| 1 | //! Local catalog document loading shared by the `/plugin marketplace add` |
| 2 | //! command and the Runtime API marketplace endpoints (#5311 surface). |
| 3 | //! |
| 4 | //! One loader so both entry points apply the same rules: LOCAL file only |
| 5 | //! (never network), bounded size, no symlink documents, strict per-format |
| 6 | //! parsing, and refusal to persist a document that parsed to nothing but |
| 7 | //! errors. Candidate→install-spec resolution also lives here so the TUI |
| 8 | //! command and the HTTP API can never disagree about what would be fetched. |
| 9 | |
| 10 | use std::io::Read; |
| 11 | use std::path::{Component, Path, PathBuf}; |
| 12 | |
| 13 | use super::parsers::{self, MarketplaceDocument}; |
| 14 | use super::store::StoredMarketplaceCatalog; |
| 15 | use super::types::{ |
| 16 | MarketplaceCandidate, MarketplaceCatalogId, MarketplaceFormat, MarketplaceInstallPlan, |
| 17 | MarketplaceSourceSpec, |
| 18 | }; |
| 19 | |
| 20 | /// Catalog documents are JSON text; four megabytes is far beyond any real |
| 21 | /// published catalog and caps the parse cost of a user-supplied file. |
| 22 | const MAX_CATALOG_BYTES: u64 = 4 * 1024 * 1024; |
| 23 | |
| 24 | /// A parsed catalog ready to store, plus the counts callers render back. |
| 25 | #[derive(Debug)] |
| 26 | pub struct LoadedCatalogDocument { |
| 27 | pub entry: StoredMarketplaceCatalog, |
| 28 | pub candidate_count: usize, |
| 29 | pub warning_count: usize, |
| 30 | } |
| 31 | |
| 32 | /// Conservative catalog name: it becomes a key, appears in candidate IDs, |
| 33 | /// and is rendered back to the operator. |
| 34 | #[must_use] |
| 35 | pub fn valid_marketplace_name(name: &str) -> bool { |
| 36 | !name.is_empty() |
| 37 | && name.len() <= 64 |
| 38 | && name |
| 39 | .chars() |
| 40 | .all(|c| c.is_ascii_alphanumeric() || c == '-' || c == '_' || c == '.') |
| 41 | } |
| 42 | |
| 43 | /// Read and parse a LOCAL catalog document. Relative paths resolve against |
| 44 | /// `workspace`. No network is touched, here or anywhere in this module. |
| 45 | pub fn load_catalog_document( |
| 46 | name: &str, |
| 47 | workspace: &Path, |
| 48 | raw_path: &str, |
| 49 | ) -> Result<LoadedCatalogDocument, String> { |
| 50 | if !valid_marketplace_name(name) { |
| 51 | return Err( |
| 52 | "Marketplace name must be 1-64 characters of letters, digits, `-`, `_`, or `.`" |
| 53 | .to_string(), |
| 54 | ); |
| 55 | } |
| 56 | let path = PathBuf::from(raw_path.trim()); |
| 57 | let path = if path.is_absolute() { |
| 58 | path |
| 59 | } else { |
| 60 | workspace.join(path) |
| 61 | }; |
| 62 | let path = canonical_document(&path)?; |
| 63 | let body = read_bounded(&path)?; |
| 64 | let root = serde_json::from_str::<serde_json::Value>(&body) |
| 65 | .map_err(|error| format!("Catalog at {} is not valid JSON: {error}", path.display()))?; |
| 66 | |
| 67 | let document = MarketplaceDocument { |
| 68 | catalog_id: MarketplaceCatalogId::new(name), |
| 69 | format: MarketplaceFormat::Auto, |
| 70 | root, |
| 71 | base: Some(path.display().to_string()), |
| 72 | }; |
| 73 | let catalog = parsers::parse_catalog(document); |
| 74 | |
| 75 | // A document-level error (unknown/ambiguous format, not-an-object) means |
| 76 | // nothing useful was parsed; do not persist it. |
| 77 | if catalog.candidates.is_empty() && catalog.error_count() > 0 { |
| 78 | return Err(format!( |
