| 1 | //! Schema sanitizer for tool `input_schema` before sending to provider APIs. |
| 2 | //! |
| 3 | //! DeepSeek's `/beta/chat/completions` strict tool mode is harsh. MCP tool |
| 4 | //! schemas frequently arrive with Pydantic-style `anyOf:[{type:"string"}, |
| 5 | //! {type:"null"}]` unions, bare `{type:"object"}` with no `properties`, or |
| 6 | //! `required` entries that don't appear in `properties`. These dirty schemas |
| 7 | //! cause silent 400s that users can't diagnose. |
| 8 | //! |
| 9 | //! The default sanitizer runs in-place on every schema returned by |
| 10 | //! `ToolRegistry::tools_for_api()` before the registry hands them off. |
| 11 | //! Provider-specific helpers below add stricter DeepSeek and OpenAI Responses |
| 12 | //! compatibility passes where their request shapes need it. |
| 13 | |
| 14 | use std::collections::HashSet; |
| 15 | |
| 16 | use serde_json::{Map, Value}; |
| 17 | |
| 18 | use crate::models::Tool; |
| 19 | |
| 20 | /// Sanitize a JSON Schema in-place for DeepSeek strict-tool compatibility. |
| 21 | /// |
| 22 | /// Applies a sequence of normalisations chosen to be semantics-preserving: |
| 23 | /// - Collapse `{"anyOf":[X, {"type":"null"}]}` → `X ∪ {"nullable": true}` |
| 24 | /// - Inject `"properties": {}` on bare-object schemas |
| 25 | /// - Prune dangling `required` entries |
| 26 | /// - Collapse single-element `oneOf` / `allOf` |
| 27 | /// - Walk recursively through all subschemas |
| 28 | pub fn sanitize(schema: &mut Value) { |
| 29 | collapse_nullable_unions(schema); |
| 30 | inject_properties_on_bare_objects(schema); |
| 31 | prune_dangling_required(schema); |
| 32 | collapse_single_element_unions(schema); |
| 33 | // Recurse into all sub-schemas |
| 34 | if let Some(obj) = schema.as_object_mut() { |
| 35 | for (_, v) in obj.iter_mut() { |
| 36 | sanitize(v); |
| 37 | } |
| 38 | } else if let Some(arr) = schema.as_array_mut() { |
| 39 | for v in arr.iter_mut() { |
| 40 | sanitize(v); |
| 41 | } |
| 42 | } |
| 43 | } |
| 44 | |
| 45 | /// Prepare a complete active tool set for DeepSeek strict function-calling. |
| 46 | /// |
| 47 | /// Each tool is evaluated independently: compatible schemas are sanitized and |
| 48 | /// marked strict, while incompatible schemas remain unchanged and non-strict. |
| 49 | /// Returns `true` only when every tool in the set can use strict mode. |
| 50 | pub fn prepare_tools_for_strict_mode(tools: &mut [Tool]) -> bool { |
| 51 | let mut all_strict = true; |
| 52 | for tool in tools { |
| 53 | if strict_schema_supported(&tool.input_schema) { |
| 54 | sanitize_for_strict(&mut tool.input_schema); |
| 55 | tool.strict = Some(true); |
| 56 | } else { |
| 57 | tool.strict = None; |
| 58 | all_strict = false; |
| 59 | } |
| 60 | } |
| 61 | all_strict |
| 62 | } |
| 63 | |
| 64 | /// Sanitize a schema for DeepSeek strict function-calling. |
| 65 | /// |
| 66 | /// This extends the general sanitizer with the official strict-mode object |
| 67 | /// rules: every object must set `additionalProperties: false`, and every |
| 68 | /// property must be listed in `required`. |
| 69 | pub fn sanitize_for_strict(schema: &mut Value) { |
| 70 | sanitize(schema); |
| 71 | enforce_strict_subset(schema); |
| 72 | } |
| 73 | |
| 74 | /// Sanitize a tool `parameters` schema for xAI chat completions. |
| 75 | /// |
| 76 | /// xAI validates that the parameters root is an object schema and rejects a |
| 77 | /// root-level `anyOf`/`oneOf` union with any non-object branch |
| 78 | /// ("tool parameter root must be an object type"). The built-in `apply_patch` |
| 79 | /// schema's `oneOf: [{required:["patch"]}, {required:["changes"]}]` trips this |
| 80 | /// with a 400 on the first tool-bearing request. The Responses-API pass |
| 81 | /// performs exactly the required normalization — merge root composition |
| 82 | /// properties, force `type: object`, drop root-only composition keywords — |
| 83 | /// so reuse it and surface the dropped constraint as a description note. |
| 84 | pub fn sanitize_for_xai_parameters(parameters: &mut Value) -> Option<String> { |
| 85 | sanitize_for_responses(parameters) |
| 86 | } |
| 87 | |
| 88 | /// Sanitize a schema for OpenAI Responses function tools. |
| 89 | /// |
| 90 | /// The Responses API requires the top-level `parameters` schema to be an object |
| 91 | /// and rejects top-level `oneOf` / `anyOf` / `allOf` / `enum` / `not`. Keep the |
| 92 | /// schema permissive rather than changing tool semantics: merge any root |
| 93 | /// alternative properties we can see, then remove the root-only composition |
| 94 | /// keywords while preserving nested schemas. |
| 95 | /// |
| 96 | /// Returns a short description note when root composition constraints with |
| 97 | /// meaningful `required` groups are dropped. |
| 98 | pub fn sanitize_for_responses(schema: &mut Value) -> Option<String> { |
| 99 | let constraint_note = schema |
| 100 | .as_object() |
| 101 | .and_then(root_composition_constraint_note); |
| 102 | |
| 103 | sanitize(schema); |
| 104 | |
| 105 | if !schema.is_object() { |
| 106 | *schema = Value::Object(Map::new()); |
| 107 | } |
| 108 | |
| 109 | let Some(obj) = schema.as_object_mut() else { |
| 110 | return constraint_note; |
| 111 | }; |
| 112 | |
| 113 | merge_root_composition_properties(obj); |
| 114 | obj.insert("type".into(), Value::String("object".to_string())); |
| 115 | obj.remove("oneOf"); |
| 116 | obj.remove("anyOf"); |
| 117 | obj.remove("allOf"); |
| 118 | obj.remove("enum"); |
| 119 | obj.remove("not"); |
| 120 | ensure_properties_object(obj); |
| 121 | prune_dangling_required(schema); |
| 122 | constraint_note |
| 123 | } |
| 124 | |
| 125 | fn strict_schema_supported(schema: &Value) -> bool { |
| 126 | let mut normalized = schema.clone(); |
| 127 | sanitize(&mut normalized); |
| 128 | !has_strict_incompatible_composition(&normalized, true) |
| 129 | } |
| 130 | |
| 131 | fn has_strict_incompatible_composition(schema: &Value, is_root: bool) -> bool { |
| 132 | if let Some(obj) = schema.as_object() { |
| 133 | if obj.contains_key("oneOf") || obj.contains_key("allOf") { |
| 134 | return true; |
| 135 | } |
| 136 | if is_root && obj.contains_key("anyOf") { |
| 137 | return true; |
| 138 | } |
| 139 | return obj |
| 140 | .values() |
| 141 | .any(|value| has_strict_incompatible_composition(value, false)); |
| 142 | } |
| 143 | schema.as_array().is_some_and(|arr| { |
| 144 | arr.iter() |
| 145 | .any(|value| has_strict_incompatible_composition(value, false)) |
| 146 | }) |
| 147 | } |
| 148 | |
| 149 | /// Collapse `{"anyOf":[X, {"type":"null"}]}` → `X ∪ {"nullable": true}`. |
| 150 | /// |
| 151 | /// Same treatment for `oneOf`. Only collapses when exactly one non-null |
| 152 | /// member and exactly one null-type member are present. |
| 153 | fn collapse_nullable_unions(schema: &mut Value) { |
| 154 | let Some(obj) = schema.as_object_mut() else { |
| 155 | return; |
| 156 | }; |
| 157 | for key in ["anyOf", "oneOf"] { |
| 158 | let members: Vec<Value> = match obj.get(key).and_then(|v| v.as_array()) { |
| 159 | Some(arr) => arr.clone(), |
| 160 | None => continue, |
| 161 | }; |
| 162 | let (nulls, nons): (Vec<_>, Vec<_>) = members.into_iter().partition(is_null_type); |
| 163 | if nulls.len() == 1 && nons.len() == 1 { |
| 164 | // `nullable` is only meaningful when it annotates a concrete |
| 165 | // schema type. Collapsing `$ref | null` or another untyped branch |
| 166 | // would manufacture an annotation-only schema and lose the |
| 167 | // terminating union needed by MFJS reference validation. |
| 168 | let has_concrete_non_null_type = nons[0] |
| 169 | .as_object() |
| 170 | .and_then(|branch| branch.get("type")) |
| 171 | .and_then(Value::as_str) |
| 172 | .is_some_and(|schema_type| schema_type != "null"); |
| 173 | if !has_concrete_non_null_type { |
| 174 | continue; |
| 175 | } |
| 176 | obj.remove(key); |
| 177 | if let Value::Object(non_obj) = nons.into_iter().next().unwrap() { |
| 178 | for (k, v) in non_obj { |
| 179 | if k != "type" || v != "null" { |
| 180 | obj.insert(k, v); |
| 181 | } |
| 182 | } |
| 183 | } |
| 184 | obj.insert("nullable".into(), Value::Bool(true)); |
| 185 | } |
| 186 | } |
| 187 | } |
| 188 | |
| 189 | fn is_null_type(v: &Value) -> bool { |
| 190 | v.as_object() |
| 191 | .and_then(|o| o.get("type")) |
| 192 | .and_then(|t| t.as_str()) |
| 193 | == Some("null") |
| 194 | } |
| 195 | |
| 196 | /// Bare `{"type": "object"}` (no `properties`, no `additionalProperties`) |
| 197 | /// → inject `"properties": {}` so DeepSeek's strict validator doesn't 400. |
| 198 | fn inject_properties_on_bare_objects(schema: &mut Value) { |
| 199 | let Some(obj) = schema.as_object_mut() else { |
| 200 | return; |
| 201 | }; |
| 202 | if obj.get("type").and_then(|t| t.as_str()) != Some("object") { |
| 203 | return; |
| 204 | } |
| 205 | if obj.contains_key("properties") || obj.contains_key("additionalProperties") { |
| 206 | return; |
| 207 | } |
| 208 | obj.insert("properties".into(), Value::Object(Map::new())); |
| 209 | } |
| 210 | |
| 211 | /// Remove entries from `required` that aren't keys in `properties`. |
| 212 | fn prune_dangling_required(schema: &mut Value) { |
| 213 | let Some(obj) = schema.as_object_mut() else { |
| 214 | return; |
| 215 | }; |
| 216 | // Collect known property names first (immutable borrow), then prune. |
| 217 | let known_keys: Vec<String> = obj |
| 218 | .get("properties") |
| 219 | .and_then(|v| v.as_object()) |
| 220 | .map(|props| props.keys().cloned().collect()) |
| 221 | .unwrap_or_default(); |
| 222 | let Some(required) = obj.get_mut("required").and_then(|v| v.as_array_mut()) else { |
| 223 | return; |
| 224 | }; |
| 225 | required.retain(|entry| { |
| 226 | entry |
| 227 | .as_str() |
| 228 | .is_some_and(|k| known_keys.iter().any(|known| known == k)) |
| 229 | }); |
| 230 | if required.is_empty() { |
| 231 | obj.remove("required"); |
| 232 | } |
| 233 | } |
| 234 | |
| 235 | /// Collapse `{"oneOf": [X]}` → X, same for `allOf`. |
| 236 | /// |
| 237 | /// Single-element unions are semantically equivalent to the element itself; |
| 238 | /// DeepSeek's strict validator doesn't always flatten them. |
| 239 | fn collapse_single_element_unions(schema: &mut Value) { |
| 240 | let Some(obj) = schema.as_object_mut() else { |
| 241 | return; |
| 242 | }; |
| 243 | for key in ["oneOf", "allOf", "anyOf"] { |
| 244 | let single = match obj.get(key).and_then(|v| v.as_array()) { |
| 245 | Some(arr) if arr.len() == 1 => arr[0].clone(), |
| 246 | _ => continue, |
| 247 | }; |
| 248 | obj.remove(key); |
| 249 | if let Value::Object(inner) = single { |
| 250 | for (k, v) in inner { |
| 251 | if !obj.contains_key(&k) { |
| 252 | obj.insert(k, v); |
| 253 | } |
| 254 | } |
| 255 | } |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | fn enforce_strict_subset(schema: &mut Value) { |
| 260 | if let Some(obj) = schema.as_object_mut() { |
| 261 | strip_unsupported_strict_keywords(obj); |
| 262 | if is_object_schema(obj) { |
| 263 | let originally_required = required_names(obj); |
| 264 | let properties = ensure_properties_object(obj); |
| 265 | let mut property_names: Vec<String> = properties.keys().cloned().collect(); |
| 266 | property_names.sort(); |
| 267 | for property_name in &property_names { |
| 268 | if !originally_required |
| 269 | .iter() |
| 270 | .any(|required| required == property_name) |
| 271 | && let Some(property_schema) = properties.get_mut(property_name) |
| 272 | { |
| 273 | mark_nullable(property_schema); |
| 274 | } |
| 275 | } |
| 276 | obj.insert( |
| 277 | "required".into(), |
| 278 | Value::Array(property_names.into_iter().map(Value::String).collect()), |
| 279 | ); |
| 280 | obj.insert("additionalProperties".into(), Value::Bool(false)); |
| 281 | } |
| 282 | |
| 283 | for value in obj.values_mut() { |
| 284 | enforce_strict_subset(value); |
| 285 | } |
| 286 | } else if let Some(arr) = schema.as_array_mut() { |
| 287 | for value in arr { |
| 288 | enforce_strict_subset(value); |
| 289 | } |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | fn strip_unsupported_strict_keywords(obj: &mut Map<String, Value>) { |
| 294 | obj.remove("patternProperties"); |
| 295 | match obj.get("type").and_then(Value::as_str) { |
| 296 | Some("string") => { |
| 297 | obj.remove("minLength"); |
| 298 | obj.remove("maxLength"); |
| 299 | } |
| 300 | Some("array") => { |
| 301 | obj.remove("minItems"); |
| 302 | obj.remove("maxItems"); |
| 303 | } |
| 304 | _ => {} |
| 305 | } |
| 306 | } |
| 307 | |
| 308 | fn is_object_schema(obj: &Map<String, Value>) -> bool { |
| 309 | obj.get("type").and_then(Value::as_str) == Some("object") || obj.contains_key("properties") |
| 310 | } |
| 311 | |
| 312 | fn ensure_properties_object(obj: &mut Map<String, Value>) -> &mut Map<String, Value> { |
| 313 | let needs_replacement = !matches!(obj.get("properties"), Some(Value::Object(_))); |
| 314 | if needs_replacement { |
| 315 | obj.insert("properties".into(), Value::Object(Map::new())); |
| 316 | } |
| 317 | obj.get_mut("properties") |
| 318 | .and_then(Value::as_object_mut) |
| 319 | .expect("properties was just ensured as object") |
| 320 | } |
| 321 | |
| 322 | fn required_names(obj: &Map<String, Value>) -> Vec<String> { |
| 323 | obj.get("required") |
| 324 | .and_then(Value::as_array) |
| 325 | .map(|required| { |
| 326 | required |
| 327 | .iter() |
| 328 | .filter_map(Value::as_str) |
| 329 | .map(ToOwned::to_owned) |
| 330 | .collect() |
| 331 | }) |
| 332 | .unwrap_or_default() |
| 333 | } |
| 334 | |
| 335 | fn mark_nullable(schema: &mut Value) { |
| 336 | if let Some(obj) = schema.as_object_mut() { |
| 337 | obj.insert("nullable".into(), Value::Bool(true)); |
| 338 | } |
| 339 | } |
| 340 | |
| 341 | fn merge_root_composition_properties(obj: &mut Map<String, Value>) { |
| 342 | let mut merged = Map::new(); |
| 343 | for key in ["oneOf", "anyOf", "allOf"] { |
| 344 | let Some(items) = obj.get(key).and_then(Value::as_array) else { |
| 345 | continue; |
| 346 | }; |
| 347 | for item in items { |
| 348 | let Some(properties) = item.get("properties").and_then(Value::as_object) else { |
| 349 | continue; |
| 350 | }; |
| 351 | for (name, schema) in properties { |
| 352 | merged.entry(name.clone()).or_insert_with(|| schema.clone()); |
| 353 | } |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | if merged.is_empty() { |
| 358 | return; |
| 359 | } |
| 360 | |
| 361 | let properties = ensure_properties_object(obj); |
| 362 | for (name, schema) in merged { |
| 363 | properties.entry(name).or_insert(schema); |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | fn root_composition_constraint_note(obj: &Map<String, Value>) -> Option<String> { |
| 368 | for (key, prefix) in [ |
| 369 | ("oneOf", "Exactly one"), |
| 370 | ("anyOf", "At least one"), |
| 371 | ("allOf", "All"), |
| 372 | ] { |
| 373 | let Some(items) = obj.get(key).and_then(Value::as_array) else { |
| 374 | continue; |
| 375 | }; |
| 376 | let mut groups: Vec<String> = items.iter().filter_map(required_group_label).collect(); |
| 377 | groups.sort(); |
| 378 | groups.dedup(); |
| 379 | if groups.len() >= 2 { |
| 380 | return Some(format!( |
| 381 | "{prefix} of these parameter groups must be provided: {}.", |
| 382 | groups.join(" | ") |
| 383 | )); |
| 384 | } |
| 385 | } |
| 386 | None |
| 387 | } |
| 388 | |
| 389 | fn required_group_label(item: &Value) -> Option<String> { |
| 390 | let mut names: Vec<String> = item |
| 391 | .get("required")? |
| 392 | .as_array()? |
| 393 | .iter() |
| 394 | .filter_map(Value::as_str) |
| 395 | .map(|name| format!("`{name}`")) |
| 396 | .collect(); |
| 397 | if names.is_empty() { |
| 398 | None |
| 399 | } else { |
| 400 | names.sort(); |
| 401 | names.dedup(); |
| 402 | Some(names.join(" + ")) |
| 403 | } |
| 404 | } |
| 405 | |
| 406 | #[cfg(test)] |
| 407 | mod tests { |
| 408 | use super::*; |
| 409 | use serde_json::json; |
| 410 | |
| 411 | fn test_tool(name: &str, input_schema: Value) -> Tool { |
| 412 | Tool { |
| 413 | tool_type: None, |
| 414 | name: name.to_string(), |
| 415 | description: name.to_string(), |
| 416 | input_schema, |
| 417 | allowed_callers: None, |
| 418 | defer_loading: None, |
| 419 | input_examples: None, |
| 420 | strict: None, |
| 421 | cache_control: None, |
| 422 | } |
| 423 | } |
| 424 | |
| 425 | #[test] |
| 426 | fn collapses_nullable_anyof() { |
| 427 | let mut schema = json!({ |
| 428 | "anyOf": [ |
| 429 | {"type": "string"}, |
| 430 | {"type": "null"} |
| 431 | ] |
| 432 | }); |
| 433 | sanitize(&mut schema); |
| 434 | assert_eq!(schema["type"], "string"); |
| 435 | assert_eq!(schema["nullable"], true); |
| 436 | assert!(schema.get("anyOf").is_none()); |
| 437 | } |
| 438 | |
| 439 | #[test] |
| 440 | fn collapses_nullable_oneof() { |
| 441 | let mut schema = json!({ |
| 442 | "oneOf": [ |
| 443 | {"type": "null"}, |
| 444 | {"type": "integer", "minimum": 0} |
| 445 | ] |
| 446 | }); |
| 447 | sanitize(&mut schema); |
| 448 | assert_eq!(schema["type"], "integer"); |
| 449 | assert_eq!(schema["minimum"], 0); |
| 450 | assert_eq!(schema["nullable"], true); |
| 451 | } |
| 452 | |
| 453 | #[test] |
| 454 | fn preserves_non_null_anyof() { |
| 455 | let original = json!({ |
| 456 | "anyOf": [ |
| 457 | {"type": "string"}, |
| 458 | {"type": "integer"} |
| 459 | ] |
| 460 | }); |
| 461 | let mut schema = original.clone(); |
| 462 | sanitize(&mut schema); |
| 463 | // Multi-typed anyOf should collapse to single element after |
| 464 | // recursive walk — but here neither is null so the collapse |
| 465 | // doesn't trigger. The anyOf array itself remains. |
| 466 | assert!(schema.get("anyOf").is_some()); |
| 467 | } |
| 468 | |
| 469 | #[test] |
| 470 | fn injects_properties_on_bare_object() { |
| 471 | let mut schema = json!({"type": "object"}); |
| 472 | sanitize(&mut schema); |
| 473 | assert!(schema.get("properties").is_some()); |
| 474 | assert_eq!(schema["properties"], json!({})); |
| 475 | } |
| 476 | |
| 477 | #[test] |
| 478 | fn does_not_inject_properties_when_present() { |
| 479 | let mut schema = json!({ |
| 480 | "type": "object", |
| 481 | "properties": {"name": {"type": "string"}} |
| 482 | }); |
| 483 | let expected = schema.clone(); |
| 484 | sanitize(&mut schema); |
| 485 | assert_eq!(schema, expected); |
| 486 | } |
| 487 | |
| 488 | #[test] |
| 489 | fn prunes_dangling_required() { |
| 490 | let mut schema = json!({ |
| 491 | "type": "object", |
| 492 | "properties": {"name": {"type": "string"}}, |
| 493 | "required": ["name", "email"] |
| 494 | }); |
| 495 | sanitize(&mut schema); |
| 496 | let required = schema["required"].as_array().unwrap(); |
| 497 | assert_eq!(required.len(), 1); |
| 498 | assert_eq!(required[0], "name"); |
| 499 | } |
| 500 | |
| 501 | #[test] |
| 502 | fn removes_required_when_all_pruned() { |
| 503 | let mut schema = json!({ |
| 504 | "type": "object", |
| 505 | "properties": {}, |
| 506 | "required": ["ghost"] |
| 507 | }); |
| 508 | sanitize(&mut schema); |
| 509 | assert!(schema.get("required").is_none()); |
| 510 | } |
| 511 | |
| 512 | #[test] |
| 513 | fn collapses_single_element_oneof() { |
| 514 | let mut schema = json!({ |
| 515 | "oneOf": [{"type": "string", "minLength": 1}] |
| 516 | }); |
| 517 | sanitize(&mut schema); |
| 518 | assert!(schema.get("oneOf").is_none()); |
| 519 | assert_eq!(schema["type"], "string"); |
| 520 | assert_eq!(schema["minLength"], 1); |
| 521 | } |
| 522 | |
| 523 | #[test] |
| 524 | fn collapses_single_element_anyof() { |
| 525 | let mut schema = json!({ |
| 526 | "anyOf": [{"type": "boolean"}] |
| 527 | }); |
| 528 | sanitize(&mut schema); |
| 529 | assert!(schema.get("anyOf").is_none()); |
| 530 | assert_eq!(schema["type"], "boolean"); |
| 531 | } |
| 532 | |
| 533 | #[test] |
| 534 | fn recursive_walk_into_properties() { |
| 535 | let mut schema = json!({ |
| 536 | "type": "object", |
| 537 | "properties": { |
| 538 | "opt_name": { |
| 539 | "anyOf": [ |
| 540 | {"type": "string"}, |
| 541 | {"type": "null"} |
| 542 | ] |
| 543 | } |
| 544 | } |
| 545 | }); |
| 546 | sanitize(&mut schema); |
| 547 | let prop = &schema["properties"]["opt_name"]; |
| 548 | assert_eq!(prop["type"], "string"); |
| 549 | assert_eq!(prop["nullable"], true); |
| 550 | } |
| 551 | |
| 552 | #[test] |
| 553 | fn recursive_walk_into_items() { |
| 554 | let mut schema = json!({ |
| 555 | "type": "array", |
| 556 | "items": { |
| 557 | "anyOf": [ |
| 558 | {"type": "integer"}, |
| 559 | {"type": "null"} |
| 560 | ] |
| 561 | } |
| 562 | }); |
| 563 | sanitize(&mut schema); |
| 564 | let items = &schema["items"]; |
| 565 | assert_eq!(items["type"], "integer"); |
| 566 | assert_eq!(items["nullable"], true); |
| 567 | } |
| 568 | |
| 569 | #[test] |
| 570 | fn nested_anyof_in_nullable_union_stays_structural() { |
| 571 | // Pydantic can nest unions: Optional[Union[str, int]]. |
| 572 | let mut schema = json!({ |
| 573 | "anyOf": [ |
| 574 | { |
| 575 | "anyOf": [ |
| 576 | {"type": "string"}, |
| 577 | {"type": "integer"} |
| 578 | ] |
| 579 | }, |
| 580 | {"type": "null"} |
| 581 | ] |
| 582 | }); |
| 583 | sanitize(&mut schema); |
| 584 | // The non-null branch has no concrete type of its own. Collapsing the |
| 585 | // outer union would manufacture an annotation-only `nullable` schema, |
| 586 | // so retain both levels and their explicit null termination. |
| 587 | assert!(schema.get("nullable").is_none()); |
| 588 | assert_eq!(schema["anyOf"][1], json!({"type": "null"})); |
| 589 | assert_eq!( |
| 590 | schema["anyOf"][0]["anyOf"], |
| 591 | json!([{"type": "string"}, {"type": "integer"}]) |
| 592 | ); |
| 593 | } |
| 594 | |
| 595 | #[test] |
| 596 | fn idempotent() { |
| 597 | let mut schema = json!({ |
| 598 | "type": "object", |
| 599 | "properties": { |
| 600 | "name": {"type": "string"}, |
| 601 | "maybe": { |
| 602 | "anyOf": [{"type": "integer"}, {"type": "null"}] |
| 603 | } |
| 604 | }, |
| 605 | "required": ["name", "missing_field"] |
| 606 | }); |
| 607 | sanitize(&mut schema); |
| 608 | let after_first = schema.clone(); |
| 609 | sanitize(&mut schema); |
| 610 | assert_eq!(schema, after_first, "sanitize must be idempotent"); |
| 611 | } |
| 612 | |
| 613 | #[test] |
| 614 | fn strict_sanitize_requires_all_object_properties_and_closes_extra_keys() { |
| 615 | let mut schema = json!({ |
| 616 | "type": "object", |
| 617 | "properties": { |
| 618 | "name": {"type": "string"}, |
| 619 | "count": {"type": "integer"} |
| 620 | }, |
| 621 | "required": ["name"], |
| 622 | "additionalProperties": {"type": "string"} |
| 623 | }); |
| 624 | |
| 625 | sanitize_for_strict(&mut schema); |
| 626 | |
| 627 | assert_eq!(schema["additionalProperties"], false); |
| 628 | assert_eq!(schema["required"], json!(["count", "name"])); |
| 629 | assert_eq!(schema["properties"]["count"]["nullable"], true); |
| 630 | assert!(schema["properties"]["name"].get("nullable").is_none()); |
| 631 | } |
| 632 | |
| 633 | #[test] |
| 634 | fn strict_sanitize_preserves_optional_properties_as_nullable() { |
| 635 | let mut schema = json!({ |
| 636 | "type": "object", |
| 637 | "properties": { |
| 638 | "path": {"type": "string"}, |
| 639 | "start_line": {"type": "integer"}, |
| 640 | "max_lines": {"type": "integer"}, |
| 641 | "options": { |
| 642 | "type": "object", |
| 643 | "properties": { |
| 644 | "encoding": {"type": "string"}, |
| 645 | "trim": {"type": "boolean"} |
| 646 | }, |
| 647 | "required": ["encoding"] |
| 648 | } |
| 649 | }, |
| 650 | "required": ["path", "options"] |
| 651 | }); |
| 652 | |
| 653 | sanitize_for_strict(&mut schema); |
| 654 | |
| 655 | assert_eq!( |
| 656 | schema["required"], |
| 657 | json!(["max_lines", "options", "path", "start_line"]) |
| 658 | ); |
| 659 | assert!(schema["properties"]["path"].get("nullable").is_none()); |
| 660 | assert!(schema["properties"]["options"].get("nullable").is_none()); |
| 661 | assert_eq!(schema["properties"]["start_line"]["nullable"], true); |
| 662 | assert_eq!(schema["properties"]["max_lines"]["nullable"], true); |
| 663 | assert_eq!( |
| 664 | schema["properties"]["options"]["required"], |
| 665 | json!(["encoding", "trim"]) |
| 666 | ); |
| 667 | assert!( |
| 668 | schema["properties"]["options"]["properties"]["encoding"] |
| 669 | .get("nullable") |
| 670 | .is_none() |
| 671 | ); |
| 672 | assert_eq!( |
| 673 | schema["properties"]["options"]["properties"]["trim"]["nullable"], |
| 674 | true |
| 675 | ); |
| 676 | } |
| 677 | |
| 678 | #[test] |
| 679 | fn strict_sanitize_applies_object_rules_recursively() { |
| 680 | let mut schema = json!({ |
| 681 | "type": "object", |
| 682 | "properties": { |
| 683 | "outer": { |
| 684 | "type": "object", |
| 685 | "properties": { |
| 686 | "inner": {"type": "string"} |
| 687 | }, |
| 688 | "required": [] |
| 689 | } |
| 690 | }, |
| 691 | "required": [] |
| 692 | }); |
| 693 | |
