VOR - add RRID identifiers link check

elifesciences · Dec 19, 2024 · 6ddd110 · 6ddd110
1 parent a594758
commit 6ddd110
Show file tree

Hide file tree

Showing 20 changed files with 1,478 additions and 37 deletions.
diff --git a/src/final-JATS-schematron.sch b/src/final-JATS-schematron.sch
@@ -7059,7 +7059,7 @@
 
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#PMCID-link-test" test="matches(.,'PMCID[:]?\p{Zs}?PMC[0-9][0-9][0-9]+') and (count(ext-link[contains(@xlink:href,'www.ncbi.nlm.nih.gov/pmc')]) = 0)" role="error" id="PMCID-link-test">[PMCID-link-test] td element containing - '<value-of select="."/>' - looks like it contains a PMCID, but it contains no link pointing to PMC, which is incorrect.</report>
 
-      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])" role="warning" id="addgene-test">[addgene-test] td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</report>
+      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])" role="warning" id="addgene-test">[addgene-test] td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</report>
 
     </rule></pattern><pattern id="colour-table-pattern"><rule context="table-wrap" id="colour-table">
       <let name="allowed-values" value="('background-color: #90CAF9;','background-color: #C5E1A5;','background-color: #FFB74D;','background-color: #FFF176;','background-color: #9E86C9;','background-color: #E57373;','background-color: #F48FB1;','background-color: #E6E6E6;')"/>
@@ -7179,13 +7179,15 @@
 
 
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#final-pmid-spacing-table" test="ends-with($lc,'pmid: ') or ends-with($lc,'pmid ')" role="warning" id="final-pmid-spacing-table">[final-pmid-spacing-table] PMID link should be preceding by 'PMID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
-    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" id="rrid-link">
+    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" id="rrid-link">
       <let name="pre-text" value="preceding-sibling::text()[1]"/>
       <let name="lc" value="lower-case($pre-text)"/>
 
 
 
       <report see="https://elifeproduction.slab.com/posts/rri-ds-5k19v560#final-rrid-spacing" test="ends-with($lc,'rrid: ') or ends-with($lc,'rrid ')" role="warning" id="final-rrid-spacing">[final-rrid-spacing] RRID link should be preceded by 'RRID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
+
+      <report test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')" role="error" id="rrid-link-format">[rrid-link-format] RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <value-of select="@xlink:href"/>.</report>
     </rule></pattern><pattern id="ref-link-mandate-pattern"><rule context="ref-list/ref" id="ref-link-mandate">
       <let name="id" value="@id"/>
 

diff --git a/src/final-JATS-schematron.xsl b/src/final-JATS-schematron.xsl
@@ -39938,16 +39938,16 @@
          </svrl:successful-report>
       </xsl:if>
       <!--REPORT warning-->
-      <xsl:if test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])">
-         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])">
+      <xsl:if test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])">
             <xsl:attribute name="id">addgene-test</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>[addgene-test] td element containing - '<xsl:text/>
                <xsl:value-of select="."/>
-               <xsl:text/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</svrl:text>
+               <xsl:text/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</svrl:text>
          </svrl:successful-report>
       </xsl:if>
       <xsl:apply-templates select="*" mode="M540"/>
@@ -40447,7 +40447,7 @@
    </xsl:template>
    <!--PATTERN rrid-link-pattern-->
    <!--RULE rrid-link-->
-   <xsl:template match="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" priority="1000" mode="M554">
+   <xsl:template match="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" priority="1000" mode="M554">
       <xsl:variable name="pre-text" select="preceding-sibling::text()[1]"/>
       <xsl:variable name="lc" select="lower-case($pre-text)"/>
       <!--REPORT warning-->
@@ -40464,6 +40464,19 @@
                <xsl:text/>'.</svrl:text>
          </svrl:successful-report>
       </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')">
+            <xsl:attribute name="id">rrid-link-format</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>[rrid-link-format] RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <xsl:text/>
+               <xsl:value-of select="@xlink:href"/>
+               <xsl:text/>.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
       <xsl:apply-templates select="*" mode="M554"/>
    </xsl:template>
    <xsl:template match="text()" priority="-1" mode="M554"/>

diff --git a/src/final-package-JATS-schematron.sch b/src/final-package-JATS-schematron.sch
@@ -7065,7 +7065,7 @@
 
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#PMCID-link-test" test="matches(.,'PMCID[:]?\p{Zs}?PMC[0-9][0-9][0-9]+') and (count(ext-link[contains(@xlink:href,'www.ncbi.nlm.nih.gov/pmc')]) = 0)" role="error" id="PMCID-link-test">td element containing - '<value-of select="."/>' - looks like it contains a PMCID, but it contains no link pointing to PMC, which is incorrect.</report>
 
-      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])" role="warning" id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</report>
+      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])" role="warning" id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</report>
 
     </rule></pattern><pattern id="colour-table-pattern"><rule context="table-wrap" id="colour-table">
       <let name="allowed-values" value="('background-color: #90CAF9;','background-color: #C5E1A5;','background-color: #FFB74D;','background-color: #FFF176;','background-color: #9E86C9;','background-color: #E57373;','background-color: #F48FB1;','background-color: #E6E6E6;')"/>
@@ -7185,13 +7185,15 @@
 
 
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#final-pmid-spacing-table" test="ends-with($lc,'pmid: ') or ends-with($lc,'pmid ')" role="warning" id="final-pmid-spacing-table">PMID link should be preceding by 'PMID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
-    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" id="rrid-link">
+    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" id="rrid-link">
       <let name="pre-text" value="preceding-sibling::text()[1]"/>
       <let name="lc" value="lower-case($pre-text)"/>
 
 
 
       <report see="https://elifeproduction.slab.com/posts/rri-ds-5k19v560#final-rrid-spacing" test="ends-with($lc,'rrid: ') or ends-with($lc,'rrid ')" role="warning" id="final-rrid-spacing">RRID link should be preceded by 'RRID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
+
+      <report test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')" role="error" id="rrid-link-format">RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <value-of select="@xlink:href"/>.</report>
     </rule></pattern><pattern id="ref-link-mandate-pattern"><rule context="ref-list/ref" id="ref-link-mandate">
       <let name="id" value="@id"/>
 

diff --git a/src/pre-JATS-schematron.sch b/src/pre-JATS-schematron.sch
@@ -7004,7 +7004,7 @@
 
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#PMCID-link-test" test="matches(.,'PMCID[:]?\p{Zs}?PMC[0-9][0-9][0-9]+') and (count(ext-link[contains(@xlink:href,'www.ncbi.nlm.nih.gov/pmc')]) = 0)" role="error" id="PMCID-link-test">[PMCID-link-test] td element containing - '<value-of select="."/>' - looks like it contains a PMCID, but it contains no link pointing to PMC, which is incorrect.</report>
 
-      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])" role="warning" id="addgene-test">[addgene-test] td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</report>
+      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])" role="warning" id="addgene-test">[addgene-test] td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</report>
 
     </rule></pattern><pattern id="colour-table-pattern"><rule context="table-wrap" id="colour-table">
       <let name="allowed-values" value="('background-color: #90CAF9;','background-color: #C5E1A5;','background-color: #FFB74D;','background-color: #FFF176;','background-color: #9E86C9;','background-color: #E57373;','background-color: #F48FB1;','background-color: #E6E6E6;')"/>
@@ -7114,13 +7114,15 @@
       <report see="https://elifeproduction.slab.com/posts/tables-3nehcouh#pre-pmid-spacing-table" test="ends-with($lc,'pmid: ') or ends-with($lc,'pmid ')" role="error" id="pre-pmid-spacing-table">[pre-pmid-spacing-table] PMID link should be preceding by 'PMID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
 
 
-    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" id="rrid-link">
+    </rule></pattern><pattern id="rrid-link-pattern"><rule context="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" id="rrid-link">
       <let name="pre-text" value="preceding-sibling::text()[1]"/>
       <let name="lc" value="lower-case($pre-text)"/>
 
       <report see="https://elifeproduction.slab.com/posts/rri-ds-5k19v560#pre-rrid-spacing" test="ends-with($lc,'rrid: ') or ends-with($lc,'rrid ')" role="error" id="pre-rrid-spacing">[pre-rrid-spacing] RRID link should be preceded by 'RRID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
 
