Skip to content

improved detectLangsFromName()#700

Closed
mottla wants to merge 1 commit into
ilius:masterfrom
mottla:master
Closed

improved detectLangsFromName()#700
mottla wants to merge 1 commit into
ilius:masterfrom
mottla:master

Conversation

@mottla

@mottla mottla commented Jan 10, 2026

Copy link
Copy Markdown

improved detectLangsFromName

and added tests

test_lang_detect_6
test_lang_detect_7

with tests that used to fail before, but now with the new detectLangsFromName pass.

Details:
Now we assume that the last occurrence of - or to, separates the target from source language. Note: This way we prevent errors that occur for example with "Neo-Persian to English.." However, we still get wrong results when it is something like "en - Neo-Persian" We cannot uniquely identify the users will when resorting to this function anyway. This way however improves results to a great extent

added tests test_lang_detect_6 and test_lang_detect_7 that used to fail before, but now pass.

Details:
Now  we assume that the last occurrence of - or to, separates the target from source language.
Note: This way we prevent errors that occur for example with "Neo-Persian to English.."
However, we still get wrong results when it is something like "en - Neo-Persian"
We cannot uniquely identify the users will when resorting to this function anyway.
This way however improves results to a great extent
@ilius

ilius commented Jan 28, 2026

Copy link
Copy Markdown
Owner

It's interesting, but I'm not sure.
Since language pair is often in the middle of name. And I think more often at the beginning than the end.
And since we get the name from file name for many formats, any space might have been replaced with -.

I rather give up on auto-detection of a few odd cases, than to "detect wrong languages", so I'd avoid trying to detect multi-word language names.

For "Na vosa vaka-Viti-deu.index", it should skip and not detect. (They should use "fij-deu.index")

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants