@contentrain/types 1.16.0 → 1.18.0

package/README.md

@@ -55,6 +55,7 @@ Core interfaces:
 
 - `FieldDef`
 - `ModelDefinition`
+- `ModelLocaleScope`
 - `ModelSummary`
 - `ContentrainConfig`
 - `Vocabulary`
@@ -127,6 +128,9 @@ Validate functions (pure, dependency-free):
 - `validateSlug(slug)` — kebab-case slug validation
 - `validateEntryId(id)` — entry ID format validation
 - `validateLocale(locale, config)` — locale format + config support check
+- `resolveModelLocales(model, config)` — the locales a model's content is expected to cover: `model.locales` when it declares one, otherwise `config.locales.supported` (the default locale alone when `i18n: false`)
+- `describeModelLocaleScope(scope)` — name a resolved scope the way a validation message should quote it
+- `validateModelLocales(model, config)` — check a `locales` declaration is a subset of `config.locales.supported`
 - `detectSecrets(value)` — detect potential secrets in field values (provider-shaped patterns, plus an `api_key = …` assignment whose tail passes `looksLikeCredential`)
 - `validateFieldValue(value, fieldDef)` — full field schema validation (type, required, min/max, pattern, select)
 
@@ -182,6 +186,9 @@ const model: ModelDefinition = {
   kind: 'collection',
   domain: 'blog',
   i18n: true,
+  // Optional. Absent means every locale in config.locales.supported; a subset
+  // narrows validation for a partially-translated model.
+  locales: ['en', 'tr'],
   fields,
 }
 
@@ -250,6 +257,8 @@ Studio (Nuxt 4, web) cannot import `@contentrain/mcp` directly because MCP depen
 | `validateSlug(slug)` | Form validation for document slugs |
 | `validateEntryId(id)` | Validate collection entry IDs |
 | `validateLocale(locale, config)` | Locale picker validation |
+| `resolveModelLocales(model, config)` | Which locales to show, or require, for one model |
+| `validateModelLocales(model, config)` | Validate a model's locale-coverage subset in a model editor |
 | `detectSecrets(value)` | Content editor secret detection warnings |
 | `validateFieldValue(value, fieldDef)` | Full field-level validation in content forms |
 | `canonicalStringify(data, fieldOrder?)` | Preview canonical JSON output |
@@ -328,14 +337,24 @@ Shared shapes for the WordPress → static-site migration pipeline. They exist h
 
 | Contract | Role |
 |---|---|
-| `RawIR` | Source-faithful extraction of a WordPress site (posts, terms, menus, comments, media, redirects) with provenance: which access rung produced it (`rest_public` → `rest_auth` → `wxr` → `bridge`). Unresolved references are kept and marked, never dropped. |
+| `RawIR` | Source-faithful extraction of a WordPress site (posts, terms, menus, comments, media, redirects, and from the bridge rung SEO and routing) with provenance: which access rung produced it (`rest_public` → `rest_auth` → `wxr` → `bridge`). Unresolved references are kept and marked, never dropped. |
 | `CapabilityManifest` | Evidence-based inventory of what the site uses (SEO, forms, comments, i18n, ACF, …) — the input for migration planning and the "what happens to X" conversation. `access.achieved` is the highest source-access rung actually reached, or `null` when none was: a site behind an access wall still gets a manifest, and a producer that writes a rung it did not reach makes the coverage figures unreadable. |
 | `ProjectIR` | The reproducible model of the site: route model, layout families, component variants, query bindings, design tokens. Not "this page's HTML" — the design system that generates unseen pages correctly. |
-| `MigrationHandoff` | What the migration hands the user: repository, per-capability dispositions, and offers for runtime capabilities (with cost comparison) — offering is this document's job; fulfilling is the receiving product's. `runtime` (`RuntimeBinding`) records where the generated site's runtime components were bound once an offer was fulfilled. |
+| `MigrationHandoff` | What the migration hands the user: repository, per-capability dispositions, and offers for runtime capabilities (with cost comparison) — offering is this document's job; fulfilling is the receiving product's. `runtime` (`RuntimeBinding`) records where the generated site's runtime components were bound once an offer was fulfilled. `content_summary.body_kinds` counts standalone pages by `PageBodyKind` — `html` (editor markup, still editable), `builder_html` (rendered by a page builder, carried as rendered), `none` (the theme renders it) — because each promises something different to the person receiving the site. |
 | `RuntimeBinding` | The provider's public API origin (`base_url`) and `project_id` — all a static site needs to mount comments and forms. Never a credential: the public endpoints are unauthenticated by design. |
 
 All are plain JSON (snake_case keys), stamped with `MIGRATION_CONTRACT_VERSION`.
 
+`RawIR` also carries five optional parts from the bridge rung:
+
+- **`seo`** (`RawSeo`) holds which plugin serves the head (`serving`: `yoast` / `rank_math` / `aioseo` / `wordpress-core`), each provider's status and settings, and each page's values keyed by `post:<id>` or `term:<taxonomy>:<id>`. `status: 'none'` is an answer, not a missing export: no SEO plugin, and WordPress core renders only the title. `resolved: true` marks values the running plugin rendered; `false` marks stored values and templates only. `robots_served` is what the page actually carries after WordPress core and the plugin reconcile it.
+- **`routing`** (`RawRouting`) states the URL rules that `RawPost.link` holds as facts: permalink structure, bases, the front and posts pages, and each post type's and taxonomy's permastruct, with `with_front` applied.
+- **`redirects_excluded`** (`RawRedirectExcluded`) lists every rule a source holds that the site does not serve as a plain redirect, with the reason (`disabled`, `source-inactive`, `conditional-match:*`, `not-a-redirect:*`, …). Served plus excluded accounts for the whole table.
+- **`hardcoded_text`** (`RawHardcodedText`) accounts for the interface text outside the content tables — theme templates, scripts, widgets, menus, options, Customizer mods and rendered pages. Each `RawTextCandidate` has one outcome: `transfer` to a `target` (`dictionary:ui-strings`, `theme-settings.<mod>`, `site.title`, `site.description`, `content:wp-menu-items`) or `exclude` with a `reason` (`TEXT_EXCLUDE_REASONS`); a source that could not be read is listed in `errors`. Candidates merge only when text, locale and context are equal, and `key` depends on text and context only. Page text that also appears in source is `rendered-from-source` with `related` pointing at the source candidate, so the same words never land twice. `totals` closes: transfer + exclude = candidates, and error = `errors.length`.
+- **`integrations`** (`RawIntegration[]`) lists the outside services the site is connected to (`INTEGRATION_CATEGORIES`: analytics, crm, newsletter, ads, comments, captcha, cdn, other), each with its evidence. Evidence names a plugin, a setting, a script host, a form's wiring or an embed host, never a value. `reconnect_required` is `false` only for an embed, which has no account. `secret_present` says a credential is set on WordPress; the credential itself is never exported. `IntegrationReconnectRequiredIssue` (`code: 'integration_reconnect_required'`) is the one intake/handoff issue that lists every service to connect again.
+
+`RawRedirect` gains `id`, `match` and `regex`. Only `match: 'url'` without `regex` is a one-to-one mapping. A consumer that writes a regex or prefix rule as a literal `from` produces the wrong redirect.
+
 Chrome markers the emitter honours: `CHROME_BODY_SLOT` (where page content goes), `CHROME_REPEAT_OPEN`/`CHROME_IF_OPEN` (per-item and conditional regions), `LIST_ITEMS_SLOT` (where a list section's items go) and `componentSlot(id)` (`<!--@@component:ID@@-->`, where a `ComponentDef` — a `comments` thread, a `form` — is mounted).
 
 `ModelDefinition` can carry runtime-owned `form` and `comments` configuration.
@@ -365,6 +384,7 @@ approval guarding it would be theatre.
 | `DeploymentTarget` | Where a build is published. Carries a `secret_ref`, never a secret — this document is written to git. |
 | `AutomationDefinition` | Reserved shape for `.contentrain/automations.json`. Nothing reads it yet; it exists so the first writer does not invent a fourth vocabulary for schedules. |
 | `SourceDeltaPlan` | The WordPress→repo delta — *not* `contentrain_reconcile`, which merges two git branches and knows nothing about WordPress. Carries explicit deletion tombstones, because `modified_after` is a filter on changed records and never reports a deletion, plus slug moves (which generate redirects) and semantic conflicts. `deletions_detectable: false` must not be read as "nothing was deleted". |
+| `SourceInventory` / `SourceInventoryRecord` | Every origin record in scope at one moment — what a `bridge_inventory` cursor's hash identifies. Deletions and moves are proven by comparing two inventories (records keyed by `wp_type` + `wp_id`, updates decided on `fingerprint`); `modified_after` only narrows what to fetch again. `SourceDeltaEntry` also carries `path_before`/`path_after` for moves a slug cannot express, and `deleted_kind` (`trashed` / `purged`). |
 
 ### The evaluator