VOR - add duplicated funding checks

elifesciences · Feb 14, 2025 · 7c73120 · 7c73120
1 parent 564ed6a
commit 7c73120
Show file tree

Hide file tree

Showing 22 changed files with 4,461 additions and 0 deletions.
diff --git a/src/final-JATS-schematron.sch b/src/final-JATS-schematron.sch
@@ -2441,6 +2441,8 @@
 
     </rule></pattern><pattern id="award-id-tests-pattern"><rule context="funding-group/award-group/award-id" id="award-id-tests">
       <let name="id" value="parent::award-group/@id"/>
+      <let name="funder-id" value="parent::award-group/descendant::institution-id[1]"/>
+      <let name="funder-name" value="parent::award-group/descendant::institution[1]"/>
 
       <report see="https://elifeproduction.slab.com/posts/funding-3sv64358#award-id-test-1" test="matches(.,',|;')" role="warning" id="award-id-test-1">[award-id-test-1] Funding entry with id <value-of select="$id"/> has a comma or semi-colon in the award id. Should this be separated out into several funding entries? - <value-of select="."/>.</report>
 
@@ -2452,6 +2454,12 @@
 
       <report test="matches(.,'http[s]?://d?x?\.?doi.org/')" role="error" id="award-id-test-5">[award-id-test-5] Award id contains a DOI link - <value-of select="."/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</report>
 
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]" role="error" id="award-id-test-6">[award-id-test-6] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]" role="error" id="award-id-test-7">[award-id-test-7] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]" role="warning" id="award-id-test-8">[award-id-test-8] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</report>
+
     </rule></pattern><pattern id="institution-wrap-tests-pattern"><rule context="article-meta//award-group//institution-wrap" id="institution-wrap-tests">
 
       <assert see="https://elifeproduction.slab.com/posts/funding-3sv64358#institution-id-test" test="institution-id[@institution-id-type='FundRef']" role="warning" id="institution-id-test">[institution-id-test] Whenever possible, a funder should have a doi - please check whether there is an appropriate doi in the open funder registry. (institution-id[@institution-id-type="FundRef"] is not present in institution-wrap).</assert>

diff --git a/src/final-JATS-schematron.xsl b/src/final-JATS-schematron.xsl
@@ -12209,6 +12209,8 @@
    <!--RULE award-id-tests-->
    <xsl:template match="funding-group/award-group/award-id" priority="1000" mode="M159">
       <xsl:variable name="id" select="parent::award-group/@id"/>
+      <xsl:variable name="funder-id" select="parent::award-group/descendant::institution-id[1]"/>
+      <xsl:variable name="funder-name" select="parent::award-group/descendant::institution[1]"/>
       <!--REPORT warning-->
       <xsl:if test="matches(.,',|;')">
          <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(.,',|;')">
@@ -12280,6 +12282,45 @@
                <xsl:text/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</svrl:text>
          </svrl:successful-report>
       </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]">
+            <xsl:attribute name="id">award-id-test-6</xsl:attribute>
+            <xsl:attribute name="role">error</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-6] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]">
+            <xsl:attribute name="id">award-id-test-7</xsl:attribute>
+            <xsl:attribute name="role">error</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-7] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
+      <!--REPORT warning-->
+      <xsl:if test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]">
+            <xsl:attribute name="id">award-id-test-8</xsl:attribute>
+            <xsl:attribute name="role">warning</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-8] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
       <xsl:apply-templates select="*" mode="M159"/>
    </xsl:template>
    <xsl:template match="text()" priority="-1" mode="M159"/>

diff --git a/src/final-package-JATS-schematron.sch b/src/final-package-JATS-schematron.sch
@@ -2447,6 +2447,8 @@
 
     </rule></pattern><pattern id="award-id-tests-pattern"><rule context="funding-group/award-group/award-id" id="award-id-tests">
       <let name="id" value="parent::award-group/@id"/>
+      <let name="funder-id" value="parent::award-group/descendant::institution-id[1]"/>
+      <let name="funder-name" value="parent::award-group/descendant::institution[1]"/>
 
       <report see="https://elifeproduction.slab.com/posts/funding-3sv64358#award-id-test-1" test="matches(.,',|;')" role="warning" id="award-id-test-1">Funding entry with id <value-of select="$id"/> has a comma or semi-colon in the award id. Should this be separated out into several funding entries? - <value-of select="."/>.</report>
 
@@ -2458,6 +2460,12 @@
 
       <report test="matches(.,'http[s]?://d?x?\.?doi.org/')" role="error" id="award-id-test-5">Award id contains a DOI link - <value-of select="."/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</report>
 
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]" role="error" id="award-id-test-6">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]" role="error" id="award-id-test-7">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]" role="warning" id="award-id-test-8">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</report>
+
     </rule></pattern><pattern id="institution-wrap-tests-pattern"><rule context="article-meta//award-group//institution-wrap" id="institution-wrap-tests">
 
       <assert see="https://elifeproduction.slab.com/posts/funding-3sv64358#institution-id-test" test="institution-id[@institution-id-type='FundRef']" role="warning" id="institution-id-test">Whenever possible, a funder should have a doi - please check whether there is an appropriate doi in the open funder registry. (institution-id[@institution-id-type="FundRef"] is not present in institution-wrap).</assert>

diff --git a/src/pre-JATS-schematron.sch b/src/pre-JATS-schematron.sch
@@ -2386,6 +2386,8 @@
 
     </rule></pattern><pattern id="award-id-tests-pattern"><rule context="funding-group/award-group/award-id" id="award-id-tests">
       <let name="id" value="parent::award-group/@id"/>
+      <let name="funder-id" value="parent::award-group/descendant::institution-id[1]"/>
+      <let name="funder-name" value="parent::award-group/descendant::institution[1]"/>
 
