Add tests from JSONTestSuite

src/main/java/dev/blaauwendraad/masker/json/KeyContainsMasker.java

@@ -52,7 +52,7 @@ public byte[] mask(byte[] input) {
             visitValue(maskingState, keyMaskingConfig);
 
             return maskingState.flushReplacementOperations();
-        } catch (ArrayIndexOutOfBoundsException e) {
+        } catch (ArrayIndexOutOfBoundsException | StackOverflowError e) {
             throw new InvalidJsonException("Invalid JSON input provided: %s".formatted(e.getMessage()), e);
         }
     }
@@ -65,6 +65,9 @@ public byte[] mask(byte[] input) {
      *                         being masked
      */
     private void visitValue(MaskingState maskingState, @Nullable KeyMaskingConfig keyMaskingConfig) {
+        if (maskingState.endOfJson()) {
+            return;
+        }
         // using switch-case over 'if'-statements to improve performance by ~20% (measured in benchmarks)
         switch (maskingState.byteAtCurrentIndex()) {
             case '[' -> visitArray(maskingState, keyMaskingConfig);
@@ -123,7 +126,7 @@ private void visitArray(MaskingState maskingState, @Nullable KeyMaskingConfig ke
 
             stepOverWhitespaceCharacters(maskingState);
             // check if we're at the end of a (non-empty) array
-            if (maskingState.byteAtCurrentIndex() == ']') {
+            if (maskingState.endOfJson() || maskingState.byteAtCurrentIndex() == ']') {
                 break;
             }
         }
@@ -185,7 +188,7 @@ private void visitObject(MaskingState maskingState, @Nullable KeyMaskingConfig p
 
             stepOverWhitespaceCharacters(maskingState);
             // check if we're at the end of a (non-empty) object
-            if (maskingState.byteAtCurrentIndex() == '}') {
+            if (maskingState.endOfJson() || maskingState.byteAtCurrentIndex() == '}') {
                 break;
             }
         }
@@ -292,7 +295,7 @@ private static void stepOverWhitespaceCharacters(MaskingState maskingState) {
     private static void stepOverNumericValue(MaskingState maskingState) {
         do {
             maskingState.next();
-        } while (maskingState.currentIndex() < maskingState.getMessage().length && AsciiJsonUtil.isNumericCharacter(maskingState.byteAtCurrentIndex()));
+        } while (!maskingState.endOfJson() && AsciiJsonUtil.isNumericCharacter(maskingState.byteAtCurrentIndex()));
     }
 
     /**

src/main/java/dev/blaauwendraad/masker/json/MaskingState.java

@@ -23,7 +23,7 @@ final class MaskingState implements ValueMaskerContext {
      * Current JSONPath is represented by a stack of segment references.
      * A stack is implemented with an array of the trie nodes that reference the end of the segment
      */
-    private KeyMatcher.@Nullable TrieNode[] currentJsonPath = null;
+    private KeyMatcher.@Nullable TrieNode @Nullable [] currentJsonPath = null;
     private int currentJsonPathHeadIndex = -1;
     private int currentValueStartIndex = -1;
 
@@ -47,7 +47,7 @@ public byte byteAtCurrentIndex() {
     }
 
     public boolean endOfJson() {
-        return currentIndex == message.length;
+        return currentIndex >= message.length;
     }
 
     public int currentIndex() {
@@ -209,7 +209,7 @@ public byte getByte(int index) {
 
     @Override
     public int byteLength() {
-        return currentIndex - getCurrentValueStartIndex();
+        return Math.min(currentIndex, message.length) - getCurrentValueStartIndex();
     }
 
     @Override
@@ -256,7 +256,7 @@ public String toString() {
         StringBuilder sb = new StringBuilder();
         sb.append(new String(message, Math.max(0, currentIndex - 10), Math.min(10, currentIndex)));
         sb.append(">");
-        if (currentIndex == message.length) {
+        if (endOfJson()) {
             sb.append("<end of json>");
         } else {
             sb.append((char) message[currentIndex]);

