Anchovy

Anchovy is a minimal, unopinionated file-processing framework equipped with a complete static website generation toolkit.

• Minimal: Anchovy’s core is around a thousand lines of code and has no mandatory dependencies. Plus, Anchovy can be used for real projects with just a few pip-installable extras, even if you want to preprocess CSS.

• Unopinionated: Anchovy offers a set of components which can be easily configured to your site’s exact requirements, without tediously ripping out or overriding entrenched behaviors. Anchovy does not assume you are building a blog or that you wish to design your templates in a specific way. You can even build things that aren’t websites! Plus, Anchovy operates on files, so it’s simple to integrate tools like imagemagick, dart-sass, or less.js if you need them.

• Complete: Anchovy comes with a dependency auditing system, allowing you to grab any component you want without installing anything but Anchovy and find out what you will need to run your build. Choose from a wealth of Steps, Anchovy’s modular file processors, for everything from rendering Jinja templates and minifying CSS to unpacking archives and thumbnailing images. Plus, add a few extra parameters or lines of configuration to get automatic intelligent minimum builds based on input checksums, and get a reproducible run artifact to boot— even if you want to fetch HTTP resources or write your own Steps. Iterate quickly by launching a lightweight development-grade web server once the build is complete.

Installation

Anchovy has no essential prerequisites and can be installed with pip install anchovy to get just the framework and a few built-in components, but for typical usage pip install anchovy[base] is recommended. This will pull in support for Jinja2 templating, markdown, minification, and Anchovy’s CSS preprocessor. A full list of available extras may be found in the pyproject.toml file.

Alternatively, Anchovy may be installed directly from source with pip install git+https://github.com/pydsigner/anchovy or the corresponding pip install git+https://github.com/pydsigner/anchovy#egg=anchovy[base].

Command Line Usage

Anchovy operates on config files written in Python, or even modules directly.

• python -m anchovy -h
• anchovy -m mypackage.anchovyconf -o ../release/
• python -m anchovy mysite/anchovy_site.py -h

If -h is supplied with a module or config file, help will be displayed for the pipeline configuration. If it is supplied without a module or config file, help will be displayed for anchovy's own run-time options, such as --audit-steps and --serve, as well as how to specify a module/config file.

Show Me

Run anchovy examples/code_index.py -s -p 8080, then open a browser to localhost:8080 (or click the link in the console). This example offers the most extensive demonstration of Anchovy’s functionality as of version 1.0.

What’s the Baseline?

Here’s minimal example performing about what the staticjinja markdown example offers:

from pathlib import Path

from anchovy import (
    DirectCopyStep,
    InputBuildSettings,
    JinjaMarkdownStep,
    OutputDirPathCalc,
    REMatcher,
    Rule,
)


# Optional, and can be overridden with CLI arguments.
SETTINGS = InputBuildSettings(
    input_dir=Path('site'),
    working_dir=Path('working'),
    output_dir=Path('build'),
    custody_cache=Path('build-cache.json'),
)
RULES = [
    # Ignore dotfiles found in either the input_dir or the working dir.
    Rule(
        (
            REMatcher(r'(.*/)*\..*', parent_dir='input_dir')
            | REMatcher(r'(.*/)*\..*', parent_dir='working_dir')
        ),
        None
    ),
    # Render markdown files, then stop processing them.
    Rule(
        REMatcher(r'.*\.md'),
        [OutputDirPathCalc('.html'), None],
        JinjaMarkdownStep()
    ),
    # Copy everything else in static/ directories through.
    Rule(
        REMatcher(r'(.*/)*static/.*', parent_dir='input_dir'),
        OutputDirPathCalc(),
        DirectCopyStep()
    ),
]

This example is very simple, but it’s legitimately enough to start with for a small website, and offers an advantage over other minimal frameworks by putting additional batteries within an arm’s reach. If we stored the configuration in config.py and added a raw site like this:

site/
    static/
        styles.css
        toolbar.js
    base.jinja.html
    index.md
    about.md
    contact.md

python -m anchovy config.py would produce output like this:

output/
    static/
        styles.css
        toolbar.js
    index.html
    about.html
    contact.html

This example can be found in runnable form as examples/basic_site.py in the source distribution. Available command line arguments can be seen by passing -h: python -m anchovy examples/basic_site.py -h.

Programmatic Usage

Anchovy is very usable from the command line, but projects desiring to customize behavior, for example by running tasks before or after pipeline execution, may utilize anchovy.cli.run_from_rules():

import time
from pathlib import Path

from anchovy.cli import run_from_rules
from anchovy.core import Context

from my_site.config import SETTINGS, RULES


class MyContext(Context):
    def find_inputs(path: Path):
        # Only process files modified in the last hour.
        hour_ago = time.time() - 3600
        for candidate in super().find_inputs(path):
            if candidate.stat().st_mtime > hour_ago:
                yield candidate


def main():
    print('Pretending to run pre-pipeline tasks...')
    run_from_rules(SETTINGS, RULES, context_cls=MyContext)
    print('Pretending to run post-pipeline tasks...')


if __name__ == '__main__':
    main()