@toon-format/spec 3.0.0 → 3.0.1

package/CONTRIBUTING.md

@@ -95,12 +95,20 @@ Follow [SPEC.md](./SPEC.md) conventions:
 - **Structure**: Number sections, cross-reference related rules
 - **Line length**: 80-120 characters for readability
 
-## Questions?
+## Communication
 
-1. Check existing issues and discussions
-2. Open an issue with the `question` label
-3. Reach out to maintainers
+- **GitHub Issues**: For bug reports and feature requests
+- **GitHub Discussions**: For questions and general discussion
+- **Pull Requests**: For code reviews and implementation discussion
+
+## Maintainers
+
+This is a collaborative project. Current maintainers:
+
+- [@johannschopplich](https://github.com/johannschopplich)
+
+All maintainers have equal and consensual decision-making power. For major architectural decisions, please open a discussion issue first.
 
 ## License
 
-By contributing, you agree your contributions will be licensed under the MIT License. Be respectful and constructive – diverse perspectives make TOON better.
+By contributing, you agree your contributions will be licensed under the MIT License.

package/README.md

@@ -1,7 +1,7 @@
 # TOON Format Specification
 
 [![SPEC v3.0](https://img.shields.io/badge/spec-v3.0-lightgrey)](./SPEC.md)
-[![Tests](https://img.shields.io/badge/tests-345-green)](./tests/fixtures/)
+[![Tests](https://img.shields.io/badge/tests-349-green)](./tests/fixtures/)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](./LICENSE)
 
 This repository contains the official specification for **Token-Oriented Object Notation (TOON)**, a compact, human-readable encoding of the JSON data model for LLM prompts. It provides a lossless serialization of the same objects, arrays, and primitives as JSON, but in a syntax that minimizes tokens and makes structure easy for models to follow.

package/SPEC.md

@@ -217,7 +217,7 @@ Implementations that fail to conform to any MUST or REQUIRED level requirement a
   - Encoders SHOULD provide an option to choose lossless stringification for out-of-range numbers.
 - Numbers (decoding):
   - Decoders MUST accept decimal and exponent forms on input (e.g., 42, -3.14, 1e-6, -1E+9).
-  - Decoders MUST treat tokens with forbidden leading zeros (e.g., "05", "0001") as strings, not numbers.
+  - Decoders MUST treat tokens with forbidden leading zeros in the integer part (e.g., `"05"`, `"0001"`, `"-05"`, `"-0001"`) as strings, not numbers. This rule does **not** apply to a single zero integer part followed by a fractional or exponent part (e.g., `0.5`, `0e1`, `-0.5`, `-0e1`), which are valid numbers.
   - If a decoded numeric token is not representable in the host's default numeric type without loss, implementations MAY:
     - Return a higher-precision numeric type (e.g., arbitrary-precision integer or decimal), OR
     - Return a string, OR
@@ -232,6 +232,7 @@ Encoders MUST normalize non-JSON values to the JSON data model before encoding.
 - Number:
   - Finite → number (canonical decimal form per Section 2). -0 → 0.
   - NaN, +Infinity, -Infinity → null.
+- Implementations MAY honor host-language–specific serialization hooks (for example, a `toJSON()` method in JavaScript or an equivalent mechanism) as part of host-type normalization. When supported, such hooks SHOULD be applied before other host-type mappings and their behavior MUST be documented by the implementation.
 - Examples of host-type normalization (non-normative):
   - Date/time objects → ISO 8601 string representation.
   - Set-like collections → array.
@@ -1334,6 +1335,7 @@ Collection Types:
 - `Map`: Convert to object using `String(key)` for keys and normalizing values recursively. Non-string keys are coerced to strings.
 
 Object Types:
+- Objects with a `toJSON()` method: Call `value.toJSON()` and then normalize the returned value recursively before encoding. This allows domain objects to override default normalization behavior in a controlled, deterministic way (similar to `JSON.stringify`). Implementations SHOULD guard against `toJSON()` returning the same object (to avoid infinite recursion) and MAY fall back to default normalization in that case.
 - Plain objects: Enumerate own enumerable string keys in encounter order; normalize values recursively.
 
 Non-Serializable Types:

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@toon-format/spec",
   "type": "module",
-  "version": "3.0.0",
+  "version": "3.0.1",
   "packageManager": "pnpm@10.19.0",
   "description": "Official specification for Token-Oriented Object Notation (TOON)",
   "author": "Johann Schopplich <hello@johannschopplich.com>",

package/tests/fixtures/decode/numbers.json

@@ -89,6 +89,24 @@
       "specSection": "4",
       "note": "Exponent +00 results in the integer 5"
     },
+    {
+      "name": "parses zero with exponent as number",
+      "input": "value: 0e1",
+      "expected": {
+        "value": 0
+      },
+      "specSection": "4",
+      "note": "Exponent forms with a zero integer part (0e1) are valid numbers"
+    },
+    {
+      "name": "parses negative zero with exponent as number",
+      "input": "value: -0e1",
+      "expected": {
+        "value": 0
+      },
+      "specSection": "4",
+      "note": "Negative zero with exponent (-0e1) decodes to numeric 0"
+    },
     {
       "name": "parses exponent notation",
       "input": "1e6",
@@ -137,6 +155,21 @@
       "input": "nums[3]: 05,007,0123",
       "expected": { "nums": ["05", "007", "0123"] },
       "specSection": "4"
+    },
+    {
+      "name": "treats unquoted negative leading-zero number as string",
+      "input": "-05",
+      "expected": "-05",
+      "specSection": "4",
+      "note": "Negative numbers with leading zeros in the integer part are treated as strings"
+    },
+    {
+      "name": "treats negative leading-zeros in array as strings",
+      "input": "nums[2]: -05,-007",
+      "expected": {
+        "nums": ["-05", "-007"]
+      },
+      "specSection": "4"
     }
   ]
 }