Skip to content

CHANGELOG.md

Latest commit

 

History

History
145 lines (86 loc) · 13.5 KB

CHANGELOG.md

File metadata and controls

145 lines (86 loc) · 13.5 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.11.0] - 2026-04-21

Changed

  • peerDependencies.emdash bumped to ^0.6.0. Picks up eager taxonomy-term hydration on getEmDashEntry results (emdash PR #626), plus the toolbar portableText fix (#685) and several cold-start / D1 replica performance improvements.

Added

  • Taxonomy keywords and section in Article schema. For content pages the plugin now calls getEmDashEntry to read the entry's eagerly-hydrated data.terms map (zero extra DB queries when the Astro template already fetched the entry in the same request). All assigned taxonomy terms are joined into the keywords field of the BlogPosting node. Terms from any taxonomy whose name starts with categor are additionally mapped to articleSection.

[0.10.0] - 2026-04-14

Changed

  • peerDependencies.emdash bumped to ^0.5.0. This release depends on emdash-cms/emdash#536, #539, and #540, all shipped in emdash@0.5.0. Older EmDash versions no longer satisfy the peer range.

Fixed

  • llms.txt and schema map now actually enumerate content. Both loops read the top-level ContentItem.slug/status/locale fields exposed by #536 + #539, and narrow at the SQL layer via where: { status: "published" } from #540 instead of filtering in userland. Before this change, both routes silently returned empty against real EmDash sites — documented as broken in the v0.9.0 README warning.
  • Fuzzy Redirects candidate list is no longer empty. The tool fetches published URLs from schema/map, so that route working fixes the candidate side of the feature for free. The other half (the 404 log capturing /404 instead of the original requested URL) is still tracked in emdash-cms/emdash#525.

Removed

  • Transient type casts. The @ts-expect-error in metadata.ts (for rel: "nlweb") and the per-call type casts in llms.ts / schema/endpoints.ts are gone — the types ship with emdash@0.5.0.
  • "Known broken" README warning — the content-enumeration features now work.

[0.9.0] - 2026-04-13

Added

  • NLWeb <link> tag. When the NLWeb endpoint URL setting is populated, the plugin contributes <link rel="nlweb" href="…"> on every rendered page so conversational agents can discover the site's chat surface. Requires the EmDash release that contains emdash-cms/emdash#523 (merged); on older versions the sandbox allowlist rejects the rel value and the contribution is silently dropped. The contribution site carries a transient @ts-expect-error against the currently-published emdash types — it auto-unblocks once the new emdash version is on npm.

Documented

  • Three earlier-shipped features (llms.txt, schema/map, Fuzzy Redirects) silently produce empty results against real EmDash sites. Root causes are upstream: ContentItem strips slug/status/locale (emdash-cms/emdash#530), and the 404 logging middleware captures the rewritten /404 instead of the original requested URL (emdash-cms/emdash#525). README now carries a prominent warning at the top with the table of affected features and the upstream tracking links. No code change — the features will start working once upstream lands the fixes.

[0.8.0] - 2026-04-13

Added

  • Fuzzy Redirects admin tool. New admin page that reads the core 404 log, pairs each missing path with the closest published URLs (from the plugin's own schema/map route), and lets you one-click create a 301 redirect for the chosen destination. Scoring combines Levenshtein distance, token overlap, and a tokenized last-segment match bonus — so moved slugs, single-character typos, and punctuation drift all surface good suggestions. A minimum-score slider tunes how aggressive the suggestions are.
  • rankCandidates() / scoreSlugMatch() exported from src/fuzzy.ts for reuse. These will plug straight into the eventual notfound hook (emdash-cms/emdash#525) once upstream lands — same matching logic, automatic trigger.

[0.7.0] - 2026-04-13

Added

  • Schema map (experimental). New public plugin route schema/map returning the list of every published URL backed by schema markup ({ items: [{ url, collection, updatedAt }] }). Wire it to /schemamap.xml at your site root with the Astro snippet in the README so agents and crawlers can enumerate structured-data URLs without scraping HTML. Per-URL schema endpoints (/schema/<slug>.json) are deferred pending an upstream helper for building page contexts from plugin routes.

[0.6.0] - 2026-04-13

Added

  • llms.txt (experimental). Generates a small-form llms.txt index of all published content across every collection that has a urlPattern, grouped by collection label. Exposed on the plugin route llms/txt; serve it from Astro by proxying that route at /llms.txt. Enabled by default — flip the llms.txt (experimental) toggle to disable. Only the plain llms.txt is implemented; the llms-full.txt variant is out of scope.
  • buildLlmsTxt() exported for consumers who want to assemble the body from getEmDashCollection() results with custom sectioning or ordering.

[0.4.2] - 2026-04-09

Changed

  • Bump @jdevalk/seo-graph-core range from ^0.2.0 to ^0.3.0 and @jdevalk/astro-seo-graph range from ^0.2.1 to ^0.2.4. This dedupes seo-graph-core in consumer node_modules — astro-seo-graph@0.2.4 pins seo-graph-core@0.3.0 as a direct dep, and the plugin's previous ^0.2.0 range excluded 0.3.0 (pre-1.0 semver), so consumers ended up with two parallel copies of seo-graph-core. They now resolve to a single copy.
  • seo-graph-core@0.3.0 ships three additive improvements this plugin doesn't currently exercise directly: no more inLanguage: 'en-US' default on piece builders, optional WebPageInput.breadcrumb, and a generic type parameter on buildOrganization. No behavioural change in this plugin's output.

[0.4.1] - 2026-04-09

Changed

  • Bump @jdevalk/seo-graph-core range from ^0.1.0 to ^0.2.0 to pick up the new core release. seo-graph-core@0.2.0 adds a generic type parameter to buildOrganization (flows schema-dts subtype autocomplete into the extra field) and makes WebPageInput.breadcrumb optional. Both improvements are additive; this plugin doesn't currently call those piece builders directly, but the bump keeps the installed seo-graph-core version in sync with what @jdevalk/astro-seo-graph@^0.2.1 pulls in, avoiding duplicate seo-graph-core copies in node_modules.

[0.4.0] - 2026-04-09

Added

  • hreflang alternates for multilingual sites. When EmDash's Astro i18n config defines multiple locales and content entries are linked via translation_group, the plugin now emits one <link rel="alternate" hreflang="…" href="…"> per published sibling plus an automatically-resolved x-default entry. Self-referential entries are included. URLs are built from each collection's urlPattern + the locale's Astro prefix rules, so every hreflang target matches the canonical URL of that page.
  • Region-tag hreflang output. BCP 47 tags are normalized on emission: fr-ca becomes fr-CA, zh-hant-hk becomes zh-Hant-HK. Sites that need fr-CA vs fr-FR as separate translations should use the code as the locale path (locales: ["en", "fr-ca", "fr-fr"] in astro.config.mjs), since EmDash core currently drops Astro's object-form { path, codes } shape.
  • buildPageUrl helper (src/urls.ts). Shared path-construction logic honoring urlPattern, prefixDefaultLocale, and the plugin's canonical-normalization rules (lowercase, collapsed slashes, trailing slash).

Changed

  • Runtime dependency on @jdevalk/astro-seo-graph. The hreflang work reuses buildAlternateLinks from astro-seo-graph for normalization, dedup, and x-default resolution. The helper is pure TypeScript, so this does not add any Astro runtime overhead — Astro is a peer dep satisfied transitively through EmDash.

Notes

Zero cost on single-locale sites: gated on isI18nEnabled() before any database call. When i18n is disabled, the hreflang path is a single boolean check and an early return.

