Backport #603 fix, update release notes

Author: fabienrenaud

release-notes/CREDITS-2.x

@@ -196,9 +196,12 @@ David Nault (dnault@github)
   (2.10.0)
 
 Fabien Renaud (fabienrenaud@github)
-  * Reported, contributed fix fir #533: UTF-8 BOM not accounted for in
+  * Reported, contributed fix for #533: UTF-8 BOM not accounted for in
     `JsonLocation.getByteOffset()`
   (2.10.0)
+  * Reported, contributed fix for #603: 'JsonParser.getCurrentLocation()`
+    byte/char offset update incorrectly for big payloads
+  (2.10.3)
 
 Todd O'Bryan (toddobryan@github)
   * Contributed fix fox #455: Jackson reports wrong locations for JsonEOFException

release-notes/VERSION-2.x

@@ -18,6 +18,8 @@ JSON library.
 
 #592: DataFormatMatcher#getMatchedFormatName throws NPE when no match exists
  (reported by Scott L)
+#603: 'JsonParser.getCurrentLocation()` byte/char offset update incorrectly for big payloads
+ (reported, fix contributed by Fabien R)
 
 2.10.2 (05-Jan-2020)
 

src/main/java/com/fasterxml/jackson/core/json/ReaderBasedJsonParser.java

@@ -244,14 +244,10 @@ protected void _loadMoreGuaranteed() throws IOException {
 
     protected boolean _loadMore() throws IOException
     {
-        final int bufSize = _inputEnd;
-
         if (_reader != null) {
             int count = _reader.read(_inputBuffer, 0, _inputBuffer.length);
             if (count > 0) {
-                _inputPtr = 0;
-                _inputEnd = count;
-
+                final int bufSize = _inputEnd;
                 _currInputProcessed += bufSize;
                 _currInputRowStart -= bufSize;
 
@@ -260,6 +256,9 @@ protected boolean _loadMore() throws IOException
                 //   in negative value, which is fine as combine value remains unchanged.
                 _nameStartOffset -= bufSize;
 
+                _inputPtr = 0;
+                _inputEnd = count;
+
                 return true;
             }
             // End of input

src/main/java/com/fasterxml/jackson/core/json/UTF8StreamJsonParser.java

@@ -210,7 +210,6 @@ public Object getInputSource() {
 
     protected final boolean _loadMore() throws IOException
     {
-        final int bufSize = _inputEnd;
         if (_inputStream != null) {
             int space = _inputBuffer.length;
             if (space == 0) { // only occurs when we've been closed
@@ -219,17 +218,19 @@ protected final boolean _loadMore() throws IOException
 
             int count = _inputStream.read(_inputBuffer, 0, space);
             if (count > 0) {
-                _inputPtr = 0;
-                _inputEnd = count;
+                final int bufSize = _inputEnd;
 
-                _currInputProcessed += _inputEnd;
-                _currInputRowStart -= _inputEnd;
+                _currInputProcessed += bufSize;
+                _currInputRowStart -= bufSize;
 
                 // 26-Nov-2015, tatu: Since name-offset requires it too, must offset
                 //   this increase to avoid "moving" name-offset, resulting most likely
                 //   in negative value, which is fine as combine value remains unchanged.
                 _nameStartOffset -= bufSize;
 
+                _inputPtr = 0;
+                _inputEnd = count;
+
                 return true;
             }
             // End of input

src/test/java/com/fasterxml/jackson/core/read/LocationOffsetsTest.java

@@ -2,6 +2,9 @@
 
 import com.fasterxml.jackson.core.*;
 
+import java.io.IOException;
+import java.util.Random;
+
 public class LocationOffsetsTest extends com.fasterxml.jackson.core.BaseTest
 {
     final JsonFactory JSON_F = new JsonFactory();
@@ -104,6 +107,45 @@ public void testOffsetWithoutInputOffset() throws Exception
         p.close();
     }
 
