Merge pull request #41 from InseeFr/acceptance

Acceptance

metadataapi-changeLog.txt

@@ -1,3 +1,4 @@
+3.0.3 : Correction des régressions sur l'encodage des retours XML (documentation sims notamment)
 3.0.2 : Correction de l'encodage des retours XML (documentation sims notamment) et logs
 3.0.1 : Changement des properties pour correspondre à ce qui a été défini en production
 3.0.0 : Ajout des services opérations statistiques pour Web4G

pom.xml

@@ -7,7 +7,7 @@
 	<packaging>war</packaging>
 	<name>Implementation of the RMéS metadata API</name>
 
-	<version>3.0.2</version>
+	<version>3.0.3</version>
 
 	<properties>
 		<title>API RM\u00e9S</title>

src/main/java/fr/insee/rmes/utils/CustomXmlEscapingWriterFactory.java

@@ -18,8 +18,7 @@ public void write(char[] cbuf, int off, int len) throws IOException {
             for (int i = off; i < len; i++) {
                 val += cbuf[i];
             }
-            String escapedStr = XmlUtils.encodeXml(escapeHtml(val)); //encode manually some xml tags
-            out.write(escapedStr);
+            out.write(val); //no transformation because of the buffer
         }
 
         @Override
@@ -34,16 +33,6 @@ public void close() throws IOException {
       };
     }
 
-
-	private String escapeHtml(String s) {
-		 return s.replace("&", "&amp;")
-				 .replace(">", "&gt;")
-				 .replace("<", "&lt;")
-				 .replace("\"", "&quot;");
-	}
-
-
-
     public Writer createEscapingWriterFor(OutputStream out, String enc) {
         throw new IllegalArgumentException("not supported");
     }

src/main/java/fr/insee/rmes/utils/ResponseUtils.java

@@ -5,6 +5,7 @@
 
 import javax.ws.rs.core.MediaType;
 
+import org.apache.commons.text.StringEscapeUtils;
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 import org.codehaus.stax2.XMLOutputFactory2;
@@ -77,7 +78,16 @@ public String produceResponse(Object obj, String header) {
         return response;
     }
 
+	private String escapeHtml(String s) {
+  		 s = StringEscapeUtils.unescapeHtml4(s);
+  		 return s.replace("&", "&")
+  				 .replace(">", ">")
+  				 .replace("<", "&lt;")
+  				 .replace("\"", "&quot;");
+  	}
+
     public String encodeXmlResponse(String response) {
+    	response = escapeHtml(response);
     	response = XmlUtils.encodeXml(response);
     	return new String(response.getBytes(), StandardCharsets.UTF_8);
     }

src/main/java/fr/insee/rmes/utils/XmlUtils.java

@@ -3,14 +3,19 @@
 public class XmlUtils {
 
 	public static String encodeXml(String s) {
-		 return s.replaceAll("&lt;([a-zA-Z]+)&gt;", "<$1>") //xml open tag
-				 .replaceAll("&lt;([a-zA-Z]+) style=(&quot;|')(.*?)(&quot;|')&gt;", "<$1 style=\"$3\">") //open tag with style
+		 return s.replaceAll("&lt;&lt;([a-zA-Z]+)&gt;&gt;", "&lt;&lt; $1 &gt;&gt;") //<<texte>> => << texte >>
+				 .replaceAll("&lt;([a-zA-Z]+)&gt;", "<$1>") //xml open tag
+		 		 .replaceAll("&lt;([a-zA-Z]+) ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|')&gt;", "<$1 $2=\"$4\">") //open tag with one attribute (text or xml:lang)
+		 		 .replaceAll("&lt;([a-zA-Z]+) ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|')&gt;", "<$1 $2=\"$4\" $8=\"$10\">") //open tag with 2 attributes (text or xml:lang)
+		 		 .replaceAll("&lt;([a-zA-Z]+) ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|')&gt;", "<$1 $2=\"$4\" $8=\"$10\" $14=\"$16\">") //open tag with 3 attributes (text or xml:lang)
+		 		 .replaceAll("&lt;([a-zA-Z]+) ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|') ([a-zA-Z:]+)?=(&quot;|')(((?!(&quot;|')).)*?)(&quot;|')&gt;", "<$1 $2=\"$4\" $8=\"$10\" $14=\"$16\" $20=\"$22\">") //open tag with 4 attributes (text or xml:lang)
 				 .replaceAll("&lt;/([a-zA-Z]+)&gt;", "</$1>") //xml close tag
 				 .replaceAll("&lt;([a-zA-Z]+)/&gt;", "<$1/>") //br
 				 .replaceAll("&lt;([a-zA-Z]+) /&gt;", "<$1 />") //br with space
-				 .replaceAll("&lt;a href=(&quot;|')(.*?)(&quot;|')&gt;", "<a href=\"$2\"> ") //a href open tag
-				 .replaceAll("&lt;a title=(&quot;|')(.*?)(&quot;|') href=(&quot;|')(.*?)(&quot;|')&gt;", "<a title=\"$2\" href=\"$5\"> ") //a title href open tag
-		 		.replaceAll("&lt;a href=(&quot;|')(.*?)(&quot;|') title=(&quot;|')(.*?)(&quot;|')&gt;", "<a title=\"$5\" href=\"$2\"> "); //a href title open tag
+
+	 			; 
+
+
 
