#1 (bbottema/outlook-message-parser/issues/17): Restored support for …

…UTF-8's legacy name (cp)65001
bbottema · Oct 22, 2019 · 20a8f47 · 20a8f47
1 parent 4d25630
commit 20a8f47
Show file tree

Hide file tree

Showing 8 changed files with 2,174 additions and 5 deletions.
diff --git a/src/main/java/org/bbottema/rtftohtml/impl/util/CharsetHelper.java b/src/main/java/org/bbottema/rtftohtml/impl/util/CharsetHelper.java
@@ -3,12 +3,18 @@
 import java.nio.charset.Charset;
 import java.nio.charset.UnsupportedCharsetException;
 
+import static java.nio.charset.StandardCharsets.UTF_8;
+
 public class CharsetHelper {
 	private static String[] CHARSET_PREFIXES = {"", "cp", "iso-", "ibm", "x-windows-", "ms"};
-	
+
 	public static final Charset WINDOWS_CHARSET = Charset.forName("CP1252");
-	
+
 	public static Charset findCharset(String rtfCodePage) {
+		return rtfCodePage.equals("65001") || rtfCodePage.equalsIgnoreCase("cp65001") ? UTF_8 : detectCharset(rtfCodePage);
+	}
+
+	private static Charset detectCharset(String rtfCodePage) {
 		for (String prefix : CHARSET_PREFIXES) {
 			try {
 				return Charset.forName(prefix + rtfCodePage);

diff --git a/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterClassicTest.java b/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterClassicTest.java
@@ -23,12 +23,20 @@ public void testComplexRtfConversion() {
 
         assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
     }
-    
+
     @Test
-    public void testChineseRtfConversion()  {
+    public void testChineseRtfConversion() {
         String html = RTF2HTMLConverterClassic.INSTANCE.rtf2html(classpathFileToString("test-messages/input/chinese-exotic-test.rtf"));
         String expectedHtml = classpathFileToString("test-messages/output/classic/chinese-exotic-test.html");
-
+
+        assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
+    }
+
+    @Test
+    public void testUnicodeRtfConversion() {
+        String html = RTF2HTMLConverterClassic.INSTANCE.rtf2html(classpathFileToString("test-messages/input/unicode-test.rtf"));
+        String expectedHtml = classpathFileToString("test-messages/output/classic/unicode-test.html");
+
         assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
     }
 }
diff --git a/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterJEditorPaneTest.java b/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterJEditorPaneTest.java
@@ -31,4 +31,12 @@ public void testChineseRtfConversion() {
 
         assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
     }
+
+    @Test
+    public void testUnicodeRtfConversion() {
+        String html = RTF2HTMLConverterClassic.INSTANCE.rtf2html(classpathFileToString("test-messages/input/unicode-test.rtf"));
+        String expectedHtml = classpathFileToString("test-messages/output/swing/unicode-test.html");
+
+        assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
+    }
 }
diff --git a/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterRFCCompliantTest.java b/src/test/java/org/bbottema/rtftohtml/impl/RTF2HTMLConverterRFCCompliantTest.java
@@ -31,4 +31,12 @@ public void testChineseRtfConversion()  {
 
         assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
     }
+
+    @Test
+    public void testUnicodeRtfConversion() {
+        String html = RTF2HTMLConverterClassic.INSTANCE.rtf2html(classpathFileToString("test-messages/input/unicode-test.rtf"));
+        String expectedHtml = classpathFileToString("test-messages/output/rfcompliant/unicode-test.html");
+
+        assertThat(normalizeText(html)).isEqualTo(normalizeText(expectedHtml));
+    }
 }
diff --git a/src/test/resources/test-messages/input/unicode-test.rtf b/src/test/resources/test-messages/input/unicode-test.rtf
diff --git a/src/test/resources/test-messages/output/classic/unicode-test.html b/src/test/resources/test-messages/output/classic/unicode-test.html
diff --git a/src/test/resources/test-messages/output/rfcompliant/unicode-test.html b/src/test/resources/test-messages/output/rfcompliant/unicode-test.html
diff --git a/src/test/resources/test-messages/output/swing/unicode-test.html b/src/test/resources/test-messages/output/swing/unicode-test.html