| 79 | "Catalog `{name}` could not be parsed as any known marketplace format (kimi, claude, codex, codewhale):\n{}", |
| 80 | render_diagnostics_inline(&catalog.diagnostics) |
| 81 | )); |
| 82 | } |
| 83 | |
| 84 | let entry = StoredMarketplaceCatalog { |
| 85 | added_at: chrono::Utc::now().to_rfc3339(), |
| 86 | source_path: path.display().to_string(), |
| 87 | catalog, |
| 88 | }; |
| 89 | Ok(LoadedCatalogDocument { |
| 90 | candidate_count: entry.catalog.total_candidates(), |
| 91 | warning_count: entry.catalog.warning_count(), |
| 92 | entry, |
| 93 | }) |
| 94 | } |
| 95 | |
| 96 | /// What installing a stored candidate would do, resolved once for every |
| 97 | /// caller. `Supported.spec` is exactly what the reviewed installer accepts. |
| 98 | pub enum CatalogInstallResolution<'a> { |
| 99 | Supported { |
| 100 | spec: String, |
| 101 | source_kind: String, |
| 102 | }, |
| 103 | /// A matching name is occupied; catalog metadata does not prove identity. |
| 104 | AlreadyPresent { |
| 105 | plugin: &'a crate::plugins::types::LoadedPlugin, |
| 106 | reason: String, |
| 107 | }, |
| 108 | Unsupported { |
| 109 | reason: String, |
| 110 | }, |
| 111 | HasErrors { |
| 112 | diagnostics: String, |
| 113 | }, |
| 114 | } |
| 115 | |
| 116 | /// Resolve a stored catalog candidate to its install spec. Relative local |
| 117 | /// paths resolve against the catalog document's own directory, not the |
| 118 | /// caller's working directory. |
| 119 | pub fn resolve_candidate_install<'a>( |
| 120 | entry: &StoredMarketplaceCatalog, |
| 121 | candidate: &MarketplaceCandidate, |
| 122 | registry: &'a crate::plugins::PluginRegistry, |
| 123 | ) -> CatalogInstallResolution<'a> { |
| 124 | if candidate.has_errors() { |
| 125 | return CatalogInstallResolution::HasErrors { |
| 126 | diagnostics: render_diagnostics_inline(&candidate.diagnostics), |
| 127 | }; |
| 128 | } |
| 129 | if let Some(plugin) = registry.get(&candidate.name) { |
| 130 | if plugin.scope == crate::plugins::types::PluginScope::Builtin { |
| 131 | // A shipped bundle (Computer Use): the published "install" |
| 132 | // command must say so, not read as a failure (B5). The catalog |
| 133 | // entry still never replaces it. |
| 134 | return CatalogInstallResolution::AlreadyPresent { |
| 135 | plugin, |
| 136 | reason: format!( |
| 137 | "'{}' is built into Codewhale, so there is nothing to install; this catalog entry does not replace it. Review it with /plugin show {}.", |
| 138 | plugin.name(), |
| 139 | plugin.id.as_str() |
| 140 | ), |
| 141 | }; |
| 142 | } |
| 143 | return CatalogInstallResolution::AlreadyPresent { |
| 144 | plugin, |
| 145 | reason: format!( |
| 146 | "A {} plugin named '{}' already exists. Review the existing bundle with /plugin show {}. Catalog metadata does not establish that it is the same bundle.", |
| 147 | plugin.scope.as_str(), |
| 148 | plugin.name(), |
| 149 | plugin.id.as_str() |
| 150 | ), |
| 151 | }; |
| 152 | } |
| 153 | match &candidate.install_plan { |
| 154 | MarketplaceInstallPlan::Supported { spec, source_kind } => match resolve_spec( |
| 155 | &entry.source_path, |
| 156 | entry.catalog.format, |
| 157 | &candidate.source, |
| 158 | spec, |
| 159 | ) { |
| 160 | Ok(spec) => CatalogInstallResolution::Supported { |
| 161 | spec, |
| 162 | source_kind: source_kind.clone(), |
| 163 | }, |
| 164 | Err(reason) => CatalogInstallResolution::Unsupported { reason }, |
