ICU4J: Throw parse error for lone surrogates

unicode-org · Sep 18, 2024 · fb0c873 · fb0c873
1 parent 5e376ba
commit fb0c873
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 1 deletion.
diff --git a/icu4j/main/core/src/main/java/com/ibm/icu/message2/InputSource.java b/icu4j/main/core/src/main/java/com/ibm/icu/message2/InputSource.java
@@ -29,6 +29,13 @@ int peekChar() {
         return buffer.charAt(cursor);
     }
 
+    // Used when checking for unpaired surrogates
+    int readChar() {
+        int ch = peekChar();
+        cursor++;
+        return (char) ch;
+    }
+
     int readCodePoint() {
         // TODO: remove this?
         // START Detect possible infinite loop

diff --git a/icu4j/main/core/src/main/java/com/ibm/icu/message2/MFParser.java b/icu4j/main/core/src/main/java/com/ibm/icu/message2/MFParser.java
@@ -113,9 +113,17 @@ private MFDataModel.PatternPart getPatternPart() throws MFParseException {
         }
     }
 
-    private String getText() {
+    private String getText() throws MFParseException {
         StringBuilder result = new StringBuilder();
         while (true) {
+            int ch = input.readChar();
+            // Check for unmatched surrogates
+            checkCondition(!(Character.isHighSurrogate((char) ch)
+                             && (input.atEnd()
+                                 || !Character.isLowSurrogate((char) input.peekChar()))),
+                           "Unpaired high surrogate");
+            checkCondition (!Character.isLowSurrogate((char) ch), "Unpaired low surrogate");
+            input.backup(1);
             int cp = input.readCodePoint();
             switch (cp) {
                 case EOF:

diff --git a/icu4j/main/core/src/test/java/com/ibm/icu/dev/test/message2/MessageFormat2Test.java b/icu4j/main/core/src/test/java/com/ibm/icu/dev/test/message2/MessageFormat2Test.java
@@ -46,6 +46,36 @@ public void test() {
                 mf2.formatToString(Args.NONE));
     }
 
+    @Test
+    public void testHighLoneSurrogate() {
+        try {
+            MessageFormatter mf2 = MessageFormatter.builder()
+                .setPattern("\uda02").build();
+            assertEquals("testHighLoneSurrogate: expected to throw, but didn't",
+                         false, true);
+        } catch (IllegalArgumentException e) {
+            // Parse error was thrown, as expected
+        } catch (Exception e) {
+            assertEquals("testHighLoneSurrogate: expected IllegalArgumentException "
+                         + "but threw " + e.toString(), false, true);
+        }
+    }
+
+    @Test
+    public void testLowLoneSurrogate() {
+        try {
+            MessageFormatter mf2 = MessageFormatter.builder()
+                .setPattern("\udc02").build();
+            assertEquals("testLowLoneSurrogate: expected to throw, but didn't",
+                         false, true);
+        } catch (IllegalArgumentException e) {
+            // Parse error was thrown, as expected
+        } catch (Exception e) {
+            assertEquals("testLowLoneSurrogate: expected IllegalArgumentException "
+                         + "but threw " + e.toString(), false, true);
+        }
+    }
+
     @Test
     public void testDateFormat() {
         Date expiration = new Date(2022 - 1900, java.util.Calendar.OCTOBER, 27);