+    public void testWithLazyStringReadStreaming() throws Exception
+    {
+        _testWithLazyStringRead(MODE_READER);
+        _testWithLazyStringRead(MODE_INPUT_STREAM);
+    }
+
+    public void testWithLazyStringReadDataInput() throws Exception
+    {
+        // DataInput-backed reader does not track column, so can not
+        // verify much; but force finishToken() regardless
+        JsonParser p = createParser(JSON_F, MODE_DATA_INPUT, "[\"text\"]");
+        assertToken(JsonToken.START_ARRAY, p.nextToken());
+        assertToken(JsonToken.VALUE_STRING, p.nextToken());
+        assertEquals(1, p.getCurrentLocation().getLineNr());
+        p.finishToken();
+        assertEquals("text", p.getText());
+        p.close();
+    }
+
+    private void _testWithLazyStringRead(int readMode) throws Exception
+    {
+        JsonParser p = createParser(JSON_F, readMode, "[\"text\"]");
+        assertToken(JsonToken.START_ARRAY, p.nextToken());
+        assertToken(JsonToken.VALUE_STRING, p.nextToken());
+        // initially location pointing to first character
+        assertEquals(3, p.getCurrentLocation().getColumnNr());
+        p.finishToken();
+        // but will move once we force reading
+        assertEquals(8, p.getCurrentLocation().getColumnNr());
+        // and no change if we call again (but is ok to call)
+        p.finishToken();
+        assertEquals(8, p.getCurrentLocation().getColumnNr());
+
+        // also just for fun, verify content
+        assertEquals("text", p.getText());
+        assertEquals(8, p.getCurrentLocation().getColumnNr());
+        p.close();
+    }
+
     // for [core#533]
     public void testUtf8Bom() throws Exception
     {
@@ -192,4 +234,75 @@ private byte[] withUtf8Bom(byte[] bytes) {
         System.arraycopy(bytes, 0, arr, 3, bytes.length);
         return arr;
     }
+
+    // [core#603]
+    public void testBigPayload() throws IOException {
+        JsonLocation loc;
+        JsonParser p;
+
+        String doc = "{\"key\":\"" + generateRandomAlpha(50000) + "\"}";
+
+        p = createParserUsingStream(JSON_F, doc, "UTF-8");
+
+        assertToken(JsonToken.START_OBJECT, p.nextToken());
+        loc = p.getTokenLocation();
+        assertEquals(0, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(1, loc.getColumnNr());
+        loc = p.getCurrentLocation();
+        assertEquals(1, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(2, loc.getColumnNr());
+
+        assertToken(JsonToken.FIELD_NAME, p.nextToken());
+        loc = p.getTokenLocation();
+        assertEquals(1, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(2, loc.getColumnNr());
+        loc = p.getCurrentLocation();
+        assertEquals(8, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(9, loc.getColumnNr());
+
+        assertToken(JsonToken.VALUE_STRING, p.nextToken());
+        loc = p.getTokenLocation();
+        assertEquals(7, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(8, loc.getColumnNr());
+        loc = p.getCurrentLocation();
+        assertEquals(8, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(9, loc.getColumnNr());
+
+        p.getTextCharacters();
+        loc = p.getTokenLocation();
+        assertEquals(7, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(8, loc.getColumnNr());
+        loc = p.getCurrentLocation();
+        assertEquals(doc.length() - 1, loc.getByteOffset());
+        assertEquals(-1L, loc.getCharOffset());
+        assertEquals(1, loc.getLineNr());
+        assertEquals(doc.length(), loc.getColumnNr());
+
+        p.close();
+    }
+
+    private String generateRandomAlpha(int length) {
+        StringBuilder sb = new StringBuilder(length);
+        Random rnd = new Random(length);
+        for (int i = 0; i < length; ++i) {
+            // let's limit it not to include surrogate pairs:
+            char ch = (char) ('A' + rnd.nextInt(26));
+            sb.append(ch);
+        }
+        return sb.toString();
+    }
 }