 	}
 

src/test/java/fr/insee/rmes/api/classifications/pseudointegrationtest/ClassificationApiTest.java

@@ -36,7 +36,7 @@ class ClassificationApiTest  {
 
 
     private static final String EXPECTED_JSON = "[{\"uri\":\"http://id.insee.fr/codes/nafr2/division/01\",\"code\":\"01\",\"uriParent\":\"http://id.insee.fr/codes/nafr2/section/A\",\"codeParent\":\"A\",\"intituleFr\":\"Intitulé FR 01\",\"intituleEn\":\"En title 01\",\"contenuLimite\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Limite</p></div>\",\"contenuCentral\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Central</p></div>\",\"contenuExclu\":\"<div xmlns='http://www.w3.org/1999/xhtml'>Exclu</div>\",\"noteGenerale\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Note générale</p></div>\"},{\"uri\":\"http://id.insee.fr/codes/nafr2/groupe/01.1\",\"code\":\"01.1\",\"uriParent\":\"http://id.insee.fr/codes/nafr2/division/01\",\"codeParent\":\"01\",\"intituleFr\":\"Titre fr 01.1\",\"intituleEn\":\"En title 01.1\",\"contenuLimite\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Limite</p></div>\",\"contenuCentral\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Central</p></div>\",\"contenuExclu\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Exclu</p></div>\",\"noteGenerale\":\"<div xmlns='http://www.w3.org/1999/xhtml'><p>Note g</p></div>\"}]";
-    private static final String EXPECTED_XML = "<Postes><Poste uri=\"http://id.insee.fr/codes/nafr2/division/01\" code=\"01\" uriParent=\"http://id.insee.fr/codes/nafr2/section/A\" codeParent=\"A\"><IntituleFr>Intitulé FR 01</IntituleFr><IntituleEn>En title 01</IntituleEn><ContenuLimite><div xmlns='http://www.w3.org/1999/xhtml'><p>Limite</p></div></ContenuLimite><ContenuCentral><div xmlns='http://www.w3.org/1999/xhtml'><p>Central</p></div></ContenuCentral><ContenuExclu><div xmlns='http://www.w3.org/1999/xhtml'>Exclu</div></ContenuExclu><NoteGenerale><div xmlns='http://www.w3.org/1999/xhtml'><p>Note générale</p></div></NoteGenerale></Poste><Poste uri=\"http://id.insee.fr/codes/nafr2/groupe/01.1\" code=\"01.1\" uriParent=\"http://id.insee.fr/codes/nafr2/division/01\" codeParent=\"01\"><IntituleFr>Titre fr 01.1</IntituleFr><IntituleEn>En title 01.1</IntituleEn><ContenuLimite><div xmlns='http://www.w3.org/1999/xhtml'><p>Limite</p></div></ContenuLimite><ContenuCentral><div xmlns='http://www.w3.org/1999/xhtml'><p>Central</p></div></ContenuCentral><ContenuExclu><div xmlns='http://www.w3.org/1999/xhtml'><p>Exclu</p></div></ContenuExclu><NoteGenerale><div xmlns='http://www.w3.org/1999/xhtml'><p>Note g</p></div></NoteGenerale></Poste></Postes>";
+    private static final String EXPECTED_XML = "<Postes><Poste uri=\"http://id.insee.fr/codes/nafr2/division/01\" code=\"01\" uriParent=\"http://id.insee.fr/codes/nafr2/section/A\" codeParent=\"A\"><IntituleFr>Intitulé FR 01</IntituleFr><IntituleEn>En title 01</IntituleEn><ContenuLimite><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Limite</p></div></ContenuLimite><ContenuCentral><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Central</p></div></ContenuCentral><ContenuExclu><div xmlns=\"http://www.w3.org/1999/xhtml\">Exclu</div></ContenuExclu><NoteGenerale><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Note générale</p></div></NoteGenerale></Poste><Poste uri=\"http://id.insee.fr/codes/nafr2/groupe/01.1\" code=\"01.1\" uriParent=\"http://id.insee.fr/codes/nafr2/division/01\" codeParent=\"01\"><IntituleFr>Titre fr 01.1</IntituleFr><IntituleEn>En title 01.1</IntituleEn><ContenuLimite><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Limite</p></div></ContenuLimite><ContenuCentral><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Central</p></div></ContenuCentral><ContenuExclu><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Exclu</p></div></ContenuExclu><NoteGenerale><div xmlns=\"http://www.w3.org/1999/xhtml\"><p>Note g</p></div></NoteGenerale></Poste></Postes>";
 
 
     @BeforeEach