Merge pull request #943 from openpreserve/fix/xml-reporting

FIX: Revert XML SAXParseException reporting
openpreserve · Aug 19, 2024 · 250f2db · 250f2db
2 parents 59cdc79 + 37d0802
commit 250f2db
Show file tree

Hide file tree

Showing 12 changed files with 54 additions and 23 deletions.
diff --git a/jhove-bbt/scripts/create-1.28-target.sh b/jhove-bbt/scripts/create-1.28-target.sh
@@ -243,3 +243,9 @@ fi
 # Patch release details of the reporting module.
 find "${targetRoot}" -type f -name "*.jhove.xml" -exec sed -i 's/jhove\/1.8\/jhove.xsd/jhove\/1.9\/jhove.xsd/' {} \;
 find "${targetRoot}" -type f -name "audit.jhove.xml" -exec sed -i 's/outputHandler release="1.9"/outputHandler release="1.10"/' {} \;
+
+# Patch XML reporting tweak differences
+find "${targetRoot}" -type f -name "valid-external.dtd.jhove.xml" -exec sed -i 's/The markup in the document preceding the root element must be well-formed. Line = 1/Line = 1/' {} \;
+find "${targetRoot}" -type f -name "valid-external.dtd.jhove.xml" -exec sed -i 's/SAXParseException/SAXParseException: The markup in the document preceding the root element must be well-formed./' {} \;
+find "${targetRoot}" -type f -name "*parsed-entity.ent.jhove.xml" -exec sed -i 's/Content is not allowed in prolog. Line = 1/Line = 1/' {} \;
+find "${targetRoot}" -type f -name "*parsed-entity.ent.jhove.xml" -exec sed -i 's/SAXParseException/SAXParseException: Content is not allowed in prolog./' {} \;
diff --git a/jhove-bbt/scripts/create-1.31-target.sh b/jhove-bbt/scripts/create-1.31-target.sh
@@ -124,6 +124,7 @@ if [[ -f "${candidateRoot}/examples/modules/TIFF-hul/peppers.tif.jhove.xml" ]];
 	cp "${candidateRoot}/examples/modules/TIFF-hul/peppers.tif.jhove.xml" "${targetRoot}/examples/modules/TIFF-hul/peppers.tif.jhove.xml"
 fi
 
+# Copy the TIFF fix affected files from the candidate to the target
 declare -a tiff_affected=("examples/modules/TIFF-hul/cramps.tif.jhove.xml"
 				"examples/modules/TIFF-hul/text.tif.jhove.xml"
 				"examples/modules/TIFF-hul/testpage-small.tif.jhove.xml")
@@ -134,6 +135,7 @@ do
 	fi
 done
 
+# Copy the XHTML fix affected files from the candidate to the target
 declare -a xhtml_affected=("errors/modules/HTML-hul/xhtml-trans-no-xml-dec.html.jhove.xml"
 				"errors/modules/HTML-hul/xhtml-strict-no-xml-dec.html.jhove.xml"
 				"errors/modules/HTML-hul/xhtml-frames-no-xml-dec.html.jhove.xml"
@@ -144,3 +146,18 @@ do
 		cp "${candidateRoot}/${filename}" "${targetRoot}/${filename}"
 	fi
 done
+
+# Copy the XML fix affected files from the candidate to the target
+declare -a xhtml_affected=("errors/modules/HTML-hul/xhtml-trans-xml-dec.html.jhove.xml"
+				"errors/modules/HTML-hul/xhtml-strict-xml-dec.html.jhove.xml"
+				"errors/modules/HTML-hul/xhtml-frames-xml-dec.html.jhove.xml"
+				"errors/modules/HTML-hul/xhtml-1-1-xml-dec.html.jhove.xml"
+				"examples/modules/XML-hul/valid-external.dtd.jhove.xml"
+				"examples/modules/XML-hul/external-unparsed-entity.ent.jhove.xml"
+				"examples/modules/XML-hul/external-parsed-entity.ent.jhove.xml")
+for filename in "${xhtml_affected[@]}"
+do
+	if [[ -f "${candidateRoot}/${filename}" ]]; then
+		cp "${candidateRoot}/${filename}" "${targetRoot}/${filename}"
+	fi
+done
diff --git a/...ve/module/xml/ErrorMessages_nl.properties → ...e/module/jpeg/ErrorMessages_nl.properties b/...ve/module/xml/ErrorMessages_nl.properties → ...e/module/jpeg/ErrorMessages_nl.properties
diff --git a/...ve/module/xml/ErrorMessages_sv.properties → ...e/module/jpeg/ErrorMessages_sv.properties b/...ve/module/xml/ErrorMessages_sv.properties → ...e/module/jpeg/ErrorMessages_sv.properties
diff --git a/jhove-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/XmlModule.java b/jhove-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/XmlModule.java
@@ -396,13 +396,7 @@ public int parse(InputStream stream, RepInfo info, int parseIndex) {
             if (handler.getSigFlag() && !_parseFromSig) {
                 info.setSigMatch(_name);
             }
-            info.setMessage(new ErrorMessage(
-                    MessageConstants.XML_HUL_1,
-                    MessageFormat.format(
-                            MessageConstants.XML_HUL_1_SUB.getMessage(),
-                            spe.getMessage(),
-                            spe.getLineNumber(),
-                            spe.getColumnNumber())));
+            info.setMessage(new ErrorMessage(MessageConstants.INSTANCE.makeSaxParseMessage(spe)));
             info.setWellFormed(false);
             return 0;
         } catch (SAXException se) {

diff --git a/...-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/xml/MessageConstants.java b/...-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/xml/MessageConstants.java
@@ -1,5 +1,9 @@
 package edu.harvard.hul.ois.jhove.module.xml;
 
