D-Net
/
dnet-hadoop
15
4
Fork 10
Code Issues 14 Pull Requests 8 Projects 2 Releases 1 Wiki Activity

base transformation rule test

Browse Source
base_stats_job_deprecated
Michele Artini 2 months ago
parent dfb05ebedb
commit 3e56b88a5f
6 changed files with 144 additions and 51 deletions
Show Stats Download Patch File Download Diff File

12
dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/collection/plugin/base/xml/base2oaf.transformationRule.xml
Unescape Escape View File

@ -188,7 +188,7 @@ base_dc:link (I used dc:identifier)
</oaf:identifier>
</xsl:for-each>
<xsl:for-each select="distinct-values(//dc:identifier[starts-with(., 'http')][not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/'))][count(index-of($varIdLdpg, .)) = 0])">
<xsl:for-each select="distinct-values(//dc:identifier[starts-with(., 'http') and (not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/')))])">
<oaf:identifier identifierType="url">
<xsl:value-of select="." />
</oaf:identifier>
@ -233,10 +233,12 @@ base_dc:link (I used dc:identifier)
<xsl:value-of select="dateCleaner:dateISO( //dc:date[1] )" />
</oaf:dateAccepted>
<xsl:if test="//dc:relation[starts-with(., 'http')] and //base_dc:oa[.='1']">
<oaf:fulltext>
<xsl:value-of select="//dc:relation[starts-with(., 'http')]" />
</oaf:fulltext>
<xsl:if test="//base_dc:oa[.='1']">
<xsl:for-each select="//dc:relation[starts-with(., 'http')]">
<oaf:fulltext>
<xsl:value-of select="normalize-space(.)" />
</oaf:fulltext>
</xsl:for-each>
</xsl:if>
<xsl:for-each select="//base_dc:collection/@ror_id">

39
dhp-workflows/dhp-aggregation/src/main/resources/eu/dnetlib/dhp/collection/plugin/base/xml/base2odf.transformationRule.xml
Unescape Escape View File