| 694 | sanitize_for_strict(&mut schema); |
| 695 | |
| 696 | assert_eq!(schema["required"], json!(["outer"])); |
| 697 | assert_eq!(schema["additionalProperties"], false); |
| 698 | assert_eq!(schema["properties"]["outer"]["required"], json!(["inner"])); |
| 699 | assert_eq!(schema["properties"]["outer"]["additionalProperties"], false); |
| 700 | } |
| 701 | |
| 702 | #[test] |
| 703 | fn strict_sanitize_removes_unsupported_string_and_array_bounds() { |
| 704 | let mut schema = json!({ |
| 705 | "type": "object", |
| 706 | "properties": { |
| 707 | "name": { |
| 708 | "type": "string", |
| 709 | "minLength": 1, |
| 710 | "maxLength": 64, |
| 711 | "pattern": "^[a-z]+$" |
| 712 | }, |
| 713 | "items": { |
| 714 | "type": "array", |
| 715 | "minItems": 1, |
| 716 | "maxItems": 5, |
| 717 | "items": {"type": "string"} |
| 718 | }, |
| 719 | "score": { |
| 720 | "type": "integer", |
| 721 | "minimum": 1, |
| 722 | "maximum": 5 |
| 723 | } |
| 724 | } |
| 725 | }); |
| 726 | |
| 727 | sanitize_for_strict(&mut schema); |
| 728 | |
| 729 | let name = &schema["properties"]["name"]; |
| 730 | assert!(name.get("minLength").is_none()); |
| 731 | assert!(name.get("maxLength").is_none()); |
| 732 | assert_eq!(name["pattern"], "^[a-z]+$"); |
| 733 | |
| 734 | let items = &schema["properties"]["items"]; |
| 735 | assert!(items.get("minItems").is_none()); |
| 736 | assert!(items.get("maxItems").is_none()); |
| 737 | |
| 738 | let score = &schema["properties"]["score"]; |
| 739 | assert_eq!(score["minimum"], 1); |
| 740 | assert_eq!(score["maximum"], 5); |
| 741 | } |
| 742 | |
| 743 | #[test] |
| 744 | fn strict_mode_applies_per_tool_in_mixed_catalog() { |
| 745 | let mut tools = vec![ |
| 746 | test_tool( |
| 747 | "lookup", |
| 748 | json!({ |
| 749 | "type": "object", |
| 750 | "properties": { |
| 751 | "query": {"type": "string"} |
| 752 | }, |
| 753 | "required": [] |
| 754 | }), |
| 755 | ), |
| 756 | test_tool( |
| 757 | "either", |
| 758 | json!({ |
| 759 | "type": "object", |
| 760 | "properties": { |
| 761 | "a": {"type": "string"}, |
| 762 | "b": {"type": "string"} |
| 763 | }, |
| 764 | "anyOf": [ |
| 765 | {"required": ["a"]}, |
| 766 | {"required": ["b"]} |
| 767 | ] |
| 768 | }), |
| 769 | ), |
| 770 | test_tool( |
| 771 | "nested", |
| 772 | json!({ |
| 773 | "type": "object", |
| 774 | "properties": { |
| 775 | "value": { |
| 776 | "oneOf": [ |
| 777 | {"type": "string"}, |
| 778 | {"type": "integer"} |
| 779 | ] |
| 780 | } |
| 781 | } |
| 782 | }), |
| 783 | ), |
| 784 | ]; |
| 785 | |
| 786 | assert!(!prepare_tools_for_strict_mode(&mut tools)); |
| 787 | assert_eq!(tools[0].strict, Some(true)); |
| 788 | assert_eq!(tools[0].input_schema["required"], json!(["query"])); |
| 789 | assert_eq!(tools[0].input_schema["additionalProperties"], false); |
| 790 | assert_eq!(tools[1].strict, None); |
| 791 | assert!(tools[1].input_schema.get("anyOf").is_some()); |
| 792 | assert_eq!(tools[2].strict, None); |
| 793 | assert!( |
| 794 | tools[2].input_schema["properties"]["value"] |
| 795 | .get("oneOf") |
| 796 | .is_some() |
| 797 | ); |
| 798 | } |
| 799 | |
| 800 | #[test] |
| 801 | fn strict_mode_rejects_nested_unsupported_composition() { |
| 802 | let mut tools = vec![Tool { |
| 803 | tool_type: None, |
| 804 | name: "nested".to_string(), |
| 805 | description: "Nested oneOf".to_string(), |
| 806 | input_schema: json!({ |
| 807 | "type": "object", |
| 808 | "properties": { |
| 809 | "value": { |
| 810 | "oneOf": [ |
| 811 | {"type": "string"}, |
| 812 | {"type": "integer"} |
| 813 | ] |
| 814 | } |
| 815 | } |
| 816 | }), |
| 817 | allowed_callers: None, |
| 818 | defer_loading: None, |
| 819 | input_examples: None, |
| 820 | strict: None, |
| 821 | cache_control: None, |
| 822 | }]; |
| 823 | |
| 824 | assert!(!prepare_tools_for_strict_mode(&mut tools)); |
| 825 | assert_eq!(tools[0].strict, None); |
| 826 | } |
| 827 | |
| 828 | #[test] |
| 829 | fn strict_mode_marks_compatible_tools_strict() { |
| 830 | let mut tools = vec![Tool { |
| 831 | tool_type: None, |
| 832 | name: "lookup".to_string(), |
| 833 | description: "Lookup".to_string(), |
| 834 | input_schema: json!({ |
| 835 | "type": "object", |
| 836 | "properties": { |
| 837 | "query": {"type": "string"} |
| 838 | }, |
| 839 | "required": [] |
| 840 | }), |
| 841 | allowed_callers: None, |
| 842 | defer_loading: None, |
| 843 | input_examples: None, |
| 844 | strict: None, |
| 845 | cache_control: None, |
| 846 | }]; |
| 847 | |
| 848 | assert!(prepare_tools_for_strict_mode(&mut tools)); |
| 849 | assert_eq!(tools[0].strict, Some(true)); |
| 850 | assert_eq!(tools[0].input_schema["required"], json!(["query"])); |
| 851 | assert_eq!(tools[0].input_schema["additionalProperties"], false); |
| 852 | } |
| 853 | |
| 854 | #[test] |
| 855 | fn responses_sanitize_removes_root_composition_from_apply_patch_shape() { |
| 856 | let mut schema = json!({ |
| 857 | "type": "object", |
| 858 | "properties": { |
| 859 | "path": {"type": "string"}, |
| 860 | "patch": {"type": "string"}, |
| 861 | "replace": { |
| 862 | "type": "array", |
| 863 | "items": { |
| 864 | "type": "object", |
| 865 | "properties": { |
| 866 | "path": {"type": "string"}, |
| 867 | "content": {"type": "string"} |
| 868 | }, |
| 869 | "required": ["path", "content"] |
| 870 | } |
| 871 | }, |
| 872 | "changes": { |
| 873 | "type": "array", |
| 874 | "items": { |
| 875 | "type": "object", |
| 876 | "properties": { |
| 877 | "path": {"type": "string"}, |
| 878 | "content": {"type": "string"} |
| 879 | }, |
| 880 | "required": ["path", "content"] |
| 881 | } |
| 882 | } |
| 883 | }, |
| 884 | "oneOf": [ |
| 885 | {"required": ["patch"]}, |
| 886 | {"required": ["replace"]}, |
| 887 | {"required": ["changes"]} |
| 888 | ] |
| 889 | }); |
| 890 | |
| 891 | let note = sanitize_for_responses(&mut schema); |
| 892 | |
| 893 | assert_eq!(schema["type"], "object"); |
| 894 | assert!(schema.get("oneOf").is_none()); |
| 895 | assert!(schema.get("anyOf").is_none()); |
| 896 | assert!(schema.get("allOf").is_none()); |
| 897 | assert!(schema.get("enum").is_none()); |
| 898 | assert!(schema.get("not").is_none()); |
| 899 | assert!(schema["properties"].get("patch").is_some()); |
| 900 | assert!(schema["properties"].get("replace").is_some()); |
| 901 | assert!(schema["properties"].get("changes").is_some()); |
| 902 | assert_eq!( |
| 903 | note.as_deref(), |
| 904 | Some( |
| 905 | "Exactly one of these parameter groups must be provided: `changes` | `patch` | `replace`." |
| 906 | ) |
| 907 | ); |
| 908 | } |
| 909 | |
| 910 | #[test] |
| 911 | fn responses_sanitize_merges_root_alternative_properties() { |
| 912 | let mut schema = json!({ |
| 913 | "anyOf": [ |
| 914 | { |
| 915 | "type": "object", |
| 916 | "properties": { |
| 917 | "path": {"type": "string"} |
| 918 | }, |
| 919 | "required": ["path"] |
| 920 | }, |
| 921 | { |
| 922 | "type": "object", |
| 923 | "properties": { |
| 924 | "url": {"type": "string"} |
| 925 | }, |
| 926 | "required": ["url"] |
| 927 | } |
| 928 | ] |
| 929 | }); |
| 930 | |
| 931 | let note = sanitize_for_responses(&mut schema); |
| 932 | |
| 933 | assert_eq!(schema["type"], "object"); |
| 934 | assert!(schema.get("anyOf").is_none()); |
| 935 | assert!(schema["properties"].get("path").is_some()); |
| 936 | assert!(schema["properties"].get("url").is_some()); |
| 937 | assert!(schema.get("required").is_none()); |
| 938 | assert_eq!( |
| 939 | note.as_deref(), |
| 940 | Some("At least one of these parameter groups must be provided: `path` | `url`.") |
| 941 | ); |
| 942 | } |
| 943 | |
| 944 | #[test] |
| 945 | fn responses_sanitize_preserves_nested_alternatives() { |
| 946 | let mut schema = json!({ |
| 947 | "type": "object", |
| 948 | "properties": { |
| 949 | "value": { |
| 950 | "anyOf": [ |
| 951 | {"type": "string"}, |
| 952 | {"type": "integer"} |
| 953 | ] |
| 954 | } |
| 955 | } |
| 956 | }); |
| 957 | |
| 958 | let note = sanitize_for_responses(&mut schema); |
| 959 | |
| 960 | assert_eq!(schema["type"], "object"); |
| 961 | assert!(schema.get("anyOf").is_none()); |
| 962 | assert!(schema["properties"]["value"].get("anyOf").is_some()); |
| 963 | assert_eq!(note, None); |
| 964 | } |
| 965 | |
| 966 | #[test] |
| 967 | fn xai_sanitize_flattens_apply_patch_root_one_of() { |
| 968 | // The exact shape that produced the live 400: |
| 969 | // "apply_patch: tool parameter root must be an object type (root |
| 970 | // schema is an anyOf/oneOf union with a non-object branch)". |
| 971 | use crate::tools::spec::ToolSpec as _; |
| 972 | let mut schema = crate::tools::apply_patch::ApplyPatchTool.input_schema(); |
| 973 | assert!(schema.get("oneOf").is_some(), "fixture must match the tool"); |
| 974 | |
| 975 | let note = sanitize_for_xai_parameters(&mut schema); |
| 976 | |
| 977 | assert_eq!(schema["type"], "object"); |
| 978 | assert!(schema.get("oneOf").is_none()); |
| 979 | assert!(schema.get("anyOf").is_none()); |
| 980 | assert!(schema["properties"].get("patch").is_some()); |
| 981 | assert!(schema["properties"].get("changes").is_some()); |
| 982 | assert_eq!( |
| 983 | note.as_deref(), |
| 984 | Some( |
| 985 | "Exactly one of these parameter groups must be provided: `changes` | `patch` | `replace`." |
| 986 | ) |
| 987 | ); |
| 988 | } |
| 989 | |
| 990 | #[test] |
| 991 | fn responses_sanitize_plain_object_has_no_constraint_note() { |
| 992 | let mut schema = json!({ |
| 993 | "type": "object", |
| 994 | "properties": { |
| 995 | "query": {"type": "string"} |
| 996 | } |
| 997 | }); |
| 998 | |
| 999 | let note = sanitize_for_responses(&mut schema); |
| 1000 | |
| 1001 | assert_eq!(schema["type"], "object"); |
| 1002 | assert_eq!(note, None); |
| 1003 | } |
| 1004 | |
| 1005 | #[test] |
| 1006 | fn responses_constraint_note_is_sorted_and_deduped() { |
| 1007 | let mut schema = json!({ |
| 1008 | "type": "object", |
| 1009 | "properties": { |
| 1010 | "a": {"type": "string"}, |
| 1011 | "b": {"type": "string"}, |
| 1012 | "c": {"type": "string"} |
| 1013 | }, |
| 1014 | "oneOf": [ |
| 1015 | {"required": ["b", "a", "a"]}, |
| 1016 | {"required": ["c"]}, |
| 1017 | {"required": ["a", "b"]} |
| 1018 | ] |
| 1019 | }); |
| 1020 | |
| 1021 | let note = sanitize_for_responses(&mut schema); |
| 1022 | |
| 1023 | assert_eq!( |
| 1024 | note.as_deref(), |
| 1025 | Some("Exactly one of these parameter groups must be provided: `a` + `b` | `c`.") |
| 1026 | ); |
| 1027 | } |
| 1028 | } |
| 1029 | |
| 1030 | /// Normalize a tool's function schema for Kimi / Moonshot API compatibility. |
| 1031 | /// |
| 1032 | /// Kimi's API enforces stricter JSON Schema validation: when a schema uses |
| 1033 | /// `anyOf` / `oneOf`, the `type` field must be placed inside each item rather |
| 1034 | /// than on the parent object. This function walks the schema root and any |
| 1035 | /// nested objects, pushing `"type": "object"` down into `anyOf` / `oneOf` |
| 1036 | /// items when present. |
| 1037 | /// |
| 1038 | /// Invariant: only mutates objects that carry a top-level `type` + an |
| 1039 | /// `anyOf` or `oneOf` array — pure schemas without conditional alternatives |
| 1040 | /// are left untouched. |
| 1041 | pub fn sanitize_for_kimi(schema: &mut serde_json::Value) { |
| 1042 | if let Some(obj) = schema.as_object_mut() { |
| 1043 | // Recurse first so a type injected into this object's alternatives is |
| 1044 | // not immediately removed again by processing that freshly-mutated item. |
| 1045 | for map_key in ["properties", "$defs"] { |
| 1046 | if let Some(children) = obj.get_mut(map_key).and_then(Value::as_object_mut) { |
| 1047 | for child in children.values_mut() { |
| 1048 | sanitize_for_kimi(child); |
| 1049 | } |
| 1050 | } |
| 1051 | } |
| 1052 | if let Some(items) = obj.get_mut("items") { |
| 1053 | sanitize_for_kimi(items); |
| 1054 | } |
| 1055 | if let Some(additional) = obj.get_mut("additionalProperties") |
| 1056 | && additional.is_object() |
| 1057 | { |
| 1058 | sanitize_for_kimi(additional); |
| 1059 | } |
| 1060 | for union_key in ["anyOf", "oneOf"] { |
| 1061 | if let Some(branches) = obj.get_mut(union_key).and_then(Value::as_array_mut) { |
| 1062 | for branch in branches { |
| 1063 | sanitize_for_kimi(branch); |
| 1064 | } |
| 1065 | } |
| 1066 | } |
| 1067 | |
| 1068 | // If this object has `type` + `anyOf`/`oneOf`, push `type` into |
| 1069 | // each item and remove it from the parent. Otherwise leave it alone. |
| 1070 | let should_push = |
