This repository contains a two Python scripts that are used for preparing the import of a project from the Wikimedia Incubator to their permanent wikis.

Download this repository (or, if you want, just the files cleaner.py and splitter.py, those are really all you'll need).

I would highly recommend that you first change the filename of the exported XML file from its default name to something identifying the wiki (for instance nqowiki.xml for the N'Ko Wikipedia or shywiktionary.xml for the Shawiya Wiktionary). The name of the file will also be used as a name for a folder in the splitter.py script.

This script is run with two obligatory arguments and two optional arguments:

• prefix needs to be the prefix used by the pages on the Wikimedia Incubator. So for a Wikipedia in Esperanto, the prefix would be Wp/eo.
• XMLfile is the filename of the XML file you got when you exported all the contents from the test wiki on the Incubator.
• (optional) --wiktionary is used if pagenames on the wiki will be case-sensitive. On Wikipedias, [[This]] and [[this]] will link to the same page, but on Wiktionaries, they will be different pages. This also applies to any other non-Wiktionary wikis with the $wgCapitalLinks option set to false, such as the Wikipedias in Lojban or Sakizaya.
• (optional) --notranslate will not translate namespace names (such as Category:, Talk:, File:, etc) into the target language, instead keeping them in English (which is the interface language on Incubator). If the wiki you're running the script for has not yet been created, you will have to use this, as the script fetches the namespace names from the wiki's API, and that will not work if the wiki doesn't exist yet.

For a Wikipedia in Esperanto you would run the script like this:

python3 cleaner.py Wp/eo eowiki.xml

And, assuming no errors, the resulting file will be named eowiki.ready.xml.

Once you are finished, you should use a diff tool such as Meld to compare the two files to make sure that everything looks good.

The script will throw errors in two cases:

1. If the filename given doesn't have a three- or four-letter extension. This is because the output filename would be malformed.
2. If, after the prefix is stripped, a pagename in the XML file would be invalid or potentially cause duplicate pagenames. This is the case if the pagename starts with a colon or a space (which can happen if there is a prefix present, but can't happen if there is no prefix). The solution to this is to (1) check in the XML file would be a duplicate (do a search for the resulting pagename without the colon or space), and then (2a) remove the page from the XML file in case it is a redirect or (2b) remove the offending characters in case there are no pagename conflicts.

Additionally, the script will print a warning in case the prefix is still present in the resulting XML file. This normally happens because of human error, like if an editor has omitted an / or the prefix is not properly inside a link or a template (e.g. if someone has put one too few square brackets around a link). The script will print the lines where this occurs, so you can go over the exported XML file and check to see what was the cause of these errors, save, and run the script again.

In this repository there is a file named testcases.txt, that you can run the script on to see what happens. The testcases file does not test for the errors mentioned above, but tests various other replacements that will be made.

Run the script with the test case file like this:

python3 cleaner.py Wp/ru testcases.txt

Then review the resulting file testcases.ready.txt. You should also try it with the optional parameters to see what the differences are.

This script is used to split the resulting XML file into smaller pieces. If you try to import a file that is too big, you will constantly run into timeouts or database errors, so the most efficient way to circumvent this has proved to be splitting the large file into smaller pieces.

The script is run with one argument, which is the name of the file you want to split:

python3 splitter.py eowiki.ready.xml

This will create a directory named eowiki, with the files within named eowiki_1.xml, eowiki_2.xml etc. The files will be maximum 2.5 MB (except if a single page within the file is much larger), or contain maximum 400 pages, whichever comes first.