| 1 | //! Structured data validation tool: `validate_data`. |
| 2 | //! |
| 3 | //! Validates JSON or TOML from inline content or a workspace file path and |
| 4 | //! returns parser errors with lightweight metadata. |
| 5 | |
| 6 | use std::fs; |
| 7 | |
| 8 | use async_trait::async_trait; |
| 9 | use serde_json::{Value, json}; |
| 10 | |
| 11 | use super::spec::{ |
| 12 | ApprovalRequirement, ToolCapability, ToolContext, ToolError, ToolResult, ToolSpec, optional_str, |
| 13 | }; |
| 14 | |
| 15 | /// Tool for validating JSON/TOML configuration data. |
| 16 | pub struct ValidateDataTool; |
| 17 | |
| 18 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 19 | enum DataFormat { |
| 20 | Auto, |
| 21 | Json, |
| 22 | Toml, |
| 23 | } |
| 24 | |
| 25 | impl DataFormat { |
| 26 | fn from_input(raw: Option<&str>) -> Result<Self, ToolError> { |
| 27 | let format = raw.unwrap_or("auto"); |
| 28 | match format { |
| 29 | "auto" => Ok(Self::Auto), |
| 30 | "json" => Ok(Self::Json), |
| 31 | "toml" => Ok(Self::Toml), |
| 32 | _ => Err(ToolError::invalid_input(format!( |
| 33 | "Unsupported format '{format}'. Expected one of: auto, json, toml" |
| 34 | ))), |
| 35 | } |
| 36 | } |
| 37 | |
| 38 | fn as_str(self) -> &'static str { |
| 39 | match self { |
| 40 | Self::Auto => "auto", |
| 41 | Self::Json => "json", |
| 42 | Self::Toml => "toml", |
| 43 | } |
| 44 | } |
| 45 | } |
| 46 | |
| 47 | #[async_trait] |
| 48 | impl ToolSpec for ValidateDataTool { |
| 49 | fn name(&self) -> &'static str { |
| 50 | "validate_data" |
| 51 | } |
| 52 | |
| 53 | fn description(&self) -> &'static str { |
| 54 | "Validate JSON or TOML content from inline input or a workspace file." |
| 55 | } |
| 56 | |
| 57 | fn input_schema(&self) -> Value { |
| 58 | json!({ |
| 59 | "type": "object", |
| 60 | "properties": { |
| 61 | "path": { |
| 62 | "type": "string", |
| 63 | "description": "Optional path to a file within the workspace." |
| 64 | }, |
| 65 | "content": { |
| 66 | "type": "string", |
| 67 | "description": "Optional inline content to validate." |
| 68 | }, |
| 69 | "format": { |
| 70 | "type": "string", |
| 71 | "enum": ["auto", "json", "toml"], |
| 72 | "default": "auto", |
| 73 | "description": "Validation format. 'auto' infers from extension then falls back to trying both." |
| 74 | } |
| 75 | }, |
| 76 | "additionalProperties": false |
| 77 | }) |
| 78 | } |
| 79 | |
| 80 | fn capabilities(&self) -> Vec<ToolCapability> { |
| 81 | vec![ToolCapability::ReadOnly, ToolCapability::Sandboxable] |
| 82 | } |
| 83 | |
| 84 | fn approval_requirement(&self) -> ApprovalRequirement { |
| 85 | ApprovalRequirement::Auto |
| 86 | } |
| 87 | |
| 88 | fn supports_parallel(&self) -> bool { |
| 89 | true |
| 90 | } |
| 91 | |
| 92 | async fn execute(&self, input: Value, context: &ToolContext) -> Result<ToolResult, ToolError> { |
| 93 | let path = optional_str(&input, "path")?; |
| 94 | let content = optional_str(&input, "content")?; |
| 95 | let requested_format = DataFormat::from_input(optional_str(&input, "format")?)?; |
| 96 | |
| 97 | let (source_name, raw_content, extension) = load_input_source(path, content, context)?; |
| 98 | if context.features.enabled(crate::features::Feature::DataHost) { |
