Fix JSON string length calculations

A JSON String length is the number of its Unicode code points; unfortunately, String's .length() counts the number of `char`s, that is the number of UTF-16 code units. But for Unicode code points outside the BMP, UTF-16 requires two code units, and String two `char`s as well. Change the length calculation by calling .codePointCount() instead. [backport from 2.1.x] Fixes: java-json-tools#92 Signed-off-by: Francis Galiegue <[email protected]>
githubcs · Apr 8, 2014 · 8e451cc · 8e451cc
1 parent a22e95c
commit 8e451cc
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 4 deletions.
diff --git a/src/main/java/com/github/fge/jsonschema/keyword/validator/common/MaxLengthValidator.java b/src/main/java/com/github/fge/jsonschema/keyword/validator/common/MaxLengthValidator.java
@@ -41,8 +41,8 @@ public void validate(final Processor<FullData, FullData> processor,
         final ProcessingReport report, final FullData data)
         throws ProcessingException
     {
-        final int size = data.getInstance().getNode().textValue()
-            .length();
+        final String value = data.getInstance().getNode().textValue();
+        final int size = value.codePointCount(0, value.length());
 
         if (size > intValue)
             report.error(newMsg(data).message(STRING_TOO_LONG)

diff --git a/src/main/java/com/github/fge/jsonschema/keyword/validator/common/MinLengthValidator.java b/src/main/java/com/github/fge/jsonschema/keyword/validator/common/MinLengthValidator.java
@@ -41,8 +41,8 @@ public void validate(final Processor<FullData, FullData> processor,
         final ProcessingReport report, final FullData data)
         throws ProcessingException
     {
-        final int size = data.getInstance().getNode().textValue()
-            .length();
+        final String value = data.getInstance().getNode().textValue();
+        final int size = value.codePointCount(0, value.length());
 
         if (size < intValue)
             report.error(newMsg(data).message(STRING_TOO_SHORT)

diff --git a/src/test/resources/keyword/validators/common/maxLength.json b/src/test/resources/keyword/validators/common/maxLength.json
@@ -18,5 +18,10 @@
         "digest": { "maxLength": 4 },
         "data": "foo",
         "valid": true
+    },
+    {
+        "digest": { "maxLength": 3 },
+        "data": "\uD83D\uDCA9\uD83D\uDCA9",
+        "valid": true
     }
 ]
diff --git a/src/test/resources/keyword/validators/common/minLength.json b/src/test/resources/keyword/validators/common/minLength.json
@@ -18,5 +18,15 @@
         "digest": { "minLength": 4 },
         "data": "foobar",
         "valid": true
+    },
+    {
+        "digest": { "minLength": 4 },
+        "data": "\uD83D\uDCA9\uD83D\uDCA9",
+        "valid": false,
+        "message": "STRING_TOO_SHORT",
+        "msgData": {
+            "minLength": 4,
+            "found": 2
+        }
     }
 ]