| 1071 | obj.contains_key("type") && (obj.contains_key("anyOf") || obj.contains_key("oneOf")); |
| 1072 | if should_push && let Some(type_val) = obj.remove("type") { |
| 1073 | for key in ["anyOf", "oneOf"] { |
| 1074 | if let Some(items) = obj.get_mut(key).and_then(|v| v.as_array_mut()) { |
| 1075 | for item in items { |
| 1076 | if let Some(item_obj) = item.as_object_mut() |
| 1077 | && !item_obj.contains_key("type") |
| 1078 | { |
| 1079 | item_obj.insert("type".to_string(), type_val.clone()); |
| 1080 | } |
| 1081 | } |
| 1082 | } |
| 1083 | } |
| 1084 | // The provider-neutral sanitizer injects an empty `properties` |
| 1085 | // map on every bare object before this provider pass. MFJS permits |
| 1086 | // only annotations beside `anyOf`, so remove that semantic no-op |
| 1087 | // after moving the object type into each branch. |
| 1088 | if obj |
| 1089 | .get("properties") |
| 1090 | .and_then(Value::as_object) |
| 1091 | .is_some_and(Map::is_empty) |
| 1092 | { |
| 1093 | obj.remove("properties"); |
| 1094 | } |
| 1095 | } |
| 1096 | } |
| 1097 | } |
| 1098 | |
| 1099 | /// A safe, provider-facing reason that Kimi parameters could not be emitted. |
| 1100 | /// |
| 1101 | /// These diagnostics deliberately never include the schema or `$ref` value: |
| 1102 | /// tool schemas can be supplied by MCP servers and may contain private data. |
| 1103 | #[derive(Debug, Clone, Copy, PartialEq, Eq, thiserror::Error)] |
| 1104 | pub enum KimiParameterSchemaError { |
| 1105 | #[error("Moonshot function parameters root must be a JSON object schema")] |
| 1106 | RootMustBeObject, |
| 1107 | #[error("Moonshot function parameters contain an unsupported root reference")] |
| 1108 | UnsupportedRootReference, |
| 1109 | #[error("Moonshot function parameters contain an unresolved internal root reference")] |
| 1110 | UnresolvedRootReference, |
| 1111 | #[error("Moonshot function parameters contain a cyclic internal root reference")] |
| 1112 | CyclicRootReference, |
| 1113 | #[error("Moonshot function parameters root reference must resolve to an object schema")] |
| 1114 | ReferencedRootMustBeObject, |
| 1115 | #[error("Moonshot function parameters contain unsupported nested allOf composition")] |
| 1116 | UnsupportedNestedAllOf, |
| 1117 | #[error("Moonshot function parameters contain conflicting nested union composition")] |
| 1118 | ConflictingNestedUnion, |
| 1119 | #[error("Moonshot function parameters contain an unsupported const literal")] |
| 1120 | UnsupportedConstLiteral, |
| 1121 | #[error("Moonshot function parameters contain conflicting literal constraints")] |
| 1122 | ConflictingLiteralConstraint, |
| 1123 | #[error("Moonshot function parameters contain an invalid nullable marker")] |
| 1124 | InvalidNullable, |
| 1125 | #[error("Moonshot function parameters contain an unsupported MFJS keyword")] |
| 1126 | UnsupportedKeyword, |
| 1127 | #[error("Moonshot function parameters contain an invalid MFJS schema node")] |
| 1128 | InvalidSchemaNode, |
| 1129 | #[error("Moonshot function parameters contain an invalid MFJS keyword value")] |
| 1130 | InvalidKeywordValue, |
| 1131 | #[error("Moonshot function parameters contain an invalid MFJS reference")] |
| 1132 | InvalidReference, |
| 1133 | #[error("Moonshot function parameters contain an MFJS schema without a concrete type")] |
| 1134 | MissingType, |
| 1135 | #[error("Moonshot function parameters exceed an MFJS resource limit")] |
| 1136 | ResourceLimitExceeded, |
| 1137 | #[error("Moonshot function parameters contain a non-terminating MFJS reference")] |
| 1138 | NonTerminatingReference, |
| 1139 | #[error("Moonshot function parameters contain an invalid MFJS default")] |
| 1140 | InvalidDefault, |
| 1141 | #[error("Moonshot function parameters contain an invalid MFJS range")] |
| 1142 | InvalidRange, |
| 1143 | } |
| 1144 | |
| 1145 | /// Normalize a complete Kimi / Moonshot `function.parameters` object. |
| 1146 | /// |
| 1147 | /// Function parameters have an additional MFJS constraint: the root must end |
| 1148 | /// as a plain `type: "object"` schema. Root composition is flattened using the |
| 1149 | /// same compatibility pass as Responses and xAI, while supported nested |
| 1150 | /// `anyOf` branches remain nested. Internal root `$ref` values are resolved and |
| 1151 | /// inlined before normalization so we never manufacture the invalid |
| 1152 | /// `type + allOf($ref)` shape rejected by MFJS. |
| 1153 | /// |
| 1154 | /// Unsupported, unresolved, cyclic, and non-object root references fail |
| 1155 | /// closed with a non-secret diagnostic instead of being sent to Moonshot. |
| 1156 | /// |
| 1157 | /// MFJS differences from JSON Schema: |
| 1158 | /// <https://github.com/MoonshotAI/walle/blob/main/docs/mfjs-walle-vs-draft-2020-12.md> |
| 1159 | pub fn sanitize_for_kimi_parameters( |
| 1160 | parameters: &mut Value, |
| 1161 | ) -> Result<Option<String>, KimiParameterSchemaError> { |
| 1162 | // Work on a clone so a rejected schema remains byte-for-byte unchanged for |
| 1163 | // callers that retain the catalog and retry against another provider. |
| 1164 | let mut candidate = parameters.clone(); |
| 1165 | let constraint_note = sanitize_kimi_parameters_candidate(&mut candidate)?; |
| 1166 | validate_mfjs_parameters(&candidate)?; |
| 1167 | *parameters = candidate; |
| 1168 | Ok(constraint_note) |
| 1169 | } |
| 1170 | |
| 1171 | fn sanitize_kimi_parameters_candidate( |
| 1172 | parameters: &mut Value, |
| 1173 | ) -> Result<Option<String>, KimiParameterSchemaError> { |
| 1174 | let Some(root) = parameters.as_object() else { |
| 1175 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1176 | }; |
| 1177 | if root |
| 1178 | .get("type") |
| 1179 | .is_some_and(|schema_type| schema_type != "object") |
| 1180 | { |
| 1181 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1182 | } |
| 1183 | |
| 1184 | inline_internal_kimi_root_ref(parameters)?; |
| 1185 | |
| 1186 | // A function schema's root cannot carry MFJS composition because it must |
| 1187 | // simultaneously be `type: object`. Flatten the actual root shape used by |
| 1188 | // apply_patch and retain its dropped required-group contract as a prompt |
| 1189 | // note for the model. |
| 1190 | let constraint_note = sanitize_for_responses(parameters); |
| 1191 | |
| 1192 | // Restore nullable unions collapsed by the registry's provider-neutral |
| 1193 | // sanitizer, translate MFJS-safe scalar const values, and normalize nested |
| 1194 | // composition. Codewhale still validates tool input before execution, so |
| 1195 | // widening oneOf to MFJS's anyOf remains safe; allOf fails closed. |
| 1196 | normalize_kimi_compatibility(parameters, true)?; |
| 1197 | |
| 1198 | // MFJS requires `type` to live inside each anyOf branch, never alongside |
| 1199 | // the union keyword. The root is composition-free at this point, so this |
| 1200 | // only adjusts valid nested unions. |
| 1201 | sanitize_for_kimi(parameters); |
| 1202 | |
| 1203 | let Some(root) = parameters.as_object() else { |
| 1204 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1205 | }; |
| 1206 | if root.get("type").and_then(Value::as_str) != Some("object") |
| 1207 | || root.contains_key("anyOf") |
| 1208 | || root.contains_key("oneOf") |
| 1209 | || root.contains_key("allOf") |
| 1210 | || root.contains_key("$ref") |
| 1211 | { |
| 1212 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1213 | } |
| 1214 | |
| 1215 | Ok(constraint_note) |
| 1216 | } |
| 1217 | |
| 1218 | fn inline_internal_kimi_root_ref(parameters: &mut Value) -> Result<(), KimiParameterSchemaError> { |
| 1219 | let document = parameters.clone(); |
| 1220 | let Some(document_root) = document.as_object() else { |
| 1221 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1222 | }; |
| 1223 | let Some(root_ref) = document_root.get("$ref") else { |
| 1224 | return Ok(()); |
| 1225 | }; |
| 1226 | let Some(mut reference) = root_ref.as_str() else { |
| 1227 | return Err(KimiParameterSchemaError::UnsupportedRootReference); |
| 1228 | }; |
| 1229 | |
| 1230 | let mut visited = HashSet::new(); |
| 1231 | let resolved = loop { |
| 1232 | if !reference.starts_with("#/") { |
| 1233 | return Err(KimiParameterSchemaError::UnsupportedRootReference); |
| 1234 | } |
| 1235 | if !visited.insert(reference.to_string()) { |
| 1236 | return Err(KimiParameterSchemaError::CyclicRootReference); |
| 1237 | } |
| 1238 | let target = document |
| 1239 | .pointer(&reference[1..]) |
| 1240 | .ok_or(KimiParameterSchemaError::UnresolvedRootReference)?; |
| 1241 | let target = target |
| 1242 | .as_object() |
| 1243 | .ok_or(KimiParameterSchemaError::ReferencedRootMustBeObject)?; |
| 1244 | if let Some(next_ref) = target.get("$ref") { |
| 1245 | reference = next_ref |
| 1246 | .as_str() |
| 1247 | .ok_or(KimiParameterSchemaError::UnsupportedRootReference)?; |
| 1248 | continue; |
| 1249 | } |
| 1250 | if target.get("type").and_then(Value::as_str) != Some("object") { |
| 1251 | return Err(KimiParameterSchemaError::ReferencedRootMustBeObject); |
| 1252 | } |
| 1253 | break target.clone(); |
| 1254 | }; |
| 1255 | |
| 1256 | let mut inlined = resolved; |
| 1257 | for (key, value) in document_root { |
| 1258 | if key != "$ref" { |
| 1259 | inlined.insert(key.clone(), value.clone()); |
| 1260 | } |
| 1261 | } |
| 1262 | *parameters = Value::Object(inlined); |
| 1263 | Ok(()) |
| 1264 | } |
| 1265 | |
| 1266 | fn normalize_kimi_compatibility( |
| 1267 | schema: &mut Value, |
| 1268 | is_root: bool, |
| 1269 | ) -> Result<(), KimiParameterSchemaError> { |
| 1270 | let Some(obj) = schema.as_object_mut() else { |
| 1271 | return Err(KimiParameterSchemaError::InvalidSchemaNode); |
| 1272 | }; |
| 1273 | |
| 1274 | if !is_root { |
| 1275 | if obj.contains_key("allOf") { |
| 1276 | return Err(KimiParameterSchemaError::UnsupportedNestedAllOf); |
| 1277 | } |
| 1278 | if let Some(one_of) = obj.remove("oneOf") { |
| 1279 | if obj.contains_key("anyOf") { |
| 1280 | return Err(KimiParameterSchemaError::ConflictingNestedUnion); |
| 1281 | } |
| 1282 | obj.insert("anyOf".to_string(), one_of); |
| 1283 | } |
| 1284 | } |
| 1285 | |
| 1286 | if let Some(constant) = obj.remove("const") { |
| 1287 | if !is_mfjs_enum_literal(&constant) { |
| 1288 | return Err(KimiParameterSchemaError::UnsupportedConstLiteral); |
| 1289 | } |
| 1290 | if !obj.contains_key("type") { |
| 1291 | let inferred = mfjs_literal_kind(&constant) |
| 1292 | .ok_or(KimiParameterSchemaError::UnsupportedConstLiteral)?; |
| 1293 | let schema_type = match inferred { |
| 1294 | MfjsLiteralKind::Integer => "integer", |
| 1295 | MfjsLiteralKind::Number => "number", |
| 1296 | MfjsLiteralKind::String => "string", |
| 1297 | }; |
| 1298 | obj.insert("type".to_string(), Value::String(schema_type.to_string())); |
| 1299 | } |
| 1300 | if let Some(existing) = obj.get("enum") { |
| 1301 | let agrees = existing |
| 1302 | .as_array() |
| 1303 | .is_some_and(|values| values.as_slice() == [constant.clone()]); |
| 1304 | if !agrees { |
| 1305 | return Err(KimiParameterSchemaError::ConflictingLiteralConstraint); |
| 1306 | } |
| 1307 | } else { |
| 1308 | obj.insert("enum".to_string(), Value::Array(vec![constant])); |
| 1309 | } |
| 1310 | } |
| 1311 | |
| 1312 | let nullable = obj.remove("nullable"); |
| 1313 | if nullable.is_some() |
| 1314 | && !obj |
| 1315 | .get("type") |
| 1316 | .is_some_and(|schema_type| schema_type.as_str().is_some_and(is_mfjs_concrete_type)) |
| 1317 | { |
| 1318 | return Err(KimiParameterSchemaError::InvalidNullable); |
| 1319 | } |
| 1320 | match nullable.as_ref().map(Value::as_bool) { |
| 1321 | None => {} |
| 1322 | Some(Some(false)) => {} |
| 1323 | Some(Some(true)) if is_root => { |
| 1324 | // Function parameters are required to be an object at the root; |
| 1325 | // null was never a valid wire instance there. |
| 1326 | } |
| 1327 | Some(Some(true)) => { |
| 1328 | let non_null = Value::Object(std::mem::take(obj)); |
| 1329 | *schema = serde_json::json!({ |
| 1330 | "anyOf": [non_null, {"type": "null"}] |
| 1331 | }); |
| 1332 | } |
| 1333 | Some(None) => return Err(KimiParameterSchemaError::InvalidNullable), |
| 1334 | } |
| 1335 | |
| 1336 | normalize_kimi_child_schemas(schema)?; |
| 1337 | Ok(()) |
| 1338 | } |
| 1339 | |
| 1340 | fn normalize_kimi_child_schemas(schema: &mut Value) -> Result<(), KimiParameterSchemaError> { |
| 1341 | let Some(obj) = schema.as_object_mut() else { |
| 1342 | return Err(KimiParameterSchemaError::InvalidSchemaNode); |
| 1343 | }; |
| 1344 | |
| 1345 | for map_key in ["properties", "$defs"] { |
| 1346 | if let Some(children) = obj.get_mut(map_key).and_then(Value::as_object_mut) { |
| 1347 | for child in children.values_mut() { |
| 1348 | normalize_kimi_compatibility(child, false)?; |
| 1349 | } |
| 1350 | } |
| 1351 | } |
| 1352 | |
| 1353 | if let Some(items) = obj.get_mut("items") { |
| 1354 | normalize_kimi_compatibility(items, false)?; |
| 1355 | } |
| 1356 | if let Some(additional) = obj.get_mut("additionalProperties") |
| 1357 | && additional.is_object() |
| 1358 | { |
| 1359 | normalize_kimi_compatibility(additional, false)?; |
| 1360 | } |
| 1361 | if let Some(branches) = obj.get_mut("anyOf").and_then(Value::as_array_mut) { |
| 1362 | for branch in branches { |
| 1363 | normalize_kimi_compatibility(branch, false)?; |
| 1364 | } |
| 1365 | } |
| 1366 | Ok(()) |
| 1367 | } |
| 1368 | |
| 1369 | fn is_mfjs_enum_literal(value: &Value) -> bool { |
| 1370 | value.is_string() || value.is_number() |
| 1371 | } |
| 1372 | |
| 1373 | /// Validate one fully normalized MFJS function-parameters schema. |
| 1374 | /// |
| 1375 | /// Every error is a fixed enum variant: schemas can originate in MCP or |
| 1376 | /// runtime tools and may contain private names or values, so diagnostics must |
| 1377 | /// never echo a keyword, property, reference, or literal from the document. |
| 1378 | pub fn validate_mfjs_parameters(parameters: &Value) -> Result<(), KimiParameterSchemaError> { |
| 1379 | let root = parameters |
| 1380 | .as_object() |
| 1381 | .ok_or(KimiParameterSchemaError::RootMustBeObject)?; |
| 1382 | if root.get("type").and_then(Value::as_str) != Some("object") |
| 1383 | || root.contains_key("anyOf") |
| 1384 | || root.contains_key("oneOf") |
| 1385 | || root.contains_key("allOf") |
| 1386 | || root.contains_key("$ref") |
| 1387 | { |
| 1388 | return Err(KimiParameterSchemaError::RootMustBeObject); |
| 1389 | } |
| 1390 | if serde_json::to_vec(parameters) |
| 1391 | .map(|encoded| encoded.len() > MFJS_MAX_SCHEMA_BYTES) |
| 1392 | .unwrap_or(true) |
| 1393 | { |
| 1394 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1395 | } |
| 1396 | |
| 1397 | let mut state = MfjsValidationState::new(parameters); |
| 1398 | state.validate_schema(parameters, true, false, 0, 0)?; |
| 1399 | |
| 1400 | let mut visiting_refs = HashSet::new(); |
| 1401 | if !mfjs_schema_can_terminate(parameters, parameters, &mut visiting_refs, 0)? { |
| 1402 | return Err(KimiParameterSchemaError::NonTerminatingReference); |
| 1403 | } |
| 1404 | |
| 1405 | validate_mfjs_expanded_depth(parameters, parameters, 0, &mut HashSet::new(), 0)?; |
| 1406 | if let Some(definitions) = root.get("$defs").and_then(Value::as_object) { |
| 1407 | for definition in definitions.values() { |
| 1408 | validate_mfjs_expanded_depth(definition, parameters, 0, &mut HashSet::new(), 0)?; |
| 1409 | } |
| 1410 | } |
| 1411 | Ok(()) |
| 1412 | } |
| 1413 | |
| 1414 | const MFJS_MAX_ANY_OF_ITEMS: usize = 10; |
| 1415 | const MFJS_MAX_OBJECT_DEPTH: usize = 5; |
| 1416 | const MFJS_MAX_TOTAL_PROPERTIES: usize = 100; |
| 1417 | const MFJS_MAX_TOTAL_ENUM_VALUES: usize = 500; |
| 1418 | const MFJS_ENUM_LENGTH_CHECK_THRESHOLD: usize = 250; |
| 1419 | const MFJS_MAX_ENUM_STRING_LENGTH: usize = 7_500; |
| 1420 | const MFJS_MAX_SCHEMA_BYTES: usize = 120_000; |
| 1421 | const MFJS_MAX_STRUCTURAL_DEPTH: usize = 64; |
| 1422 | const MFJS_MAX_SAFE_INTEGER: f64 = 9_007_199_254_740_991.0; |
| 1423 | |
| 1424 | struct MfjsValidationState<'a> { |
| 1425 | document: &'a Value, |
| 1426 | total_properties: usize, |
| 1427 | total_enum_values: usize, |
| 1428 | } |
| 1429 | |
| 1430 | impl<'a> MfjsValidationState<'a> { |
| 1431 | fn new(document: &'a Value) -> Self { |
| 1432 | Self { |
| 1433 | document, |
| 1434 | total_properties: 0, |
| 1435 | total_enum_values: 0, |
| 1436 | } |
| 1437 | } |
| 1438 | |
| 1439 | fn validate_schema( |
| 1440 | &mut self, |
| 1441 | schema: &Value, |
| 1442 | is_root: bool, |
| 1443 | allow_empty: bool, |
| 1444 | property_depth: usize, |
| 1445 | structural_depth: usize, |
| 1446 | ) -> Result<(), KimiParameterSchemaError> { |
| 1447 | if structural_depth > MFJS_MAX_STRUCTURAL_DEPTH { |
| 1448 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1449 | } |
| 1450 | let obj = schema |
| 1451 | .as_object() |
| 1452 | .ok_or(KimiParameterSchemaError::InvalidSchemaNode)?; |
| 1453 | if obj.is_empty() { |
| 1454 | return if allow_empty { |
| 1455 | Ok(()) |
| 1456 | } else { |
| 1457 | Err(KimiParameterSchemaError::MissingType) |
| 1458 | }; |
| 1459 | } |
| 1460 | |
| 1461 | const ALLOWED_KEYWORDS: &[&str] = &[ |
| 1462 | "$id", |
| 1463 | "$ref", |
| 1464 | "$defs", |
| 1465 | "anyOf", |
| 1466 | "properties", |
| 1467 | "additionalProperties", |
| 1468 | "items", |
| 1469 | "type", |
| 1470 | "enum", |
| 1471 | "required", |
| 1472 | "maxLength", |
| 1473 | "minLength", |
| 1474 | "maximum", |
| 1475 | "minimum", |
| 1476 | "maxItems", |
| 1477 | "minItems", |
| 1478 | "title", |
| 1479 | "description", |
| 1480 | "default", |
| 1481 | ]; |
| 1482 | if obj |
| 1483 | .keys() |
| 1484 | .any(|keyword| !ALLOWED_KEYWORDS.contains(&keyword.as_str())) |
| 1485 | { |
| 1486 | return Err(KimiParameterSchemaError::UnsupportedKeyword); |
| 1487 | } |
| 1488 | |
| 1489 | for annotation in ["title", "description"] { |
| 1490 | if obj.get(annotation).is_some_and(|value| !value.is_string()) { |
| 1491 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1492 | } |
| 1493 | } |
| 1494 | if obj.get("$id").is_some_and(|value| !value.is_string()) |
| 1495 | || (!is_root && obj.contains_key("$id")) |
| 1496 | { |
| 1497 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1498 | } |
| 1499 | |
| 1500 | if let Some(definitions) = obj.get("$defs") { |
| 1501 | if !is_root { |
| 1502 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1503 | } |
| 1504 | let definitions = definitions |
| 1505 | .as_object() |
| 1506 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1507 | for (name, definition) in definitions { |
| 1508 | if name.is_empty() || name.contains('/') { |
| 1509 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1510 | } |
| 1511 | self.validate_schema(definition, false, false, 0, structural_depth + 1)?; |
| 1512 | } |
| 1513 | } |
| 1514 | |
| 1515 | if let Some(reference) = obj.get("$ref") { |
| 1516 | let reference = reference |
| 1517 | .as_str() |
| 1518 | .ok_or(KimiParameterSchemaError::InvalidReference)?; |
| 1519 | if resolve_mfjs_reference(self.document, reference).is_none() { |
| 1520 | return Err(KimiParameterSchemaError::InvalidReference); |
| 1521 | } |
| 1522 | let allowed_ref_sibling = |key: &str| { |
| 1523 | matches!(key, "$ref" | "title" | "description") |
| 1524 | || (is_root && matches!(key, "$defs" | "$id")) |
| 1525 | }; |
| 1526 | if obj.keys().any(|key| !allowed_ref_sibling(key)) { |
| 1527 | return Err(KimiParameterSchemaError::InvalidReference); |
| 1528 | } |
| 1529 | return Ok(()); |
| 1530 | } |
| 1531 | |
| 1532 | if let Some(any_of) = obj.get("anyOf") { |
| 1533 | let branches = any_of |
| 1534 | .as_array() |
| 1535 | .filter(|branches| !branches.is_empty() && branches.len() <= MFJS_MAX_ANY_OF_ITEMS) |
| 1536 | .ok_or(KimiParameterSchemaError::ResourceLimitExceeded)?; |
| 1537 | let allowed_union_sibling = |key: &str| { |
| 1538 | matches!(key, "anyOf" | "title" | "description") |
| 1539 | || (is_root && matches!(key, "$defs" | "$id")) |
| 1540 | }; |
| 1541 | if obj.keys().any(|key| !allowed_union_sibling(key)) { |
| 1542 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1543 | } |
| 1544 | for branch in branches { |
| 1545 | self.validate_schema(branch, false, false, property_depth, structural_depth + 1)?; |
| 1546 | } |
| 1547 | return Ok(()); |
| 1548 | } |
| 1549 | |
| 1550 | let schema_type = obj |
| 1551 | .get("type") |
| 1552 | .and_then(Value::as_str) |
| 1553 | .filter(|schema_type| is_mfjs_concrete_type(schema_type)) |
| 1554 | .ok_or(KimiParameterSchemaError::MissingType)?; |
| 1555 | if obj |
| 1556 | .keys() |
| 1557 | .any(|keyword| !mfjs_keyword_allowed_for_type(keyword, schema_type, is_root)) |
| 1558 | { |
| 1559 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1560 | } |
| 1561 | |
| 1562 | if let Some(values) = obj.get("enum") { |
| 1563 | validate_mfjs_enum(values, schema_type, self)?; |
| 1564 | } |
| 1565 | if let Some(default) = obj.get("default") { |
| 1566 | validate_mfjs_default(default, schema_type)?; |
| 1567 | } |
| 1568 | |
| 1569 | if let Some(properties) = obj.get("properties") { |
| 1570 | let properties = properties |
| 1571 | .as_object() |
| 1572 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1573 | self.total_properties = self |
| 1574 | .total_properties |
| 1575 | .checked_add(properties.len()) |
| 1576 | .ok_or(KimiParameterSchemaError::ResourceLimitExceeded)?; |
| 1577 | if self.total_properties > MFJS_MAX_TOTAL_PROPERTIES { |
| 1578 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1579 | } |
| 1580 | for (name, property) in properties { |
| 1581 | if name.is_empty() |
| 1582 | || matches!( |
| 1583 | name.as_str(), |
| 1584 | "$defs" | "$ref" | "anyOf" | "required" | "additionalProperties" |
| 1585 | ) |
| 1586 | { |
| 1587 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1588 | } |
| 1589 | let child_depth = property_depth |
| 1590 | .checked_add(1) |
| 1591 | .ok_or(KimiParameterSchemaError::ResourceLimitExceeded)?; |
| 1592 | if child_depth > MFJS_MAX_OBJECT_DEPTH { |
| 1593 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1594 | } |
| 1595 | self.validate_schema(property, false, false, child_depth, structural_depth + 1)?; |
| 1596 | } |
| 1597 | } |
| 1598 | |
| 1599 | if let Some(required) = obj.get("required") { |
| 1600 | let required = required |
| 1601 | .as_array() |
| 1602 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1603 | let properties = obj |
| 1604 | .get("properties") |
| 1605 | .and_then(Value::as_object) |
| 1606 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1607 | let mut seen = HashSet::new(); |
| 1608 | for name in required { |
| 1609 | let name = name |
| 1610 | .as_str() |
| 1611 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1612 | if name.is_empty() || !properties.contains_key(name) || !seen.insert(name) { |
| 1613 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1614 | } |
| 1615 | } |
| 1616 | } |
| 1617 | |
| 1618 | if let Some(additional) = obj.get("additionalProperties") { |
| 1619 | match additional { |
| 1620 | Value::Bool(_) => {} |
| 1621 | Value::Object(_) => self.validate_schema( |
| 1622 | additional, |
| 1623 | false, |
| 1624 | true, |
| 1625 | property_depth, |
| 1626 | structural_depth + 1, |
| 1627 | )?, |
| 1628 | _ => return Err(KimiParameterSchemaError::InvalidKeywordValue), |
| 1629 | } |
| 1630 | } |
| 1631 | |
| 1632 | if let Some(items) = obj.get("items") { |
| 1633 | self.validate_schema(items, false, false, property_depth, structural_depth + 1)?; |
| 1634 | } |
| 1635 | |
| 1636 | validate_mfjs_bounds(obj, schema_type)?; |
| 1637 | Ok(()) |
| 1638 | } |
| 1639 | } |
| 1640 | |
| 1641 | fn is_mfjs_concrete_type(schema_type: &str) -> bool { |
| 1642 | matches!( |
| 1643 | schema_type, |
| 1644 | "null" | "boolean" | "object" | "array" | "number" | "integer" | "string" |
| 1645 | ) |
| 1646 | } |
| 1647 | |
| 1648 | fn mfjs_keyword_allowed_for_type(keyword: &str, schema_type: &str, is_root: bool) -> bool { |
| 1649 | if is_root && matches!(keyword, "$defs" | "$id") { |
| 1650 | return true; |
| 1651 | } |
| 1652 | if matches!(keyword, "type" | "title" | "description") { |
| 1653 | return true; |
| 1654 | } |
| 1655 | match schema_type { |
| 1656 | "object" => matches!(keyword, "properties" | "required" | "additionalProperties"), |
| 1657 | "array" => matches!(keyword, "items" | "minItems" | "maxItems"), |
| 1658 | "string" => matches!(keyword, "enum" | "default" | "minLength" | "maxLength"), |
| 1659 | "number" | "integer" => { |
| 1660 | matches!(keyword, "enum" | "default" | "minimum" | "maximum") |
| 1661 | } |
| 1662 | "boolean" | "null" => keyword == "default", |
| 1663 | _ => false, |
| 1664 | } |
| 1665 | } |
| 1666 | |
| 1667 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 1668 | enum MfjsLiteralKind { |
| 1669 | Integer, |
| 1670 | Number, |
