path: root/doc/haunt.texi

\input texinfo
@c -*-texinfo-*-

@c %**start of header
@setfilename haunt.info
@documentencoding UTF-8
@settitle Haunt Reference Manual
@c %**end of header

@include version.texi

@copying
Copyright @copyright{} 2015 David Thompson@*

Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.  A
copy of the license is included in the section entitled ``GNU Free
Documentation License''.
@end copying

@dircategory Web development
@direntry
* Haunt: (haunt).       Haunt, the functional static site generator.
@end direntry

@titlepage
@title Haunt Reference Manual
@subtitle Using Haunt
@author The Haunt Developers

@page
@vskip 0pt plus 1filll
Edition @value{EDITION} @*
@value{UPDATED} @*

@insertcopying
@end titlepage

@contents

@node Top
@top Haunt

This document describes Haunt version @value{VERSION}, an extensible,
functional static site generator.

@menu
* Introduction::                About Haunt.
* Installation::                Installing Haunt.
* Tutorial::                    How to get started quickly.
* Command-line Interface::      Using Haunt from the command-line.
* Programming Interface::       Using the Haunt API in Scheme.
* Contributing::                How to contribute to Haunt.

* GNU Free Documentation License::  The license of this manual.
* Concept Index::               Concepts.
* Programming Index::           Data types, procedures, syntax, and variables.

@detailmenu
 --- The Detailed Node Listing ---

Installation

* Downloading::                 Downloading the source code.
* Requirements::                Software needed to build and run Haunt.
* Building::                    Building from source code.

@end detailmenu
@end menu

@node Introduction
@chapter Introduction

Haunt is a hackable static site generator written in Guile Scheme.  A
static site generator assists an author with generating the HTML pages
that they publish on the web.  Unlike ``content management systems''
such as Wordpress or Drupal, static site generators are not dynamic
web applications (complete with a relational database) that build
pages on-the-fly.  Instead, web pages are built in advance, on the
author's computer, and copied to a web server when it is time to
publish changes.  The consequence of this design is that the web
server no longer needs to run a complex, potentially insecure web
application that connects to a database to retrieve data.  Static
files can be served easily by any generic web server.  Since there is
no web application or database server to deal with, static websites
are easier to maintain, more secure, and resistant to high web traffic
(``slashdotting.'')  Furthermore, the entire website is stored in
plain text, which allows the files to be version-controlled rather
than kept in a relational database with no concept of history that
needs to be backed up regularly.

At the time that Haunt was conceived, there existed literally hundreds
of other static site generators.  Why add another one?  Haunt
differentiates itself from most other static site generators in that
it aspires to the Emacs philosophy of ``practical software freedom.''
Not only is the source code available under a Free Software license,
as most static site generators are, it is designed to be easily hacked
and extended without altering the core source code.  Haunt
purposefully blurs the line between document and program, author and
programmer, by embracing the notion of data as code.  A Haunt-based
website is not simply data, but a computer program.  This design
strategy encourages authors to automate repetitive tasks and empowers
them to extend the software with their own ideas.

To make such a system work well, a general-purpose, extensible
programming language is needed.  A traditional configuration file
format simply will not do.  The programming language that we feel is
best suited to this task is Scheme, a clean and elegant dialect of
Lisp.  We believe that by giving authors the full expressive power of
Scheme, they will be able to produce better websites and make better
use of their time than with less programmable systems and less capable
programming languages.  Authors should feel empowered to hack the
system to make it do what they want rather than what some programmer
decided they should want.  And perhaps most importantly, building
websites with Haunt should be @emph{fun}.

Websites written in Haunt are described as purely functional programs
that accept ``posts'', text documents containing arbitrary metadata,
as input and transform them into complete HTML pages using Scheme
procedures.  Haunt has no opinion about what markup language authors
should use to write their posts and will happily work with any format
for which a ``reader'' procedure exists.  Likewise, Haunt also has no
opinion about how authors should structure their sites, but has sane
defaults.  Currently, there exist helpful ``builder'' procedures that
do common tasks such as generating a blog or Atom feed.  While the
built-in features may be enough for some, they surely will not be
enough for all.  Haunt's Scheme API empowers authors to easily tweak
existing components, write replacements, or add entirely new features
that do things no else has thought to do yet.

