From 94453bbd031d04a3c8777dfc6dcbbeafd84bfe48 Mon Sep 17 00:00:00 2001
From: Stefan Schmiedl <stefan@xss.de>
Date: Mon, 8 Jul 2019 22:25:48 +0200
Subject: [PATCH] skip over BOM in XML input

---
 .../ZUGFeRD/ZUGFeRDImporter.java              | 31 ++++++++++++++++++-
 1 file changed, 30 insertions(+), 1 deletion(-)

diff --git a/src/main/java/org/mustangproject/ZUGFeRD/ZUGFeRDImporter.java b/src/main/java/org/mustangproject/ZUGFeRD/ZUGFeRDImporter.java
index a33ea2e3..be95f293 100644
--- a/src/main/java/org/mustangproject/ZUGFeRD/ZUGFeRDImporter.java
+++ b/src/main/java/org/mustangproject/ZUGFeRD/ZUGFeRDImporter.java
@@ -180,11 +180,40 @@ public class ZUGFeRDImporter {
 		DocumentBuilderFactory xmlFact = DocumentBuilderFactory.newInstance();
 		xmlFact.setNamespaceAware(false);
 		DocumentBuilder builder = xmlFact.newDocumentBuilder();
-		Document doc = builder.parse(new ByteArrayInputStream(rawXML));
+		ByteArrayInputStream is = new ByteArrayInputStream(rawXML);
+		is.skip(guessBOMSize(is));
+		Document doc = builder.parse(is);
 		//prettyPrint(doc);
 		return doc;
 	}
 
+	/**
+	 * Skips over a BOM at the beginning of the given ByteArrayInputStream, if one exists.
+	 * @param is the ByteArrayInputStream used
+	 * @throws IOException
+	 * @see <a href="https://www.w3.org/TR/xml/#sec-guessing">Autodetection of Character Encodings</a>
+	 */
+	private int guessBOMSize(ByteArrayInputStream is) throws IOException {
+		byte[] pad = new byte[4];
+		is.read(pad);
+		is.reset();
+		int test2 = ((pad[0] & 0xFF) << 8) | (pad[1] & 0xFF);
+		int test3 = ((test2 & 0xFFFF) << 8) | (pad[2] & 0xFF);
+		int test4 = ((test3 & 0xFFFFFF) << 8) | (pad[3] & 0xFF);
+		//
+		if (test4 == 0x0000FEFF || test4 == 0xFFFE0000 || test4 == 0x0000FFFE || test4 == 0xFEFF0000) {
+			// UCS-4: BOM takes 4 bytes
+			return 4;
+		} else if (test3 == 0xEFBBFF) {
+			// UTF-8: BOM takes 3 bytes
+			return 3;
+		} else if (test2 == 0xFEFF || test2 == 0xFFFE) {
+			// UTF-16: BOM takes 2 bytes
+			return 2;
+		}
+		return 0;
+	}
+
 	private String extractString(String xpathStr) {
 		if (!containsMeta) {
 			throw new ZUGFeRDExportException("No suitable data/ZUGFeRD file could be found.");