fix: handle \r in windows

Signed-off-by: Hiroshi Miura <[email protected]>
omegat-org · Jun 5, 2024 · a5ea8d4 · a5ea8d4
1 parent 0363a9e
commit a5ea8d4
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 5 deletions.
diff --git a/src/org/omegat/util/TmxEscapingWriterFactory.java b/src/org/omegat/util/TmxEscapingWriterFactory.java
@@ -62,6 +62,7 @@ public static class EscapeWriter extends Writer {
             q['<'] = 1;
             q['>'] = 1;
             q['&'] = 1;
+            q['\r'] = (byte) (Platform.isWindows ? 0 : 1);
             QUOTABLE_TEXT_CHARS = q;
         }
 

diff --git a/test/src/org/omegat/util/TMXWriterTest.java b/test/src/org/omegat/util/TMXWriterTest.java
@@ -24,6 +24,7 @@
  **************************************************************************/
 package org.omegat.util;
 
+import static org.assertj.core.api.Assertions.assertThat;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertTrue;
 
@@ -178,7 +179,7 @@ public void testEOLwrite() throws Exception {
                 text.append(buffer, 0, len);
             }
         }
-        assertTrue(text.toString().contains("tar" + eol + "get"));
+        assertThat(text.toString()).as("Preserve EOL mark in text.").contains("tar" + eol + "get");
 
         final List<String> trs = new ArrayList<>();
         load(null, trs, true, false);

diff --git a/test/src/org/omegat/util/TmxEscapingWriterTest.java b/test/src/org/omegat/util/TmxEscapingWriterTest.java
@@ -48,31 +48,81 @@ public void setUp() throws UnsupportedEncodingException {
         writer = factory.createEscapingWriterFor(outputStream, null);
     }
 
+    /**
+     * Test basic characters, as-is.
+     *
+     * @throws IOException I/O error happened.
+     */
     @Test
     public void escapeBasic() throws IOException {
         writer.write("Hello world!\n");
         writer.flush();
         assertThat(outputStream.toString()).isEqualTo("Hello world!\n");
     }
 
+    /**
+     * Test signs to be escaped.
+     * @throws IOException I/O error happened.
+     */
     @Test
     public void escapeToEntities() throws IOException {
-        writer.write("\"<escape>\"");
+        writer.write("'<escape>\"");
         writer.flush();
-        assertThat(outputStream.toString()).isEqualTo("\"&lt;escape&gt;\"");
+        assertThat(outputStream.toString())
+                .as("Check escape of < & and > signs and as-is for single/double quote")
+                .isEqualTo("'&lt;escape&gt;\"");
     }
 
+    /**
+     * Test escape of NBSP, no-escape.
+     * @throws IOException I/O error happened.
+     */
+    @Test
+    public void testNBSP() throws IOException {
+        writer.write("\u00a0");
+        writer.flush();
+        assertThat(outputStream.toString())
+                .as("Check NBSP is not escaped.")
+                .isEqualTo("\u00a0");
+    }
+
+    /**
+     * Test Control character No-Break-Here, escape.
+     *
+     * @throws IOException I/O error happened.
+     */
+    @Test
+    public void testNBH() throws IOException {
+        writer.write("\u0083");
+        writer.flush();
+        assertThat(outputStream.toString())
+                .as("Check NO_BREAK_HERE control character to be escaped")
+                .isEqualTo("&#x83;");
+    }
+
+    /**
+     * Test emoji and flag, surrogate pair, escape.
+     * @throws IOException I/O error happened.
+     */
     @Test
     public void testSurrogatePair() throws IOException {
         writer.write("\uD83D\uDE00");
         writer.flush();
-        assertThat(outputStream.toString()).isEqualTo("\uD83D\uDE00");
+        assertThat(outputStream.toString())
+                .as("Check emoji and flag that requires surrogate pair for encode.")
+                .isEqualTo("\uD83D\uDE00");
     }
 
+    /**
+     * Test Invalid character, BOM flag, escape.
+     * @throws IOException I/O error happened.
+     */
     @Test
     public void testInvalidChar() throws IOException {
         writer.write((char) 0xFFFE);
         writer.flush();
-        assertThat(outputStream.toString()).isEqualToIgnoringCase("&#xfffe;");
+        assertThat(outputStream.toString())
+                .as("check BOM mark to be escaped when appeared in TEXT.")
+                .isEqualToIgnoringCase("&#xfffe;");
     }
 }