+import java.text.MessageFormat;
+
+import org.xml.sax.SAXParseException;
+
 import edu.harvard.hul.ois.jhove.messages.JhoveMessage;
 import edu.harvard.hul.ois.jhove.messages.JhoveMessageFactory;
 import edu.harvard.hul.ois.jhove.messages.JhoveMessages;
@@ -63,4 +67,15 @@ public enum MessageConstants {
 			.getMessage("XML-HUL-12");
 	public static final JhoveMessage XML_HUL_13 = messageFactory
 			.getMessage("XML-HUL-13");
+
+    public final JhoveMessage makeSaxParseMessage(final SAXParseException spe) {
+        return JhoveMessages.getMessageInstance(
+                MessageConstants.XML_HUL_1.getId(),
+                MessageFormat.format(MessageConstants.XML_HUL_1.getMessage(),
+                        spe.getMessage()),
+                MessageFormat.format(
+                        MessageConstants.XML_HUL_1_SUB.getMessage(),
+                        spe.getLineNumber(),
+                        spe.getColumnNumber()));
+    }
 }
diff --git a/...-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/xml/XmlModuleHandler.java b/...-modules/xml-hul/src/main/java/edu/harvard/hul/ois/jhove/module/xml/XmlModuleHandler.java
@@ -13,7 +13,12 @@
 import java.net.URL;
 import java.net.URLConnection;
 import java.text.MessageFormat;
-import java.util.*;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
 
 import org.xml.sax.Attributes;
 import org.xml.sax.InputSource;
@@ -392,9 +397,7 @@ public void unparsedEntityDecl(String name, String publicId,
 	 */
 	@Override
 	public void warning(SAXParseException spe) {
-		_messages.add(new InfoMessage(
-				MessageConstants.XML_HUL_1,
-				spe.getMessage()));
+		_messages.add(new InfoMessage(MessageConstants.INSTANCE.makeSaxParseMessage(spe)));
 	}
 
 	/**
@@ -413,14 +416,7 @@ public void error(SAXParseException spe) {
 							MAXERRORS));
 			_messages.add(new InfoMessage(message));
 		} else if (_nErrors < MAXERRORS) {
-			_messages.add(new ErrorMessage(
-					MessageConstants.XML_HUL_1,
-					MessageFormat.format(
-							MessageConstants.XML_HUL_1_SUB.getMessage(),
-							spe.getMessage(),
-							spe.getLineNumber(),
-							spe.getColumnNumber()
-					)));
+			_messages.add(new ErrorMessage(MessageConstants.INSTANCE.makeSaxParseMessage(spe)));
 		}
 		++_nErrors;
 	}

diff --git a/.../xml-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages.properties b/.../xml-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages.properties
@@ -1,5 +1,5 @@
-XML-HUL-1 = SAXParseException
-XML-HUL-1-SUB = {0} Line = {1,number,integer}, Column = {2,number,integer}.
+XML-HUL-1 = SAXParseException: {0}
+XML-HUL-1-SUB = Line = {0,number,integer}, Column = {1,number,integer}.
 XML-HUL-2 = Error messages in excess of {0,number,integer} not reported.
 XML-HUL-3 = SAXException: {0}
 XML-HUL-4 = Not able to determine end-of-line type.

diff --git a/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_da.properties b/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_da.properties
@@ -1,4 +1,5 @@
 XML-HUL-1 = SaxParseException: {0}
+XML-HUL-1-SUB = Line = {0,number,integer}, Column = {1,number,integer}.
 XML-HUL-2 = Fejlmeddelelser flere end {0} ikke rapporteret
 XML-HUL-3 = SaxParseException: {0}
 XML-HUL-4 = Ikke i stand til at bestemme typen af linjeslutning

diff --git a/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_en.properties b/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_en.properties
@@ -1,5 +1,5 @@
-XML-HUL-1 = SAXParseException
-XML-HUL-1-SUB = {0} Line = {1,number,integer}, Column = {2,number,integer}.
+XML-HUL-1 = SAXParseException: {0}
+XML-HUL-1-SUB = Line = {0,number,integer}, Column = {1,number,integer}.
 XML-HUL-2 = Error messages in excess of {0,number,integer} not reported.
 XML-HUL-3 = SAXException: {0}
 XML-HUL-4 = Not able to determine end-of-line type.

diff --git a/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_fr.properties b/...l-hul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_fr.properties
@@ -1,4 +1,5 @@
 XML-HUL-1 = SaxParseException : {0}
+XML-HUL-1-SUB = Line = {0,number,integer}, Column = {1,number,integer}.
 XML-HUL-2 = {0} messages d'erreur supplémentaires non signalés
 XML-HUL-3 = SaxParseException : {0}
 XML-HUL-4 = Impossible de déterminer le type de fin de ligne 

diff --git a/...ul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_pt_BR.properties b/...ul/src/main/resources/edu/harvard/hul/ois/jhove/module/xml/ErrorMessages_pt_BR.properties
@@ -1,4 +1,5 @@
 XML-HUL-1 = SaxParseException: {0}
+XML-HUL-1-SUB = Line = {0,number,integer}, Column = {1,number,integer}.
 XML-HUL-2 = As mensagens de erro acima de {0} não reportadas
 XML-HUL-3 = SaxParseException: {0}
 XML-HUL-4 = Não foi possível determinar o tipo de fim de linha