Fix #1046: implement max-doc-len limits

Author: cowtowncoder

release-notes/VERSION-2.x

@@ -31,6 +31,8 @@ a pure JSON library.
 #1041: Start using AssertJ in unit tests
 #1042: Allow configuring spaces before and/or after the colon in `DefaultPrettyPrinter`
  (contributed by @digulla)
+#1046: Add configurable limit for the maximum number of bytes/chars
+  of content to parse before failing
 #1047: Add configurable limit for the maximum length of Object property names
   to parse before failing (default max: 50,000 chars)
  (contributed by @pjfanning)

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

@@ -273,11 +273,14 @@ protected void _loadMoreGuaranteed() throws IOException {
     protected boolean _loadMore() throws IOException
     {
         if (_reader != null) {
+            final int bufSize = _inputEnd;
+            _currInputProcessed += bufSize;
+            _currInputRowStart -= bufSize;
+            // 06-Sep-2023, tatu: [core#1046] Enforce max doc length limit
+            streamReadConstraints().validateDocumentLength(_currInputProcessed);
+
             int count = _reader.read(_inputBuffer, 0, _inputBuffer.length);
             if (count > 0) {
-                final int bufSize = _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
@@ -289,6 +292,7 @@ protected boolean _loadMore() throws IOException
 
                 return true;
             }
+            _inputPtr = _inputEnd = 0;
             // End of input
             _closeInput();
             // Should never return 0, so let's fail

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

@@ -255,12 +255,14 @@ protected final boolean _loadMore() throws IOException
                 return false;
             }
 
+            final int bufSize = _inputEnd;
+            _currInputProcessed += bufSize;
+            _currInputRowStart -= bufSize;
+            // 06-Sep-2023, tatu: [core#1046] Enforce max doc length limit
+            streamReadConstraints().validateDocumentLength(_currInputProcessed);
+
             int count = _inputStream.read(_inputBuffer, 0, space);
             if (count > 0) {
-                final int bufSize = _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
@@ -272,6 +274,7 @@ protected final boolean _loadMore() throws IOException
 
                 return true;
             }
+            _inputPtr = _inputEnd = 0;
             // End of input
             _closeInput();
             // Should never return 0, so let's fail

src/test/java/com/fasterxml/jackson/core/constraints/LargeDocReadTest.java

@@ -33,7 +33,7 @@ public void testLargeNameChars() throws Exception {
 
     public void testLargeNameWithSmallLimitBytes() throws Exception
     {
-        final String doc = generateJSON(15_000);
+        final String doc = generateJSON(12_000);
         try (JsonParser p = createParserUsingStream(JSON_F_DOC_10K, doc, "UTF-8")) {
             consumeTokens(p);
             fail("expected StreamConstraintsException");
@@ -44,7 +44,7 @@ public void testLargeNameWithSmallLimitBytes() throws Exception
 
     public void testLargeNameWithSmallLimitChars() throws Exception
     {
-        final String doc = generateJSON(15_000);
+        final String doc = generateJSON(12_000);
         try (JsonParser p = createParserUsingReader(JSON_F_DOC_10K, doc)) {
             consumeTokens(p);
             fail("expected StreamConstraintsException");
@@ -55,7 +55,7 @@ public void testLargeNameWithSmallLimitChars() throws Exception
 
     public void testLargeNameWithSmallLimitAsync() throws Exception
     {
-        final byte[] doc = utf8Bytes(generateJSON(25_000));
+        final byte[] doc = utf8Bytes(generateJSON(12_000));
 
         // first with byte[] backend
         try (AsyncReaderWrapper p = asyncForBytes(JSON_F_DOC_10K, 1000, doc, 1)) {