| 1 | //! Catalog parsers, one per real published schema. |
| 2 | //! |
| 3 | //! Each parser consumes only fields its format documents. Unknown fields |
| 4 | //! produce visible warnings, never silent acceptance and never invented |
| 5 | //! fallbacks: a field the format does not define is not parsed as some |
| 6 | //! plausible equivalent. There is no network access at this layer. |
| 7 | |
| 8 | pub mod claude; |
| 9 | pub mod codewhale; |
| 10 | pub mod codex; |
| 11 | pub mod kimi; |
| 12 | |
| 13 | use serde_json::Value; |
| 14 | |
| 15 | use crate::plugins::manifest::{PluginCompatibility, PluginInventory}; |
| 16 | |
| 17 | use super::types::{ |
| 18 | MarketplaceCatalog, MarketplaceCatalogId, MarketplaceDiagnostic, MarketplaceFormat, |
| 19 | }; |
| 20 | |
| 21 | /// Catalog-side compatibility. Unlike an installed bundle's reviewed |
| 22 | /// inventory, catalog declarations do not state MCP transport, so any |
| 23 | /// declared `mcp_servers` count is treated as a supported-capable |
| 24 | /// declaration; the install-time review decides the real activation |
| 25 | /// policy binding. |
| 26 | pub(super) fn declared_catalog_compatibility(declared: &PluginInventory) -> PluginCompatibility { |
| 27 | let supported = declared.skills > 0 || declared.mcp_servers > 0; |
| 28 | let unsupported = declared.commands > 0 |
| 29 | || declared.agents > 0 |
| 30 | || declared.hooks > 0 |
| 31 | || declared.lsp > 0 |
| 32 | || declared.native > 0 |
| 33 | || !declared.filesystem_roots.is_empty() |
| 34 | || !declared.network_hosts.is_empty() |
| 35 | || declared.lifecycle_mutation; |
| 36 | match (supported, unsupported) { |
| 37 | (true, false) => PluginCompatibility::Full, |
| 38 | (true, true) => PluginCompatibility::Partial, |
| 39 | (false, true) => PluginCompatibility::Unsupported, |
| 40 | (false, false) => PluginCompatibility::Full, |
| 41 | } |
| 42 | } |
| 43 | |
| 44 | /// Input to catalog parsing. `base` is where the document was read from |
| 45 | /// (path or URL) — kept for install-time resolution of relative sources; |
| 46 | /// the parser itself never touches it. |
| 47 | pub struct MarketplaceDocument { |
| 48 | pub catalog_id: MarketplaceCatalogId, |
| 49 | pub format: MarketplaceFormat, |
| 50 | pub root: Value, |
| 51 | pub base: Option<String>, |
| 52 | } |
| 53 | |
| 54 | /// Parse one catalog document. `MarketplaceFormat::Auto` detects the |
| 55 | /// format from documented structural markers only, and reports ambiguity |
| 56 | /// as an error diagnostic rather than guessing. |
| 57 | pub fn parse_catalog(document: MarketplaceDocument) -> MarketplaceCatalog { |
| 58 | let format = match document.format { |
| 59 | MarketplaceFormat::Auto => match detect_format(&document.root) { |
| 60 | Ok(format) => format, |
| 61 | Err(diagnostic) => { |
| 62 | let name = document.catalog_id.as_str().to_string(); |
| 63 | return MarketplaceCatalog { |
| 64 | id: document.catalog_id, |
| 65 | format: MarketplaceFormat::Auto, |
| 66 | name, |
| 67 | display_name: None, |
| 68 | description: None, |
| 69 | version: None, |
| 70 | base: document.base, |
| 71 | provenance: super::types::CatalogProvenance::default(), |
| 72 | candidates: Vec::new(), |
| 73 | diagnostics: vec![diagnostic], |
| 74 | }; |
| 75 | } |
| 76 | }, |
| 77 | explicit => explicit, |
| 78 | }; |
| 79 | |
| 80 | let mut catalog = match format { |
| 81 | MarketplaceFormat::Kimi => kimi::parse_kimi_catalog(document), |
| 82 | MarketplaceFormat::Claude => claude::parse_claude_catalog(document), |