Happy haunting!

@node Installation
@chapter Installation

@menu
* Downloading::                 Downloading the source code.
* Requirements::                Software needed to build and run Haunt.
* Building::                    Building from source code.
@end menu

@node Downloading
@section Downloading

Official Haunt source code release tarballs can be found on the
@url{http://haunt.dthompson.us/downloads.html, downloads page} of
Haunt's website, along with their associated checksums.

@node Requirements
@section Requirements

Haunt depends on the following packages:

@itemize
@item
@url{https://gnu.org/software/guile, GNU Guile} version 2.0.11 or
later
@end itemize

The following dependencies are optional:

@itemize
@item
@url{http://www.nongnu.org/guile-reader/, Guile-Reader} version 0.6 or
later (for Skribe support)
@item
@url{https://github.com/OrangeShark/guile-commonmark, guile-commonmark}
version 0.1 or later (for CommonMark support)
@end itemize

@node Building
@section Building

Haunt uses the standard GNU build system, so the basic installation
procedure looks like this:

@example
./configure
make
make install
@end example

@node Tutorial
@chapter Tutorial

The goal of this tutorial is to quickly create a barebones blog with
Haunt in order to demonstrate the basic workflow and key concepts.

First, create a directory for the new site:

@example
mkdir haunt-tutorial
cd haunt-tutorial
@end example

Next, create the site configuration file @file{haunt.scm}.  This is
where all of the code for building the website will go.

Here's what a simple Haunt configuration looks like:

@example
(use-modules (haunt asset)
             (haunt site)
             (haunt builder blog)
             (haunt builder atom)
             (haunt reader skribe))

(site #:title "My First Haunt Site"
      #:domain "example.com"
      #:default-metadata
      '((author . "Eva Luator")
        (email  . "eva@@example.com"))
      #:readers (list skribe-reader)
      #:builders (list (blog)
                       (atom-feed)
                       (atom-feeds-by-tag)))
@end example

Haunt represents the full configuration of the website using the
@code{site} procedure.  Site objects specify things like the site
title, the default metadata to use for posts, which markup formats are
supported, and which builders are used to generate web pages.

With the above code saved into the @file{haunt.scm} file, the next
step is to create a @file{posts} directory and populate it with
articles to publish.  Put the text below into a file named
@file{posts/hello.skr}:

@example
(post
 :title "Hello, World!"
 :date (make-date* 2015 10 15)
 :tags '("hello")

 (h1 [Hello, World!])

 (p [This is my very first Skribe document!]))
@end example

This is a
@url{http://www.nongnu.org/skribilo/doc/user-3.html#skribe-syntax,
Skribe} document.  Skribe is one of the built-in languages that Haunt
knows how to work with.  It's basically Scheme, but with support for
writing literal text without quoting it all by enclosing it in square
brackets.  The code above defines a post named ``Hello, World!'' with
a publishing date of 2015-10-15, whose contents are just a single
heading and a paragraph.

To build the site, run @command{haunt build} to compile all of the
HTML pages.  To view the results, run @command{haunt serve} and visit
@url{http://localhost:8080} in a web browser.  @command{haunt serve}
is a handy utility that serves the contents of the website using
Guile's built-in HTTP server.  Since the blog builder was specified in
@file{haunt.scm}, the default index page is a simple listing of all
posts, which for now is a single post.  Clicking on the post title
will display a page with only that post's contents.

In addition to the basic blog builder, the @file{haunt.scm} file
specifies two additional builders for Atom feeds.  The
@code{atom-feed} builder creates a feed of all posts located at
@url{http://localhost:8080/feed.xml}.  The @code{atom-feeds-by-tag}
builder creates one feed for each unique tag specified in the post
metadata.  There's only one tag right now, ``hello'', and its feed is
located at @url{http://localhost/feeds/tags/hello.xml}.

Tweaking a post, rebuilding the site, and viewing the results in a web
browser is the typical Haunt workflow.  However, having to run
@command{haunt build} every after each edit is tedious.  To address
this, run @command{haunt serve --watch}.  The Haunt web server, in
addition to serving web pages, will now watch for changes to important
files and automatically rebuild the site when they are edited.  This
streamlines the workflow into an edit, save, view loop.

