Build your ZIM file
There are many possibilities to obtain your own ZIM file.
There are many reasons why you might want your own ZIM: the content is not already ZIMed, you want a special selection of the content, support for a new language, a custom UI inside the ZIM, ...
Solutions are ranging from straightforward (but obviously limited) solutions to complex (but mostly unlimited, sky is the limit) ones.
Not everyone has tech skills, and even when you have it is usually important to keep things as simple and straightforward as possible.
Request a ZIM file
If your content match openZIM publishing policies, you may ask the Kiwix team to create a ZIM file for you.
This main limitations is that you have no control on the timeline when the ZIM will be available.
Kiwix does its best to create ZIMs in a timely manner, but being a free service the resources are limited.
It is also possible to pay Kiwix to create ZIMs, and in such a situation the service will of course be much quicker and responsive.
To request such a ZIM, simply follow the process described in the zim-requests Github repository.
Youzim.it is an online website where you can request an automated system to create a ZIM of any online website.
Once the ZIM is produced, a download link will be provided to your email address.
This automated system rely on Zimit scraper, which is a very versatile solution to scrape any website.
Unfortunately, Zimit scraper has some limitations, at least for now.
First, Zimit 1.x is relying on a technical solution named Service Workers which limits the readers capable to display the ZIM produced (only kiwix-android, kiwix-serve and kiwix-js ; note that kiwix-desktop on Windows and Linux can start a kiwix-serve for you).
Zimit 2.x (planned for 2024) should overcome this limitation, but it is not here yet.
Second, Zimit is a scraper, and as such many website will quickly identify it as a bot and refuse to serve any content or serve rubbish (e.g. a captcha).
It should also be noted that for fair usage, Youzim.it imposes some limitations: your request will be stopped after 2 hours of processing or if the content retrieved exceeds 4GB.
If your use-case match all these limitations, it is clearly the quickest solution to get a ZIM (even if the processing capabilities are limited and your job might end-up in a waiting queue for few hours).
It should be noted for now advanced use of Youzim.it requires some technical skills and expert knowledge to configure the advanced options. This process should be enhanced in 2024 to provide more explanations and guide the user in the configuration process.
If you have some "Ops" skills, mainly around using a command-line tool, you can benefit from a range of existing tools to create your own ZIM on your own machine
Tools created and used by openZIM/Kiwix teams are open-sourced on Github openZIM or Kiwix organizations.
This means that every tool used to create ZIMs available online in Kiwix library are readily available for anyone to use.
MWoffliner is a tool which allows to "dump" a Wikimedia project (Wikipedia, Wiktionary, ...) to a local storage. It should also work for any Mediawiki instance. It goes through all articles (or a selection if specified) of the project and write HTML/pictures to your local filesystem as plain HTML/JS/CSS/... files or in a ZIM file.
If you are a developer, you can download it directly from its git repository.
A virtual machine with zimwriterfs is provided here.
If you have developments skills, you can create your own tool to create a ZIM from your content. This is what is called it a scraper, even if most of them do not "scrape" a website but used specific techniques like APIs or exported databases.
The libzim library (openZIM implementation of the ZIM specification, to read and write ZIM files, written in C++) has bindings available for many programming languages: Python, Node.JS, Java.
Since most openZIM scraper are written in Python, there is even a python-scraperlib library providing higher level functions to simplify common scraper tasks.
There are some older tools that might interest you. They are however not maintained anymore, sometimes even archived, and probably do not work anymore without at least some tinkering.
The following description is based on the notes published by the original author of Zimbalaka, as they're no longer available on the site they were published on. An archived copy is available on archive.org https://web.archive.org/web/20150531004251/http://www.arunmozhi.in:80/blog/zimbalaka-an-openzim-creator/#content
Zimbalaka, is designed as a web hosted tool which enables #Wikipedia ZIM files to be created based on articles selections.
It accepts two types of inputs: a list of pages or a Wikipedia category. Then Zimbalaka downloads those pages, removes all the clutter such as: sidebars, toolbox, edit links, etc., and provides a cleaned version as a ZIM file for download. It can be opened in Kiwix, etc.
The ZIM is created with a simple welcome page with all the pages as a list of links.
Zimbalaka has multilingual and multi-site support. That is, you can create a ZIM file from pages of any language of the 280+ existing Wikipedias, and also from sites like WikiBooks, Wiktionary, Wikiversity and such. You can even input any custom url like (http://sub.domain.com/), Zimblaka would add (/wiki/Page_title) to it and download the pages.
A small pain point is that, Zimbalaka also strips the external references that occur at the end of the Wikipedia articles, as the original author didn’t find these useful content intended to be used in an offline environment.
You cannot add a custom Welcome page in the zim file. Not a very big priority. The current file does its work of listing all the pages.
You cannot include pages from multiple sites as a single zim file. The workaround is to create multiple files or use a tool called zimwriterfs, which has to be compiled from source (this is used by zimbalaka behind the scenes).
This tool is written using Flask – A simple Python web framework for the backend, Bootstrap as the frontend and uses the zimwriterfs compiled binary as the workhorse. The zimming tasks are run by Celery, which has been automated by supervisord. All the co-ordination and message passing happens via Redis.
zimwriterdb is part of the openZIM project. This binary uses a pre-filled Postgres Database and create the corresponding ZIM file; the schema for the database is linked on the main zimwriterdb page.
Wiki2html can be used to prepare static HTML files from a running Mediawiki instance.
The zimmer package allows creating a ZIM dump from a Mediawiki-based wiki. This package is relatively easy to install and supports both old and new versions of Mediawiki. It is a kind of an alternative to MWoffliner.
The package consists of two Node.js scripts:
- wikizimmer.js -- creates static HTML files from the wiki's articles. It requires public access both to the normal web interface and to the wiki's API interface. Unlike mwoffliner, this script does not require Redis
- zimmer.js -- creates a ZIM file from the static HTML files (without requiring the libzim).