@ -66,7 +66,7 @@
<datacite:alternateIdentifiers>
<xsl:for-each
select="distinct-values(//dc:identifier[starts-with(., 'http')][not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/'))][count(index-of($varIdLdpg, .)) = 0])">
select="distinct-values(//dc:identifier[starts-with(., 'http') and (not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/')))])">
<datacite:identifier alternateIdentifierType="url">
<xsl:value-of select="." />
</datacite:identifier>
@ -110,7 +110,7 @@
<datacite:creatorName>
<xsl:value-of select="$author" />
</datacite:creatorName>
<xsl:for-each select="//base_dc:authod_id[normalize-space(./base_dc:creator_name = $author]/base_dc:creator_id ">
<xsl:for-each select="//base_dc:authod_id[normalize-space(./base_dc:creator_name) = $author]/base_dc:creator_id ">
<xsl:if test="contains(.,'https://orcid.org/')">
<nameIdentifier schemeURI="https://orcid.org/" nameIdentifierScheme="ORCID">
<xsl:value-of select="substring-after(., 'https://orcid.org/')" />
@ -175,14 +175,12 @@
</datacite:language>
<oaf:accessrights>
<xsl:choose>
<xsl:if test="//base_dc:oa[.='1']">
<datacite:rights rightsURI="http://purl.org/coar/access_right/c_abf2">open access</datacite:rights>
</xsl:if>
<xsl:for-each select="//dc:rights|//base_dc:rightsnorm">
<datacite:rights><xsl:value-of select="vocabulary:clean(., 'dnet:access_modes')" /></datacite:rights>
</xsl:for-each>
</xsl:choose>
<xsl:if test="//base_dc:oa[.='1']">
<datacite:rights rightsURI="http://purl.org/coar/access_right/c_abf2">open access</datacite:rights>
</xsl:if>
<xsl:for-each select="//dc:rights|//base_dc:rightsnorm">
<datacite:rights><xsl:value-of select="vocabulary:clean(., 'dnet:access_modes')" /></datacite:rights>
</xsl:for-each>
</oaf:accessrights>
</datacite:resource>
@ -209,18 +207,14 @@
<oaf:accessrights>
<xsl:choose>
<xsl:when test="//base_dc:oa[.='1']">
OPEN
</xsl:when>
<xsl:when test="//base_dc:oa[.='1']">OPEN</xsl:when>
<xsl:when test="//base_dc:rightsnorm">
<xsl:value-of select="vocabulary:clean(//base_dc:rightsnorm, 'dnet:access_modes')" />
</xsl:when>
<xsl:when test="//dc:rights">
<xsl:value-of select="vocabulary:clean( //dc:rights, 'dnet:access_modes')" />
</xsl:when>
<xsl:otherwise>
UNKNOWN
</xsl:otherwise>
<xsl:otherwise>UNKNOWN</xsl:otherwise>
</xsl:choose>
</oaf:accessrights>
@ -231,7 +225,7 @@
</xsl:for-each>
<xsl:for-each
select="distinct-values(//dc:identifier[starts-with(., 'http')][not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/'))][count(index-of($varIdLdpg, .)) = 0])">
select="distinct-values(//dc:identifier[starts-with(., 'http') and ( not(contains(., '://dx.doi.org/') or contains(., '://doi.org/') or contains(., '://hdl.handle.net/')))])">
<oaf:identifier identifierType="url">
<xsl:value-of select="." />
</oaf:identifier>
@ -276,10 +270,13 @@
<xsl:value-of select="dateCleaner:dateISO( //dc:date[1] )" />
</oaf:dateAccepted>
<xsl:if test="//dc:relation[starts-with(., 'http')] and //base_dc:oa[.='1']">
<oaf:fulltext>
<xsl:value-of select="//dc:relation[starts-with(., 'http')]" />
</oaf:fulltext>
<xsl:if test="//base_dc:oa[.='1']">
<xsl:for-each select="//dc:relation[starts-with(., 'http')]">
<oaf:fulltext>
<xsl:value-of select="normalize-space(.)" />
</oaf:fulltext>
</xsl:for-each>
</xsl:if>
<xsl:for-each select="//base_dc:collection/@ror_id">

16
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/collection/plugin/base/BaseTransfomationTest.java
Unescape Escape View File

@ -7,6 +7,7 @@ import org.apache.spark.SparkConf;
import org.apache.spark.util.LongAccumulator;
import org.dom4j.io.SAXReader;
import org.junit.jupiter.api.BeforeEach;
import org.junit.jupiter.api.Disabled;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.ExtendWith;
import org.mockito.junit.jupiter.MockitoExtension;
@ -18,6 +19,7 @@ import eu.dnetlib.dhp.schema.mdstore.Provenance;
import eu.dnetlib.dhp.transformation.xslt.XSLTTransformationFunction;
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpException;
@Disabled
@ExtendWith(MockitoExtension.class)
public class BaseTransfomationTest extends AbstractVocabularyTest {
@ -47,6 +49,20 @@ public class BaseTransfomationTest extends AbstractVocabularyTest {
System.out.println(result.getBody());
}
@Test
void testBase2OAF() throws Exception {
final MetadataRecord mr = new MetadataRecord();
mr.setProvenance(new Provenance("DSID", "DSNAME", "PREFIX"));
mr.setBody(IOUtils.toString(getClass().getResourceAsStream("record.xml")));
final XSLTTransformationFunction tr = loadTransformationRule("xml/base2oaf.transformationRule.xml");
final MetadataRecord result = tr.call(mr);
System.out.println(result.getBody());
}
private XSLTTransformationFunction loadTransformationRule(final String path) throws Exception {
final String xslt = new SAXReader()
.read(this.getClass().getResourceAsStream(path))

73
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/collection/plugin/base/record.xml
Unescape Escape View File