| 165 | }, |
| 166 | MarketplaceInstallPlan::Unsupported { reason, .. } => { |
| 167 | CatalogInstallResolution::Unsupported { |
| 168 | reason: reason.clone(), |
| 169 | } |
| 170 | } |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | /// A catalog's local source must name a directory inside the catalog's own |
| 175 | /// tree. An absolute path, a root or drive prefix, or any `..` component would |
| 176 | /// let catalog metadata point the installer at an arbitrary directory on this |
| 177 | /// machine, so such an entry is not installable from the catalog; installing a |
| 178 | /// directory outside it stays an explicit `/plugin install path:...`. |
| 179 | /// |
| 180 | /// Known limitation: the check is textual. It inspects the source's path |
| 181 | /// components and never resolves the filesystem, so a symlink inside the |
| 182 | /// catalog tree (`plugins/foo -> /elsewhere`) still passes, and the install |
| 183 | /// follows it out of the catalog directory. |
| 184 | fn resolve_spec( |
| 185 | source_path: &str, |
| 186 | format: MarketplaceFormat, |
| 187 | source: &MarketplaceSourceSpec, |
| 188 | spec: &str, |
| 189 | ) -> Result<String, String> { |
| 190 | if let MarketplaceSourceSpec::LocalPath { path } = source |
| 191 | && !path |
| 192 | .components() |
| 193 | .all(|part| matches!(part, Component::Normal(_) | Component::CurDir)) |
| 194 | { |
| 195 | return Err(format!( |
| 196 | "Local catalog source `{}` leaves the catalog directory; only paths inside it (no absolute paths or `..`) are installable from a catalog", |
| 197 | path.display() |
| 198 | )); |
| 199 | } |
| 200 | if let MarketplaceSourceSpec::LocalPath { path } = source |
| 201 | && let Some(dir) = Path::new(source_path).parent() |
| 202 | { |
| 203 | // Claude keeps its catalog in a manifest-only metadata directory; |
| 204 | // relative sources are rooted at the marketplace repository. |
| 205 | let dir = if format == MarketplaceFormat::Claude |
| 206 | && dir.file_name().is_some_and(|name| name == ".claude-plugin") |
| 207 | { |
| 208 | dir.parent().unwrap_or(dir) |
| 209 | } else { |
| 210 | dir |
| 211 | }; |
| 212 | return Ok(format!("path:{}", dir.join(path).display())); |
| 213 | } |
| 214 | // Every other source kind (GitHub, archive/tarball URL) is remote. The |
| 215 | // installer parses the spec again, and any value it does not recognise |
| 216 | // as remote is a local directory there — so an archive `url` of `/etc` |
| 217 | // or `../..` would bypass the containment check above. Only a spec the |
| 218 | // installer's own parser reads as remote is installable from a catalog. |
| 219 | match crate::plugins::install::PluginInstallSource::parse(spec) { |
| 220 | Ok(crate::plugins::install::PluginInstallSource::Remote(_)) => Ok(spec.to_string()), |
| 221 | _ => Err(format!( |
| 222 | "Catalog source `{spec}` is not a remote (github: or http(s)://) source; a catalog may name a local directory only as a path inside the catalog" |
| 223 | )), |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | /// Resolve a user-supplied document path to an existing regular file without |
| 228 | /// following a final symlink (the document is untrusted input). |
| 229 | fn canonical_document(path: &Path) -> Result<PathBuf, String> { |
| 230 | let metadata = std::fs::symlink_metadata(path) |
