From 2e1e241bd96cf63f1302d7c908306c265615d605 Mon Sep 17 00:00:00 2001
From: Mark Rotteveel <mark@lawinegevaar.nl>
Date: Fri, 27 Oct 2023 12:45:19 +0200
Subject: [PATCH] #770 Fix FBStringField.fixPadding to work with codepoints
 instead of char - backport #760 to Jaybird 5.0.3

---
 src/docs/asciidoc/release_notes.adoc          |  3 ++
 .../firebirdsql/jdbc/field/FBStringField.java |  5 ++-
 .../jdbc/field/FBStringFieldTest.java         | 40 +++++++++++++++++++
 3 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/src/docs/asciidoc/release_notes.adoc b/src/docs/asciidoc/release_notes.adoc
index 031ae7027..6c7556ec0 100644
--- a/src/docs/asciidoc/release_notes.adoc
+++ b/src/docs/asciidoc/release_notes.adoc
@@ -37,6 +37,9 @@ Previously, Jaybird rejected attempts to read blobs with blob id `0` (not on all
 Formally, blob id `0` is not a valid blob id, but in practice they can occur (e.g. due to bugs, or access components/drivers explicitly setting a blob column to id `0`).
 Other drivers and tools simply send the requests for blob id `0` to the server, which then treats it as an empty blob.
 For consistency, we decided to let Jaybird handle it the same.
+* Fixed: on `CHAR` fields, a too short value could be returned if the string contained one or more codepoints represented by surrogate pairs and the string length in `char` exceeded the maximum string length -- backported from Jaybird 6 (https://github.com/FirebirdSQL/jaybird/issues/770[#770])
++
+We now truncate the returned string if the codepoint count exceeds the maximum string length.
 
 [#jaybird-5-0-2-changelog]
 === Jaybird 5.0.2
diff --git a/src/main/org/firebirdsql/jdbc/field/FBStringField.java b/src/main/org/firebirdsql/jdbc/field/FBStringField.java
index 515e78dac..d6bd7ffb9 100644
--- a/src/main/org/firebirdsql/jdbc/field/FBStringField.java
+++ b/src/main/org/firebirdsql/jdbc/field/FBStringField.java
@@ -197,8 +197,9 @@ private String fixPadding(String result) {
         // NOTE: For Firebird 3.0 and earlier, this prevents access to oversized CHAR(n) CHARACTER SET UNICODE_FSS.
         // We accept that limitation because the workaround is to cast to VARCHAR, and because Firebird 4.0 no longer
         // allows storing oversized UNICODE_FSS values
-        if (result.length() > possibleCharLength) {
-            return result.substring(0, possibleCharLength);
+        if (result != null && result.length() > possibleCharLength
+            && result.codePointCount(0, result.length()) > possibleCharLength) {
+            return result.substring(0, result.offsetByCodePoints(0, possibleCharLength));
         }
         return result;
     }
diff --git a/src/test/org/firebirdsql/jdbc/field/FBStringFieldTest.java b/src/test/org/firebirdsql/jdbc/field/FBStringFieldTest.java
index 9670f34f3..a64381af7 100644
--- a/src/test/org/firebirdsql/jdbc/field/FBStringFieldTest.java
+++ b/src/test/org/firebirdsql/jdbc/field/FBStringFieldTest.java
@@ -790,6 +790,46 @@ void getObject_java_time_ZonedDateTime() throws SQLException {
         assertEquals(expectedZonedDateTime, field.getObject(ZonedDateTime.class));
     }
 
+    @Test
+    void setString_surrogatePairs_maxLength() throws Exception {
+        fieldDescriptor = rowDescriptorBuilder
+                .setType(ISCConstants.SQL_TEXT)
+                .setLength(4 /* bytes */)
+                .setSubType(4 /* UTF8 */)
+                .toFieldDescriptor();
+        field = new FBStringField(fieldDescriptor, fieldData, Types.CHAR);
+        datatypeCoder = fieldDescriptor.getDatatypeCoder();
+        Encoding encoding = datatypeCoder.getEncoding();
+        assertEquals("UTF-8", encoding.getCharsetName(), "Unexpected charset for field");
+
+        // character Smiling Face with Open Mouth; uses surrogate pairs
+        String surrogatePairsValue = Character.toString(0x1f603);
+        assertEquals(2, surrogatePairsValue.length(), "Expected string with 2 characters (surrogate pairs)");
+        field.setString(surrogatePairsValue);
+
+        verifySetString(surrogatePairsValue, encoding);
+    }
+
+    @Test
+    void getString_surrogatePairs_maxLength() throws Exception {
+        fieldDescriptor = rowDescriptorBuilder
+                .setType(ISCConstants.SQL_TEXT)
+                .setLength(4 /* bytes */)
+                .setSubType(4 /* UTF8 */)
+                .toFieldDescriptor();
+        field = new FBStringField(fieldDescriptor, fieldData, Types.CHAR);
+        datatypeCoder = fieldDescriptor.getDatatypeCoder();
+        Encoding encoding = datatypeCoder.getEncoding();
+        assertEquals("UTF-8", encoding.getCharsetName(), "Unexpected charset for field");
+
+        // character Smiling Face with Open Mouth; uses surrogate pairs
+        String surrogatePairsValue = Character.toString(0x1f603);
+        assertEquals(2, surrogatePairsValue.length(), "Expected string with 2 characters (surrogate pairs)");
+        toReturnStringExpectations(surrogatePairsValue, encoding);
+
+        assertEquals(surrogatePairsValue, field.getString(), "Unexpected value for string with surrogate pairs");
+    }
+
     @Override
     String getNonNullObject() {
         return TEST_STRING_SHORT;