@ -1,31 +1,58 @@
<record>
<header xmlns="http://www.openarchives.org/OAI/2.0/">
<identifier>ftterritoryanzac:oai:www.territorystories.nt.gov.au:10070/74188</identifier>
<datestamp>2015-09-07T22:26:28Z</datestamp>
<identifier>ftdoajarticles:oai:doaj.org/article:e2d5b5126b2d4e479933cc7f9a9ae0c1</identifier>
<datestamp>2022-12-31T11:48:55Z</datestamp>
</header>
<metadata xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:base_dc="http://oai.base-search.net/base_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:dc="http://purl.org/dc/elements/1.1/">
<base_dc:dc xsi:schemaLocation="http://oai.base-search.net/base_dc/ http://oai.base-search.net/base_dc/base_dc.xsd">
<base_dc:global_id>ftterritoryanzac:oai:www.territorystories.nt.gov.au:10070/74188</base_dc:global_id>
<base_dc:continent>cau</base_dc:continent>
<base_dc:country>au</base_dc:country>
<base_dc:collection opendoar_id="1234">ftterritoryanzac</base_dc:collection>
<base_dc:collname>Territory Stories (Northern Territory Government, Australia)</base_dc:collname>
<dc:title>W. Wardle</dc:title>
<dc:creator>Wardle, W.</dc:creator>
<dc:description>"NX 112520 Bdr. W. Wardle Darwin Coast Art. 1943-45" ; NX 112520. Bombardier W. Wardle. Darwin Coast Artillery. 1943-1945.</dc:description>
<dc:date>2007-12-18T01:19:36Z</dc:date>
<base_dc:year>2007</base_dc:year>
<dc:type>Image</dc:type>
<dc:type>Reference</dc:type>
<base_dc:typenorm>51</base_dc:typenorm>
<dc:format>Cropped 408w X 338h Size 26.62k</dc:format>
<dc:identifier>http://hdl.handle.net/10070/74188</dc:identifier>
<base_dc:link>http://hdl.handle.net/10070/74188</base_dc:link>
<dc:relation>Darwin Commemorative Wall Quilt</dc:relation>
<dc:relation>http://www.ww2roll.gov.au/script/veteran.asp?ServiceID=A&amp;VeteranID=220307</dc:relation>
<dc:relation>http://hdl.handle.net/10070/74188</dc:relation>
<base_dc:global_id>ftdoajarticles:oai:doaj.org/article:e2d5b5126b2d4e479933cc7f9a9ae0c1</base_dc:global_id>
<base_dc:continent>cww</base_dc:continent>
<base_dc:country>org</base_dc:country>
<base_dc:collection opendoar_id="1234" ror_id="ror1234">ftdoajarticles</base_dc:collection>
<base_dc:collname>TEST REPO</base_dc:collname>
<dc:title>Assessment of cultural heritage: the legislative and methodological framework of Russian Federation</dc:title>
<dc:creator>ALBU, Svetlana</dc:creator>
<dc:creator>LEȘAN, Anna</dc:creator>
<dc:subject>architectural heritage</dc:subject>
<dc:subject>evaluation of architectural heritage</dc:subject>
<dc:subject>types of values</dc:subject>
<dc:subject>experience of russian federation</dc:subject>
<dc:subject>Social Sciences</dc:subject>
<dc:subject>H</dc:subject>
<dc:description>Architectural heritage is the real estate inheritance by population of a country becoming an extremely valuable and specific category, preserving and capitalizing on those assets requires considerable effort. The state does not have sufficient means to maintain and preserve cultural heritage, as a result it is included in the civil circuit. The transfer of property right or of some partial rights over the architectural patrimony is accompanied by the necessity to estimate the value of goods. In this article, the authors examine the experience of Russian Federation (one of the largest countries with a huge architectural heritage) on the legislative framework of architectural and methodological heritage of architectural heritage assessment. The particularities of cultural assets valuation compared to other categories of real estate are examined, as well as the methodological aspects (types of values, methods applied in valuation, approaches according to the purpose of valuation) regarding the valuation of real estate with architectural value in Russian Federation.</dc:description>
<dc:publisher>Technical University of Moldova</dc:publisher>
<dc:date>2020-09-01T00:00:00Z</dc:date>
<base_dc:year>2020</base_dc:year>
<dc:type>article</dc:type>
<base_dc:typenorm>121</base_dc:typenorm>
<dc:identifier>https://doi.org/10.5281/zenodo.3971988</dc:identifier>
<dc:identifier>https://doaj.org/article/e2d5b5126b2d4e479933cc7f9a9ae0c1</dc:identifier>
<base_dc:link>https://doi.org/10.5281/zenodo.3971988</base_dc:link>
<dc:source>Journal of Social Sciences, Vol 3, Iss 3, Pp 134-143 (2020)</dc:source>
<dc:language>EN</dc:language>
<dc:language>FR</dc:language>
<dc:language>RO</dc:language>
<dc:relation>http://ibn.idsi.md/sites/default/files/imag_file/JSS-3-2020_134-143.pdf</dc:relation>
<dc:relation>https://doaj.org/toc/2587-3490</dc:relation>
<dc:relation>https://doaj.org/toc/2587-3504</dc:relation>
<dc:relation>doi:10.5281/zenodo.3971988</dc:relation>
<dc:relation>2587-3490</dc:relation>
<dc:relation>2587-3504</dc:relation>
<dc:relation>https://doaj.org/article/e2d5b5126b2d4e479933cc7f9a9ae0c1</dc:relation>
<base_dc:autoclasscode type="ddc">720</base_dc:autoclasscode>
<base_dc:authod_id>
<base_dc:creator_name>ALBU, Svetlana</base_dc:creator_name>
<base_dc:creator_id>https://orcid.org/0000-0002-8648-950X</base_dc:creator_id>
</base_dc:authod_id>
<base_dc:authod_id>
<base_dc:creator_name>LEȘAN, Anna</base_dc:creator_name>
<base_dc:creator_id>https://orcid.org/0000-0003-3284-0525</base_dc:creator_id>
</base_dc:authod_id>
<base_dc:doi>https://doi.org/10.5281/zenodo.3971988</base_dc:doi>
<base_dc:oa>1</base_dc:oa>
<base_dc:lang>unknown</base_dc:lang>
<base_dc:lang>eng</base_dc:lang>
<base_dc:lang>fre</base_dc:lang>
<base_dc:lang>rum</base_dc:lang>
</base_dc:dc>
</metadata>
</record>
</record>

