--- title: Ampoule source-url: https://roundabout-host.com/roundabout/ampoule topics: ["software", "web", "python", "ampoule", "jinja2", "docs", "gpl"] --- Ampoule is a lightweight, simple yet flexible, static site generator written in Python. It uses Jinja2 for templating. This site was generated using Ampoule. Features -------- * **_Extremely_ simple and small**, only a few hundred lines of code. * _Only_ depends on Jinja2, Ruamel YAML, bs4, and colorama. * **Jinja2 templating** will be familiar to Flask users. Now you can use the same templates for both dynamic and static sites. * More of **a framework**. Sites are generated by a short **Python script** that you write to customise what **pages** it loads, which **templates** it uses, and what **data** it passes to them, or create custom **filters**, **tests** and more. * Supports **YAML front matter** for pages. It can be accessed using indexing syntax. * **Indexes** can be sorted using a function, iterated and can index any directory, recursively or not. They can also **transform URLs** to make them end in ".html". * **Object-oriented** design. The same objects used in that script can also be passed to the templates. * Any **markup language** can be used, as long as it can be converted to HTML. You just need to configure a filter for it. You can even mix multiple markup languages in the same site. * Ships with a light **markdown** implementation. * Easy to use for **_both_ programmers and non-programmers**. While you do need a script, you can also use an off-the-shelf one. * **Themes** can be *exactly how you want*. * Keeping **static files** is easy, because indexes can be static. * Static files are always **binary** and not templated. The same happens for files that can't be decoded. * **URL**-based definitions. Pages are added using the URL that will be used to access them. * Reinforces the **web** as a **publishing medium**. Static sites are not for everyone, but if you want to **publish** something, it's the best way. * And GitHub will give you **free hosting**, because it's static and *very cheap to serve*. Roundabout-host now also offers free hosting for static sites and will soon offer a way to generate them using CI and the generator you prefer. * It's **free software** and available under the **GPLv3**. * **No JavaScript** is required, but it can of course be used if you want. * Decently **fast**: even if you've got a huge site, it should not take more than *30 seconds*. Local rebuilding will also be added. And it's still much faster than any dynamic site. * Beautiful logging thanks to colorama. * Great for educational use; you can learn **Python**, **HTML**, **CSS**, **JavaScript**, and **Jinja2** all at once. * You can **make your site** in *an hour*, and then it's time to focus on writing what you want to publish. * If you see fit, it's easy to **convert** to a dynamic site. A **Flask implementation** is planned. * Clear and **magic-free**. You can see exactly what's happening and why. No magic, no configuration files, no hidden behaviour. The code is so short you can read it. Minimal example --------------- ```python import string from datetime import datetime import string import ampoule_ssg as ampoule from ampoule_ssg import markdown # Create a site object. This is where we are adding pages to. The argument is the directory # where the site will be built. site = ampoule.Site("my_site") # Use this as "| markdown" in Jinja2 templates to convert any Markdown source to HTML. @site.filter("markdown") def markdown_filter(text): return markdown.markdown2html(text) # Make the URLs web-friendly and make it end in ".html" so it will be correctly formatted # by dumb servers. def article_url(url): url = url.lower().rpartition(".")[0] new_url = "" for i in url: if i in string.ascii_lowercase: new_url += i elif i in string.whitespace: new_url += "-" return new_url + ".html" # Set context that will be passed to all templates. You can still override this. site.context["timestamp"] = datetime.now() site.context["ampoule"] = ampoule # Add the index of articles. In the template, we're looping over it to list them all. articles = ampoule.Index("articles", url_transform=article_url, sort_by=lambda x: x.date) # This makes it take all indexed files and put them under the /articles URL, keeping the # index's URL transformation and placing all of them in the article.html template. This # will be passed as "document" to the template. site.add_from_index(articles, "/articles", "article.html") # Create the main page which has access to the index so it can list all articles. main_page = ampoule.Page(site, "home.html", articles=articles) # Add the page. Note how we're binding it to a path; it will automatically be set as # index.html in that directory, and the URL is site-relative, not the OS root. site.add_page("/", main_page) # Add static files using a recursive static index. It will add all files in the static # directory and all its subdirectories, without putting them into templates. You could # still use them in templates, so you can make a photo gallery or something. site.add_from_index( # We're excluding Markdown files because we're using them as licence information # for when the site is distributed together