| 83 | MarketplaceFormat::Codex => codex::parse_codex_catalog(document), |
| 84 | MarketplaceFormat::Codewhale => codewhale::parse_codewhale_catalog(document), |
| 85 | MarketplaceFormat::Auto => unreachable!("resolved above"), |
| 86 | }; |
| 87 | flag_duplicate_names(&mut catalog); |
| 88 | catalog |
| 89 | } |
| 90 | |
| 91 | /// A candidate's name is its identity: `/plugin show`, install, and registry |
| 92 | /// lookups all resolve by it, and a lookup returns the first match. Entries |
| 93 | /// sharing a name (after each format's normalization) are ambiguous, so every |
| 94 | /// one of them carries an error and none installs from the catalog. |
| 95 | fn flag_duplicate_names(catalog: &mut super::types::MarketplaceCatalog) { |
| 96 | let mut counts: std::collections::HashMap<String, usize> = std::collections::HashMap::new(); |
| 97 | for candidate in &catalog.candidates { |
| 98 | *counts.entry(candidate.name.clone()).or_default() += 1; |
| 99 | } |
| 100 | for candidate in &mut catalog.candidates { |
| 101 | let count = counts.get(&candidate.name).copied().unwrap_or(0); |
| 102 | if count > 1 { |
| 103 | candidate.diagnostics.push(MarketplaceDiagnostic::error( |
| 104 | "DUPLICATE_NAME", |
| 105 | format!( |
| 106 | "{count} catalog entries are named `{}`; the entry's identity is ambiguous", |
| 107 | candidate.name |
| 108 | ), |
| 109 | Some(candidate.name.clone()), |
| 110 | None, |
| 111 | )); |
| 112 | } |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | /// Detection uses only markers each format's own documentation defines: |
| 117 | /// |
| 118 | /// - **Kimi**: `plugins[]` whose entries carry `id` + `source` (Kimi uses |
| 119 | /// `id`; the Claude-family formats use `name`). |
| 120 | /// - **Claude**: top-level `owner` object, a `plugins[]` entry with a |
| 121 | /// `source` object using the Claude discriminators (`github`, `url`, |
| 122 | /// `git-subdir`, `npm`, `archive`, `command`), or `metadata.pluginRoot`. |
| 123 | /// - **Codex**: a `plugins[]` entry with a `policy` object, a `source` |
| 124 | /// object with the `local` discriminator, or a top-level `interface` |
| 125 | /// object. |
| 126 | /// - **Codewhale**: `plugins[]` entries with `name` + a string `source` |
| 127 | /// that is a Codewhale install spec (`github:`, `path:`, URL). |
| 128 | /// |
| 129 | /// Documents matching no documented marker are ambiguous, not guessed. |
| 130 | fn detect_format(root: &Value) -> Result<MarketplaceFormat, MarketplaceDiagnostic> { |
| 131 | let Some(obj) = root.as_object() else { |
| 132 | return Err(MarketplaceDiagnostic::error( |
| 133 | "NOT_AN_OBJECT", |
| 134 | "marketplace catalog must be a JSON object", |
| 135 | None, |
| 136 | None, |
| 137 | )); |
| 138 | }; |
| 139 | let Some(entries) = obj.get("plugins").and_then(Value::as_array) else { |
| 140 | return Err(MarketplaceDiagnostic::error( |
| 141 | "UNKNOWN_FORMAT", |
| 142 | "catalog has no documented marker: expected Kimi `plugins` with `id` entries, \ |
| 143 | Claude `owner`/`plugins`, Codex `policy`/`interface`, or Codewhale `plugins` \ |
| 144 | with install-spec sources", |
| 145 | None, |
| 146 | None, |
| 147 | )); |
| 148 | }; |
| 149 | |
| 150 | // Claude: `owner` and `metadata.pluginRoot` are documented top-level |
| 151 | // fields no other format defines. |
| 152 | if obj.contains_key("owner") |
| 153 | || obj |
| 154 | .get("metadata") |
| 155 | .and_then(|m| m.get("pluginRoot")) |
| 156 | .is_some() |
| 157 | { |
| 158 | return Ok(MarketplaceFormat::Claude); |