28
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/transform/synonyms.txt
Unescape Escape View File

@ -1496,4 +1496,30 @@ cnr:institutes @=@ __CDS131__ @=@ IBE - Istituto per la BioEconomia
cnr:institutes @=@ https://ror.org/0263zy895 @=@ CDS132
cnr:institutes @=@ https://ror.org/0263zy895 @=@ SCITEC - Istituto di Scienze e Tecnologie Chimiche \"Giulio Natta\"
cnr:institutes @=@ __CDS133__ @=@ CDS133
cnr:institutes @=@ __CDS133__ @=@ STEMS - Istituto di Scienze e Tecnologie per l'Energia e la Mobilità Sostenibili
cnr:institutes @=@ __CDS133__ @=@ STEMS - Istituto di Scienze e Tecnologie per l'Energia e la Mobilità Sostenibili
base:normalized_types @=@ Text @=@ 1
base:normalized_types @=@ Book @=@ 11
base:normalized_types @=@ Book part @=@ 111
base:normalized_types @=@ Journal/Newspaper @=@ 12
base:normalized_types @=@ Article contribution @=@ 121
base:normalized_types @=@ Other non-article @=@ 122
base:normalized_types @=@ Conference object @=@ 13
base:normalized_types @=@ Report @=@ 14
base:normalized_types @=@ Review @=@ 15
base:normalized_types @=@ Course material @=@ 16
base:normalized_types @=@ Lecture @=@ 17
base:normalized_types @=@ Thesis @=@ 18
base:normalized_types @=@ Bachelor's thesis @=@ 181
base:normalized_types @=@ Master's thesis @=@ 182
base:normalized_types @=@ Doctoral and postdoctoral thesis @=@ 183
base:normalized_types @=@ Manuscript @=@ 19
base:normalized_types @=@ Patent @=@ 1A
base:normalized_types @=@ Musical notation @=@ 2
base:normalized_types @=@ Map @=@ 3
base:normalized_types @=@ Audio @=@ 4
base:normalized_types @=@ Image/Video @=@ 5
base:normalized_types @=@ Still image @=@ 51
base:normalized_types @=@ Moving image/Video @=@ 52
base:normalized_types @=@ Software @=@ 6
base:normalized_types @=@ Dataset @=@ 7
base:normalized_types @=@ Unknown @=@ F