| 1671 | String, |
| 1672 | } |
| 1673 | |
| 1674 | fn mfjs_literal_kind(value: &Value) -> Option<MfjsLiteralKind> { |
| 1675 | if value.is_string() { |
| 1676 | return Some(MfjsLiteralKind::String); |
| 1677 | } |
| 1678 | let number = value.as_number()?; |
| 1679 | if number.is_i64() || number.is_u64() { |
| 1680 | Some(MfjsLiteralKind::Integer) |
| 1681 | } else { |
| 1682 | Some(MfjsLiteralKind::Number) |
| 1683 | } |
| 1684 | } |
| 1685 | |
| 1686 | fn validate_mfjs_enum( |
| 1687 | value: &Value, |
| 1688 | schema_type: &str, |
| 1689 | state: &mut MfjsValidationState<'_>, |
| 1690 | ) -> Result<(), KimiParameterSchemaError> { |
| 1691 | let values = value |
| 1692 | .as_array() |
| 1693 | .filter(|values| !values.is_empty()) |
| 1694 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1695 | state.total_enum_values = state |
| 1696 | .total_enum_values |
| 1697 | .checked_add(values.len()) |
| 1698 | .ok_or(KimiParameterSchemaError::ResourceLimitExceeded)?; |
| 1699 | if state.total_enum_values > MFJS_MAX_TOTAL_ENUM_VALUES { |
| 1700 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1701 | } |
| 1702 | |
| 1703 | let values_match_type = match schema_type { |
| 1704 | "string" => values.iter().all(Value::is_string), |
| 1705 | "integer" => values |
| 1706 | .iter() |
| 1707 | .all(|value| mfjs_integer_value(value).is_some()), |
| 1708 | "number" => values.iter().all(Value::is_number), |
| 1709 | _ => false, |
| 1710 | }; |
| 1711 | if !values_match_type { |
| 1712 | return Err(KimiParameterSchemaError::InvalidKeywordValue); |
| 1713 | } |
| 1714 | |
| 1715 | if values.len() > MFJS_ENUM_LENGTH_CHECK_THRESHOLD { |
| 1716 | let encoded_length = values.iter().try_fold(0usize, |total, value| { |
| 1717 | let literal_length = value |
| 1718 | .as_str() |
| 1719 | .map(str::len) |
| 1720 | .unwrap_or_else(|| value.to_string().len()); |
| 1721 | total.checked_add(literal_length) |
| 1722 | }); |
| 1723 | if encoded_length.is_none_or(|length| length > MFJS_MAX_ENUM_STRING_LENGTH) { |
| 1724 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1725 | } |
| 1726 | } |
| 1727 | Ok(()) |
| 1728 | } |
| 1729 | |
| 1730 | fn validate_mfjs_default(value: &Value, schema_type: &str) -> Result<(), KimiParameterSchemaError> { |
| 1731 | let valid = match schema_type { |
| 1732 | "boolean" => value.is_boolean(), |
| 1733 | "number" => value.is_number(), |
| 1734 | "integer" => mfjs_integer_value(value).is_some(), |
| 1735 | "string" => value.is_string(), |
| 1736 | "null" => value.is_null(), |
| 1737 | _ => false, |
| 1738 | }; |
| 1739 | if valid { |
| 1740 | Ok(()) |
| 1741 | } else { |
| 1742 | Err(KimiParameterSchemaError::InvalidDefault) |
| 1743 | } |
| 1744 | } |
| 1745 | |
| 1746 | fn mfjs_integer_value(value: &Value) -> Option<f64> { |
| 1747 | let value = value.as_number()?.as_f64()?; |
| 1748 | (value.is_finite() && value.fract() == 0.0 && value.abs() <= MFJS_MAX_SAFE_INTEGER) |
| 1749 | .then_some(value) |
| 1750 | } |
| 1751 | |
| 1752 | fn validate_mfjs_bounds( |
| 1753 | obj: &Map<String, Value>, |
| 1754 | schema_type: &str, |
| 1755 | ) -> Result<(), KimiParameterSchemaError> { |
| 1756 | validate_mfjs_unsigned_range(obj, schema_type, "string", "minLength", "maxLength")?; |
| 1757 | validate_mfjs_unsigned_range(obj, schema_type, "array", "minItems", "maxItems")?; |
| 1758 | |
| 1759 | let minimum = obj.get("minimum"); |
| 1760 | let maximum = obj.get("maximum"); |
| 1761 | if minimum.is_some() || maximum.is_some() { |
| 1762 | let parse_bound = |value: &Value| match schema_type { |
| 1763 | "integer" => mfjs_integer_value(value), |
| 1764 | "number" => value.as_number().and_then(serde_json::Number::as_f64), |
| 1765 | _ => None, |
| 1766 | }; |
| 1767 | let minimum = minimum |
| 1768 | .map(|value| parse_bound(value).ok_or(KimiParameterSchemaError::InvalidRange)) |
| 1769 | .transpose()?; |
| 1770 | let maximum = maximum |
| 1771 | .map(|value| parse_bound(value).ok_or(KimiParameterSchemaError::InvalidRange)) |
| 1772 | .transpose()?; |
| 1773 | if minimum.zip(maximum).is_some_and(|(min, max)| min > max) { |
| 1774 | return Err(KimiParameterSchemaError::InvalidRange); |
| 1775 | } |
| 1776 | } |
| 1777 | Ok(()) |
| 1778 | } |
| 1779 | |
| 1780 | fn validate_mfjs_unsigned_range( |
| 1781 | obj: &Map<String, Value>, |
| 1782 | schema_type: &str, |
| 1783 | expected_type: &str, |
| 1784 | minimum_keyword: &str, |
| 1785 | maximum_keyword: &str, |
| 1786 | ) -> Result<(), KimiParameterSchemaError> { |
| 1787 | let minimum = obj.get(minimum_keyword); |
| 1788 | let maximum = obj.get(maximum_keyword); |
| 1789 | if minimum.is_none() && maximum.is_none() { |
| 1790 | return Ok(()); |
| 1791 | } |
| 1792 | if schema_type != expected_type { |
| 1793 | return Err(KimiParameterSchemaError::InvalidRange); |
| 1794 | } |
| 1795 | let minimum = minimum |
| 1796 | .map(|value| value.as_u64().ok_or(KimiParameterSchemaError::InvalidRange)) |
| 1797 | .transpose()?; |
| 1798 | let maximum = maximum |
| 1799 | .map(|value| value.as_u64().ok_or(KimiParameterSchemaError::InvalidRange)) |
| 1800 | .transpose()?; |
| 1801 | if minimum.zip(maximum).is_some_and(|(min, max)| min > max) { |
| 1802 | return Err(KimiParameterSchemaError::InvalidRange); |
| 1803 | } |
| 1804 | Ok(()) |
| 1805 | } |
| 1806 | |
| 1807 | fn resolve_mfjs_reference<'a>(document: &'a Value, reference: &str) -> Option<&'a Value> { |
| 1808 | if reference == "#" { |
| 1809 | return Some(document); |
| 1810 | } |
| 1811 | let name = reference.strip_prefix("#/$defs/")?; |
| 1812 | if name.is_empty() || name.contains('/') { |
| 1813 | return None; |
| 1814 | } |
| 1815 | document |
| 1816 | .pointer(&reference[1..]) |
| 1817 | .filter(|target| target.is_object()) |
| 1818 | } |
| 1819 | |
| 1820 | fn mfjs_schema_can_terminate( |
| 1821 | schema: &Value, |
| 1822 | document: &Value, |
| 1823 | visiting_refs: &mut HashSet<String>, |
| 1824 | structural_depth: usize, |
| 1825 | ) -> Result<bool, KimiParameterSchemaError> { |
| 1826 | if structural_depth > MFJS_MAX_STRUCTURAL_DEPTH { |
| 1827 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1828 | } |
| 1829 | let obj = schema |
| 1830 | .as_object() |
| 1831 | .ok_or(KimiParameterSchemaError::InvalidSchemaNode)?; |
| 1832 | |
| 1833 | if let Some(reference) = obj.get("$ref").and_then(Value::as_str) { |
| 1834 | if !visiting_refs.insert(reference.to_string()) { |
| 1835 | return Ok(false); |
| 1836 | } |
| 1837 | let target = resolve_mfjs_reference(document, reference) |
| 1838 | .ok_or(KimiParameterSchemaError::InvalidReference)?; |
| 1839 | let terminates = |
| 1840 | mfjs_schema_can_terminate(target, document, visiting_refs, structural_depth + 1)?; |
| 1841 | visiting_refs.remove(reference); |
| 1842 | return Ok(terminates); |
| 1843 | } |
| 1844 | |
| 1845 | if let Some(branches) = obj.get("anyOf").and_then(Value::as_array) { |
| 1846 | for branch in branches { |
| 1847 | let mut branch_refs = visiting_refs.clone(); |
| 1848 | if mfjs_schema_can_terminate(branch, document, &mut branch_refs, structural_depth + 1)? |
| 1849 | { |
| 1850 | return Ok(true); |
| 1851 | } |
| 1852 | } |
| 1853 | return Ok(false); |
| 1854 | } |
| 1855 | |
| 1856 | match obj.get("type").and_then(Value::as_str) { |
| 1857 | Some("object") => { |
| 1858 | let Some(required) = obj.get("required").and_then(Value::as_array) else { |
| 1859 | return Ok(true); |
| 1860 | }; |
| 1861 | if required.is_empty() { |
| 1862 | return Ok(true); |
| 1863 | } |
| 1864 | let properties = obj |
| 1865 | .get("properties") |
| 1866 | .and_then(Value::as_object) |
| 1867 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1868 | for name in required { |
| 1869 | let property = name |
| 1870 | .as_str() |
| 1871 | .and_then(|name| properties.get(name)) |
| 1872 | .ok_or(KimiParameterSchemaError::InvalidKeywordValue)?; |
| 1873 | let mut property_refs = visiting_refs.clone(); |
| 1874 | if !mfjs_schema_can_terminate( |
| 1875 | property, |
| 1876 | document, |
| 1877 | &mut property_refs, |
| 1878 | structural_depth + 1, |
| 1879 | )? { |
| 1880 | return Ok(false); |
| 1881 | } |
| 1882 | } |
| 1883 | Ok(true) |
| 1884 | } |
| 1885 | Some("array") => { |
| 1886 | if obj.get("minItems").and_then(Value::as_u64).unwrap_or(0) == 0 { |
| 1887 | return Ok(true); |
| 1888 | } |
| 1889 | let items = obj |
| 1890 | .get("items") |
| 1891 | .ok_or(KimiParameterSchemaError::InvalidSchemaNode)?; |
| 1892 | mfjs_schema_can_terminate(items, document, visiting_refs, structural_depth + 1) |
| 1893 | } |
| 1894 | Some(schema_type) if is_mfjs_concrete_type(schema_type) => Ok(true), |
| 1895 | _ => Err(KimiParameterSchemaError::MissingType), |
| 1896 | } |
| 1897 | } |
| 1898 | |
| 1899 | fn validate_mfjs_expanded_depth( |
| 1900 | schema: &Value, |
| 1901 | document: &Value, |
| 1902 | property_depth: usize, |
| 1903 | visiting_refs: &mut HashSet<String>, |
| 1904 | structural_depth: usize, |
| 1905 | ) -> Result<(), KimiParameterSchemaError> { |
| 1906 | if property_depth > MFJS_MAX_OBJECT_DEPTH || structural_depth > MFJS_MAX_STRUCTURAL_DEPTH { |
| 1907 | return Err(KimiParameterSchemaError::ResourceLimitExceeded); |
| 1908 | } |
| 1909 | let obj = schema |
| 1910 | .as_object() |
| 1911 | .ok_or(KimiParameterSchemaError::InvalidSchemaNode)?; |
| 1912 | |
| 1913 | if let Some(reference) = obj.get("$ref").and_then(Value::as_str) { |
| 1914 | if !visiting_refs.insert(reference.to_string()) { |
| 1915 | return Ok(()); |
| 1916 | } |
| 1917 | let target = resolve_mfjs_reference(document, reference) |
| 1918 | .ok_or(KimiParameterSchemaError::InvalidReference)?; |
| 1919 | let result = validate_mfjs_expanded_depth( |
| 1920 | target, |
| 1921 | document, |
| 1922 | property_depth, |
| 1923 | visiting_refs, |
| 1924 | structural_depth + 1, |
| 1925 | ); |
| 1926 | visiting_refs.remove(reference); |
| 1927 | return result; |
| 1928 | } |
| 1929 | |
| 1930 | if let Some(properties) = obj.get("properties").and_then(Value::as_object) { |
| 1931 | for property in properties.values() { |
| 1932 | validate_mfjs_expanded_depth( |
| 1933 | property, |
| 1934 | document, |
| 1935 | property_depth + 1, |
| 1936 | visiting_refs, |
| 1937 | structural_depth + 1, |
| 1938 | )?; |
| 1939 | } |
| 1940 | } |
| 1941 | if let Some(items) = obj.get("items") { |
| 1942 | validate_mfjs_expanded_depth( |
| 1943 | items, |
| 1944 | document, |
| 1945 | property_depth, |
| 1946 | visiting_refs, |
| 1947 | structural_depth + 1, |
| 1948 | )?; |
| 1949 | } |
| 1950 | if let Some(additional) = obj |
| 1951 | .get("additionalProperties") |
| 1952 | .filter(|additional| additional.is_object()) |
| 1953 | { |
| 1954 | validate_mfjs_expanded_depth( |
| 1955 | additional, |
| 1956 | document, |
| 1957 | property_depth, |
| 1958 | visiting_refs, |
| 1959 | structural_depth + 1, |
| 1960 | )?; |
| 1961 | } |
| 1962 | if let Some(branches) = obj.get("anyOf").and_then(Value::as_array) { |
| 1963 | for branch in branches { |
| 1964 | validate_mfjs_expanded_depth( |
| 1965 | branch, |
| 1966 | document, |
| 1967 | property_depth, |
| 1968 | visiting_refs, |
| 1969 | structural_depth + 1, |
| 1970 | )?; |
| 1971 | } |
| 1972 | } |
| 1973 | Ok(()) |
| 1974 | } |
| 1975 | |
| 1976 | #[cfg(test)] |
| 1977 | mod kimi_tests { |
| 1978 | use super::*; |
| 1979 | use crate::tools::apply_patch::ApplyPatchTool; |
| 1980 | use crate::tools::spec::ToolSpec; |
| 1981 | use serde_json::json; |
| 1982 | |
| 1983 | #[test] |
| 1984 | fn kimi_sanitize_pushes_type_into_anyof_items() { |
| 1985 | let mut schema = json!({ |
| 1986 | "type": "object", |
| 1987 | "properties": { |
| 1988 | "handle": { |
| 1989 | "type": "object", |
| 1990 | "anyOf": [ |
| 1991 | {"type": "string"}, |
| 1992 | {"type": "null"} |
| 1993 | ] |
| 1994 | } |
| 1995 | } |
| 1996 | }); |
| 1997 | sanitize_for_kimi(&mut schema); |
| 1998 | let handle = &schema["properties"]["handle"]; |
| 1999 | assert!( |
| 2000 | !handle.as_object().unwrap().contains_key("type"), |
| 2001 | "root type should be removed" |
| 2002 | ); |
| 2003 | let any_of = handle["anyOf"].as_array().unwrap(); |
| 2004 | assert_eq!(any_of[0]["type"], "string"); |
| 2005 | assert_eq!(any_of[1]["type"], "null"); |
| 2006 | } |
| 2007 | |
| 2008 | #[test] |
| 2009 | fn kimi_sanitize_injects_missing_anyof_item_types() { |
| 2010 | let mut schema = json!({ |
| 2011 | "type": "object", |
| 2012 | "anyOf": [ |
| 2013 | {"properties": {"path": {"type": "string"}}}, |
