                                    <xsl:when test=".[contains(., 'researcherid.com/rid/')]">
670
                                        <xsl:value-of select="substring-after(., 'researcherid.com/rid/')"/>
671
                                    </xsl:when>
672
                                    <xsl:otherwise>
673
                                        <xsl:value-of select="."/>
674
                                    </xsl:otherwise>
675
                                </xsl:choose>
676
                            </datacite:nameIdentifier>
677
                        </xsl:template>
678
                        <xsl:template match="//*[local-name() = 'header']">
679
                            <header xmlns="HTTP://www.openarchives.org/OAI/2.0/">
680
                                <xsl:apply-templates select="node()|@*"/>
681
                                <xsl:element name="dr:dateOfTransformation">
682
                                    <xsl:value-of select="$transDate"/>
683
                                </xsl:element>
684
                            </header>
685
                        </xsl:template>
686
                    </xsl:stylesheet>
687
                </CODE>
688
            </SCRIPT>
689
        </CONFIGURATION>
690
        <STATUS/>
691
        <SECURITY_PARAMETERS/>
692
    </BODY>
693
</RESOURCE_PROFILE>
(61-61/64)