       <report see="https://elifeproduction.slab.com/posts/funding-3sv64358#award-id-test-1" test="matches(.,',|;')" role="warning" id="award-id-test-1">[award-id-test-1] Funding entry with id <value-of select="$id"/> has a comma or semi-colon in the award id. Should this be separated out into several funding entries? - <value-of select="."/>.</report>
 
@@ -2397,6 +2399,12 @@
 
       <report test="matches(.,'http[s]?://d?x?\.?doi.org/')" role="error" id="award-id-test-5">[award-id-test-5] Award id contains a DOI link - <value-of select="."/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</report>
 
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]" role="error" id="award-id-test-6">[award-id-test-6] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]" role="error" id="award-id-test-7">[award-id-test-7] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]" role="warning" id="award-id-test-8">[award-id-test-8] Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</report>
+
     </rule></pattern><pattern id="institution-wrap-tests-pattern"><rule context="article-meta//award-group//institution-wrap" id="institution-wrap-tests">
 
       <assert see="https://elifeproduction.slab.com/posts/funding-3sv64358#institution-id-test" test="institution-id[@institution-id-type='FundRef']" role="warning" id="institution-id-test">[institution-id-test] Whenever possible, a funder should have a doi - please check whether there is an appropriate doi in the open funder registry. (institution-id[@institution-id-type="FundRef"] is not present in institution-wrap).</assert>

diff --git a/src/pre-JATS-schematron.xsl b/src/pre-JATS-schematron.xsl
@@ -12054,6 +12054,8 @@
    <!--RULE award-id-tests-->
    <xsl:template match="funding-group/award-group/award-id" priority="1000" mode="M157">
       <xsl:variable name="id" select="parent::award-group/@id"/>
+      <xsl:variable name="funder-id" select="parent::award-group/descendant::institution-id[1]"/>
+      <xsl:variable name="funder-name" select="parent::award-group/descendant::institution[1]"/>
       <!--REPORT warning-->
       <xsl:if test="matches(.,',|;')">
          <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(.,',|;')">
@@ -12125,6 +12127,45 @@
                <xsl:text/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</svrl:text>
          </svrl:successful-report>
       </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]">
+            <xsl:attribute name="id">award-id-test-6</xsl:attribute>
+            <xsl:attribute name="role">error</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-6] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]">
+            <xsl:attribute name="id">award-id-test-7</xsl:attribute>
+            <xsl:attribute name="role">error</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-7] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
+      <!--REPORT warning-->
+      <xsl:if test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]">
+            <xsl:attribute name="id">award-id-test-8</xsl:attribute>
+            <xsl:attribute name="role">warning</xsl:attribute>
+            <xsl:attribute name="location">
+               <xsl:apply-templates select="." mode="schematron-select-full-path"/>
+            </xsl:attribute>
+            <svrl:text>[award-id-test-8] Funding entry has an award id - <xsl:text/>
+               <xsl:value-of select="."/>
+               <xsl:text/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
       <xsl:apply-templates select="*" mode="M157"/>
    </xsl:template>
    <xsl:template match="text()" priority="-1" mode="M157"/>

diff --git a/src/schematron.sch b/src/schematron.sch
@@ -3541,6 +3541,8 @@
 
     <rule context="funding-group/award-group/award-id" id="award-id-tests">
       <let name="id" value="parent::award-group/@id"/>
+      <let name="funder-id" value="parent::award-group/descendant::institution-id[1]"/>
+      <let name="funder-name" value="parent::award-group/descendant::institution[1]"/>
 
       <report see="https://elifeproduction.slab.com/posts/funding-3sv64358#award-id-test-1" 
         test="matches(.,',|;')" 
@@ -3566,6 +3568,18 @@
         role="error" 
         id="award-id-test-5">Award id contains a DOI link - <value-of select="."/>. If the award ID is for a grant DOI it should contain the DOI without the https://... protocol (e.g. 10.37717/220020477).</report>
 
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution-id[1] = $funder-id]" 
+        role="error" 
+        id="award-id-test-6">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder ID. This must be incorrect. Either the funder ID or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group/descendant::institution[1] = $funder-name]" 
+        role="error" 
+        id="award-id-test-7">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with the same funder name. This must be incorrect. Either the funder name or the award ID is wrong, or it is a duplicate that should be removed.</report>
+
+      <report test=". = preceding::award-id[parent::award-group[not(descendant::institution[1] = $funder-name) and not(descendant::institution-id[1] = $funder-id)]]" 
+        role="warning" 
+        id="award-id-test-8">Funding entry has an award id - <value-of select="."/> - which is also used in another funding entry with a different funder. Has there been a mistake with the award id? If the grant was awarded jointly by two funders, then this capture is correct and should be retained.</report>
+
     </rule>
 
     <rule context="article-meta//award-group//institution-wrap" id="institution-wrap-tests">

diff --git a/test/tests/gen/award-id-tests/award-id-test-1/award-id-test-1.sch b/test/tests/gen/award-id-tests/award-id-test-1/award-id-test-1.sch
@@ -1363,6 +1363,8 @@
   <pattern id="article-metadata">
     <rule context="funding-group/award-group/award-id" id="award-id-tests">
       <let name="id" value="parent::award-group/@id"/>
+      <let name="funder-id" value="parent::award-group/descendant::institution-id[1]"/>
+      <let name="funder-name" value="parent::award-group/descendant::institution[1]"/>
       <report see="https://elifeproduction.slab.com/posts/funding-3sv64358#award-id-test-1" test="matches(.,',|;')" role="warning" id="award-id-test-1">Funding entry with id <value-of select="$id"/> has a comma or semi-colon in the award id. Should this be separated out into several funding entries? - <value-of select="."/>.</report>
     </rule>
   </pattern>