| 2014 | {"required": ["url"], "properties": {"url": {"type": "string"}}} |
| 2015 | ] |
| 2016 | }); |
| 2017 | |
| 2018 | sanitize_for_kimi(&mut schema); |
| 2019 | |
| 2020 | assert!( |
| 2021 | !schema.as_object().unwrap().contains_key("type"), |
| 2022 | "parent type should be removed" |
| 2023 | ); |
| 2024 | let any_of = schema["anyOf"].as_array().unwrap(); |
| 2025 | assert_eq!(any_of[0]["type"], "object"); |
| 2026 | assert_eq!(any_of[1]["type"], "object"); |
| 2027 | } |
| 2028 | |
| 2029 | #[test] |
| 2030 | fn kimi_sanitize_preserves_type_injected_into_nested_anyof_item() { |
| 2031 | let mut schema = json!({ |
| 2032 | "type": "object", |
| 2033 | "anyOf": [ |
| 2034 | { |
| 2035 | "anyOf": [ |
| 2036 | {"properties": {"path": {"type": "string"}}} |
| 2037 | ] |
| 2038 | } |
| 2039 | ] |
| 2040 | }); |
| 2041 | |
| 2042 | sanitize_for_kimi(&mut schema); |
| 2043 | |
| 2044 | let outer_item = &schema["anyOf"][0]; |
| 2045 | assert_eq!(outer_item["type"], "object"); |
| 2046 | assert!( |
| 2047 | !schema.as_object().unwrap().contains_key("type"), |
| 2048 | "outer parent type should be removed" |
| 2049 | ); |
| 2050 | } |
| 2051 | |
| 2052 | #[test] |
| 2053 | fn kimi_sanitize_leaves_pure_object_untouched() { |
| 2054 | let original = json!({ |
| 2055 | "type": "object", |
| 2056 | "properties": {"x": {"type": "string"}}, |
| 2057 | "required": ["x"] |
| 2058 | }); |
| 2059 | let mut schema = original.clone(); |
| 2060 | sanitize_for_kimi(&mut schema); |
| 2061 | assert_eq!(schema, original); |
| 2062 | } |
| 2063 | |
| 2064 | #[test] |
| 2065 | fn kimi_parameters_add_type_to_empty_root() { |
| 2066 | let mut schema = json!({}); |
| 2067 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2068 | assert_eq!(schema, json!({"type": "object", "properties": {}})); |
| 2069 | } |
| 2070 | |
| 2071 | #[test] |
| 2072 | fn kimi_parameters_add_type_to_properties_root_without_corrupting_properties_map() { |
| 2073 | let mut schema = json!({ |
| 2074 | "properties": { |
| 2075 | "path": {"type": "string"} |
| 2076 | }, |
| 2077 | "required": ["path"] |
| 2078 | }); |
| 2079 | |
| 2080 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2081 | |
| 2082 | assert_eq!(schema["type"], "object"); |
| 2083 | assert_eq!(schema["properties"]["path"]["type"], "string"); |
| 2084 | assert!(schema["properties"].get("type").is_none()); |
| 2085 | } |
| 2086 | |
| 2087 | // Function parameters must end as a plain object root. Composition stays |
| 2088 | // available only in valid nested anyOf positions. |
| 2089 | |
| 2090 | #[test] |
| 2091 | fn kimi_parameters_add_type_to_anyof_root() { |
| 2092 | let mut schema = json!({ |
| 2093 | "anyOf": [ |
| 2094 | {"type": "object", "properties": {"path": {"type": "string"}}}, |
| 2095 | {"type": "null"} |
| 2096 | ] |
| 2097 | }); |
| 2098 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2099 | assert_eq!(schema["type"], "object"); |
| 2100 | assert!(schema.get("anyOf").is_none()); |
| 2101 | assert_eq!(schema["properties"]["path"]["type"], "string"); |
| 2102 | } |
| 2103 | |
| 2104 | #[test] |
| 2105 | fn kimi_parameters_add_type_to_allof_root() { |
| 2106 | let mut schema = json!({ |
| 2107 | "allOf": [ |
| 2108 | {"type": "object", "properties": {"name": {"type": "string"}}} |
| 2109 | ] |
| 2110 | }); |
| 2111 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2112 | assert_eq!(schema["type"], "object"); |
| 2113 | assert!(schema.get("allOf").is_none()); |
| 2114 | assert_eq!(schema["properties"]["name"]["type"], "string"); |
| 2115 | } |
| 2116 | |
| 2117 | #[test] |
| 2118 | fn kimi_parameters_add_type_to_oneof_root() { |
| 2119 | let mut schema = json!({ |
| 2120 | "oneOf": [ |
| 2121 | {"type": "object", "properties": {"id": {"type": "integer"}}}, |
| 2122 | {"type": "object", "properties": {"name": {"type": "string"}}} |
| 2123 | ] |
| 2124 | }); |
| 2125 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2126 | assert_eq!(schema["type"], "object"); |
| 2127 | assert!(schema.get("oneOf").is_none()); |
| 2128 | assert_eq!(schema["properties"]["id"]["type"], "integer"); |
| 2129 | assert_eq!(schema["properties"]["name"]["type"], "string"); |
| 2130 | } |
| 2131 | |
| 2132 | #[test] |
| 2133 | fn kimi_parameters_flattens_actual_apply_patch_root_and_returns_constraint_note() { |
| 2134 | let mut schema = ApplyPatchTool.input_schema(); |
| 2135 | |
| 2136 | let note = sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2137 | |
| 2138 | assert_eq!(schema["type"], "object"); |
| 2139 | assert!(schema.get("oneOf").is_none()); |
| 2140 | assert!(schema.get("anyOf").is_none()); |
| 2141 | assert!(schema.get("allOf").is_none()); |
| 2142 | assert_eq!(schema["properties"]["patch"]["type"], "string"); |
| 2143 | assert_eq!(schema["properties"]["replace"]["type"], "array"); |
| 2144 | assert_eq!(schema["properties"]["changes"]["type"], "array"); |
| 2145 | assert_eq!( |
| 2146 | note.as_deref(), |
| 2147 | Some( |
| 2148 | "Exactly one of these parameter groups must be provided: `changes` | `patch` | `replace`." |
| 2149 | ) |
| 2150 | ); |
| 2151 | } |
| 2152 | |
| 2153 | #[test] |
| 2154 | fn kimi_parameters_preserves_nested_anyof_branches() { |
| 2155 | let mut schema = json!({ |
| 2156 | "type": "object", |
| 2157 | "properties": { |
| 2158 | "selector": { |
| 2159 | "type": "object", |
| 2160 | "anyOf": [ |
| 2161 | {"properties": {"path": {"type": "string"}}}, |
| 2162 | {"properties": {"id": {"type": "integer"}}} |
| 2163 | ] |
| 2164 | } |
| 2165 | } |
| 2166 | }); |
| 2167 | |
| 2168 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2169 | |
| 2170 | assert_eq!(schema["type"], "object"); |
| 2171 | let selector = &schema["properties"]["selector"]; |
| 2172 | assert!(selector.get("type").is_none()); |
| 2173 | let branches = selector["anyOf"].as_array().unwrap(); |
| 2174 | assert_eq!(branches.len(), 2); |
| 2175 | assert!(branches.iter().all(|branch| branch["type"] == "object")); |
| 2176 | } |
| 2177 | |
| 2178 | #[test] |
| 2179 | fn kimi_parameters_converts_nested_oneof_to_supported_anyof() { |
| 2180 | let mut schema = json!({ |
| 2181 | "type": "object", |
| 2182 | "properties": { |
| 2183 | "selector": { |
| 2184 | "type": "object", |
| 2185 | "oneOf": [ |
| 2186 | {"properties": {"path": {"type": "string"}}}, |
| 2187 | {"properties": {"id": {"type": "integer"}}} |
| 2188 | ] |
| 2189 | } |
| 2190 | } |
| 2191 | }); |
| 2192 | |
| 2193 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2194 | |
| 2195 | let selector = &schema["properties"]["selector"]; |
| 2196 | assert!(selector.get("oneOf").is_none()); |
| 2197 | assert!(selector["anyOf"].is_array()); |
| 2198 | assert!(selector.get("type").is_none()); |
| 2199 | } |
| 2200 | |
| 2201 | #[test] |
| 2202 | fn kimi_parameters_restores_registry_collapsed_nullable_anyof() { |
| 2203 | let mut schema = json!({ |
| 2204 | "type": "object", |
| 2205 | "properties": { |
| 2206 | "query": { |
| 2207 | "anyOf": [ |
| 2208 | {"type": "string"}, |
| 2209 | {"type": "null"} |
| 2210 | ] |
| 2211 | } |
| 2212 | } |
| 2213 | }); |
| 2214 | |
| 2215 | // Exercise the exact two-stage production path: ToolRegistry applies |
| 2216 | // the provider-neutral pass before the Moonshot request adapter sees |
| 2217 | // the schema. |
| 2218 | sanitize(&mut schema); |
| 2219 | assert_eq!(schema["properties"]["query"]["nullable"], true); |
| 2220 | assert!(schema["properties"]["query"].get("anyOf").is_none()); |
| 2221 | |
| 2222 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2223 | |
| 2224 | let query = &schema["properties"]["query"]; |
| 2225 | assert!(query.get("nullable").is_none(), "{query}"); |
| 2226 | assert_eq!( |
| 2227 | query["anyOf"], |
| 2228 | json!([{"type": "string"}, {"type": "null"}]) |
| 2229 | ); |
| 2230 | validate_mfjs_parameters(&schema).unwrap(); |
| 2231 | } |
| 2232 | |
| 2233 | #[test] |
| 2234 | fn kimi_parameters_recursively_translates_safe_const_to_enum() { |
| 2235 | let mut schema = json!({ |
| 2236 | "type": "object", |
| 2237 | "properties": { |
| 2238 | "envelope": { |
| 2239 | "type": "object", |
| 2240 | "properties": { |
| 2241 | "items": { |
| 2242 | "type": "array", |
| 2243 | "items": { |
| 2244 | "type": "object", |
| 2245 | "properties": { |
| 2246 | "kind": {"type": "string", "const": "var_handle"} |
| 2247 | }, |
| 2248 | "required": ["kind"] |
| 2249 | } |
| 2250 | } |
| 2251 | } |
| 2252 | } |
| 2253 | } |
| 2254 | }); |
| 2255 | |
| 2256 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2257 | |
| 2258 | let kind = schema |
| 2259 | .pointer("/properties/envelope/properties/items/items/properties/kind") |
| 2260 | .expect("nested kind schema"); |
| 2261 | assert!(kind.get("const").is_none(), "{kind}"); |
| 2262 | assert_eq!(kind["enum"], json!(["var_handle"])); |
| 2263 | } |
| 2264 | |
| 2265 | #[test] |
| 2266 | fn kimi_parameters_rejects_unsafe_const_without_mutating_or_leaking() { |
| 2267 | let mut schema = json!({ |
| 2268 | "type": "object", |
| 2269 | "properties": { |
| 2270 | "private-toggle-8172": {"type": "boolean", "const": true} |
| 2271 | } |
| 2272 | }); |
| 2273 | let original = schema.clone(); |
| 2274 | |
| 2275 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2276 | |
| 2277 | assert_eq!(error, KimiParameterSchemaError::UnsupportedConstLiteral); |
| 2278 | assert!(!error.to_string().contains("private-toggle-8172")); |
| 2279 | assert_eq!(schema, original, "a rejected schema must remain reusable"); |
| 2280 | } |
| 2281 | |
| 2282 | #[test] |
| 2283 | fn kimi_parameters_validator_fails_closed_without_echoing_schema_values() { |
| 2284 | let mut schema = json!({ |
| 2285 | "type": "object", |
| 2286 | "properties": { |
| 2287 | "private-field-4921": { |
| 2288 | "type": "string", |
| 2289 | "pattern": "private-pattern-value-7395" |
| 2290 | } |
| 2291 | } |
| 2292 | }); |
| 2293 | let original = schema.clone(); |
| 2294 | |
| 2295 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2296 | let diagnostic = error.to_string(); |
| 2297 | |
| 2298 | assert_eq!(error, KimiParameterSchemaError::UnsupportedKeyword); |
| 2299 | assert!(!diagnostic.contains("private-field-4921")); |
| 2300 | assert!(!diagnostic.contains("private-pattern-value-7395")); |
| 2301 | assert_eq!(schema, original, "failed validation must be transactional"); |
| 2302 | } |
| 2303 | |
| 2304 | #[test] |
| 2305 | fn kimi_parameters_rejects_untyped_schema_and_nullable_transactionally() { |
| 2306 | for (mut schema, expected, sentinels) in [ |
| 2307 | ( |
| 2308 | json!({ |
| 2309 | "type": "object", |
| 2310 | "properties": { |
| 2311 | "private-missing-type-1207": { |
| 2312 | "description": "private-description-1208" |
| 2313 | } |
| 2314 | } |
| 2315 | }), |
| 2316 | KimiParameterSchemaError::MissingType, |
| 2317 | ["private-missing-type-1207", "private-description-1208"], |
| 2318 | ), |
| 2319 | ( |
| 2320 | json!({ |
| 2321 | "type": "object", |
| 2322 | "properties": { |
| 2323 | "private-nullable-1209": { |
| 2324 | "nullable": true, |
| 2325 | "description": "private-nullable-description-1210" |
| 2326 | } |
| 2327 | } |
| 2328 | }), |
| 2329 | KimiParameterSchemaError::InvalidNullable, |
| 2330 | ["private-nullable-1209", "private-nullable-description-1210"], |
| 2331 | ), |
| 2332 | ] { |
| 2333 | let original = schema.clone(); |
| 2334 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2335 | assert_eq!(error, expected); |
| 2336 | for sentinel in sentinels { |
| 2337 | assert!(!error.to_string().contains(sentinel)); |
| 2338 | } |
| 2339 | assert_eq!(schema, original, "rejection must be transactional"); |
| 2340 | } |
| 2341 | } |
| 2342 | |
| 2343 | #[test] |
| 2344 | fn kimi_parameters_infers_safe_types_for_untyped_const() { |
| 2345 | let mut schema = json!({ |
| 2346 | "type": "object", |
| 2347 | "properties": { |
| 2348 | "kind": {"const": "var_handle"}, |
| 2349 | "count": {"const": 7}, |
| 2350 | "ratio": {"const": 1.25} |
| 2351 | } |
| 2352 | }); |
| 2353 | |
| 2354 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2355 | |
| 2356 | assert_eq!(schema["properties"]["kind"]["type"], "string"); |
| 2357 | assert_eq!(schema["properties"]["kind"]["enum"], json!(["var_handle"])); |
| 2358 | assert_eq!(schema["properties"]["count"]["type"], "integer"); |
| 2359 | assert_eq!(schema["properties"]["count"]["enum"], json!([7])); |
| 2360 | assert_eq!(schema["properties"]["ratio"]["type"], "number"); |
