06 — Blog, taxonomy and SEO#

6.1 Posts are pages#

A post is a row in pages with type = 'post'. It uses the same editor, the same block tree, the same version history and the same publish pipeline. There is no separate posts table and no separate editor. What posts add:

The right panel in the editor gains a Post tab holding exactly these. Nothing about the content editing experience changes.

6.2 Permalinks#

The pattern lives in sites.options.permalink_pattern, default blog/{slug}.

The resolved value is written to pages.path, so public lookup stays one indexed query regardless of the pattern. Changing the pattern is an explicit action in Settings -> SEO that rewrites every post path and writes a 301 for each one; the screen states how many URLs will change before it runs, and it runs in chunks via cron if there are more than 200 posts.

6.3 Taxonomy#

• Categories nest (categories.parent_id), are ordered (position), and have a name, slug, description, image and their own SEO fields. A post may have several; primary_category_id decides breadcrumbs and {category} permalinks.
• Tags are flat and created inline from the editor (type a name, press enter). Tag creation from the editor requires the taxonomy.manage permission; a Writer without it can only pick existing tags.
• Deleting a category asks explicitly what happens to its children (move to parent, move to another category, delete them too) and to the posts in it (posts are never deleted — they lose the assignment). Deleting a tag just detaches it.
• Merging: Merge into... on a category or tag moves assignments and writes a 301 from the old archive URL to the new one.

Archive URLs:

Archive pages are rendered by a designable layout: Design -> Archives holds a template part tree for archive, so a non-technical owner can change the card design without touching Blade. The default archive template is a single section containing one post_list block pre-filtered by the archive context.

6.4 The post_list block#

In an archive context the block inherits the archive filter automatically (Source: archive context), which is how one block serves the index, category, tag and author pages.

Pagination uses ?page=N. A paginated URL is cached per page number and is noindex past page 1 by default (configurable), with correct rel=prev/next.

6.5 Scheduled publishing#

• Status scheduled with published_at in the future.
• php artisan schedule:run (the single cron entry) runs kodepress:publish-due every minute; it publishes every due page, clears each page from the cache, and marks the sitemap dirty.
• Timezone: comparisons use the site timezone (sites.timezone, default Asia/Dhaka) converted to UTC for storage. The admin UI always shows and accepts site-local time, labelled with the zone. Storage is UTC, display is site-local, and both go through the one shared date helper.
• If the cron has not run for a while, the next run publishes everything overdue in one pass and the Dashboard shows a warning when the last cron tick is older than 5 minutes — a missing cron entry is the single most common shared-hosting mistake and it must be visible, not silent.

6.6 Per-page SEO#

pages.seo keys, all optional, all editable in an SEO tab in the editor:

The SEO tab shows a live Google-style preview and a character counter with soft limits (60 / 160). It warns, it never blocks.

Emitted per page: <title>, description, canonical, og:*, twitter:card, hreflang alternates for translations, and JSON-LD — WebSite + Organization on the homepage, Article on posts (headline, image, datePublished, dateModified, author), BreadcrumbList where a hierarchy exists.

6.7 sitemap.xml and robots.txt#

• /sitemap.xml is generated and written to disk; it is regenerated by cron when marked dirty, and at most once per minute. It includes published pages and posts, category, tag and author archives, with lastmod from updated_at. It excludes noindex pages, the search page and anything behind auth.
• More than 5,000 URLs switches to a sitemap index with per-type child sitemaps.
• /robots.txt is generated from a template in Settings -> SEO, always containing the sitemap line. In maintenance mode it serves Disallow: / — and the Dashboard says so loudly, because shipping that to production by accident is a classic disaster.
• Settings -> SEO also holds: default title template, default description, default OG image, verification meta tags, and the analytics snippets (head/body) which are injected outside the cached body so changing them does not require a cache rebuild.

6.8 Redirects#

Matching order on a 404-bound request: exact from_path match first, then regex rules in id order. from_path is stored lower-cased without a leading slash; matching ignores the query string but preserves it on the redirect target. Loops are refused at creation time (A -> B when B -> A exists) and the resolver follows at most 3 hops before giving up with a 404.

The redirect screen shows hits and last-hit time, so dead rules can be pruned, and offers a CSV import/export.

6.9 Search#

Public search is MySQL-based: a LIKE-plus-relevance query over pages.title, pages.excerpt and a plain-text projection of the published tree stored in page_versions at publish time (a search_text column is not in the schema — the projection is stored in the file cache keyed by page and rebuilt on publish; see ADR-0006).

Bengali matters here: MySQL full-text with the default parser does not tokenise Bengali usefully, so the search normalises the query and the projection the same way (NFC, fold digits, strip zero-width joiners) and matches on substrings. Results are ranked title > excerpt > body. The search page is never cached and is noindex.

6.10 Tests for this document#

KodePress documentation · generated from the Markdown sources by tools/build-docs-site.py · internal preview, not indexed.