Now that we've introduced the basic utilities and concepts, continue
reading this manual to learn more about Haunt's command-line and
programming interfaces.

@node Command-line Interface
@chapter Command-line Interface

@menu
* Invoking haunt build::        Build the website.
* Invoking haunt serve::        Serve the website over HTTP.
@end menu

The Haunt command-line interface is composed of many subcommands.  The
general syntax for all Haunt commands is:

@example
haunt @var{subcommand} @var{options}@dots{}
@end example

@node Invoking haunt build
@section Invoking @command{haunt build}

The @command{haunt build} command realizes a Haunt site configuration
file by compiling web pages and copying static assets to the output
directory.  For details on how to configure a Haunt site,
@pxref{Sites}.

Example:

@example
haunt build --config=haunt.scm
@end example

@table @code

@item --config=@var{configuration-file}
@itemx -c @var{configuration-file}
Load the Haunt site declaration from @var{configuration-file}.

@end table

@node Invoking haunt serve
@section Invoking @command{haunt serve}

The @command{haunt serve} command allows one to quickly view a local
preview of the generated website before publishing the finished
product to a remote web server.  When @command{haunt serve} runs, a
local HTTP server is spawned.  Visit the server using a web browser to
inspect the results of the build.  By default, the web server listens
on port 8080, so the URL to visit would be
@url{http://localhost:8080}.

While developing, it is common to rebuild the site frequently to view
the results of incremental changes.  Rather than manually running
@command{haunt build} (@ref{Invoking haunt build}) each time changes
are made, the @code{--watch} flag can be used to automatically rebuild
the site when a source file changes on disk.

@table @code

@item --config=@var{configuration-file}
@itemx -c @var{configuration-file}
Load the Haunt site declaration from @var{configuration-file}.

@item --port=@var{port}
@itemx -p @var{port}

Listen for HTTP requests on @var{port}.

@item --watch
@itemx -w

Automatically rebuild the site when source files change.

@end table

@node Programming Interface
@chapter Programming Interface

@menu
* Sites::                       Description of the site and how to build it.
* Posts::                       Articles, prose, blog posts, etc.
* Readers::                     Post interpreters.
* Pages::                       HTML/XML pages.
* Assets::                      Images, stylesheets, etc.
* Builders::                    Web page builders.
@end menu

Haunt is a fully-programmable system composed of several Guile Scheme
modules.  This section documents the public API.

@node Sites
@section Sites

@example
(use-modules (haunt site))
@end example

A site object defines all of the properties for a Haunt website: The
site name, domain name, where blog posts are found, what post formats
are understood, which procedures are used to build the site, where the
output files are written to, etc.

@deffn {Scheme Procedure} site [#:title "This Place is Haunted"] @
       [#:domain "example.com"] [#:posts-directory "posts"] @
       [#:file-filter @code{default-file-filter}] @
       [#:build-directory "site"] [#:default-metadata '()] @
       [#:make-slug @code{post-slug}] [#:readers '()] @
       [#:builders '()]
Create a new site object.  All arguments are optional:

@table @var

@item title
The name of the site.

@item posts-directory
The directory where posts are found.

@item file-filter
A predicate procedure that returns @code{#f} when a post file should
be ignored, and @code{#t} otherwise.  Emacs temporary files are
ignored by default.

@item build-directory
The directory that generated pages are stored in.

@item default-metadata
An alist of arbitrary default metadata for posts whose keys are
symbols.

@item make-slug
A procedure generating a file name slug from a post.

@item readers
A list of reader objects for processing posts.

@item builders
A list of procedures for building pages from posts.

@end table

@end deffn

@deffn {Scheme Procedure} site? @var{obj}
Return @code{#t} if @var{obj} is a site object.
@end deffn

@deffn {Scheme Procedure} site-title @var{site}
Return the title of @var{site}.
@end deffn

@deffn {Scheme Procedure} site-domain @var{site}
Return the domain of @var{site}.
@end deffn

@deffn {Scheme Procedure} site-posts-directory @var{site}
Return the posts directory for @var{site}.
@end deffn

@deffn {Scheme Procedure} site-file-filter @var{site}
Return the file filter procedure for @var{site}.
@end deffn

@deffn {Scheme Procedure} site-build-directory @var{site}
Return the build directory of @var{site}.
@end deffn

@deffn {Scheme Procedure} site-make-slug @var{site}
Return the slug constructor for @var{site}.
@end deffn

@deffn {Scheme Procedure} site-readers @var{site}
Return the list of reader procedures for @var{site}.
@end deffn

@deffn {Scheme Procedure} site-builders @var{site}
Return the list of builder procedures for @var{site}.
@end deffn

@node Posts
@section Posts

@example
(use-modules (haunt post))
@end example

Posts represent the articles that are kept in a site's post directory
and written in a markup format that Haunt can
understand. @pxref{Readers}) for how files on disk can be transformed
into posts.

@deffn {Scheme Procedure} make-post @var{file-name} @var{metadata} @var{sxml}
Create a new post object that represents the contents of the file
@var{file-name}.  The body of the post, @var{sxml}, is represented as
an SXML tree (@pxref{SXML, SXML,, guile, GNU Guile Reference Manual})
and the metadata is an association list (@pxref{Association Lists,
Association Lists,, guile, GNU Guile Reference Manual}) of arbitrary
key/value pairs.
@end deffn

@deffn {Scheme Procedure} post? @var{object}
Return @code{#t} if @var{object} is a post.
@end deffn

@deffn {Scheme Procedure} post-file-name @var{post}
Return the file name for @var{post}.
@end deffn

@deffn {Scheme Procedure} post-metadata @var{post}
Return the metadata association list for @var{post}.
@end deffn

@deffn {Scheme Procedure} post-sxml @var{post}
Return the SXML tree for @var{post}.
@end deffn

@deffn {Scheme Procedure} post-ref @var{post} @var{key}
Return the metadata value corresponding to @var{key} within
@var{post}.
@end deffn

@deffn {Scheme Procedure} post-slug @var{post}
Transform the title of @var{post} into a URL slug suitable for the
file name of a web page.
@end deffn

@defvr {Scheme Variable} %default-date
The default date of a post when no other date is specified in the
metadata association list.
@end defvr

@deffn {Scheme Procedure} post-date @var{post}
Return the date for @var{post}, or @code{%default-date} if no date is
specified.
@end deffn

@deffn {Scheme Procedure} posts/reverse-chronological @var{posts}
Sort @var{posts}, a list of posts, in reverse chronological order.
@end deffn

@deffn {Scheme Procedure} posts/group-by-tag @var{posts}
Create an association list of tags mapped to the posts in the list
@var{posts} that used them.
@end deffn

@node Readers
@section Readers

@example
(use-modules (haunt reader))
@end example

The purpose of a reader is to translate the markup within a post file
into an SXML tree representing the HTML structure and associate some
metadata with it.

@deffn {Scheme Procedure} make-reader @var{matcher} @var{proc}
Create a new reader.  The reader is to be activated when
@var{matcher}, a procedure that accepts a file name as its only
argument, returns @code{#t}.  When a post file matches, the procedure
@var{proc}, which also accepts a file name as its only argument, reads
the contents and returns a post object (@pxref{Posts}).
@end deffn

@deffn {Scheme Procedure} reader? @var{object}
Return @code{#t} if @var{object} is a reader.
@end deffn

@deffn {Scheme Procedure} reader-matcher @var{reader}
Return the match procedure for @var{reader}.
@end deffn

@deffn {Scheme Procedure} reader-proc @var{reader}
Return the read procedure for @var{reader}.
@end deffn

@deffn {Scheme Procedure} reader-match? @var{reader} @var{file-name}
Return @code{#t} if @var{file-name} is a file supported by
@var{reader}.
@end deffn

@deffn {Scheme Procedure} read-post @var{reader} @var{file-name} [@var{default-metadata}]
Read a post object from @var{file-name} using @var{reader}, merging
its metadata with @var{default-metadata}, or the empty list if not
specified.
@end deffn

@deffn {Scheme Procedure} read-posts @var{directory} @var{keep?} @var{readers} [@var{default-metadata}]
Read all of the files in @var{directory} that match @var{keep?} as
post objects.  The @var{readers} list must contain a matching reader
for every post.
@end deffn

@deffn {Scheme Procedure} make-file-extension-matcher @var{ext}
Create a procedure that returns @code{#t} when a file name ends with
``.ext''.
@end deffn

@defvr {Scheme Procedure} sxml-reader
A basic reader for posts written as Scheme code that evaluates to an
an association list.  The special key @code{content} contains the post
body as an SXML tree.

