csg

csg is the "complicated (static) site generator". It's a somewhat simple static site generator, written in a shell script, with many "poweruser" features thanks to awk. The shell script and underlying awk scripts are POSIX compatible.

The overall design is heavily inspired by Roman Zolotarev's ssg, however the codebase is completely different, with the added bonus of:

• arbitrary file modifications via awk
• git-based "created" and "edited" timestamps
• support for (multiple) description and title meta tags
• support for custom title format

How to use (for the layman)

Inside a "source" folder you put your content, structured in any way. Your webpages are written in Markdown and must end with .md. You'll also need to have a _header.html and _footer.html, both of those will be slapped on the top and bottom of your Markdown pages. Then run:

csg './source' './destination' '& - You' 'https://you.com'

• './source' is your source folder
• './destination' is where you want your generated website to go
• '& - You' is the format of your title, & takes the first h1 from the page and puts it in, with \\& you can put an actual ampersand
• 'https://you.com' is your website

If you want to use awk scripts for extended functionality, put them in a folder and reference it at the end of the command:

csg './source' './destination' '& - You' 'https://you.com' './awk-scripts'

Demo

This repository comes with a small demo setup. To test it out, run:

./csg ./demo/src ./demo/website '& - Demo' ./demo/awk-edits

You will find your website, ready to be served, inside ./demo/website.

Regarding awk

Additional features are available with awk. By default, the title, description, word count, date created and date edited are taken from every Markdown file. The title and description will be put in place of [TITLE] and [DESCRIPTION] inside _header.html. You can also create/use your own awk scripts for further page modification (all variables will also be available inside those scripts).

Writing your own scripts

There are some things you must consider while creating an awk script:

1. It will be ran on the generated HTML pages
2. It's name must start with a capital letter - You can also create so-called "libraries", mainly awk scripts with only function definitions - libraries must start with a # character in their name, and are included with every script
3. The current line mustn't be printed (unless you want to add a line to the file) - With awk we can modify only certain lines, but the easiest way to get the whole file with the modified lines, is to just print every line. Such a rule is "appended" to the end of your script. - Do modifications to $0 instead of print (fancy expression) - If you don't want to print a line, do $0 = 0
4. Every script must begin with a # regex line - For optimisation purposes, a script is ran only on the files which match the given regular expression - The regex is on the whole path, so you could make changes (for example) to specific pages or all pages inside folders - The regular expression is fed to find -regex, so for exact support view find's documentation
5. Script variables mustn't collide and you mustn't use exit or nextfile - Also for optimisation purposes, scripts with the same regex are combined into one
6. Inside your scripts you'll have the title, description, words, created and edited variables available The last two will be dates in the format "%d %B %Y", and are only available if your website uses git.

Usage of the script

csg comes with a help message, which contains everything you might want to know for running the script:

Usage: csg [-h | -? | --help]
       csg <source_dir> <destination_dir> <title_format> <base_url> [<scripts_dir>]

Options:
  -h, -?, --help    Print this help message (also available when given no
                    arguments)

Arguments:
  <source_dir>         Folder which contains source files
  <destination_dir>    Folder in which the generated site will be put
  <title_format>       Format string for page title
                       Accepts any text or HTML, & is replaced with contents
                       title (first h1 header).
  <base_url>           Base URL of your website (scheme + authority), without
                       trailing "/", used for sitemap generation.
                       Example: https://google.com
  <scripts_dir>        Optional, folder which contains awk script for editing of
                       generated webpages.

Variables (you can predefine):
  CSG_TIMESTAMP_FILE    File path for git-based timestamps (when git exists)
                        Default: "./timestamps.txt"
                        Disable: ""
  CSG_TMP_DIR           Directory to put temporary files
                        Default: "/tmp"
  CSG_AWK               awk executable and flags
                        Default: searches for gawk
  CSG_MD_TO_HTML        Program to convert Markdown to HTML and it'\''s flags
                        Default: searches for lowdown

Dependencies (required):
  bash, printf, find, cpio


csg - complicated (static) site generator
Created by Syndamia, https://gitlab.com/Syndamia/csg

Future features

1. The way awk scripts are integrated and ordered is somewhat messy and needs major rethinking. Ideally you should be able to run an awk script on the markdown, on the markdown generated html (as is already done) and on all final HTML pages (so you could create a list of latest articles with certain excerpts from them, for example). There is also the need to be able to reorder script execution easily, without script modifications (so when using someone else's script, you dont need to change it, hence you can include it with a git submodule or sync it with rsync). - The regex on top of awk scrits is for optimization purposes, but maybe this new system should be able to figure out when a script needs to be ran?
2. Right now the whole website is regenerated every time, it's important to add a system for updating only the currently changed pages. I'm not entirely sure how to do that, since we remove the source Markdown after it is converted to HTML, current thinking leads me to a file containing lines of "filepath hash", but we'll see.
3. Timestamps.txt but based on created/edited file stats
4. Need to add a _ignore file to not convert certain folders
5. Add native support for more awk parsers and markdown to html converters

Raison d'etre

Comparing to other "portable" generators:

(bash)

Nanoblogger - ?

Statix - requires GNU grep with certain compile flags - limited, template language

Ultra Simle Site Maker - ocaml?

utterson + drafts - way too complicated

vee + gopher support + editing last x - very limited abilities

simsalabash - way too bash-oriented

bashblog + drafts + tags - template "customisation" - not really extensible

dsw + arbitrary markup language

(awk)

zodiac - template language - template syntax makes source way too "ugly"

simple-static - no customization

mksite - templating

(c)

HSC - way too ugly formatter, more like fancy html

GSL iMatrix - way to big an heavy and fancy

blogc - way too big