staticpipes 0.0.1

StaticPipes, the flexible and extendable static site website generator in Python

StaticPipes, the flexible and extendable static site website generator in Python

Most static website generators have technologies, conventions and source code layout requirements that you have to follow.

Instead this is a framework and a collection of pipelines to process your source files. Use only the pipelines you want and configure them as you need.

If you are a python programmer and need something different, then write a python class that extends our base class and write what you need.

Install

• pip install staticpipes[allbuild] - if you just want to build a website
• pip install staticpipes[allbuild,dev] - if you want to develop a website

If you are developing the actual tool, check it out from git, create a virtual environment and run python3 -m pip install --upgrade pip && pip install -e .[allbuild,dev,staticpipesdev]

Getting started

Configure this tool with a simple Python file.py in the root of your site. This copies files with these extensions into the _site directory:

from staticpipes.config import Config
from staticpipes.pipelines.copy import PipeCopy

import os

config = Config(
    pipes=[
        PipeCopy(extensions=["html", "css", "js"]),
    ],
)

if __name__ == "__main__":
    from staticpipes.cli import cli
    cli(config, os.path.join(os.path.dirname(os.path.realpath(__file__))), "_site")

Then run with:

python file.py build
python file.py watch

Use Jinja2 templates for html files:

from staticpipes.pipes.jinja2 import PipeJinja2

config = Config(
    pipes=[
        PipeCopy(extensions=["css", "js"]),
        PipeJinja2(extensions=["html"]),
    ],
    context={
        "title": "An example website",
    }
)

If you like putting your CSS and JS in a assets directory in your source, you can do:

config = Config(
    pipes=[
        PipeCopy(extensions=["css", "js"], source_sub_directory="assets"),
        PipeJinja2(extensions=["html"]),
    ],
    context={
        "title": "An example website",
    }
)

(Now assets/css/main.css will appear in css/main.css)

Version your assets:

from staticpipes.pipes.copy_with_versioning import PipeCopyWithVersioning

config = Config(
    pipes=[
        PipeCopyWithVersioning(extensions=["css", "js"]),
        PipeJinja2(extensions=["html"]),
    ]
)

(files like js/main.ceba641cf86025b52dfc12a1b847b4d8.js will be created, and that string will be available in Jinja2 variables so you can load them.)

Exclude library files like _layouts/base.html templates:

from staticpipes.pipes.exclude_underscore_directories import PipeExcludeUnderscoreDirectories

config = Config(
    pipes=[
        PipeExcludeUnderscoreDirectories(),
        PipeCopyWithVersioning(extensions=["css", "js"]),
        PipeJinja2(extensions=["html"]),
    ],
)

Minify your JS:

from staticpipes.pipes.javascript_minifier import ProcessJavascriptMinifier

config = Config(
    pipes=[
        PipeExcludeUnderscoreDirectories(),
        PipeJavascriptMinifier(),
        PipeCopyWithVersioning(extensions=["css"]),
        PipeJinja2(extensions=["html"]),
    ],
)

Use the special Process pipeline to chain together processes, so the same source file goes through multiple steps before being published. This minifies then versions JS, putting new filenames in the context for templates to use:

from staticpipes.pipes.process import PipeProcess
from staticpipes.processes.version import ProcessVersion
from staticpipes.processes.javascript_minifier import ProcessJavascriptMinifier

config = Config(
    pipes=[
        PipeExcludeUnderscoreDirectories(),
        PipeProcess(extensions=["js"], processors=[ProcessJavascriptMinifier(), ProcessVersion()]),
        PipeCopyWithVersioning(extensions=["css"]),
        PipeJinja2(extensions=["html"]),
    ],
)

Or write your own pipeline! For instance, if you want your robots.txt to block AI crawlers here's all you need:

from staticpipes.pipes.pipe_base.py import BasePipe

class PipeNoAIRobots(BasePipe):
    def start_build(self, current_info) -> None:
        r = requests.get("https://raw.githubusercontent.com/ai-robots-txt/ai.robots.txt/refs/heads/main/robots.txt")
        self.build_directory.write("/", "robots.txt", r.text)

config = Config(
    pipes=[
        PipeNoAIRobots(),
    ],
)

How it works

Instances of pipeline classes are created and passed to the config. The same instance is used throughout. This means if a pipeline wants to store information early on to use later, it can do. Pipelines classes should extend the staticpipes.pipe_base.BasePipe class.