with the fonts. You can exclude any # file you want using regex. ampoule.Index("static", recursive=True, exclude=r"\.md$", static=True), "/static", # There is no template, because the index is static. ) # Makes Ampoule take all pages and put them in a directory. site.build() ``` More information ---------------- ### Name origin An ampoule is smaller than a flask. Because it is related to Flask (it uses Jinja2) but is a much smaller static version of it, the name makes sense. ### What about the other static site generators? There are many static site generators out there, but they all have their own problems. In particular, I haven't seen one that uses code to describe the site, rather than a configuration file. This makes it much more flexible and powerful. Also, Ampoule is familiar to Python programmers, because it's written in Python and uses Jinja2, a templating engine that is also used in Flask. It's even the smallest static site generator: 1. Hugo: written in Go, uses go html/template, and it has 133k lines of Go, not counting comments or blanks. 2. Jekyll: written in Ruby, uses Liquid, and it has 17300 lines of Ruby, not counting comments or blanks. Interestingly, it's got more Markdown than Ruby. 3. Gatsby: they call it a framework, and rightfully so, because it's overkill for actually static (i.e. for publishing content) sites, even though JS people use it for precisely that purpose. It's written in JavaScript, uses React, and it's git 380k lines of JavaScript and TypeScript combined. (For comparison, it's over 1/100 of Linux itself, which is HUGE considering it uses a high-level language and only has to do so much.) 4. Pelican: written in Python, uses Jinja2, and it has 12400 lines of Python, not counting comments or blanks. 5. Docusaurus: written in TypeScript, uses React (of course, because it's made by Facebook), and it has 140k lines of TypeScript and JavaScript combined. 6. VuePress: written in JavaScript, uses Vue, and it has 11k lines of JavaScript, Vue and TypeScript combined. 7. Zola: written in Rust, uses Tera, and it has 17k lines of Rust, not counting comments or blanks. Also, it's designed to be monolithic and not extensible at all. Whereas I have only got 750 lines of Python, not counting comments or blanks. Add the script to generate the site, and it's still under 1000 lines. I don't want to criticise other static site generators, they all do some things well, but they're not what I want. I want a simple, small, flexible and versatile static site generator that is low-maintenance and easy to use. I don't know about you, but maybe you want the same thing. The JS-based ones are particularly unsuitable for most people, because they're slow, bloated, hard to install, and most often actually generate an SPA, which is not what you want for a blog or documentation or web book or anything like that. ### Why generated static sites? If you don't want generated static sites, you've got two other options. #### Dynamic sites * bloated; * slow; * requires smart server; * requires maintenance; * requires security; * requires a database; * hard to post content; * databases can't be managed with git; * hard to import content; * no free hosting; #### Static sites * hard to manage layouts; * hard to list the content; * hard to update indexes; * no support for metadata; * markup languages must be manually converted; With a *generated* static site, you get the best of both worlds. It's the best publishing platform, because it's just files, but it still provides the convenience of just writing content and having it magically appear on the site and formatted correctly. How to install -------------- Please note that this is not yet available on PyPI. For now you'll need to download the code (ideally using git) and install it with `pip` as a local package by giving it the path to the directory containing `setup.py`. Full documentation ------------------ To demonstrate just how easy it is, the docs can all fit on one page. ### class `ampoule_ssg.Site` `Site` is the main class of Ampoule; it represents a single website. It is responsible for handling added pages, the template engine and features, as well as building it. #### def `__init__(self, build_dir: typing.Union[str, bytes, os.PathLike], template_dir: typing.Union[str, bytes, os.PathLike] = "templates")` Create a new site object. `build_dir` is the directory where the site will be built. `template_dir` is the directory where the templates are stored. Both are relative to the script current working directory. #### def `add_page(self, location: typing.Union[str, bytes, os.PathLike], page: typing.union[Static, Page])` Add a page object to the site at the server-relative URL `location`. The page object can be either a `Static` or a `Page`. #### def `add_from_index(self, index: Index, location: typing.Union[str, bytes, os.PathLike], template: str = None, **kwargs)` Add all pages from an index to the site with the root at the server-relative URL `location`. The pages will be rendered with the template `template` and the context `kwargs`. will be passed to all of them. If the index is static, the pages will not be rendered with a template, but rather copied as-is. For each page, the `document` object found