| 159 | } |
| 160 | |
| 161 | let entry_markers: Vec<MapMarker> = entries |
| 162 | .iter() |
| 163 | .map(|entry| { |
| 164 | let entry_obj = entry.as_object(); |
| 165 | MapMarker { |
| 166 | has_id: entry_obj.is_some_and(|o| o.contains_key("id")), |
| 167 | has_name: entry_obj.is_some_and(|o| o.contains_key("name")), |
| 168 | source_kind: entry_obj.and_then(|o| o.get("source")).map(source_marker), |
| 169 | has_policy: entry_obj.is_some_and(|o| o.contains_key("policy")), |
| 170 | } |
| 171 | }) |
| 172 | .collect(); |
| 173 | |
| 174 | // Codex: `policy` blocks or `local` sources are Codex-only markers. |
| 175 | if entry_markers |
| 176 | .iter() |
| 177 | .any(|m| m.has_policy || m.source_kind == Some(SourceMarker::Local)) |
| 178 | || obj.contains_key("interface") |
| 179 | { |
| 180 | return Ok(MarketplaceFormat::Codex); |
| 181 | } |
| 182 | |
| 183 | // Claude source discriminators other than `local`. |
| 184 | if entry_markers.iter().any(|m| { |
| 185 | matches!( |
| 186 | m.source_kind, |
| 187 | Some(SourceMarker::Github) |
| 188 | | Some(SourceMarker::Url) |
| 189 | | Some(SourceMarker::GitSubdir) |
| 190 | | Some(SourceMarker::Npm) |
| 191 | | Some(SourceMarker::Archive) |
| 192 | | Some(SourceMarker::Command) |
| 193 | ) |
| 194 | }) { |
| 195 | return Ok(MarketplaceFormat::Claude); |
| 196 | } |
| 197 | |
| 198 | // Kimi: `id`-keyed entries with string sources. |
| 199 | if entry_markers.iter().any(|m| { |
| 200 | m.has_id |
| 201 | && matches!( |
| 202 | m.source_kind, |
| 203 | Some(SourceMarker::String | SourceMarker::UrlString) |
| 204 | ) |
| 205 | }) { |
| 206 | return Ok(MarketplaceFormat::Kimi); |
| 207 | } |
| 208 | |
| 209 | // Codewhale: `name` entries with install-spec string sources. |
| 210 | if entry_markers.iter().any(|m| { |
| 211 | m.has_name |
| 212 | && matches!( |
| 213 | m.source_kind, |
| 214 | Some(SourceMarker::InstallSpec | SourceMarker::UrlString) |
| 215 | ) |
| 216 | }) { |
| 217 | return Ok(MarketplaceFormat::Codewhale); |
| 218 | } |
| 219 | |
| 220 | Err(MarketplaceDiagnostic::error( |
| 221 | "AMBIGUOUS_FORMAT", |
| 222 | "catalog markers match no documented format uniquely; pass the format explicitly", |
| 223 | None, |
| 224 | None, |
| 225 | )) |
| 226 | } |
| 227 | |
| 228 | struct MapMarker { |
| 229 | has_id: bool, |
| 230 | has_name: bool, |
| 231 | source_kind: Option<SourceMarker>, |
| 232 | has_policy: bool, |
| 233 | } |
| 234 | |
| 235 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 236 | enum SourceMarker { |
| 237 | String, |
| 238 | UrlString, |
| 239 | InstallSpec, |
| 240 | Github, |
| 241 | Url, |
| 242 | GitSubdir, |
| 243 | Npm, |
| 244 | Archive, |
| 245 | Command, |
| 246 | Local, |
| 247 | OtherObject, |
| 248 | } |
| 249 | |
| 250 | fn source_marker(source: &Value) -> SourceMarker { |
| 251 | match source { |
| 252 | Value::String(s) => { |
| 253 | if s.starts_with("github:") || s.starts_with("path:") { |
| 254 | SourceMarker::InstallSpec |
| 255 | } else if s.starts_with("https://") || s.starts_with("http://") { |
| 256 | SourceMarker::UrlString |
| 257 | } else { |
| 258 | SourceMarker::String |
| 259 | } |
| 260 | } |
| 261 | Value::Object(o) => match o.get("source").and_then(Value::as_str) { |
| 262 | Some("github") => SourceMarker::Github, |
| 263 | Some("url") => SourceMarker::Url, |
| 264 | Some("git-subdir") => SourceMarker::GitSubdir, |
| 265 | Some("npm") => SourceMarker::Npm, |
| 266 | Some("archive") => SourceMarker::Archive, |
