Skip to content

fix: normalize German sharp s in German morphology - #4836

Open
sanikolaev wants to merge 1 commit into
mainfrom
fix/4827-german-sharp-s
Open

fix: normalize German sharp s in German morphology#4836
sanikolaev wants to merge 1 commit into
mainfrom
fix/4827-german-sharp-s

Conversation

@sanikolaev

Copy link
Copy Markdown
Collaborator

Related issue: #4827


The issue had two parts: non_cont converted ß to a single s before morphology, and German AOT did not treat ß/ẞ and ss as equivalent.

The fix:

  • adds charset_table = non_cont, german to preserve German characters;
  • folds ß/ẞ to ss before lemmatize_de and lemmatize_de_all;
  • keeps original spellings available for exact-word queries;
  • versions the morphology fingerprint so old tables warn that reindexing is required.

Add the `german` charset alias so tables can preserve German umlauts and
sharp s with `charset_table=non_cont,german`.

Treat ß, capital ẞ, and ss as search equivalents for German AOT
lemmatization while retaining exact-word distinctions.

Version the German morphology fingerprint, warn when existing plain or RT
tables need rebuilding, and add plain/RT regression coverage for both
German lemmatization modes.

Related issue: #4827
@sanikolaev

Copy link
Copy Markdown
Collaborator Author

Previously discussed in the issue:

And I strongly suggest to make the new treatment of "ß" the default. It's what normal users would expect.

@tomatolog: "we can not just change the defaults as users using the defaults \ existed charset table would get the different settings \ data indexed for new tables \ RT table disk chunks and would complain that after upgrade old tables \ RT disk chunks would produce warnings or will not return the same results as the new tables \ disk chunks"

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.

1 participant