| 99 | return validate_host( |
| 100 | &raw_content, |
| 101 | &source_name, |
| 102 | requested_format, |
| 103 | extension.as_deref(), |
| 104 | context, |
| 105 | ) |
| 106 | .await; |
| 107 | } |
| 108 | match requested_format { |
| 109 | DataFormat::Json => validate_json(&raw_content, &source_name), |
| 110 | DataFormat::Toml => validate_toml(&raw_content, &source_name), |
| 111 | DataFormat::Auto => validate_auto(&raw_content, &source_name, extension.as_deref()), |
| 112 | } |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | fn load_input_source( |
| 117 | path: Option<&str>, |
| 118 | content: Option<&str>, |
| 119 | context: &ToolContext, |
| 120 | ) -> Result<(String, String, Option<String>), ToolError> { |
| 121 | match (path, content) { |
| 122 | (Some(_), Some(_)) => Err(ToolError::invalid_input( |
| 123 | "Provide either 'path' or 'content', but not both.", |
| 124 | )), |
| 125 | (None, None) => Err(ToolError::missing_field("path or content")), |
| 126 | (Some(path), None) => { |
| 127 | let resolved = context.resolve_path(path)?; |
| 128 | let raw_content = if context.features.enabled(crate::features::Feature::DataHost) { |
| 129 | use std::io::Read; |
| 130 | let mut content = String::new(); |
| 131 | fs::File::open(&resolved) |
| 132 | .and_then(|file| file.take(1024 * 1024 + 1).read_to_string(&mut content)) |
| 133 | .map_err(|e| { |
| 134 | ToolError::execution_failed(format!( |
| 135 | "Failed to read {}: {e}", |
| 136 | resolved.display() |
| 137 | )) |
| 138 | })?; |
| 139 | if content.len() > 1024 * 1024 { |
| 140 | return Err(ToolError::invalid_input("Host data input exceeds 1 MiB")); |
| 141 | } |
| 142 | content |
| 143 | } else { |
| 144 | fs::read_to_string(&resolved).map_err(|e| { |
| 145 | ToolError::execution_failed(format!( |
| 146 | "Failed to read {}: {e}", |
| 147 | resolved.display() |
| 148 | )) |
| 149 | })? |
| 150 | }; |
| 151 | let extension = resolved |
| 152 | .extension() |
| 153 | .and_then(|ext| ext.to_str()) |
| 154 | .map(|s| s.to_ascii_lowercase()); |
| 155 | Ok((path.to_string(), raw_content, extension)) |
| 156 | } |
| 157 | (None, Some(content)) => Ok(("inline".to_string(), content.to_string(), None)), |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | /// Preserve the existing Rust parsers and their exact diagnostics. The host |
| 162 | /// selects the format and owns summary/result presentation from minimal facts. |
| 163 | async fn validate_host( |
| 164 | raw: &str, |
| 165 | source: &str, |
| 166 | format: DataFormat, |
| 167 | extension: Option<&str>, |
| 168 | context: &ToolContext, |
| 169 | ) -> Result<ToolResult, ToolError> { |
| 170 | if raw.len() > 1024 * 1024 { |
| 171 | return Err(ToolError::invalid_input("Host data input exceeds 1 MiB")); |
| 172 | } |
| 173 | let json = match serde_json::from_str::<Value>(raw) { |
| 174 | Ok(parsed) => json!({"ok":true,"value":json_descriptor(&parsed)}), |
| 175 | Err(error) => json!({"ok":false,"error":error.to_string()}), |
| 176 | }; |
| 177 | let toml = match toml::from_str::<toml::Value>(raw) { |
| 178 | Ok(parsed) => json!({"ok":true,"value":toml_descriptor(&parsed)}), |
| 179 | Err(error) => json!({"ok":false,"error":error.to_string()}), |
