From: James Ahlborn <jtahlborn@yahoo.com>
Date: Sun, 15 Sep 2013 02:05:11 +0000 (+0000)
Subject: implement nicer ascii text conversion for ole package paths
X-Git-Tag: jackcess-2.0.1~21
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=08c5914dc92523a8edb986a178d38abdb5f8db28;p=jackcess.git

implement nicer ascii text conversion for ole package paths

git-svn-id: https://svn.code.sf.net/p/jackcess/code/jackcess/trunk@801 f203690c-595d-4dc9-a70b-905162fa7fd2
---

diff --git a/src/main/java/com/healthmarketscience/jackcess/impl/OleUtil.java b/src/main/java/com/healthmarketscience/jackcess/impl/OleUtil.java
index b82ded2..4471da5 100644
--- a/src/main/java/com/healthmarketscience/jackcess/impl/OleUtil.java
+++ b/src/main/java/com/healthmarketscience/jackcess/impl/OleUtil.java
@@ -30,8 +30,10 @@ import java.nio.charset.Charset;
 import java.sql.Blob;
 import java.sql.SQLException;
 import java.sql.SQLFeatureNotSupportedException;
+import java.text.Normalizer;
 import java.util.EnumSet;
 import java.util.Set;
+import java.util.regex.Pattern;
 
 import com.healthmarketscience.jackcess.DataType;
 import com.healthmarketscience.jackcess.util.OleBlob;
@@ -78,6 +80,10 @@ public class OleUtil
     0x00, 0x00, 0x01, (byte)0xAD, 0x05, (byte)0xFE
   };
 
+  // regex pattern which matches all the crazy extra stuff in unicode
+  private static final Pattern UNICODE_ACCENT_PATTERN = 
+    Pattern.compile("[\\p{InCombiningDiacriticalMarks}\\p{IsLm}\\p{IsSk}]+");
+
   private static final CompoundPackageFactory COMPOUND_FACTORY;
 
   static {
@@ -433,6 +439,17 @@ public class OleUtil
   }
 
   private static byte[] getZeroTermStrBytes(String str) {
+    // since we are converting to ascii, try to make "nicer" versions of crazy
+    // chars (e.g. convert "u with an umlaut" to just "u").  this may not
+    // ultimately help anything but it is what ms access does.
+
+    // decompose complex chars into combos of char and accent
+    str = Normalizer.normalize(str, Normalizer.Form.NFD);
+    // strip the accents
+    str = UNICODE_ACCENT_PATTERN.matcher(str).replaceAll("");
+    // (re)normalize what is left
+    str = Normalizer.normalize(str, Normalizer.Form.NFC);
+
     return (str + '\0').getBytes(OLE_CHARSET);
   }