Refine string quote check to more closely match emitter in snakeyaml

Signed-off-by: Michael Edgar <michael@xlate.io>

src/main/java/io/xlate/yamljson/StringQuotingChecker.java

@@ -29,7 +29,8 @@ boolean needToQuoteValue(String value) {
     }
 
     boolean needToQuote(String value, boolean quoteNumeric) {
-        return isReservedKeyword(value) ||
+        return value.isEmpty() ||
+                isReservedKeyword(value) ||
                 hasQuoteableCharacter(value) ||
                 (quoteNumeric && YamlNumbers.isNumeric(value));
     }
@@ -39,7 +40,6 @@ boolean needToQuote(String value, boolean quoteNumeric) {
      * <ul>
      * <li>YAML 1.2 keyword representing boolean</li>
      * <li>YAML 1.2 keyword representing null value</li>
-     * <li>empty String (length 0)</li></li> and returns {@code true} if so.
      *
      * @param value
      *            String to check
@@ -48,10 +48,6 @@ boolean needToQuote(String value, boolean quoteNumeric) {
      *         (as per YAML 1.2 specification) or empty String
      */
     boolean isReservedKeyword(String value) {
-        if (value.length() == 0) {
-            return true;
-        }
-
         switch (value.charAt(0)) {
         // First, reserved name starting chars:
         case 'f': // false
@@ -70,23 +66,23 @@ boolean isReservedKeyword(String value) {
     }
 
     /**
-     * As per YAML <a href="https://yaml.org/spec/1.2/spec.html#id2788859">Plain
+     * As per YAML <a href="https://yaml.org/spec/1.2.2/#733-plain-style">Plain
      * Style</a>unquoted strings are restricted to a reduced charset and must be
      * quoted in case they contain one of the following characters or character
      * combinations.
      */
     boolean hasQuoteableCharacter(String inputStr) {
+        if (quotableLeadingCharacter(inputStr)) {
+            return true;
+        }
+
         final int end = inputStr.length();
-        for (int i = 0; i < end; ++i) {
-            switch (inputStr.charAt(i)) {
-            case '[':
-            case ']':
-            case '{':
-            case '}':
-            case ',':
-                return true;
+
+        for (int i = 1; i < end; ++i) {
+            int current = inputStr.charAt(i);
+            switch (current) {
             case '#':
-                if (precededByBlank(inputStr, i)) {
+                if (isBlank(inputStr.charAt(i - 1))) {
                     return true;
                 }
                 break;
@@ -96,17 +92,57 @@ boolean hasQuoteableCharacter(String inputStr) {
                 }
                 break;
             default:
+                if (current < 0x20) {
+                    // Control character
+                    return true;
+                }
                 break;
             }
         }
         return false;
     }
 
-    boolean precededByBlank(String inputStr, int offset) {
-        if (offset == 0) {
+    boolean quotableLeadingCharacter(String inputStr) {
+        final int first = inputStr.charAt(0);
+
+        switch (first) {
+        case ' ':
+            // Leading space
+            return true;
+        case '#':
+        case ',':
+        case '[':
+        case ']':
+        case '{':
+        case '}':
+        case '&':
+        case '*':
+        case '!':
+        case '|':
+        case '>':
+        case '"':
+        case '%':
+        case '@':
+        case '`':
+            // Leading indicators
             return true;
+        case '?':
+        case ':':
+        case '-':
+            // Leading indicators not followed by non-space "safe" character
+            if (followedByBlank(inputStr, 0)) {
+                return true;
+            }
+            break;
+        default:
+            if (first < 0x20) {
+                // Control character
+                return true;
+            }
+            break;
         }
-        return isBlank(inputStr.charAt(offset - 1));
+
+        return false;
     }
 
     boolean followedByBlank(String inputStr, int offset) {

src/test/java/io/xlate/yamljson/YamlGeneratorTest.java

@@ -245,7 +245,7 @@ void testMappingOfValuesWithQuotesMinimized(String version) {
                 + "  DoubleQuote: Contains only \"\n"
                 + "  \"100\": Numeric key\n"
                 + "  empty: \"\"\n"
-                + "  blank: ' '\n"
+                + "  blank: \" \"\n"
                 + "  positiveInfinity: .inf\n"
                 + "  negativeInfinity: -.inf\n"
                 + "  NaN: .nan\n",
@@ -390,7 +390,7 @@ void testSpecialStringsQuoted(String version) {
 
         try (JsonGenerator generator = createGenerator(config, writer)) {
             generator.writeStartObject()
-                .write("#keywithhash", "value with: colon")
+                .write("#keywithhash", "value with:\tcolon")
                 .write("#anotherwithhash", "value with:colon but the :is not followed by a space")
                 .write("key with spaces", "ends with colon:")
                 .write("key\twith\ttabs", "ends with hash (preceded by space) #")
@@ -399,14 +399,16 @@ void testSpecialStringsQuoted(String version) {
                 .write(".inf", "Key is infinite")
                 .write(".NAN", "Key is not a number!")
                 .write("false", "Key is reserved word")
-                .write("array[]", "Key has indicators")
+                .write("array[]", "Key has indicators, but not inside of a flow collection")
+                .write("? question", "Key has leading indicator followed by space")
+                .write("\ttab first", "Key with leading tab (special character) is quoted")
                 .writeEnd();
 
             writer.flush();
         }
 
         assertEquals(""
-                + "\"#keywithhash\": \"value with: colon\"\n"
+                + "\"#keywithhash\": \"value with:\\tcolon\"\n"
                 + "\"#anotherwithhash\": value with:colon but the :is not followed by a space\n"
                 + "key with spaces: \"ends with colon:\"\n"
                 // snakeyaml* adds quotes due to `\t`
@@ -416,7 +418,9 @@ void testSpecialStringsQuoted(String version) {
                 + "\".inf\": Key is infinite\n"
                 + "\".NAN\": Key is not a number!\n"
                 + "\"false\": Key is reserved word\n"
-                + "\"array[]\": Key has indicators\n",
+                + "array[]: Key has indicators, but not inside of a flow collection\n"
+                + "\"? question\": Key has leading indicator followed by space\n"
+                + "\"\\ttab first\": Key with leading tab (special character) is quoted\n",
                 writer.toString());
     }