Add optIdentifiers option

Author: BitoLabs

CHANGELOG.md

@@ -3,25 +3,30 @@ Change log
 
 This project follows [Semantic Versioning](http://semver.org/).
 
-1.3.0 (IN PROGRESS)
+1.3.0 (2025-12-19)
 ==================
 
 ### New features
 
-* New optHex option (OFF by default). It allows integers in hex
-  format (like: 0x1A, 0X2b).
+* The `Json::parse()` and `parse_in_place()` methods now have
+  an additional `options` parameter, which can turn ON/OFF
+  additional syntax that is not present in JSON standard.
+* New `optHex` parsing option (OFF by default). It allows integers
+  in hex format (like: 0x1A, 0X2b).
+* New `optIdentifiers` parsing option (OFF by default). It allows
+  object member names as C identifiers without quotes: {foo: "bar"}.
 * Added parsing options that are ON by default.
   Previous version supported these features, but these could not
   be turned off. Note that JSON standard doesn't allow these:
 
-  - optUniqueMembers: disallow duplicate member names within the
+  - `optUniqueMembers`: disallow duplicate member names within the
     object. Apparently standard JSON allows (but not recommends)
     duplicates, see https://datatracker.ietf.org/doc/html/rfc8259#section-4
-  - optTrailingComma: allow one trailing comma in arrays and objects,
+  - `optTrailingComma`: allow one trailing comma in arrays and objects,
     like [1,2,].
-  - optEmptyFraction: Allow floating point numbers with no digits after
+  - `optEmptyFraction`: Allow floating point numbers with no digits after
     decimal point (like 1. or 2.e3).
-  - optLineCommentC: allow // comments.
+  - `optLineCommentC`: allow // comments.
 
 ### Changes
 

README.md

@@ -42,8 +42,17 @@ Summary of features
   from the input buffer rather allocating each such token in the heap.
 * The input is in UTF-8 format. The string tokens can contain escape sequences with
   [UTF-16 code points].
-* `//` comments are supported. Note that these are not part of the JSON standard.
 * Number values can be fetched as integers (32 or 64 bit) or as 64-bit float values.
+* Optional extended syntax that is not part of JSON standard. These can
+  be turned ON/OFF:
+    - `//` comments.
+    - Check that member names are unique within the object and raise
+      an error if not. When this feature is OFF, duplicates are allowed
+      as per JSON standard.
+    - Allow trailing commas in arrays and objects: `{"a": [1, 2, ], "b": 42, }`.
+    - Allow no digits after the decimal point: `[0., -1., 2.e10]`.
+    - Allow integers in hex format like: `[0x1A, 0X2b]`.
+    - Allow object member names as C identifiers without quotes: `{foo: "bar"}`.
 * Exceptions are used to handle the errors. See [error handling].
 * Value validation is easy and doesn't require a JSON schema. See:
     - [Number range checking].

ujson.cpp

@@ -598,6 +598,26 @@ class Parser
         return v;
     }
 
+    const char* parse_member_name()
+    {
+        char* end = nullptr;
+        const char* name = parse_str();
+        if (!name && (m_options & optIdentifiers)) {
+            name = parse_identifier(&end);
+        }
+        if (!name) {
+            throw ErrSyntax("invalid object syntax: expected member name or '}'", m_line_count);
+        }
+        skip_white_space();
+        if (!skip_text(":")) {
+            throw ErrSyntax("invalid object syntax: expected ':' after member name", m_line_count);
+        }
+        if (end) {
+            *end = 0; // ensure identifier is null terminated
+        }
+        return name;
+    }
+
     ObjImpl* parse_val_obj(ArrImpl* parent)
     {
         ObjImpl* obj = nullptr;
@@ -606,14 +626,7 @@ class Parser
         skip_white_space();
         if (skip_text("}")) return obj;
         while (true) {
-            const char* name = parse_str();
-            if (nullptr == name) {
-                throw ErrSyntax("invalid object syntax: expected member name or '}'", m_line_count);
-            }
-            skip_white_space();
-            if (!skip_text(":")) {
-                throw ErrSyntax("invalid object syntax: expected ':' after member name", m_line_count);
-            }
+            const char* name = parse_member_name();
             skip_white_space();
             size_t idx = obj->get_len();
             ValImpl* v = parse_val(obj);
@@ -904,6 +917,22 @@ class Parser
         return code;
     }
 
+    const char* parse_identifier(char** end)
+    {
+        char* name = nullptr;
+        char* p = m_next;
+        uint8_t ch = *p;
+        if ('_' != ch && (ch < 'A' || ch > 'Z') && (ch < 'a' || ch > 'z')) return nullptr;
+        while (true) {
+            ch = *(++p);
+            if ('_' != ch && (ch < 'A' || ch > 'Z') && (ch < 'a' || ch > 'z') && (ch < '0' || ch > '9')) break;
+        }
+        name = m_next;
+        m_next = p;
+        *end = p;
+        return name;
+    }
+
     bool skip_text(const char* str)
     {
         size_t len = strlen(str);

ujson.h

@@ -30,21 +30,18 @@ class Obj;
 
 enum Options : uint32_t // Parse options
 {
-    optUniqueMembers    = 1 << 0, // Member names must be unique within the object.
-                                  // On duplicates throw ErrSyntax.
-                                  // If not set, duplicates are accepted, but only
-                                  // the first value can be accessed by name, while other
-                                  // values can be accessed only by index.
-
-    optTrailingComma    = 1 << 1, // One comma allowed after the last array element or object member.
-                                  // Example: { "a": [1, 2, ], "b": 42, }
-
-    optEmptyFraction    = 1 << 2, // Allow floating point numbers with no digits after decimal point.
-                                  // Example: [0., -1., 2.e10]
-
-    optLineCommentC     = 1 << 3, // C style single line comment: //
-
-    optHex              = 1 << 6, // Allow integers in hex format like: 0x1A, 0X2b
+    optUniqueMembers    = 1 << 0, // * Member names must be unique within the object.
+                                  //   On duplicates throw ErrSyntax.
+                                  //   If not set, duplicates are accepted, but only
+                                  //   the first value can be accessed by name, while other
+                                  //   values can be accessed only by index.
+    optTrailingComma    = 1 << 1, // * One comma allowed after the last array element or object member.
+                                  //   Example: { "a": [1, 2, ], "b": 42, }
+    optEmptyFraction    = 1 << 2, // * Allow floating point numbers with no digits after decimal point.
+                                  //   Example: [0., -1., 2.e10]
+    optLineCommentC     = 1 << 3, // * C style single line comment: //
+    optHex              = 1 << 6, // * Allow integers in hex format like: 0x1A, 0X2b
+    optIdentifiers      = 1 << 7, // * Allow object member names as C identifiers without quotes: {foo: "bar"}
 
     optStandard         = 0,      // Conforms to JSON standard, no extra features are allowed.
     optDefault          = optUniqueMembers | optTrailingComma | optEmptyFraction | optLineCommentC,