src/main/java/dev/blaauwendraad/masker/json/ValueMaskers.java

@@ -429,7 +429,7 @@ the byte encoded unicode character (\\u + 4 hex bytes) to prevent possible Array
                                                 decodedBytes[decodedIndex++] = (byte) (0x80 | ((unicodeHexBytesAsChar >> 6) & 0x3f));
                                                 decodedBytes[decodedIndex++] = (byte) (0x80 | (unicodeHexBytesAsChar & 0x3f));
                                             }
-                                        } catch (IllegalArgumentException e) {
+                                        } catch (IllegalArgumentException | IndexOutOfBoundsException e) {
                                             throw context.invalidJson(Objects.requireNonNull(e.getMessage()), valueStartIndex);
                                         }
                                     }

src/test/JSONTestSuite/masked/i_number_double_huge_neg_exp.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_huge_exp.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_neg_int_huge_exp.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_pos_double_huge_exp.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_real_neg_overflow.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_real_pos_overflow.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_real_underflow.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_too_big_neg_int.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_too_big_pos_int.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_number_very_big_negative_int.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/i_object_key_lone_2nd_surrogate.json

@@ -0,0 +1 @@
+{"\uDFAA":"###"}

src/test/JSONTestSuite/masked/i_string_1st_surrogate_but_2nd_missing.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_1st_valid_surrogate_2nd_invalid.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_UTF-8_invalid_sequence.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_UTF8_surrogate_U+D800.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_incomplete_surrogate_and_escape_valid.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_incomplete_surrogate_pair.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_incomplete_surrogates_escape_valid.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_invalid_lonely_surrogate.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_invalid_surrogate.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_invalid_utf-8.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_inverted_surrogates_U+1D11E.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_iso_latin_1.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_lone_second_surrogate.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_lone_utf8_continuation_byte.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_not_in_unicode_range.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_overlong_sequence_2_bytes.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_overlong_sequence_6_bytes.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_overlong_sequence_6_bytes_null.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_string_truncated-utf-8.json

@@ -0,0 +1 @@
+["***"]

src/test/JSONTestSuite/masked/i_structure_500_nested_arrays.json

@@ -0,0 +1 @@
+[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]

src/test/JSONTestSuite/masked/i_structure_UTF-8_BOM_empty_object.json

@@ -0,0 +1 @@
+{}

src/test/JSONTestSuite/masked/n_array_1_true_without_comma.json

@@ -0,0 +1 @@
+["###" true]

src/test/JSONTestSuite/masked/n_array_a_invalid_utf8.json

@@ -0,0 +1 @@
+[aå]

src/test/JSONTestSuite/masked/n_array_colon_instead_of_comma.json

@@ -0,0 +1 @@
+["***": "###"]

src/test/JSONTestSuite/masked/n_array_comma_after_close.json

@@ -0,0 +1 @@
+["***"],

src/test/JSONTestSuite/masked/n_array_comma_and_number.json

@@ -0,0 +1 @@
+[,"###"]

src/test/JSONTestSuite/masked/n_array_double_comma.json

@@ -0,0 +1 @@
+["###",,"###"]

src/test/JSONTestSuite/masked/n_array_double_extra_comma.json

@@ -0,0 +1 @@
+["***",,]

src/test/JSONTestSuite/masked/n_array_extra_close.json

@@ -0,0 +1 @@
+["***"]]

src/test/JSONTestSuite/masked/n_array_extra_comma.json

@@ -0,0 +1 @@
+["***",]

src/test/JSONTestSuite/masked/n_array_incomplete.json

@@ -0,0 +1 @@
+["***"

src/test/JSONTestSuite/masked/n_array_incomplete_invalid_value.json

@@ -0,0 +1 @@
+[x

src/test/JSONTestSuite/masked/n_array_inner_array_no_comma.json

@@ -0,0 +1 @@
+["###"["###"]]

src/test/JSONTestSuite/masked/n_array_invalid_utf8.json

@@ -0,0 +1 @@
+[ÿ]

src/test/JSONTestSuite/masked/n_array_items_separated_by_semicolon.json

@@ -0,0 +1 @@
+["###":"###"]

src/test/JSONTestSuite/masked/n_array_just_comma.json

@@ -0,0 +1 @@
+[,]

src/test/JSONTestSuite/masked/n_array_just_minus.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_array_missing_value.json

