D-Net
/
dnet-hadoop
13
4
Fork 10
Code Issues 14 Pull Requests 8 Projects 2 Releases 1 Wiki Activity

xslt

This commit is contained in:
Michele Artini 2024-03-07 13:06:41 +01:00
parent a059747f16
commit 6648d710a3
1 changed files with 14 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/collection/plugin/base/tr/base-2-oaf.xml
View File

@ -12,7 +12,7 @@
<SINK_METADATA_FORMAT name="odf_hbase" />
<IMPORTED />
<SCRIPT>
<TITLE>xslt_dc2oaf_base_hadoop</TITLE>
<TITLE>xslt_base2oaf_hadoop</TITLE>
<CODE>
<xsl:stylesheet xmlns:oaire="http://namespace.openaire.eu/schema/oaire/" xmlns:dateCleaner="http://eu/dnetlib/transform/dateISO"
xmlns:base_dc="http://oai.base-search.net/base_dc/"
@ -51,7 +51,7 @@ base_dc:collection/text()
base_dc:collection/@ror_id
base_dc:continent
base_dc:country
base_dc:year (I used dc:date)
base_dc:year (I used dc:date)
dc:coverage
dc:language (I used base_dc:lang)
base_dc:link (I used dc:identifier)
@ -86,7 +86,7 @@ base_dc:link (I used dc:identifier)
<!-- TODO: I'm not sure if this is the correct encoding -->
<xsl:for-each select="//base_dc:classcode|//base_dc:autoclasscode">
<dc:subject><xsl:value-of select="concat(@type, ':', .)" /></dc:subject>
<dc:subject><xsl:value-of select="concat(@type, ':', .)" /></dc:subject>
</xsl:for-each>
<!-- END TODO -->
@ -168,20 +168,20 @@ base_dc:link (I used dc:identifier)
<oaf:hostedBy>
<xsl:attribute name="name">
<xsl:value-of select="//base_dc:collname" />
</xsl:attribute>
<xsl:value-of select="//base_dc:collname" />
</xsl:attribute>
<xsl:attribute name="id">
<xsl:value-of select="concat('opendoar____::', //base_dc:collection/@opendoar_id)" />
</xsl:attribute>
<xsl:value-of select="concat('opendoar____::', //base_dc:collection/@opendoar_id)" />
</xsl:attribute>
</oaf:hostedBy>
<oaf:collectedFrom>
<xsl:attribute name="name">
<xsl:value-of select="$varOfficialName" />
</xsl:attribute>
<xsl:value-of select="$varOfficialName" />
</xsl:attribute>
<xsl:attribute name="id">
<xsl:value-of select="$varDataSourceId" />
</xsl:attribute>
<xsl:value-of select="$varDataSourceId" />
</xsl:attribute>
</oaf:collectedFrom>
<oaf:dateAccepted>
@ -213,7 +213,7 @@ base_dc:link (I used dc:identifier)
<xsl:variable name="varRefereedConvt"
select="for $i in (//dc:type, //dc:description, //oai:setSpec) return vocabulary:clean( normalize-space($i), 'dnet:review_levels')" />
select="for $i in (//dc:type, //dc:description, //oai:setSpec) return vocabulary:clean( normalize-space($i), 'dnet:review_levels')" />
<xsl:variable name="varRefereedIdntf"
select="(//*[string(node-name(.)) = 'dc:identifier' and matches(lower-case(.), '(^|.*[\.\-_/\s\(\)%\d#])pre[\.\-_/\s\(\)%\d#]?prints?([\.\-_/\s\(\)%\d#].*)?$')][count(//dc:identifier) = 1]/'0002', //*[string(node-name(.)) = 'dc:identifier' and matches(lower-case(.), '(^|.*[\.\-_/\s\(\)%\d#])refereed([\.\-_/\s\(\)\d%\d#].*)?$')]/'0001', //*[string(node-name(.)) = 'dc:identifier' and matches(lower-case(.), '.*-peer-reviewed-(fulltext-)?article-.*')]/'0001')" />
<xsl:variable name="varRefereedSourc"
@ -221,9 +221,9 @@ base_dc:link (I used dc:identifier)
<xsl:variable name="varRefereedDescr"
select="(//dc: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', //dc:description[matches(., '^version\s*(préliminaire.*|preliminary.*|0$)')]/'0002')" />
<xsl:variable name="varRefereedTitle"
select="(//dc:title[matches(lower-case(.), '.*\[.*peer[\s\-\._]*review\s*:.*\]\s*$')]/'0001', //dc:title[matches(lower-case(.), '.*\(\s*pre[\s\-\._]*prints?\s*\)\s*$')]/'0002')" />
select="(//dc:title[matches(lower-case(.), '.*\[.*peer[\s\-\._]*review\s*:.*\]\s*$')]/'0001', //dc:title[matches(lower-case(.), '.*\(\s*pre[\s\-\._]*prints?\s*\)\s*$')]/'0002')" />
<xsl:variable name="varRefereedSubjt"
select="(//dc:subject[matches(lower-case(.), '^\s*refereed\s*$')][//oaf:datasourceprefix = 'narcis______']/'0001', //dc:subject[matches(lower-case(.), '^\s*no[nt].{0,3}refereed\s*$')][//oaf:datasourceprefix = 'narcis______']/'0002')" />
select="(//dc:subject[matches(lower-case(.), '^\s*refereed\s*$')][//oaf:datasourceprefix = 'narcis______']/'0001', //dc:subject[matches(lower-case(.), '^\s*no[nt].{0,3}refereed\s*$')][//oaf:datasourceprefix = 'narcis______']/'0002')" />
<xsl:variable name="varRefereed" select="($varRefereedConvt, $varRefereedIdntf, $varRefereedSourc, $varRefereedDescr, $varRefereedTitle, $varRefereedSubjt)" />
@ -245,7 +245,6 @@ base_dc:link (I used dc:identifier)
for DOIs reduced here - letters like less-than and quotation marks don't work in matches, use identiferExtract when enabled -->
<xsl:variable name="varIdHdl" select="distinct-values(//dc:identifier[starts-with(., 'http') and contains(., '://hdl.handle.net/')]/substring-after(., 'hdl.handle.net/'))" />
<xsl:for-each select="$varIdHdl">
<oaf:identifier>