27
dhp-workflows/dhp-aggregation/src/test/resources/eu/dnetlib/dhp/transform/terms.txt
Unescape Escape View File

@ -1210,4 +1210,29 @@ cnr:institutes @=@ cnr:institutes @=@ __CDS130__ @=@ __CDS130__
cnr:institutes @=@ cnr:institutes @=@ __CDS131__ @=@ __CDS131__
cnr:institutes @=@ cnr:institutes @=@ https://ror.org/0263zy895 @=@ https://ror.org/0263zy895
cnr:institutes @=@ cnr:institutes @=@ __CDS133__ @=@ __CDS133__
base:normalized_types @=@ base:normalized_types @=@ Text @=@ Text
base:normalized_types @=@ base:normalized_types @=@ Book @=@ Book
base:normalized_types @=@ base:normalized_types @=@ Book part @=@ Book part
base:normalized_types @=@ base:normalized_types @=@ Journal/Newspaper @=@ Journal/Newspaper
base:normalized_types @=@ base:normalized_types @=@ Article contribution @=@ Article contribution
base:normalized_types @=@ base:normalized_types @=@ Other non-article @=@ Other non-article
base:normalized_types @=@ base:normalized_types @=@ Conference object @=@ Conference object
base:normalized_types @=@ base:normalized_types @=@ Report @=@ Report
base:normalized_types @=@ base:normalized_types @=@ Review @=@ Review
base:normalized_types @=@ base:normalized_types @=@ Course material @=@ Course material
base:normalized_types @=@ base:normalized_types @=@ Lecture @=@ Lecture
base:normalized_types @=@ base:normalized_types @=@ Thesis @=@ Thesis
base:normalized_types @=@ base:normalized_types @=@ Bachelor's thesis @=@ Bachelor's thesis
base:normalized_types @=@ base:normalized_types @=@ Master's thesis @=@ Master's thesis
base:normalized_types @=@ base:normalized_types @=@ Doctoral and postdoctoral thesis @=@ Doctoral and postdoctoral thesis
base:normalized_types @=@ base:normalized_types @=@ Manuscript @=@ Manuscript
base:normalized_types @=@ base:normalized_types @=@ Patent @=@ Patent
base:normalized_types @=@ base:normalized_types @=@ Musical notation @=@ Musical notation
base:normalized_types @=@ base:normalized_types @=@ Map @=@ Map
base:normalized_types @=@ base:normalized_types @=@ Audio @=@ Audio
base:normalized_types @=@ base:normalized_types @=@ Image/Video @=@ Image/Video
base:normalized_types @=@ base:normalized_types @=@ Still image @=@ Still image
base:normalized_types @=@ base:normalized_types @=@ Moving image/Video @=@ Moving image/Video
base:normalized_types @=@ base:normalized_types @=@ Software @=@ Software
base:normalized_types @=@ base:normalized_types @=@ Dataset @=@ Dataset
base:normalized_types @=@ base:normalized_types @=@ Unknown @=@ Unknown
Write Preview
Loading…
Cancel
Save