Remove BOM from UTF-8 encoded file (String) if present

-SITE-3007 ETT GG "Validator broke" - UTF-8-BOM issue
onc-healthit · Jun 6, 2019 · b2dce3c · b2dce3c
1 parent 970bce6
commit b2dce3c
Show file tree

Hide file tree

Showing 4 changed files with 39,910 additions and 560 deletions.
diff --git a/src/main/java/org/sitenv/referenceccda/services/ReferenceCCDAValidationService.java b/src/main/java/org/sitenv/referenceccda/services/ReferenceCCDAValidationService.java
@@ -109,7 +109,13 @@ private List<RefCCDAValidationResult> runValidators(String validationObjective,
         InputStream ccdaFileInputStream = null;
         try {
             ccdaFileInputStream = ccdaFile.getInputStream();
-            String ccdaFileContents = IOUtils.toString(new BOMInputStream(ccdaFileInputStream));
+            BOMInputStream bomInputStream = new BOMInputStream(ccdaFileInputStream);
+            if(bomInputStream.hasBOM()) {
+				logger.warn(
+						"The C-CDA file has a BOM which is supposed to be removed by BOMInputStream - encoding w/o BOM: "
+								+ bomInputStream.getBOMCharsetName());
+            }
+            String ccdaFileContents = IOUtils.toString(bomInputStream, "UTF-8");
 
             List<RefCCDAValidationResult> mdhtResults = doMDHTValidation(validationObjective, referenceFileName, ccdaFileContents);
             if(mdhtResults != null && !mdhtResults.isEmpty()) {

diff --git a/src/main/java/org/sitenv/referenceccda/validators/BaseCCDAValidator.java b/src/main/java/org/sitenv/referenceccda/validators/BaseCCDAValidator.java
@@ -1,23 +1,36 @@
 package org.sitenv.referenceccda.validators;
 
+import java.io.IOException;
+import java.io.StringReader;
+
+import org.apache.log4j.Logger;
 import org.xml.sax.InputSource;
 import org.xml.sax.SAXException;
 import org.xml.sax.XMLReader;
 import org.xml.sax.helpers.XMLReaderFactory;
 
-import java.io.IOException;
-import java.io.StringReader;
-
 public abstract class BaseCCDAValidator {
-    protected static void trackXPathsInXML(XPathIndexer xpathIndexer, String xmlString) throws SAXException{
+    public static final String UTF8_BOM = "\uFEFF";
+    private static Logger logger = Logger.getLogger(BaseCCDAValidator.class);
+
+    protected static void trackXPathsInXML(XPathIndexer xpathIndexer, String xmlString) throws SAXException{    	    	
         XMLReader parser = XMLReaderFactory.createXMLReader();
         parser.setContentHandler(xpathIndexer);
         try {
+        	xmlString = ifHasUtf8BomThenRemove(xmlString);
             InputSource inputSource = new InputSource(new StringReader(xmlString));
             parser.parse(inputSource);
         } catch (IOException e) {
             e.printStackTrace();
-            System.out.println("Error In Line Number Routine: Bad filename, path or invalid document.");
+            logger.error("Error In Line Number Routine: Bad filename, path or invalid document.");
+        }
+    }
+
+    private static String ifHasUtf8BomThenRemove(String xml) {
+        if (xml.startsWith(UTF8_BOM)) {
+        	logger.warn("Found UTF-8 BOM, removing...");
+        	xml = xml.substring(1);
         }
+        return xml;
     }
 }