A PieCrust website is really just a directory with a special structure. The very
basic structure is a directory with a
config.yml file in it – but that would
be just an empty website that uses the default theme!
<root> +--- config.yml
You can create that directory with the configuration file in it yourself, but
chef init command can also do it for you.
Once you’ve started filling it up with contents and assets – and assuming a default website configuration – it will look a bit like this:
<root> |---- assets/ | |--- css/ | |--- img/ | +--- js/ |---- pages | |--- about.md | |--- blog.md | +--- projects/ | |--- bar.md | +--- foo.md |---- posts | |--- 2014-10-23_new-foo-project.md | +--- 2014-10-29_some-stuff-about-something.md |---- templates | |--- default.html | |--- blog.html | +--- post.html +---- config.yml
The config file
The main thing that differentiates any directory from a PieCrust website is the
config.yml file that’s in it. That’s what
chef looks for in order to know
where your site root is – and it will look in parent directories too, which
means it will work also if you’re in a sub-directory of your website.
config.yml file is found,
chef will return and print an error about
A PieCrust website has a couple of special directories that you probably don’t want to mess around with (unless you know what you’re doing).
_cache: this directory contains cached information and intermediate files that allow PieCrust to run faster. The
chef purgecommand will delete it, which can be necessary if it has been corrupted, or if you want to start fresh.
_counter: this is the default output directory for a bake (
chef bakecommand). That’s where the static version of your site would be generated, for you to upload to your public server. Like the
_cachedirectory, you can also safely delete
_counter, and PieCrust will just generate it again. And of course, you can always pass a different output directory to
chef bakeso that you never see
PieCrust will only look for content (mostly pages) in directories that you point it to.
All of this is configurable, of course, but by default, these are:
pages: that’s where PieCrust will look for all your pages. Any file in there with a
.textextension (among others) will be turned into a page.
posts: just like
pages, PieCrust will look for all your posts in this directory, but it will expect filenames to be formatted a certain way – namely,
YYYY-MM-DD_title-of-post. In PieCrust terms, those 2 directories are not really different, as they’re both “page sources” with different source types. For more information on this, see Content Model.
templates: Where pages and posts define the contents of said pages and posts, files found in the
templatesdirectory define the layouts and other re-usable bits of markup that are used by those pages and posts.
All those directories are configurable through the website configuration. This means that if, instead of “pages” and “posts” (which is a content model suitable for a blog) you wanted “products”, “updates”, and “pages” (which would be suitable for a company’s website), you can totally do that.
PieCrust will look for assets in the aptly named
favicon.ico, etc. PieCrust comes with a very capable built-in asset pipeline,
and, in most cases, putting that stuff in there will “just work”.
However, if you have advanced requirements for how you want your assets to be
processed, or if you just prefer using another asset pipeline (like Grunt or
Gulp), there are also simple ways to not have any
assets directory, so that
PieCrust effectively only takes care of your actual content. Again, see the
website configuration page.
Any other directory or file will be ignored by PieCrust. If you customized your website configuration on that front, anything not specifically mentioned in it will be ignored.
This means that directories like
node_modues are free to
co-exist with PieCrust.