Build stage

During building, the start_build method is called on each pipeline. Methods are always called on each pipeline in the order the pipelines are passed to the config.

Then, the build_file method is called on each pipeline for each file in the source directory. The order of files in the directory is not set and should not be relied on.

The end_build method is called on each pipeline.

A pipeline should deal with the file completely or not at all. Either it ignores it or it does something that ends with a method on self.build_directory being called to write some content to the site.

A pipeline can write zero to many files to the site for a single source file. For instance, a image processing pipeline could write multiple files at different resolutions for every image in the source.

A current_info object is passed to all methods. This contains information and can be used to set information.

A pipeline can mark a file as excluded (by setting current_info.current_file_excluded) , which means that later pipelines won't have build_file called for that file. However, they will have file_excluded_during_build called for each excluded file.

A context is maintained on current_info.context. This is a dictionary of values that is initially set in the configuration object but pipelines can read and modify. For example, an earlier pipeline might version a CSS file at a particular location and store the location in the context. A later pipeline might build Jinja2 templates with the context as temple variables so the html can actually load the CSS.

Prepare stage

Before the build stage is started a prepare stage is done. start_prepare is called on each pipeline, then prepare_file for each file, then end_prepare. This can be used to collect info before building. For example, see the PipeCopyWithVersioning pipeline that works out the filename for any file it will work with in the prepare stage. This ensures information about the new file name is already in the context before a single build method is called.

It's not possible to exclude any files during the prepare stage.

Watch mode

In watch mode, a normal build is done first. The start_watch method is then called on each pipeline. Then every time a file is changed, the file_changed_during_watch or file_changed_but_excluded_during_watch method is called for that file. There is no end_watch method, as the watch stage is ended by the user forcibly quitting the program.

Writing pipelines for watch mode can be more complicated than writing pipelines for build mode. This is due to the idea of dependencies. If the process of building source file A depends in some way on building source file B, when source file B changes then both files A and B must be rebuilt.

Dependencies are left up to each pipeline to handle. Generally the pipeline should build up dependency information during the prepare or build stage and cache it for use during the watch stage. During prepare and build stage a flag current_info.watch is set if watch will be called afterwards. This means pipelines can avoid doing any extra work tracking dependencies for the watch stage if it isn't going to be called.

If a pipeline has no possible interactions with dependencies it can usually use the same code for building. Just add this to the pipeline:

    def file_changed_during_watch(self, dir, filename, current_info):
        self.build_file(dir, filename, current_info)

If a pipeline does not overwrite the file_changed_during_watch method then it is considered not to support watch mode and the user will see a warning when using watch mode.

Multiple processes for each source file

If you want to set up a situation where every source file can go through more than one process you will want to use the special process pipeline. Pass this as a pipeline to the config and also pass instances of the processes for each file.

from staticpipes.pipes.process import PipeProcess
from staticpipes.processes.version import ProcessVersion
from staticpipes.processes.javascript_minifier import ProcessJavascriptMinifier

config = Config(
    pipes=[
        PipeProcess(extensions=["js"], processors=[ProcessJavascriptMinifier(), ProcessVersion()]),
    ],
)

Again, processes are class instances and the same class instance is used all the time. They should extend the staticpipes.pipes.process.BaseProcessor class. When that pipeline is called, the process_file method is called for every file. The process_current_info parameter has directory, filename and contents attributes and these should be changed as needed.

At the end of calling all the processes, the file will be written to the site. During the prepare stage, process_file is called but the results are not written to the build site.

This has the limitation that one source file must produce exactly one destination file.

Misc

Generally, the pipeline API is designed to be as easy to write pipelines for as possible while maintaining flexibility and power. Extend the base classes and overwrite as little or as many methods as you need.

More information and feedback

• https://github.com/StaticPipes/StaticPipes