Example:

@example
(use-modules (haunt utils))

`((title . "Hello, world!")
  (date . ,(string->date* "2015-04-10 23:00"))
  (tags "foo" "bar")
  (summary . "Just a test")
  (content
   ((h2 "Hello!")
    (p "This is Haunt.  A static site generator for GNU Guile."))))
@end example

@end defvr

@defvr {Scheme Procedure} html-reader
A basic reader for posts written in plain ol' HTML.  Metadata is
encoded as the @code{key: value} pairs, one per line, at the beginning
of the file.  A line with the @code{---} sentinel marks the end of the
metadata section and the rest of the file is encoded as HTML.

Example:

@example
title: A Foo Walks Into a Bar
date: 2015-04-11 20:00
tags: bar
---
<p>
  This is an example using raw HTML, because Guile doesn't have a
  Markdown parser.
</p>
@end example

@end defvr

@node Pages
@section Pages

@example
(use-modules (haunt page))
@end example

Page objects represent files that have yet to be written to disk.
Their contents may be any arbitrary object that their writer procedure
knows how to serialize.  In practice, pages are almost always written
to disk as HTML or XML.

@deffn {Scheme Procedure} make-page @var{file-name} @var{contents} @var{writer}
Create a new page object.  The string @var{file-name} specifies where
the page should be written to in the file system.  The procedure
@var{writer} is responsible for serializing @var{contents}.
@end deffn

@deffn {Scheme Procedure} page? @var{object}
Return @code{#t} if @var{object} is a page object.
@end deffn

@deffn {Scheme Procedure} page-file-name @var{page}
Return the file name string for @var{page}.
@end deffn

@deffn {Scheme Procedure} page-contents @var{page}
Return the contents of @var{page}.
@end deffn

@deffn {Scheme Procedure} page-writer @var{page}
Return the writer procedure @var{page}.
@end deffn

@deffn {Scheme Procedure} write-page @var{page} @var{output-directory}
Write @var{page} to @var{output-directory}.
@end deffn

@node Assets
@section Assets

@example
(use-modules (haunt asset))
@end example

Assets represent files on disk that should be copied verbatim to a
site's output directory.  Common types of assets include CSS,
JavaScript, images, and fonts.

@deffn {Scheme Procedure} make-asset @var{source} @var{target}
Create a new asset object.  The @var{source} and @var{target}
arguments are file names that are relative to a site source and target
directory, respectively (@pxref{Sites}).
@end deffn

@deffn {Scheme Procedure} asset? @var{object}
Return @code{#t} if @var{object} is an asset object.
@end deffn

@deffn {Scheme Procedure} asset-source @var{asset}
Return the source file name for @var{asset}.
@end deffn

@deffn {Scheme Procedure} asset-target @var{asset}
Return the target file name for @var{asset}.
@end deffn

@deffn {Scheme Procedure} install-asset @var{asset} @var{prefix}
Install the source file of @var{asset} into the target directory
within @var{prefix}.
@end deffn

@deffn {Scheme Procedure} directory-assets @var{directory} @var{keep?} @var{dest}
Create a list of asset objects to be stored within @var{dest} for all
files in @var{directory} that match @var{keep?}, recursively.
@end deffn

@node Builders
@section Builders

@menu
* Static Assets::              Images, CSS, JavaScript, etc.
* Blog::                       Dear diary...
* Atom::                       Atom feeds.
@end menu

Builders are procedures that return one or more page objects
(@pxref{Pages}) when applied.  A builder accepts two arguments: A site
(@pxref{Sites} and a list of posts (@pxref{Posts}).