| 180 | }; |
| 181 | let result = crate::extension_host::manager().execute_stock( |
| 182 | crate::extension_host::StockOperation::ValidateData, |
| 183 | json!({"format":format.as_str(),"source":source,"extension":extension,"json":json,"toml":toml}), |
| 184 | context, |
| 185 | std::time::Duration::from_secs(120), |
| 186 | ).await?; |
| 187 | let metadata = result |
| 188 | .metadata |
| 189 | .as_ref() |
| 190 | .ok_or_else(|| ToolError::execution_failed("Host data result omitted metadata"))?; |
| 191 | if metadata.get("valid").and_then(Value::as_bool) != Some(result.success) |
| 192 | || metadata.get("source").and_then(Value::as_str) != Some(source) |
| 193 | { |
| 194 | return Err(ToolError::execution_failed( |
| 195 | "Host data result changed its captured source", |
| 196 | )); |
| 197 | } |
| 198 | if result.success { |
| 199 | if metadata.get("summary").is_none() { |
| 200 | return Err(ToolError::execution_failed( |
| 201 | "Host data result changed its captured source", |
| 202 | )); |
| 203 | } |
| 204 | ToolResult::json(metadata).map_err(|error| ToolError::execution_failed(error.to_string())) |
| 205 | } else { |
| 206 | Ok(result) |
| 207 | } |
| 208 | } |
| 209 | fn json_descriptor(value: &Value) -> Value { |
| 210 | match value { |
| 211 | Value::Object(map) => json!({"kind":"object","keys":map.keys().collect::<Vec<_>>()}), |
| 212 | Value::Array(array) => json!({"kind":"array","entries":array.len()}), |
| 213 | Value::String(_) => json!({"kind":"string"}), |
| 214 | Value::Number(_) => json!({"kind":"number"}), |
| 215 | Value::Bool(_) => json!({"kind":"boolean"}), |
| 216 | Value::Null => json!({"kind":"null"}), |
| 217 | } |
| 218 | } |
| 219 | fn toml_descriptor(value: &toml::Value) -> Value { |
| 220 | match value { |
| 221 | toml::Value::Table(table) => { |
| 222 | json!({"kind":"table","keys":table.keys().collect::<Vec<_>>()}) |
| 223 | } |
| 224 | toml::Value::Array(array) => json!({"kind":"array","entries":array.len()}), |
| 225 | toml::Value::String(_) => json!({"kind":"string"}), |
| 226 | toml::Value::Integer(_) => json!({"kind":"integer"}), |
| 227 | toml::Value::Float(_) => json!({"kind":"float"}), |
| 228 | toml::Value::Boolean(_) => json!({"kind":"boolean"}), |
| 229 | toml::Value::Datetime(_) => json!({"kind":"datetime"}), |
| 230 | } |
| 231 | } |
| 232 | |
| 233 | fn validate_auto( |
| 234 | raw_content: &str, |
| 235 | source_name: &str, |
| 236 | extension: Option<&str>, |
| 237 | ) -> Result<ToolResult, ToolError> { |
| 238 | let hint = match extension { |
| 239 | Some("json") => Some(DataFormat::Json), |
| 240 | Some("toml") => Some(DataFormat::Toml), |
| 241 | _ => None, |
| 242 | }; |
| 243 | |
| 244 | if let Some(format_hint) = hint { |
| 245 | return match format_hint { |
| 246 | DataFormat::Json => validate_json(raw_content, source_name), |
| 247 | DataFormat::Toml => validate_toml(raw_content, source_name), |
| 248 | DataFormat::Auto => unreachable!(), |
| 249 | }; |
| 250 | } |
| 251 | |
| 252 | let json_result = serde_json::from_str::<serde_json::Value>(raw_content); |
| 253 | if let Ok(parsed) = &json_result { |
| 254 | return build_success_result(DataFormat::Json, source_name, summarize_json(parsed)); |
| 255 | } |
| 256 | |
| 257 | let toml_result = toml::from_str::<toml::Value>(raw_content); |
| 258 | if let Ok(parsed) = &toml_result { |
| 259 | return build_success_result(DataFormat::Toml, source_name, summarize_toml(parsed)); |
| 260 | } |
| 261 | |
| 262 | let json_error = json_result.err().map(|e| e.to_string()).unwrap_or_default(); |
| 263 | let toml_error = toml_result.err().map(|e| e.to_string()).unwrap_or_default(); |
| 264 | |
| 265 | Ok( |
| 266 | ToolResult::error( |
| 267 | "Validation failed in auto mode: content is neither valid JSON nor TOML.", |
| 268 | ) |
| 269 | .with_metadata(json!({ |
| 270 | "valid": false, |
| 271 | "format": DataFormat::Auto.as_str(), |
| 272 | "source": source_name, |
| 273 | "json_error": json_error, |
| 274 | "toml_error": toml_error, |
| 275 | })), |
| 276 | ) |
| 277 | } |
| 278 | |
| 279 | fn validate_json(raw_content: &str, source_name: &str) -> Result<ToolResult, ToolError> { |
| 280 | match serde_json::from_str::<serde_json::Value>(raw_content) { |
| 281 | Ok(parsed) => build_success_result(DataFormat::Json, source_name, summarize_json(&parsed)), |
| 282 | Err(err) => Ok( |
| 283 | ToolResult::error(format!("Invalid JSON: {err}")).with_metadata(json!({ |
| 284 | "valid": false, |
| 285 | "format": DataFormat::Json.as_str(), |
| 286 | "source": source_name, |
| 287 | "error": err.to_string(), |
| 288 | })), |
| 289 | ), |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | fn validate_toml(raw_content: &str, source_name: &str) -> Result<ToolResult, ToolError> { |
| 294 | match toml::from_str::<toml::Value>(raw_content) { |
| 295 | Ok(parsed) => build_success_result(DataFormat::Toml, source_name, summarize_toml(&parsed)), |
| 296 | Err(err) => Ok( |
| 297 | ToolResult::error(format!("Invalid TOML: {err}")).with_metadata(json!({ |
| 298 | "valid": false, |
| 299 | "format": DataFormat::Toml.as_str(), |
| 300 | "source": source_name, |
| 301 | "error": err.to_string(), |
| 302 | })), |
| 303 | ), |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | fn build_success_result( |
| 308 | format: DataFormat, |
| 309 | source_name: &str, |
| 310 | summary: Value, |
| 311 | ) -> Result<ToolResult, ToolError> { |
| 312 | ToolResult::json(&json!({ |
| 313 | "valid": true, |
| 314 | "format": format.as_str(), |
| 315 | "source": source_name, |
| 316 | "summary": summary, |
| 317 | })) |
| 318 | .map_err(|e| ToolError::execution_failed(e.to_string())) |
| 319 | } |
| 320 | |
| 321 | fn summarize_json(value: &serde_json::Value) -> Value { |
| 322 | match value { |
| 323 | serde_json::Value::Object(map) => json!({ |
| 324 | "top_level": "object", |
| 325 | "entries": map.len(), |
| 326 | "keys_preview": map.keys().take(10).collect::<Vec<_>>(), |
| 327 | }), |
| 328 | serde_json::Value::Array(arr) => json!({ |
| 329 | "top_level": "array", |
| 330 | "entries": arr.len(), |
| 331 | }), |
| 332 | serde_json::Value::String(_) => json!({ "top_level": "string" }), |
| 333 | serde_json::Value::Number(_) => json!({ "top_level": "number" }), |
| 334 | serde_json::Value::Bool(_) => json!({ "top_level": "boolean" }), |
| 335 | serde_json::Value::Null => json!({ "top_level": "null" }), |
| 336 | } |
| 337 | } |
| 338 | |