in the index will be passed to the template under that name. #### def `filter(self, name: str)` A decorator that registers a filter function with the site. The function should take at least one argument, the value to be filtered, and return the filtered value. #### def `test(self, name: str)` A decorator that registers a test function with the site. The function should take at least one argument, the value to be tested, and return a boolean. #### def `build(self, dont_delete: typing.Optional[list[str]] = None)` Build (save) the site to the build directory it was constructed with. This will create the directory if it does not exist, clear it (but not delete it) and then write all the pages. You can set `dont_delete` to a list of files that should not be deleted when the directory is cleared, for example, the `.git`. #### `context: dict[str, typing.Any]` A dictionary containing names that are available to all pages. It can be overriden by the page's context or modified at any time. ### class `ampoule_ssg.Page(str)` `Page` is a class that represents a single page on the site. A page is composed of a template, a document and a context. #### def `__new__(cls, site: Site, template: str, document: Document = None, **kwargs)` Create a new page object. `site` is the site object that the page belongs to. `template` is the template the document will be put in. `document` is the document object that will be passed to the template. `kwargs` are names that will be available to the template for additional context. If there's no document, it will not be available to the template. This is useful for single pages with fully static content, like a contact page. ### class `ampoule_ssg.Static(bytes)` `Static` is a class that represents a single static file on the site. A static file is just the content, in binary format, and it doesn't use templating. #### def `__new__(cls, site: Site, document: Document)` Create a new static object. `site` is the site object that the static file belongs to. `document` is the document object that will be written to the file; it can contain any encoding, even text, and will be written as-is. ### class `ampoule_ssg.Index` An index is a collection of documents that can be iterated over or added to a site using a common template (see `ampoule_ssg.Site.add_from_index`). #### def `__init__(self, directory: typing.Union[str, bytes, os.PathLike], recursive: bool = False, url_transform: typing.Callable = lambda x: x, sort_by: typing.Callable = lambda x: x.file_name, exclude: typing.Union[str, NoneType] = None, static: bool = False)` Create a new index. `directory` is the directory to get content from. If `recursive` is true, the whole tree of that directory will be indexed. `url_transform` is a function that will be applied to the file name to get the new file name. Generally you want to set it so it makes them end in `.html` so dumb servers can serve them correctly. However, for static files you most likely will not set it. `sort_by` is the key after which to sort the documents after they are indexed; by default it is the file name. `exclude` is a regular expression that will be used to exclude files from the index. If the index is `static`, all documents will be parsed as-is, without removing front matter. #### def `__iter__(self)` Return an iterator for the index. #### def `__next__(self)` Get the next document in the index. #### def `__repr__(self)` Return a string representation of the index. It contains the directory and the names of the documents in it. #### def `__len__(self)` Return the number of documents in the index, that is, its length. ### class `ampoule_ssg.Document` A document is a file, not rendered, but available for use. It is what is passed to the template as `document` for processing. Generally, you won't create these yourself, but rather use them as they are returned by an index. However, if you do need one, you can create it manually and pass it to a page. Documents will parse YAML front matter for textual files, unless disabled. The front matter is available as an attribute of the document, and can be accessed using indexing syntax. #### def `__init__(self, file_name: typing.Union[str, bytes, os.PathLike], url_transform: typing.Callable = lambda x: x, front_matter_enabled: bool = True)` Create a new document. `file_name` is the name of the file. `url_transform` is a function that will be applied to the file name to get the new file name; it has the same meaning as in the `Index`. `front_matter_enabled` is a boolean that determines whether the document will parse YAML front matter. #### def `__repr__(self)` Return a string containing `Document` and the file name. #### def `__getitem__(self, item: str)` Access the document's front matter. If front matter is disabled or not available, this will never work. #### def `__setitem__(self, item: str, value: typing.Any)` Change the document's front matter. It works even if it wasn't parsed, because YAML behaves like a dictionary. #### def `__delitem__(self, item: str)` Delete an item from the document's front matter. #### def `__contains__(self, item: str)` Check if an item is in the document's front matter. Licence ------- This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see .