| 231 | .map_err(|e| format!("Cannot read catalog at {}: {e}", path.display()))?; |
| 232 | if metadata.is_symlink() { |
| 233 | return Err(format!( |
| 234 | "Catalog path {} is a symlink; marketplace documents must be regular files", |
| 235 | path.display() |
| 236 | )); |
| 237 | } |
| 238 | if !metadata.is_file() { |
| 239 | return Err(format!( |
| 240 | "Catalog path {} is not a regular file", |
| 241 | path.display() |
| 242 | )); |
| 243 | } |
| 244 | Ok(path.to_path_buf()) |
| 245 | } |
| 246 | |
| 247 | fn read_bounded(path: &Path) -> Result<String, String> { |
| 248 | // Validate the opened handle, not just the path checked before opening. |
| 249 | let file = crate::plugins::registry::open_existing_regular_file(path, false)? |
| 250 | .ok_or_else(|| format!("Cannot read catalog at {}: file is missing", path.display()))?; |
| 251 | let mut text = String::new(); |
| 252 | let mut limited = file.take(MAX_CATALOG_BYTES + 1); |
| 253 | limited |
| 254 | .read_to_string(&mut text) |
| 255 | .map_err(|e| format!("Cannot read catalog at {}: {e}", path.display()))?; |
| 256 | // Check bytes actually read: the file can grow after its metadata is read. |
| 257 | if text.len() as u64 > MAX_CATALOG_BYTES { |
| 258 | return Err(format!( |
| 259 | "Catalog at {} exceeds the {} byte limit", |
| 260 | path.display(), |
| 261 | MAX_CATALOG_BYTES |
| 262 | )); |
| 263 | } |
| 264 | Ok(text) |
| 265 | } |
| 266 | |
| 267 | fn render_diagnostics_inline(diagnostics: &[super::types::MarketplaceDiagnostic]) -> String { |
| 268 | use crate::plugins::types::PluginDiagnosticLevel; |
| 269 | diagnostics |
| 270 | .iter() |
| 271 | .map(|d| { |
| 272 | format!( |
| 273 | "{} {}: {}", |
| 274 | match d.level { |
| 275 | PluginDiagnosticLevel::Error => "error", |
| 276 | PluginDiagnosticLevel::Warning => "warning", |
| 277 | }, |
| 278 | d.code, |
| 279 | d.message |
| 280 | ) |
| 281 | }) |
| 282 | .collect::<Vec<_>>() |
| 283 | .join("; ") |
| 284 | } |
| 285 | |
| 286 | #[cfg(test)] |
| 287 | mod tests { |
| 288 | use super::*; |
| 289 | |
| 290 | #[test] |
| 291 | fn marketplace_names_are_conservative() { |
| 292 | assert!(valid_marketplace_name("official")); |
| 293 | assert!(valid_marketplace_name("My-Catalog_2.beta")); |
| 294 | assert!(!valid_marketplace_name("")); |
| 295 | assert!(!valid_marketplace_name("has space")); |
| 296 | assert!(!valid_marketplace_name("a".repeat(65).as_str())); |
| 297 | } |
| 298 | |
| 299 | #[test] |
| 300 | fn catalog_read_enforces_actual_byte_limit() { |
| 301 | use std::io::Write as _; |
| 302 | let dir = tempfile::tempdir().unwrap(); |
| 303 | let path = dir.path().join("catalog.json"); |
| 304 | let body = " ".repeat(MAX_CATALOG_BYTES as usize); |
| 305 | std::fs::write(&path, &body).unwrap(); |
| 306 | assert_eq!(read_bounded(&path).unwrap(), body); |
| 307 | let checked = canonical_document(&path).unwrap(); |
| 308 | let mut file = std::fs::OpenOptions::new() |
| 309 | .append(true) |
| 310 | .open(&path) |
| 311 | .unwrap(); |
| 312 | file.write_all(b" ").unwrap(); |
| 313 | assert!(read_bounded(&checked).unwrap_err().contains("byte limit")); |
| 314 | } |
| 315 | |
| 316 | #[cfg(unix)] |
| 317 | #[test] |
| 318 | fn catalog_read_refuses_symlink_substituted_after_path_check() { |
| 319 | let dir = tempfile::tempdir().unwrap(); |
| 320 | let path = dir.path().join("catalog.json"); |
| 321 | let other = dir.path().join("other.json"); |