@@ -0,0 +1 @@
+[   , "***"]

src/test/JSONTestSuite/masked/n_array_newlines_unclosed.json

@@ -0,0 +1,3 @@
+["***",
+"###"
+,"###",

src/test/JSONTestSuite/masked/n_array_number_and_comma.json

@@ -0,0 +1 @@
+["###",]

src/test/JSONTestSuite/masked/n_array_number_and_several_commas.json

@@ -0,0 +1 @@
+["###",,]

src/test/JSONTestSuite/masked/n_array_spaces_vertical_tab_formfeed.json

@@ -0,0 +1 @@
+["***"\"&&&"

src/test/JSONTestSuite/masked/n_array_star_inside.json

@@ -0,0 +1 @@
+[*]

src/test/JSONTestSuite/masked/n_array_unclosed.json

@@ -0,0 +1 @@
+["***"

src/test/JSONTestSuite/masked/n_array_unclosed_trailing_comma.json

@@ -0,0 +1 @@
+["###",

src/test/JSONTestSuite/masked/n_array_unclosed_with_new_lines.json

@@ -0,0 +1,3 @@
+["###",
+"###"
+,"###"

src/test/JSONTestSuite/masked/n_array_unclosed_with_object_inside.json

@@ -0,0 +1 @@
+[{}

src/test/JSONTestSuite/masked/n_incomplete_false.json

@@ -0,0 +1 @@
+["&&&"

src/test/JSONTestSuite/masked/n_incomplete_null.json

@@ -0,0 +1 @@
+[nul]

src/test/JSONTestSuite/masked/n_incomplete_true.json

@@ -0,0 +1 @@
+["&&&"

src/test/JSONTestSuite/masked/n_number_++.json

@@ -0,0 +1 @@
+[++"###"]

src/test/JSONTestSuite/masked/n_number_+1.json

@@ -0,0 +1 @@
+[+"###"]

src/test/JSONTestSuite/masked/n_number_+Inf.json

@@ -0,0 +1 @@
+[+Inf]

src/test/JSONTestSuite/masked/n_number_-01.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_-1.0..json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_-2..json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_-NaN.json

@@ -0,0 +1 @@
+["###"NaN]

src/test/JSONTestSuite/masked/n_number_.-1.json

@@ -0,0 +1 @@
+[."###"]

src/test/JSONTestSuite/masked/n_number_.2e-3.json

@@ -0,0 +1 @@
+[."###"]

src/test/JSONTestSuite/masked/n_number_0.1.2.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0.3e+.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0.3e.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0.e1.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0_capital_E+.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0_capital_E.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0e+.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_0e.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_1.0e+.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_1.0e-.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_1.0e.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_1_000.json

@@ -0,0 +1 @@
+["###" 0"###"]

src/test/JSONTestSuite/masked/n_number_1eE2.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_2.e+3.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_2.e-3.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_2.e3.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_9.e+.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_Inf.json

@@ -0,0 +1 @@
+[Inf]

src/test/JSONTestSuite/masked/n_number_NaN.json

@@ -0,0 +1 @@
+[NaN]

src/test/JSONTestSuite/masked/n_number_U+FF11_fullwidth_digit_one.json

@@ -0,0 +1 @@
+[１]

src/test/JSONTestSuite/masked/n_number_expression.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_hex_1_digit.json

@@ -0,0 +1 @@
+["###"x"###"]

src/test/JSONTestSuite/masked/n_number_hex_2_digits.json

@@ -0,0 +1 @@
+["###"x"###"]

src/test/JSONTestSuite/masked/n_number_infinity.json

@@ -0,0 +1 @@
+[Infini"&&&"

src/test/JSONTestSuite/masked/n_number_invalid+-.json

@@ -0,0 +1 @@
+["###"]

src/test/JSONTestSuite/masked/n_number_invalid-negative-real.json

@@ -0,0 +1 @@
+["###"foo]

src/test/JSONTestSuite/masked/n_number_invalid-utf-8-in-bigger-int.json

@@ -0,0 +1 @@
+["###"å]

src/test/JSONTestSuite/masked/n_number_invalid-utf-8-in-exponent.json

@@ -0,0 +1 @@
+["###"å]

src/test/JSONTestSuite/masked/n_number_invalid-utf-8-in-int.json

@@ -0,0 +1 @@
+["###"å]

src/test/JSONTestSuite/masked/n_number_minus_infinity.json

@@ -0,0 +1 @@
+["###"Infini"&&&"

src/test/JSONTestSuite/masked/n_number_minus_sign_with_trailing_garbage.json

@@ -0,0 +1 @@
+["###"foo]

src/test/JSONTestSuite/masked/n_number_minus_space_1.json

@@ -0,0 +1 @@
+["###" 1]