The ideas behind PhlatPage.
PhlatPage is built on a few ideas that hold everywhere: pages are folders, content is JSON, and every save is a git commit. Learn them once and the rest of the system follows.
PhlatPage is built on a few ideas that hold everywhere: pages are folders, content is JSON, and every save is a git commit. Learn them once and the rest of the system follows.
Pages, files and history: what a site is made of and how you find it.
A blog post, a gallery item and the dash are the same thing: a page in a tree, pointing at a view. There is no separate concept of a route, a content type or a dash URL. The dash is an addon mounted as ordinary pages, so learning pages covers the whole system.
A page is a folder. Its content is JSON next to the files it uses, so you can cd into a page, edit _data.json in any editor, or read the git log of the folder. What is on disk is what is served.
site/pages/
about/
_index.json # uuid, view, metadata
_data.json # title, body, whatever fields you add
Pages live in a git repository of their own, and every write is a commit, whether it comes from the dash, the CLI or your own code. Editors get history, restore and sync to a remote without touching git. Developers get a normal repository to branch, diff and deploy.
A view holds a template, a controller, a router and a field schema in one folder. Define "article" once and every article page gets the same layout, request handling and fields. A view can also bind a PHP class, so logic that belongs to the page lives on the page.
A SQLite registry maps pages to paths, views and fields, so a query does not walk the filesystem. It filters, sorts, pages and runs ranked full-text search. The registry is a cache: delete it and one command rebuilds it from the pages on disk.
How a page reaches a visitor, and who is allowed to see it.
Every route returns HTML, and HTMX drives each interaction over that same HTML. There is no JSON layer, no client-side router and no state to keep in sync. Any block of a page can be fetched on its own at a :blockname URL and cached.
Four independent settings decide whether a page is live, listed, open to everyone or locked against change. Roles and dot-notation permissions guard routes and templates. Editors work on autosaved drafts, and visitors never see a page until it is published.
Resize, crop, filter and draw text or shapes over an image with a short option string. Variants are generated on the first request and served as static files afterward. Files of an unpublished or gated page stay private and are served through PHP behind the page's access check.
{$page->cover->src('w=800,h=400,crop')}
{$page->cover->srcset(['thumbnail', 'card', 'hero'])}
Start from what ships, then replace it one piece at a time.
PhlatPage ships a dash, login and roles, ui elements, icons, fonts and a render template for every field. Put them together and a prototype or a small web app runs without custom CSS or routes. When the project grows, replace one piece at a time with your own views, routers, fieldtypes and markup, until the whole site is custom or only the parts that need to be.
{* a working page from shipped parts *}
<h1>{$page->title}</h1>
{$page->body|render}
{ui list, items: $page->children() /}
{* the same page, fully yours *}
{* site/views/@ui/list.latte replaces the shipped list *}
Everything in phlat/ is a default. A file at the same path under site/ replaces it, and an override can include the original and add to it. Addons add views, fields, routes and services without touching framework code.
Create users, scaffold views and pages, check and repair the registry, back up and restore from the terminal. The project ships a Deployer config. State is two folders to back up.
Nothing heavy. Nothing hidden.