[0.3.0] - 2026-04-09

Added

  • BreadcrumbList schema. Every non-homepage, non-404 page now emits a BreadcrumbList entity in the schema graph, with a matching breadcrumb: { "@id": ... } back-reference on the WebPage node. Two override layers:
    • Segment labels — a settings-level segment → display label map (editable in the admin UI under Breadcrumbs → Segment labels). Overrides the default title-cased segment name wherever that segment appears in a path. Example: blog → Blog relabels /blog/ on every post under it.
    • Page type rules — a settings-level pageType → ordered crumb list map (editable in the admin UI under Breadcrumbs → Page type rules). Advanced, JSON-edited. Each crumb is { label, href? } where label may contain the {title} placeholder and an omitted href resolves to the current canonical URL.
  • Default path derivation — when no rule matches, breadcrumbs are derived from page.path by walking segments, applying the label map or title-casing dashes, skipping numeric year/month segments (/2025/) and page/N pagination, and always using page.title for the final crumb.

Changed

  • Breaking: @id scheme migrated to makeIds() from @jdevalk/seo-graph-core. All entity @id values now match joost.blog's scheme for consistency across the SEO graph ecosystem. Notable shifts:
    • WebPage @id is now the canonical URL itself (previously ${url}#webpage).
    • WebSite @id is ${site}/#/schema.org/WebSite (previously ${site}/#website).
    • Person @id is ${personUrl}#/schema.org/Person (previously a name-hashed path). Set the personUrl field to relocate to e.g. /about-me/.
    • Organization @id is ${site}/#/schema.org/Organization/${slug} where slug is derived from the configured org name.
    • isPartOf/mainEntityOfPage references on Article updated to point at the new WebPage @id.
  • buildAuthorPerson and buildSiteEntity now share ids.person for Person sites, collapsing what were two near-identical nodes into one via assembleGraph's first-wins dedupe.
  • The schema orchestrator constructs a single IdFactory once per page and threads it through every piece builder, making @id generation testable and consistent.

Notes

This is a breaking change in the emitted JSON-LD @id values. Consumers that depend on specific @id strings (custom schema consumers, analytics that index by @id) will need to update. The overall graph shape is unchanged — only the identifier strings shift.

[0.2.0] - 2026-04-09

Added

  • Runtime dependency on @jdevalk/seo-graph-core — the shared schema.org graph infrastructure also used by joost.blog via @jdevalk/astro-seo-graph. Both consumers now emit structurally-identical graphs even though they build pieces from very different runtime contexts (EmDash's PublicPageContext here, Astro's content collections there).

Changed

  • src/schema/index.ts now uses assembleGraph from @jdevalk/seo-graph-core to wrap the graph, rather than manually constructing the { @context, @graph } envelope. This means both consumers share the same first-wins deduplication semantics out of the box. The individual piece builders (buildSiteEntity, buildWebSite, buildWebPage, buildArticle, buildAuthorPerson) remain EmDash-specific for now because EmDash's @id scheme, Organization-or-Person site entity dispatch, and WebSite SearchAction shape don't map cleanly onto the core's opinionated piece builders.
  • Piece builder return types aligned with core's GraphEntity type for type consistency across the ecosystem.

Notes

This release is a minor bump rather than a patch because introducing a runtime dependency is a material change in the plugin's install footprint, even though the JSON-LD output for existing pages is byte-identical to 0.1.3.

0.1.0 - 2026-04-03

Added

  • Meta description generation with configurable fallback chain
  • Meta robots directives with max-snippet, max-image-preview, and max-video-preview; noindex for search/utility pages
  • Canonical URL generation — absolute, normalized, with trailing slash and pagination support
  • Open Graph tags — og:title, og:type, og:image, og:url, og:description, og:site_name, og:locale
  • Twitter Card tags — summary_large_image with site handle
  • JSON-LD schema graph with Person/Organization, WebSite with SearchAction, WebPage/CollectionPage, and Article with author
  • Admin settings UI auto-generated from settingsSchema
  • Configurable site identity (Person or Organization), social profiles, title separator, and default description