introduce mutation variants to generic token analyser

[nominatim.git] / settings / icu_tokenizer.yaml

diff --git a/settings/icu_tokenizer.yaml b/settings/icu_tokenizer.yaml
index a3c62e67b8d2a500baee37223f56b151bce14324..c6601faf0645186740c06b99eaec124744aba0e0 100644 (file)
--- a/settings/icu_tokenizer.yaml
+++ b/settings/icu_tokenizer.yaml
@@ -59,6 +59,13 @@ token-analysis:
       mode: variant-only
       variants:
           - !include icu-rules/variants-de.yaml
       mode: variant-only
       variants:
           - !include icu-rules/variants-de.yaml

+      mutations:
+          - pattern: ä
+            replacements: ["ä", "ae"]
+          - pattern: ö
+            replacements: ["ö", "oe"]
+          - pattern: ü
+            replacements: ["ü", "ue"]

     - id: el
       analyzer: generic
       mode: variant-only
     - id: el
       analyzer: generic
       mode: variant-only