| 339 | fn summarize_toml(value: &toml::Value) -> Value { |
| 340 | match value { |
| 341 | toml::Value::Table(table) => json!({ |
| 342 | "top_level": "table", |
| 343 | "entries": table.len(), |
| 344 | "keys_preview": table.keys().take(10).collect::<Vec<_>>(), |
| 345 | }), |
| 346 | toml::Value::Array(arr) => json!({ |
| 347 | "top_level": "array", |
| 348 | "entries": arr.len(), |
| 349 | }), |
| 350 | toml::Value::String(_) => json!({ "top_level": "string" }), |
| 351 | toml::Value::Integer(_) => json!({ "top_level": "integer" }), |
| 352 | toml::Value::Float(_) => json!({ "top_level": "float" }), |
| 353 | toml::Value::Boolean(_) => json!({ "top_level": "boolean" }), |
| 354 | toml::Value::Datetime(_) => json!({ "top_level": "datetime" }), |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | #[cfg(test)] |
| 359 | mod tests { |
| 360 | use super::*; |
| 361 | use tempfile::tempdir; |
| 362 | |
| 363 | #[tokio::test] |
| 364 | async fn validate_json_content_succeeds() { |
| 365 | let tmp = tempdir().expect("tempdir"); |
| 366 | let ctx = ToolContext::new(tmp.path()); |
| 367 | |
| 368 | let result = ValidateDataTool |
| 369 | .execute( |
| 370 | json!({"content": "{\"name\":\"deepseek\"}", "format": "json"}), |
| 371 | &ctx, |
| 372 | ) |
| 373 | .await |
| 374 | .expect("execute"); |
| 375 | assert!(result.success); |
| 376 | let content: Value = serde_json::from_str(&result.content).expect("validation json"); |
| 377 | assert_eq!(content.get("valid").and_then(Value::as_bool), Some(true)); |
| 378 | } |
| 379 | |
| 380 | #[tokio::test] |
| 381 | async fn validate_toml_file_succeeds() { |
| 382 | let tmp = tempdir().expect("tempdir"); |
| 383 | let ctx = ToolContext::new(tmp.path()); |
| 384 | let config = tmp.path().join("config.toml"); |
| 385 | fs::write(&config, "name = \"deepseek\"\n").expect("write"); |
| 386 | |
| 387 | let result = ValidateDataTool |
| 388 | .execute(json!({"path": "config.toml", "format": "toml"}), &ctx) |
| 389 | .await |
| 390 | .expect("execute"); |
| 391 | assert!(result.success); |
| 392 | let content: Value = serde_json::from_str(&result.content).expect("validation json"); |
| 393 | assert_eq!(content.get("format").and_then(Value::as_str), Some("toml")); |
| 394 | } |
| 395 | |
| 396 | #[tokio::test] |
| 397 | async fn validate_auto_reports_error_for_invalid_content() { |
| 398 | let tmp = tempdir().expect("tempdir"); |
| 399 | let ctx = ToolContext::new(tmp.path()); |
| 400 | |
| 401 | let result = ValidateDataTool |
| 402 | .execute(json!({"content": "not-valid-data"}), &ctx) |
| 403 | .await |
| 404 | .expect("execute"); |
| 405 | assert!(!result.success); |
| 406 | assert!(result.content.contains("Validation failed in auto mode")); |
| 407 | } |
| 408 | |
| 409 | #[tokio::test] |
| 410 | async fn validate_rejects_path_and_content_together() { |
| 411 | let tmp = tempdir().expect("tempdir"); |
| 412 | let ctx = ToolContext::new(tmp.path()); |
| 413 | |
| 414 | let err = ValidateDataTool |
| 415 | .execute(json!({"path": "a.toml", "content": "x=1"}), &ctx) |
| 416 | .await |
| 417 | .expect_err("should fail"); |
| 418 | assert!(matches!(err, ToolError::InvalidInput { .. })); |
| 419 | } |
| 420 | } |
| 421 |