| 322 | std::fs::write(&path, "{}").unwrap(); |
| 323 | std::fs::write(&other, "synthetic unrelated content").unwrap(); |
| 324 | let checked = canonical_document(&path).unwrap(); |
| 325 | std::fs::rename(&path, dir.path().join("original.json")).unwrap(); |
| 326 | std::os::unix::fs::symlink(&other, &path).unwrap(); |
| 327 | assert!(read_bounded(&checked).is_err()); |
| 328 | assert_eq!( |
| 329 | std::fs::read_to_string(&other).unwrap(), |
| 330 | "synthetic unrelated content" |
| 331 | ); |
| 332 | } |
| 333 | |
| 334 | #[cfg(unix)] |
| 335 | #[test] |
| 336 | fn catalog_read_refuses_fifo_without_waiting_for_a_writer() { |
| 337 | const CHILD_PATH: &str = "CODEWHALE_TEST_CATALOG_FIFO"; |
| 338 | if let Some(path) = std::env::var_os(CHILD_PATH) { |
| 339 | assert!(read_bounded(Path::new(&path)).is_err()); |
| 340 | return; |
| 341 | } |
| 342 | // Isolate a regressed blocking open so the test can stop it safely. |
| 343 | let dir = tempfile::tempdir().unwrap(); |
| 344 | let path = dir.path().join("catalog.json"); |
| 345 | std::fs::write(&path, "{}").unwrap(); |
| 346 | let checked = canonical_document(&path).unwrap(); |
| 347 | std::fs::rename(&path, dir.path().join("original.json")).unwrap(); |
| 348 | assert!( |
| 349 | std::process::Command::new("mkfifo") |
| 350 | .arg(&path) |
| 351 | .status() |
| 352 | .unwrap() |
| 353 | .success() |
| 354 | ); |
| 355 | let mut child = std::process::Command::new(std::env::current_exe().unwrap()) |
| 356 | .args(["--exact", "plugins::marketplace::document::tests::catalog_read_refuses_fifo_without_waiting_for_a_writer"]) |
| 357 | .env(CHILD_PATH, checked) |
| 358 | .stdout(std::process::Stdio::null()) |
| 359 | .stderr(std::process::Stdio::null()) |
| 360 | .spawn().unwrap(); |
| 361 | let deadline = std::time::Instant::now() + std::time::Duration::from_secs(5); |
| 362 | loop { |
| 363 | if let Some(status) = child.try_wait().unwrap() { |
| 364 | assert!(status.success()); |
| 365 | break; |
| 366 | } |
| 367 | if std::time::Instant::now() >= deadline { |
| 368 | let _ = child.kill(); |
| 369 | let _ = child.wait(); |
| 370 | panic!("catalog read waited for a FIFO writer"); |
| 371 | } |
| 372 | std::thread::sleep(std::time::Duration::from_millis(10)); |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | #[cfg(unix)] |
| 377 | #[test] |
| 378 | fn load_refuses_symlink_documents() { |
| 379 | let dir = tempfile::tempdir().unwrap(); |
| 380 | let real = dir.path().join("real.json"); |
| 381 | std::fs::write(&real, "{}").unwrap(); |
| 382 | let link = dir.path().join("link.json"); |
| 383 | std::os::unix::fs::symlink(&real, &link).unwrap(); |
| 384 | |
| 385 | let error = load_catalog_document("test", dir.path(), link.to_str().unwrap()) |
| 386 | .expect_err("symlink document must be refused"); |
| 387 | assert!(error.contains("symlink"), "{error}"); |
| 388 | } |
| 389 | |
| 390 | #[test] |
| 391 | fn load_refuses_unknown_format_documents() { |
| 392 | let dir = tempfile::tempdir().unwrap(); |
| 393 | let doc = dir.path().join("catalog.json"); |
| 394 | std::fs::write(&doc, r#"{"totally":"unknown"}"#).unwrap(); |
| 395 | |
| 396 | let error = load_catalog_document("test", dir.path(), doc.to_str().unwrap()) |
| 397 | .expect_err("unknown format must be refused"); |
| 398 | assert!(error.contains("could not be parsed"), "{error}"); |
| 399 | } |
| 400 | } |
| 401 |