| 267 | Some("command") => SourceMarker::Command, |
| 268 | Some("local") => SourceMarker::Local, |
| 269 | _ => SourceMarker::OtherObject, |
| 270 | }, |
| 271 | _ => SourceMarker::OtherObject, |
| 272 | } |
| 273 | } |
| 274 | |
| 275 | /// Shared helper: read a documented string field; a wrong-typed value is |
| 276 | /// a per-entry warning and treated as absent, not a guess. |
| 277 | pub(super) fn str_field<'a>( |
| 278 | entry: &'a serde_json::Map<String, Value>, |
| 279 | field: &str, |
| 280 | ) -> (Option<&'a str>, Option<MarketplaceDiagnostic>) { |
| 281 | match entry.get(field) { |
| 282 | None | Some(Value::Null) => (None, None), |
| 283 | Some(Value::String(s)) => (Some(s.as_str()), None), |
| 284 | Some(other) => ( |
| 285 | None, |
| 286 | Some(MarketplaceDiagnostic::warning( |
| 287 | "FIELD_TYPE", |
| 288 | format!("field `{field}` must be a string, got {}", type_name(other)), |
| 289 | None, |
| 290 | None, |
| 291 | )), |
| 292 | ), |
| 293 | } |
| 294 | } |
| 295 | |
| 296 | /// Shared helper: read a documented string-array field. |
| 297 | pub(super) fn str_array_field( |
| 298 | entry: &serde_json::Map<String, Value>, |
| 299 | field: &str, |
| 300 | ) -> (Vec<String>, Option<MarketplaceDiagnostic>) { |
| 301 | match entry.get(field) { |
| 302 | None | Some(Value::Null) => (Vec::new(), None), |
| 303 | Some(Value::Array(items)) => { |
| 304 | let mut out = Vec::new(); |
| 305 | let mut skipped = false; |
| 306 | for item in items { |
| 307 | match item { |
| 308 | Value::String(s) => out.push(s.clone()), |
| 309 | _ => skipped = true, |
| 310 | } |
| 311 | } |
| 312 | let diag = skipped.then(|| { |
| 313 | MarketplaceDiagnostic::warning( |
| 314 | "FIELD_TYPE", |
| 315 | format!( |
| 316 | "field `{field}` must be an array of strings; non-string items skipped" |
| 317 | ), |
| 318 | None, |
| 319 | None, |
| 320 | ) |
| 321 | }); |
| 322 | (out, diag) |
| 323 | } |
| 324 | Some(other) => ( |
| 325 | Vec::new(), |
| 326 | Some(MarketplaceDiagnostic::warning( |
| 327 | "FIELD_TYPE", |
| 328 | format!( |
| 329 | "field `{field}` must be an array of strings, got {}", |
| 330 | type_name(other) |
| 331 | ), |
| 332 | None, |
| 333 | None, |
| 334 | )), |
| 335 | ), |
| 336 | } |
| 337 | } |
| 338 | |
| 339 | /// Shared helper: warn on fields this format does not document. Unknown |
| 340 | /// fields are preserved in diagnostics so catalog authors see them; they |
| 341 | /// are never silently reinterpreted. |
| 342 | pub(super) fn unknown_fields_warning( |
| 343 | entry: &serde_json::Map<String, Value>, |
| 344 | known: &[&str], |
| 345 | ) -> Option<MarketplaceDiagnostic> { |
| 346 | let unknown: Vec<&String> = entry |
| 347 | .keys() |
| 348 | .filter(|k| !known.contains(&k.as_str())) |
| 349 | .collect(); |
| 350 | if unknown.is_empty() { |
| 351 | return None; |
| 352 | } |
| 353 | Some(MarketplaceDiagnostic::warning( |
| 354 | "UNKNOWN_FIELD", |
| 355 | format!( |
| 356 | "undeclared field(s) ignored: {}", |
| 357 | unknown |
| 358 | .iter() |
| 359 | .map(|s| s.as_str()) |
| 360 | .collect::<Vec<_>>() |
| 361 | .join(", ") |
| 362 | ), |
| 363 | None, |
| 364 | None, |
| 365 | )) |
| 366 | } |
| 367 | |
| 368 | fn type_name(value: &Value) -> &'static str { |
| 369 | match value { |
| 370 | Value::Null => "null", |
| 371 | Value::Bool(_) => "a boolean", |
| 372 | Value::Number(_) => "a number", |
| 373 | Value::String(_) => "a string", |
| 374 | Value::Array(_) => "an array", |
| 375 | Value::Object(_) => "an object", |
| 376 | } |
| 377 | } |
| 378 |