app rssg

A rusty static-site generator leveraging shell commands

8 stable releases

1.8.0 Nov 15, 2023
1.7.3 Sep 17, 2023
1.7.2 Mar 30, 2023

#70 in Template engine

MIT license

56KB
1.5K SLoC

RSSG

A rusty static-site generator

Contents

Usage

An example website can be found in the example-site directory.

Use rssg -i path-to-new-site to create a new site.


Command-line usage

rssg [options]
     -h |    --help : print this help dialog
     -c | --compile : compile the site
     -i |    --init : create a new site
     -v | --verbose : include debug output
     -f |   --force : force recompilation
                      rebuilds cache
          --content : set source directory
                      defaults to `content`
           --output : set output directory
                      defaults to `output`
           --public : set public directory
                      defaults to `public`
            --clean : cleans the `output`
                      and `temp` directories

File structure

Websites use a structure to make compilation simpler. The structure looks like this:

my_site/
├── content/
│   └── ...
├── templates/
│   └── ...
├── public/
│   └── ...
├── output/
│   └── ...
└── rules.toml

All your pages go into the content directory. These files will be processed by the rules.toml (more about that below) which might run them through templates in the templates directory. The results are placed into the output directory. Any and all files in the public directory get copied into the output without any modifications, preserving directory structure.


rules.toml

The core of the generator is the rules.toml. It dictates what happens to everything in the content directory. It's made up of two things: filters and rules.

Filters

Filters take an input file, run a command (probably to change the file), and output a file. A command might look like pandoc {full} -o {outfile}. In this example, two substitutions are made: first, {full} gets replaced with the input files' full path; second, {outfile} gets replaced with the output file of the filter. After substitution, it might look something like this: pandoc content/index.md -o temp/<hash>/index.html.

All filters require an output file. This gets substituted, then substituted for {outfile} in the command. The valid substitutions are as follows:

  • {full}: The full path to the input file.
  • {dir}: The parent directories of the input file.
  • {name}: The filename of the input file (minus extension).
  • {ext}: The extension of the input file.

In the rules.toml, filters can be in a list at the top-level of the file. Here's an example that runs the file through pandoc, then outputs it without changing it's path (but updating the extension):

[[filters]]
# The name of the filter, to use in rules
name = "markdown"

# The command to run, with substitutions
# example/path.html -> `pandoc example/path.html -o temp/<..>/path.html`
command = "pandoc {full} -o {outfile}"

# The resulting file, with substitutions
# !!! NOT neccessarily where the file will be in the final output
# example/path.html -> <hash of command + filepath>/path.html
outfile = "{dir}/{name}.html"

You can also specify "inline" filters inside of a rule specification (see below), like so:

# ...
filters = [
    {command = "pandoc {full} -o {outfile}", outfile = "{dir}/{name}.html"}
]
# ...

NOTE: Filter outfiles are stored in the temp directory during generation, with unique directory names. This is irrelevant for site development.

Filters can also omit the outfile property. Filters like this do not output any information; as far as the other filters are concerned, they never existed. It is possible for such a filter to directly mutate the output from a previous filter, but this is inadvisable. These filters are intended for things like logging and generating sitemaps.

Filters, by default, never see the raw source file. Even first-layer filters only ever see a YAML-filtered version. However, if you have a filter without an outfile, you can specify give_original = true in order to get the unchanged source file path. Never use this to modify the source file, unless you have an exceptional reason.

Templates

Templates are files that you can use encapsulate other files. For example, you might have a default.html template that contains a header and footer to wrap your page content in. They reside in the templates directory, and have some substitution rules of their own:

  • {{data}}: The full data of the page you are embedding.
  • {{version}}: The version of rssg used to compile the page.
  • {{data.<key>}}: Data from the content file's frontmatter.

Note that, unlike command substitutions, these are enclosed in double brackets. Content files can have YAML frontmatter, to use in these substitutions. For example, you might have a title key in each page, and a title element in the template that uses the key. Frontmatter is enclosed on both sides by triple-dashes (---) and must be at the start. Here's an example (assuming a very basic markdown-to-html filter):

input.md

---
title: Homepage
---

# Hello, World!

template.html

<html>
    <head>
        <title>{{data.title}}</title>
    </head>
    <body>
        {{data}}
    </body>
</html>

output.html

<html>
    <head>
        <title>Homepage</title>
    </head>
    <body>
        <h1>Hello, World!</h1>
    </body>
</html>

Rules

Filters do nothing on their own; they have to be used inside of rules. Rules are composed of four components: a regex pattern (rule), a list of filters, (filters), a list of templates (templates), and an output (output). When they are applied to a file, first they apply each filter to it in sequence. Then they apply each template to it in sequence. The result is stored in the output path, prefixed with output. Once one rule has matched a file, no other rule can.

NOTE: Rules, unlike filters, store their output files directly in the output directory.

This rule matches all files ending in .md or .markdown, translates them to HTML, applies a template, then saves it in a directory named like it but with the filename index.html (this turns the url example.com/contact.html into example.com/contact/):

[[rules]]
# starts with any characters, ends in .md or .markdown
rule = ".*\\.(md|markdown)"

# Can also use inline filters, see [filters](#filters)
filters = ["markdown"]

# -> templates/default.html
templates = ["default.html"]

# example/path.html -> output/example/path/index.html
output = "{dir}/{name}/index.html"

Rules, like filters, can omit the output property. In this case, no templates will be applied, and no files/directories created. The same warning goes for rules as for filters; you really shouldn't mutate data from inside a no-output rule.

These are just the recommended style guidelines. Any other way to create a TOML list called rules, or filters, will work. This is just the cleanest way. If you need to change it up for whatever reason, check out the official TOML website.

Pre- and post-commands

In your rules.toml, you can add arbitrary commands to run before and after building your site. You can list your pre-commands in the root-level pre_commands list, and your post-commands in the post_commands list. Neither pre- nor post-commands undergo any substitution, they are run as-is and will cause a build to fail on a non-zero exit code.

Contributing

First, thank you for even considering contributing to the project!

There are several things that need improving right now. First of all, unit tests should really get made. Several other little changes would be nice as well; a watch-mode for development (watch the files and recompile when something changes), a more configurable log system, better logging, documentation, and general style improvements. This readme is indicative of the rest of the project; functional, but flawed.

Whatever you do, TEST it first! Use the provided example (or your own) to ensure correctness. If you add a feature, add a test for it.

Please run cargo clippy and cargo fmt before making any pull requests. These not only help with style and performance issues, but clippy can also inadvertently catch some bad bugs in your code. However, as long as your contributions are helpful and functional, I won't be a stickler for formatting.

Unless you specifically state otherwise, all contributions are licensed under the project license (MIT).

Never submit incomplete code (todo!(), unimplemented!(), etc.)

Copyright (c) 2023 Kyllingene, MIT license.

Dependencies

~6–8.5MB
~157K SLoC