 
+
+      <report test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')" role="error" id="rrid-link-format">[rrid-link-format] RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <value-of select="@xlink:href"/>.</report>
     </rule></pattern><pattern id="ref-link-mandate-pattern"><rule context="ref-list/ref" id="ref-link-mandate">
       <let name="id" value="@id"/>
 

diff --git a/src/pre-JATS-schematron.xsl b/src/pre-JATS-schematron.xsl
@@ -39649,16 +39649,16 @@
          </svrl:successful-report>
       </xsl:if>
       <!--REPORT warning-->
-      <xsl:if test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])">
-         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])">
+      <xsl:if test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])">
             <xsl:attribute name="id">addgene-test</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>[addgene-test] td element containing - '<xsl:text/>
                <xsl:value-of select="."/>
-               <xsl:text/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</svrl:text>
+               <xsl:text/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</svrl:text>
          </svrl:successful-report>
       </xsl:if>
       <xsl:apply-templates select="*" mode="M538"/>
@@ -40300,7 +40300,7 @@
    </xsl:template>
    <!--PATTERN rrid-link-pattern-->
    <!--RULE rrid-link-->
-   <xsl:template match="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" priority="1000" mode="M551">
+   <xsl:template match="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" priority="1000" mode="M551">
       <xsl:variable name="pre-text" select="preceding-sibling::text()[1]"/>
       <xsl:variable name="lc" select="lower-case($pre-text)"/>
       <!--REPORT error-->
@@ -40317,6 +40317,19 @@
                <xsl:text/>'.</svrl:text>
          </svrl:successful-report>
       </xsl:if>
+      <!--REPORT error-->
+      <xsl:if test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')">
+         <svrl:successful-report xmlns:svrl="http://purl.oclc.org/dsdl/svrl" test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')">
+            <xsl:attribute name="id">rrid-link-format</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>[rrid-link-format] RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <xsl:text/>
+               <xsl:value-of select="@xlink:href"/>
+               <xsl:text/>.</svrl:text>
+         </svrl:successful-report>
+      </xsl:if>
       <xsl:apply-templates select="*" mode="M551"/>
    </xsl:template>
    <xsl:template match="text()" priority="-1" mode="M551"/>

diff --git a/src/schematron.sch b/src/schematron.sch
@@ -13737,9 +13737,9 @@ else self::*/local-name() = $allowed-p-blocks"
         id="PMCID-link-test">td element containing - '<value-of select="."/>' - looks like it contains a PMCID, but it contains no link pointing to PMC, which is incorrect.</report>
 
       <report
-        test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])" 
+        test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])" 
         role="warning" 
-        id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</report>
+        id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</report>
 
     </rule>
 
@@ -14019,7 +14019,7 @@ else self::*/local-name() = $allowed-p-blocks"
         id="final-pmid-spacing-table">PMID link should be preceding by 'PMID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
     </rule>
 
-    <rule context="ext-link[contains(@xlink:href,'scicrunch.org/resolver') and not(ancestor::sub-article)]" id="rrid-link">
+    <rule context="ext-link[contains(lower-case(@xlink:href),'identifiers.org/rrid') and not(ancestor::sub-article)]" id="rrid-link">
       <let name="pre-text" value="preceding-sibling::text()[1]"/>
       <let name="lc" value="lower-case($pre-text)"/>
 
@@ -14032,6 +14032,10 @@ else self::*/local-name() = $allowed-p-blocks"
         test="ends-with($lc,'rrid: ') or ends-with($lc,'rrid ')" 
         role="warning" 
         id="final-rrid-spacing">RRID link should be preceded by 'RRID:' with no space but instead it is preceded by '<value-of select="concat(substring($pre-text,string-length($pre-text)-15),.)"/>'.</report>
+
+      <report test="matches(@xlink:href,'identifiers\.org/RRID/RRID:')" 
+        role="error" 
+        id="rrid-link-format">RRID links should be in the format 'https://identifiers.org/RRID:XXXX', but this one is not - <value-of select="@xlink:href"/>.</report>
     </rule>
 
     <rule context="ref-list/ref" id="ref-link-mandate">

diff --git a/test/tests/gen/KRT-td-checks/addgene-test/addgene-test.sch b/test/tests/gen/KRT-td-checks/addgene-test/addgene-test.sch
@@ -1362,7 +1362,7 @@
   </xsl:function>
   <pattern id="house-style">
     <rule context="table-wrap[@id='keyresource']//td" id="KRT-td-checks">
-      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID/.*')])" role="warning" id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID/RRID:addgene_{number} link?</report>
+      <report test="matches(lower-case(.),'addgene\p{Zs}?#?\p{Zs}?\d') and not(ext-link[matches(@xlink:href,'identifiers\.org/RRID:.*')])" role="warning" id="addgene-test">td element containing - '<value-of select="."/>' - looks like it contains an addgene number. Should this be changed to an RRID with a https://identifiers.org/RRID:addgene_{number} link?</report>
     </rule>
   </pattern>
   <pattern id="root-pattern">