Question

Is there a way to stem Roman numerals in sphinx, so that Roman numerals are synonymous with Arab numerals?

For example, the query “Mafia 2” would be the same as “Mafia II”, and vice versa.

Answer 1

You can make it with wordforms.
Just write some script, translating Arab nums to Roman and make text file:

walks > walk
walker > walk
1 > I
2 > II
3 > III
4 > IV

and so on… Then save in to /absolute/path/wordforms.txt and in your config in index section write:

wordforms = /absolute/path/wordforms.txt

That’s it. After reindexing if you search Mafia 2 sphinx find Mafia II for you. But it’s only one way translation (not so good for your question, but still useful) and better be used for things like this example from documentation:

core 2 duo > c2d
e6600 > c2d
core 2duo > c2d

Round 2

Better way is to prepare search query before the search starts and stay away from indexer 🙂

Your query:

mafia 2|mafia II

or

gta 4|gta IV

That’s all. You only need to write some arab to roman convert function. You can find it googling something like

“arab to roman (your programming
language)”