Haunt comes with a few convenient builders to help users who want to
create a simple blog with an Atom feed.

@node Static Assets
@subsection Static Assets

@example
(use-modules (haunt builder assets))
@end example

@deffn {Scheme Procedure} static-directory @var{directory} @
       [@var{dest} @var{directory}]

Create a builder procedure that recursively copies all of the files in
@var{directory}, a file names relative to a site's source directory,
and copies them into @var{dest}, a prefix relative to a site's target
output directory.  By default, @var{dest} is @var{directory}.
@end deffn

@node Blog
@subsection Blog

@example
(use-modules (haunt builder blog))
@end example

@deffn {Scheme Procedure} theme [#:name #:layout #:post-template #:collection-template]
Create a new theme named @var{name}.

The procedure @var{layout} accepts three arguments: a site, a page
title string, and an SXML tree.  Its purpose is to wrap the contents
of a post with the theme's header/footer and return the complete SXML
tree for a web page.

The procedure @var{post-template} accepts a single argument: a post.
Its purpose is to return an SXML tree containing the contents of the
post, applying any desired post-processing operations.  The values
returned from this procedure will be wrapped in the theme's layout.

The procedure @var{collection-template} accepts four arguments: a
site, a title string, a list of posts, and a URL prefix string.  Its
purpose is to return an SXML tree containing the body of the
collection page.  The values returned from this procedure will be
wrapped in the theme's layout.
@end deffn

@deffn {Scheme Procedure} theme? @var{object}
Return @code{#t} if @var{object} is a theme object.
@end deffn

@deffn {Scheme Procedure} blog [#:theme #:prefix #:collections]
Create a builder procedure that transforms a list of posts into pages
decorated by @var{theme}, a theme object, whose URLs start with
@var{prefix}.

Additionally, this builder creates pages that aggregate previews of
many posts corresponding to what is specified in the list
@var{collections}.  Each collection is a three element list in the
form @code{(title file-name filter)}.

@table @var

@item title
The human readable name of the collection.

@item file-name
The HTML file that will contain the rendered collection.

@item filter
A procedure that accepts a list of posts as its only argument and
returns a new list of posts.  The filter procedure is used to remove
and/or sort the posts into the desired form for the collection.  For
example, a filter could sort posts in reverse chronological order or
select all posts that are written by a particular author.

@end table

By default, a single collection is created that lists posts in reverse
chronological order and writes to @file{index.html}.

The default theme is intended only for testing purposes.

@end deffn

@node Atom
@subsection Atom

@example
(use-modules (haunt builder atom))
@end example

@deffn {Scheme Procedure} atom-feed [#:file-name #:subtitle #:filter #:max-entries #:blog-prefix]
Return a builder procedure that renders a site's posts as an Atom
feed.  All arguments are optional:

@table @var

@item file-name:
The page file name.  The default is @file{feed.xml}.

@item subtitle
The feed subtitle.  The default is ``Recent Posts''.

@item filter
The procedure called to manipulate the posts list before rendering.
The default is to keep all posts and sort them in reverse
chronological order.

@item max-entries
The maximum number of posts to render in the feed.  The default is 20.
@end table

@end deffn

@deffn {Scheme Procedure} atom-feeds-by-tag [#:prefix #:filter #:max-entries #:blog-prefix]
Return a builder procedure that renders an atom feed for every tag
used in a post.  All arguments are optional:

@table @var

@item prefix
The directory in which to write the feeds.  The default is
@file{feeds/tags}.

@item filter
The procedure called to manipulate the posts list before rendering.
The default is to keep all posts and sort them in reverse
chronological order.

@item max-entries
The maximum number of posts to render in each feed.  The default is
20.

@end table

@end deffn

@node Contributing
@chapter Contributing

@c *********************************************************************
@node GNU Free Documentation License
@appendix GNU Free Documentation License

@include fdl-1.3.texi

@c *********************************************************************
@node Concept Index
@unnumbered Concept Index
@printindex cp

@node Programming Index
@unnumbered Programming Index
@syncodeindex tp fn
@syncodeindex vr fn
@printindex fn

@bye

@c Local Variables:
@c ispell-local-dictionary: "american";
@c End: