Email address format validation (#312)

* Email address format validation * Update regx for email format * Allowing empty email in email format validation to avoid schematron validation duplication
metadata101 · Mar 27, 2023 · 8aa2267 · 8aa2267
1 parent 0a60118
commit 8aa2267
Show file tree

Hide file tree

Showing 8 changed files with 106 additions and 14 deletions.
diff --git a/src/main/java/ca/gc/schema/iso19139hnap/util/XslUtilHnap.java b/src/main/java/ca/gc/schema/iso19139hnap/util/XslUtilHnap.java
@@ -44,6 +44,8 @@
 import java.util.Calendar;
 import java.util.Date;
 import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 import java.util.stream.Collectors;
 
 public class XslUtilHnap {
@@ -254,4 +256,21 @@ String removeFromUrl(ArrayList<NodeInfo> parametersToRemove, String url) {
 
         return result;
     }
+
+    /**
+     * Validate if email format. Empty email is allowed based on parameter of allowEmptyEmail.
+     * See more info: https://html.spec.whatwg.org/multipage/input.html#valid-e-mail-address
+     * @param emailAddress email address to validate the pattern
+     * @param allowEmptyEmail allow empty email input. The empty email was handled within the schematron already. This flag allows to ignore repeat checking.
+     * @return boolean
+     */
+    public static boolean isEmailFormat(String emailAddress, boolean allowEmptyEmail) {
+        if (allowEmptyEmail && org.apache.commons.lang3.StringUtils.isEmpty(emailAddress)) {
+            return true;
+        }
+
+        Pattern pattern = Pattern.compile("^[a-zA-Z0-9.!#$%&'*+\\/=?^_`{|}~-]+@[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*$");
+        Matcher matcher = pattern.matcher(emailAddress);
+        return matcher.matches();
+    }
 }
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-multilingual.xml b/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-multilingual.xml
@@ -83,4 +83,6 @@
   <ECThesaurusOrg>Thesaurus cited responsible party organisation is required</ECThesaurusOrg>
   <ECThesaurusRole>Thesaurus cited responsible party role is required</ECThesaurusRole>
   <ECThesaurusEmail>Thesaurus cited responsible party email is required</ECThesaurusEmail>
+
+  <ElectronicMailFormat>Electronic mail address format is invalid (for example [email protected])</ElectronicMailFormat>
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-non-multilingual.xml b/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-non-multilingual.xml
@@ -56,4 +56,6 @@
   <ResourceDescriptionContentType>Online resource description content type is not valid. Valid values are: Web Service,Service Web,Dataset,Données,API,Application,Supporting Document,Document de soutien</ResourceDescriptionContentType>
   <ResourceDescriptionFormat>Online resource description format is not valid. Valid values are:</ResourceDescriptionFormat>
   <ResourceDescriptionLanguage>Online resource description language is not valid. Should be a comma separated values of ISO-LANG-3 codes</ResourceDescriptionLanguage>
+
+  <ElectronicMailFormat>Electronic mail address format is invalid (for example [email protected])</ElectronicMailFormat>
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-multilingual.xml b/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-multilingual.xml
@@ -84,4 +84,6 @@
   <ECThesaurusOrg>Le nom de l’organisation responsable du thésaurus est requis</ECThesaurusOrg>
   <ECThesaurusRole>Le rôle du responsable du thesaurus est requis</ECThesaurusRole>
   <ECThesaurusEmail>Le courriel du responsable du thésaurus est requis</ECThesaurusEmail>
+
+  <ElectronicMailFormat>Le format de l'adresse électronique n'est pas valide (par exemple [email protected])</ElectronicMailFormat>
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-non-multilingual.xml b/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-non-multilingual.xml
@@ -56,4 +56,6 @@
   <ResourceDescriptionContentType>Le type de contenu dans la description de la ressource en ligne n'est pas valide. Les valeurs valides sont : Web Service,Service Web,Dataset,Données,API,Application,Supporting Document,Document de soutien</ResourceDescriptionContentType>
   <ResourceDescriptionFormat>Le format dans la description de la ressource en ligne n’est pas valide. Les valeurs valides sont : </ResourceDescriptionFormat>
   <ResourceDescriptionLanguage>La langue dans la description de la ressource en ligne n’est pas valide. Devrait être des valeurs de code ISO-LANG-3 séparées par des virgules</ResourceDescriptionLanguage>
+
+  <ElectronicMailFormat>Le format de l'adresse électronique n'est pas valide (par exemple [email protected])</ElectronicMailFormat>
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-multilingual.sch b/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-multilingual.sch
@@ -164,6 +164,7 @@
     />
   </xsl:function>
 
+
   <!-- =============================================================
   EC schematron rules for multilingual validation in metadata editor:
   ============================================================= -->
@@ -348,17 +349,23 @@
 
     <!-- Contact - Electronic Mail -->
     <sch:rule context="//gmd:contact/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
+      <sch:let name="emailAddressOtherLang" value="string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)])" />
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
+      <sch:let name="missing" value="not($emailAddress)
                 or (@gco:nilReason)" />
 
-      <sch:let name="missingOtherLang" value="not(string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)]))" />
+      <sch:let name="missingOtherLang" value="not($emailAddressOtherLang)" />
+
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
+      <sch:let name="isOtherLangEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddressOtherLang, true())"/>
 
       <sch:assert
         test="not($missing) and not($missingOtherLang)"
 
       >$loc/strings/ContactElectronicMail</sch:assert>
 
+      <sch:assert test="string($isEmailAddressFormat) ='true' and string($isOtherLangEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
     </sch:rule>
 
 
@@ -570,17 +577,24 @@
     <sch:rule context="//gmd:identificationInfo/*/gmd:citation/*/gmd:citedResponsibleParty/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress
                       |//*[@gco:isoType='gmd:MD_DataIdentification']/gmd:citation/*/gmd:citedResponsibleParty/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress
                       |//*[@gco:isoType='srv:SV_ServiceIdentification']/gmd:citation/*/gmd:citedResponsibleParty/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
+      <sch:let name="emailAddressOtherLang" value="string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)])" />
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
+      <sch:let name="missing" value="not($emailAddress)
                 or (@gco:nilReason)" />
 
-      <sch:let name="missingOtherLang" value="not(string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)]))" />
+      <sch:let name="missingOtherLang" value="not($emailAddressOtherLang)" />
+
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
+      <sch:let name="isOtherLangEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddressOtherLang, true())"/>
 
       <sch:assert
         test="not($missing) and not($missingOtherLang)"
 
       >$loc/strings/CitedResponsiblePartyElectronicMail</sch:assert>
 
+      <sch:assert test="string($isEmailAddressFormat) ='true' and string($isOtherLangEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
 
 
@@ -673,14 +687,22 @@
 
       <sch:let name="emailPresent" value="count(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress) > 0" />
 
-      <sch:let name="missingEmail" value="not(string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gco:CharacterString))
+      <sch:let name="emailAddress" value="string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gco:CharacterString)" />
+      <sch:let name="emailAddressOtherLang" value="string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)])" />
+
+      <sch:let name="missingEmail" value="not($emailAddress)
               or (@gco:nilReason)" />
 
-      <sch:let name="missingEmailOtherLang" value="not(string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)]))" />
+      <sch:let name="missingEmailOtherLang" value="not($emailAddressOtherLang)" />
+
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
+      <sch:let name="isOtherLangEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddressOtherLang, true())"/>
 
       <sch:assert
         test="not($thesaurusNamePresent) or ($thesaurusNamePresent and (not($emailPresent) or ($emailPresent and not($missingEmail) and not($missingEmailOtherLang))))"
       >$loc/strings/ECThesaurusEmail</sch:assert>
+
+      <sch:assert test="string($isEmailAddressFormat) ='true' and string($isOtherLangEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
     </sch:rule>
 
     <!-- Supplemental information -->
@@ -973,17 +995,24 @@
 
     <!-- Distributor - Electronic Mail -->
     <sch:rule context="//gmd:distributionInfo/*/gmd:distributor/gmd:MD_Distributor/gmd:distributorContact/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
+      <sch:let name="emailAddressOtherLang" value="string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)])" />
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
+      <sch:let name="missing" value="not($emailAddress)
                 or (@gco:nilReason)" />
 
-      <sch:let name="missingOtherLang" value="not(string(gmd:PT_FreeText/gmd:textGroup/gmd:LocalisedCharacterString[@locale=concat('#', $altLanguageId)]))" />
+      <sch:let name="missingOtherLang" value="not($emailAddressOtherLang)" />
+
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
+   	  <sch:let name="isOtherLangEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddressOtherLang, true())"/>
 
       <sch:assert
         test="not($missing) and not($missingOtherLang)"
 
       >$loc/strings/DistributorElectronicMail</sch:assert>
 
+      <sch:assert test="string($isEmailAddressFormat) ='true' and string($isOtherLangEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
 
 

diff --git a/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-non-multilingual.sch b/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-non-multilingual.sch
@@ -160,6 +160,7 @@
     />
   </xsl:function>
 
+
   <!-- =============================================================
   EC schematron rules for multilingual validation in metadata editor:
   ============================================================= -->
@@ -220,15 +221,20 @@
 
     <!-- Contact - Electronic Mail -->
     <sch:rule context="//gmd:contact/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
+      <sch:let name="missing" value="not($emailAddress)
                 or (@gco:nilReason)" />
 
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
+
       <sch:assert
         test="not($missing)"
 
       >$loc/strings/ContactElectronicMail</sch:assert>
 
+      <sch:assert test="string($isEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
   </sch:pattern>
 
@@ -318,15 +324,17 @@
                       |//*[@gco:isoType='gmd:MD_DataIdentification']/gmd:citation/*/gmd:citedResponsibleParty/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress
                       |//*[@gco:isoType='srv:SV_ServiceIdentification']/gmd:citation/*/gmd:citedResponsibleParty/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
-                or (@gco:nilReason)" />
-
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
+      <sch:let name="missing" value="not($emailAddress) or (@gco:nilReason)" />
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
 
       <sch:assert
         test="not($missing)"
 
       >$loc/strings/CitedResponsiblePartyElectronicMail</sch:assert>
 
+      <sch:assert test="string($isEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
 
     <!-- Keywords -->
@@ -394,12 +402,17 @@
 
       <sch:let name="emailPresent" value="count(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress) > 0" />
 
-      <sch:let name="missingEmail" value="not(string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gco:CharacterString))
+      <sch:let name="emailAddress" value="string(gmd:MD_Keywords/gmd:thesaurusName/gmd:CI_Citation/gmd:citedResponsibleParty/gmd:CI_ResponsibleParty/gmd:contactInfo/gmd:CI_Contact/gmd:address/gmd:CI_Address/gmd:electronicMailAddress/gco:CharacterString)" />
+      <sch:let name="missingEmail" value="not($emailAddress)
               or (@gco:nilReason)" />
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
 
       <sch:assert
         test="not($thesaurusNamePresent) or ($thesaurusNamePresent and (not($emailPresent) or ($emailPresent and not($missingEmail))))"
       >$loc/strings/ECThesaurusEmail</sch:assert>
+
+      <sch:assert test="string($isEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
 
     <!-- Note (Other constraints) -->
@@ -555,14 +568,19 @@
     <!-- Distributor - Electronic Mail -->
     <sch:rule context="//gmd:distributionInfo/*/gmd:distributor/gmd:MD_Distributor/gmd:distributorContact/*/gmd:contactInfo/*/gmd:address/gmd:CI_Address/gmd:electronicMailAddress">
 
-      <sch:let name="missing" value="not(string(gco:CharacterString))
+      <sch:let name="emailAddress" value="string(gco:CharacterString)" />
+      <sch:let name="missing" value="not($emailAddress)
                   or (@gco:nilReason)" />
+      <sch:let name="isEmailAddressFormat" value="XslUtilHnap:isEmailFormat($emailAddress, true())"/>
 
       <sch:assert
         test="not($missing)"
 
       >$loc/strings/DistributorElectronicMail</sch:assert>
 
+
+      <sch:assert test="string($isEmailAddressFormat) ='true'">$loc/strings/ElectronicMailFormat</sch:assert>
+
     </sch:rule>
 
   </sch:pattern>

diff --git a/src/test/java/ca/gc/schema/iso19139hnap/util/XslUtilHnapTest.java b/src/test/java/ca/gc/schema/iso19139hnap/util/XslUtilHnapTest.java
@@ -31,6 +31,8 @@
 import java.util.ArrayList;
 
 import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertFalse;
+import static org.junit.Assert.assertTrue;
 
 /**
  * Tests for {@link XslUtilHnap} class.
@@ -133,6 +135,22 @@ public void removeFromUrl() {
 
     }
 
+    @Test
+    public void testIsEmailAddress() {
+        assertTrue(XslUtilHnap.isEmailFormat("o'brian.test@localhost", false));
+        assertTrue(XslUtilHnap.isEmailFormat("o'[email protected]", false));
+
+        assertFalse(XslUtilHnap.isEmailFormat("o'[email protected].", false));
+        assertFalse(XslUtilHnap.isEmailFormat("o'brian.test@localhost.", false));
+        assertFalse(XslUtilHnap.isEmailFormat("o'brian.test@", false));
+        assertFalse(XslUtilHnap.isEmailFormat("o'brian.test", false));
+
+        assertFalse(XslUtilHnap.isEmailFormat("",false));
+        assertTrue(XslUtilHnap.isEmailFormat("",true));
+        assertTrue(XslUtilHnap.isEmailFormat(null,true));
+
+    }
+
     class MyTinyNodeImpl extends TinyNodeImpl {
 
         private String value;