| 2361 | assert_eq!(schema["properties"]["ratio"]["enum"], json!([1.25])); |
| 2362 | } |
| 2363 | |
| 2364 | #[test] |
| 2365 | fn kimi_parameters_allows_only_the_documented_empty_schema_exception() { |
| 2366 | let mut schema = json!({ |
| 2367 | "type": "object", |
| 2368 | "additionalProperties": {} |
| 2369 | }); |
| 2370 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2371 | assert_eq!(schema["additionalProperties"], json!({})); |
| 2372 | |
| 2373 | let mut invalid = json!({ |
| 2374 | "type": "object", |
| 2375 | "properties": {"value": {}} |
| 2376 | }); |
| 2377 | assert_eq!( |
| 2378 | sanitize_for_kimi_parameters(&mut invalid).unwrap_err(), |
| 2379 | KimiParameterSchemaError::MissingType |
| 2380 | ); |
| 2381 | } |
| 2382 | |
| 2383 | #[test] |
| 2384 | fn kimi_parameters_rejects_required_direct_and_mutual_recursion() { |
| 2385 | let fixtures = [ |
| 2386 | json!({ |
| 2387 | "type": "object", |
| 2388 | "properties": { |
| 2389 | "private-root-node-2201": {"$ref": "#/$defs/private-node-2202"} |
| 2390 | }, |
| 2391 | "required": ["private-root-node-2201"], |
| 2392 | "$defs": { |
| 2393 | "private-node-2202": { |
| 2394 | "type": "object", |
| 2395 | "properties": { |
| 2396 | "private-next-2203": {"$ref": "#/$defs/private-node-2202"} |
| 2397 | }, |
| 2398 | "required": ["private-next-2203"] |
| 2399 | } |
| 2400 | } |
| 2401 | }), |
| 2402 | json!({ |
| 2403 | "type": "object", |
| 2404 | "properties": { |
| 2405 | "private-root-a-2204": {"$ref": "#/$defs/private-a-2205"} |
| 2406 | }, |
| 2407 | "required": ["private-root-a-2204"], |
| 2408 | "$defs": { |
| 2409 | "private-a-2205": { |
| 2410 | "type": "object", |
| 2411 | "properties": { |
| 2412 | "private-to-b-2206": {"$ref": "#/$defs/private-b-2207"} |
| 2413 | }, |
| 2414 | "required": ["private-to-b-2206"] |
| 2415 | }, |
| 2416 | "private-b-2207": { |
| 2417 | "type": "object", |
| 2418 | "properties": { |
| 2419 | "private-to-a-2208": {"$ref": "#/$defs/private-a-2205"} |
| 2420 | }, |
| 2421 | "required": ["private-to-a-2208"] |
| 2422 | } |
| 2423 | } |
| 2424 | }), |
| 2425 | ]; |
| 2426 | |
| 2427 | for mut schema in fixtures { |
| 2428 | let original = schema.clone(); |
| 2429 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2430 | assert_eq!(error, KimiParameterSchemaError::NonTerminatingReference); |
| 2431 | for sentinel in ["private-root", "private-node", "private-to"] { |
| 2432 | assert!(!error.to_string().contains(sentinel)); |
| 2433 | } |
| 2434 | assert_eq!(schema, original, "recursive rejection must be atomic"); |
| 2435 | } |
| 2436 | } |
| 2437 | |
| 2438 | #[test] |
| 2439 | fn kimi_parameters_preserves_optional_and_nullable_recursive_termination() { |
| 2440 | let mut optional = json!({ |
| 2441 | "type": "object", |
| 2442 | "properties": { |
| 2443 | "node": {"$ref": "#/$defs/Node"} |
| 2444 | }, |
| 2445 | "$defs": { |
| 2446 | "Node": { |
| 2447 | "type": "object", |
| 2448 | "properties": { |
| 2449 | "next": {"$ref": "#/$defs/Node"} |
| 2450 | }, |
| 2451 | "required": ["next"] |
| 2452 | } |
| 2453 | } |
| 2454 | }); |
| 2455 | sanitize_for_kimi_parameters(&mut optional).unwrap(); |
| 2456 | |
| 2457 | let mut nullable = json!({ |
| 2458 | "type": "object", |
| 2459 | "properties": { |
| 2460 | "node": { |
| 2461 | "anyOf": [ |
| 2462 | {"$ref": "#/$defs/Node"}, |
| 2463 | {"type": "null"} |
| 2464 | ] |
| 2465 | } |
| 2466 | }, |
| 2467 | "required": ["node"], |
| 2468 | "$defs": { |
| 2469 | "Node": { |
| 2470 | "type": "object", |
| 2471 | "properties": { |
| 2472 | "next": { |
| 2473 | "anyOf": [ |
| 2474 | {"$ref": "#/$defs/Node"}, |
| 2475 | {"type": "null"} |
| 2476 | ] |
| 2477 | } |
| 2478 | }, |
| 2479 | "required": ["next"] |
| 2480 | } |
| 2481 | } |
| 2482 | }); |
| 2483 | sanitize_for_kimi_parameters(&mut nullable).unwrap(); |
| 2484 | } |
| 2485 | |
| 2486 | #[test] |
| 2487 | fn kimi_parameters_enforces_anyof_and_aggregate_resource_limits() { |
| 2488 | let mut too_many_branches = json!({ |
| 2489 | "type": "object", |
| 2490 | "properties": { |
| 2491 | "choice": { |
| 2492 | "anyOf": (0..11) |
| 2493 | .map(|_| json!({"type": "string"})) |
| 2494 | .collect::<Vec<_>>() |
| 2495 | } |
| 2496 | } |
| 2497 | }); |
| 2498 | assert_eq!( |
| 2499 | sanitize_for_kimi_parameters(&mut too_many_branches).unwrap_err(), |
| 2500 | KimiParameterSchemaError::ResourceLimitExceeded |
| 2501 | ); |
| 2502 | |
| 2503 | let string_property = || json!({"type": "string"}); |
| 2504 | let mut root_properties = Map::new(); |
| 2505 | for index in 0..51 { |
| 2506 | root_properties.insert(format!("root_{index}"), string_property()); |
| 2507 | } |
| 2508 | let mut definition_properties = Map::new(); |
| 2509 | for index in 0..50 { |
| 2510 | definition_properties.insert(format!("definition_{index}"), string_property()); |
| 2511 | } |
| 2512 | let mut too_many_properties = json!({ |
| 2513 | "type": "object", |
| 2514 | "properties": Value::Object(root_properties), |
| 2515 | "$defs": { |
| 2516 | "Holder": { |
| 2517 | "type": "object", |
| 2518 | "properties": Value::Object(definition_properties) |
| 2519 | } |
| 2520 | } |
| 2521 | }); |
| 2522 | assert_eq!( |
| 2523 | sanitize_for_kimi_parameters(&mut too_many_properties).unwrap_err(), |
| 2524 | KimiParameterSchemaError::ResourceLimitExceeded |
| 2525 | ); |
| 2526 | |
| 2527 | let enum_values = |start: usize, count: usize| { |
| 2528 | Value::Array((start..start + count).map(|value| json!(value)).collect()) |
| 2529 | }; |
| 2530 | let mut too_many_enum_values = json!({ |
| 2531 | "type": "object", |
| 2532 | "properties": { |
| 2533 | "first": {"type": "integer", "enum": enum_values(0, 250)}, |
| 2534 | "second": {"type": "integer", "enum": enum_values(250, 251)} |
| 2535 | } |
| 2536 | }); |
| 2537 | assert_eq!( |
| 2538 | sanitize_for_kimi_parameters(&mut too_many_enum_values).unwrap_err(), |
| 2539 | KimiParameterSchemaError::ResourceLimitExceeded |
| 2540 | ); |
| 2541 | } |
| 2542 | |
| 2543 | #[test] |
| 2544 | fn kimi_parameters_enforces_depth_and_enum_text_limits() { |
| 2545 | let mut too_deep = json!({"type": "string"}); |
| 2546 | for index in (0..6).rev() { |
| 2547 | let mut properties = Map::new(); |
| 2548 | properties.insert(format!("level_{index}"), too_deep); |
| 2549 | too_deep = json!({ |
| 2550 | "type": "object", |
| 2551 | "properties": Value::Object(properties) |
| 2552 | }); |
| 2553 | } |
| 2554 | assert_eq!( |
| 2555 | sanitize_for_kimi_parameters(&mut too_deep).unwrap_err(), |
| 2556 | KimiParameterSchemaError::ResourceLimitExceeded |
| 2557 | ); |
| 2558 | |
| 2559 | let long_values = Value::Array( |
| 2560 | (0..251) |
| 2561 | .map(|index| Value::String(format!("private-enum-{index:04}-xxxxxxxxxxxxxx"))) |
| 2562 | .collect(), |
| 2563 | ); |
| 2564 | let mut too_much_enum_text = json!({ |
| 2565 | "type": "object", |
| 2566 | "properties": { |
| 2567 | "choice": {"type": "string", "enum": long_values} |
| 2568 | } |
| 2569 | }); |
| 2570 | assert_eq!( |
| 2571 | sanitize_for_kimi_parameters(&mut too_much_enum_text).unwrap_err(), |
| 2572 | KimiParameterSchemaError::ResourceLimitExceeded |
| 2573 | ); |
| 2574 | } |
| 2575 | |
| 2576 | #[test] |
| 2577 | fn kimi_parameters_validates_default_type_and_placement_without_leaks() { |
| 2578 | let mut valid = json!({ |
| 2579 | "type": "object", |
| 2580 | "properties": { |
| 2581 | "enabled": {"type": "boolean", "default": true}, |
| 2582 | "count": {"type": "integer", "default": 3}, |
| 2583 | "ratio": {"type": "number", "default": 1.5}, |
| 2584 | "label": {"type": "string", "default": "default-label"}, |
| 2585 | "empty": {"type": "null", "default": null} |
| 2586 | } |
| 2587 | }); |
| 2588 | sanitize_for_kimi_parameters(&mut valid).unwrap(); |
| 2589 | |
| 2590 | for (mut invalid, expected) in [ |
| 2591 | ( |
| 2592 | json!({ |
| 2593 | "type": "object", |
| 2594 | "properties": { |
| 2595 | "private-default-3301": { |
| 2596 | "type": "integer", |
| 2597 | "default": "private-default-value-3302" |
| 2598 | } |
| 2599 | } |
| 2600 | }), |
| 2601 | KimiParameterSchemaError::InvalidDefault, |
| 2602 | ), |
| 2603 | ( |
| 2604 | json!({ |
| 2605 | "type": "object", |
| 2606 | "properties": { |
| 2607 | "private-untyped-default-3303": {"default": 1} |
| 2608 | } |
| 2609 | }), |
| 2610 | KimiParameterSchemaError::MissingType, |
| 2611 | ), |
| 2612 | ( |
| 2613 | json!({ |
| 2614 | "type": "object", |
| 2615 | "properties": { |
| 2616 | "private-object-default-3304": { |
| 2617 | "type": "object", |
| 2618 | "default": {} |
| 2619 | } |
| 2620 | } |
| 2621 | }), |
| 2622 | KimiParameterSchemaError::InvalidKeywordValue, |
| 2623 | ), |
| 2624 | ] { |
| 2625 | let original = invalid.clone(); |
| 2626 | let error = sanitize_for_kimi_parameters(&mut invalid).unwrap_err(); |
| 2627 | assert_eq!(error, expected); |
| 2628 | assert!(!error.to_string().contains("private-default")); |
| 2629 | assert_eq!(invalid, original); |
| 2630 | } |
| 2631 | } |
| 2632 | |
| 2633 | #[test] |
| 2634 | fn kimi_parameters_rejects_inverted_and_fractional_integer_bounds() { |
| 2635 | for mut schema in [ |
| 2636 | json!({ |
| 2637 | "type": "object", |
| 2638 | "properties": { |
| 2639 | "value": {"type": "string", "minLength": 5, "maxLength": 4} |
| 2640 | } |
| 2641 | }), |
| 2642 | json!({ |
| 2643 | "type": "object", |
| 2644 | "properties": { |
| 2645 | "value": {"type": "array", "minItems": 3, "maxItems": 2} |
| 2646 | } |
| 2647 | }), |
| 2648 | json!({ |
| 2649 | "type": "object", |
| 2650 | "properties": { |
| 2651 | "value": {"type": "number", "minimum": 10, "maximum": 9} |
| 2652 | } |
| 2653 | }), |
| 2654 | json!({ |
| 2655 | "type": "object", |
| 2656 | "properties": { |
| 2657 | "value": {"type": "integer", "minimum": 1.5} |
| 2658 | } |
| 2659 | }), |
| 2660 | json!({ |
| 2661 | "type": "object", |
| 2662 | "properties": { |
| 2663 | "value": {"type": "integer", "maximum": 2.5} |
| 2664 | } |
| 2665 | }), |
| 2666 | ] { |
| 2667 | assert_eq!( |
| 2668 | sanitize_for_kimi_parameters(&mut schema).unwrap_err(), |
| 2669 | KimiParameterSchemaError::InvalidRange |
| 2670 | ); |
| 2671 | } |
| 2672 | } |
| 2673 | |
| 2674 | #[test] |
| 2675 | fn kimi_parameters_inlines_valid_internal_object_root_ref() { |
| 2676 | let mut schema = json!({ |
| 2677 | "$ref": "#/$defs/FileArgs", |
| 2678 | "$defs": { |
| 2679 | "FileArgs": { |
| 2680 | "type": "object", |
| 2681 | "properties": {"path": {"type": "string"}}, |
| 2682 | "required": ["path"] |
| 2683 | } |
| 2684 | }, |
| 2685 | "description": "File arguments" |
| 2686 | }); |
| 2687 | |
| 2688 | sanitize_for_kimi_parameters(&mut schema).unwrap(); |
| 2689 | |
| 2690 | assert_eq!(schema["type"], "object"); |
| 2691 | assert_eq!(schema["properties"]["path"]["type"], "string"); |
| 2692 | assert_eq!(schema["required"], json!(["path"])); |
| 2693 | assert_eq!(schema["description"], "File arguments"); |
| 2694 | assert!(schema["$defs"].is_object()); |
| 2695 | assert!(schema.get("$ref").is_none()); |
| 2696 | assert!(schema.get("allOf").is_none()); |
| 2697 | } |
| 2698 | |
| 2699 | #[test] |
| 2700 | fn kimi_parameters_rejects_unresolved_root_ref_without_leaking_it() { |
| 2701 | let mut schema = json!({ |
| 2702 | "$ref": "#/$defs/private-schema-name-9217", |
| 2703 | "$defs": {} |
| 2704 | }); |
| 2705 | let original = schema.clone(); |
| 2706 | |
| 2707 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2708 | |
| 2709 | assert_eq!(error, KimiParameterSchemaError::UnresolvedRootReference); |
| 2710 | assert!(!error.to_string().contains("private-schema-name-9217")); |
| 2711 | assert_eq!(schema, original, "a rejected schema must never be emitted"); |
| 2712 | } |
| 2713 | |
| 2714 | #[test] |
| 2715 | fn kimi_parameters_rejects_non_object_root_ref_without_leaking_it() { |
| 2716 | let mut schema = json!({ |
| 2717 | "$ref": "#/$defs/private-scalar-name-4831", |
| 2718 | "$defs": { |
| 2719 | "private-scalar-name-4831": {"type": "string"} |
| 2720 | } |
| 2721 | }); |
| 2722 | let original = schema.clone(); |
| 2723 | |
| 2724 | let error = sanitize_for_kimi_parameters(&mut schema).unwrap_err(); |
| 2725 | |
| 2726 | assert_eq!(error, KimiParameterSchemaError::ReferencedRootMustBeObject); |
| 2727 | assert!(!error.to_string().contains("private-scalar-name-4831")); |
| 2728 | assert_eq!(schema, original, "a rejected schema must never be emitted"); |
| 2729 | } |
| 2730 | } |
| 2731 |