settings/icu_tokenizer.yaml

   1 normalization:
   2     - ":: lower ()"
   3     - ":: Hans-Hant"
   4     - !include icu-rules/unicode-digits-to-decimal.yaml
   5     - "'№' > 'no'"
   6     - "'n°' > 'no'"
   7     - "'nº' > 'no'"
   8     - "ª > a"
   9     - "º > o"
  10     - "[[:Punctuation:][:Symbol:]]  > ' '"
  11     - "ß > 'ss'" # German szet is unimbigiously equal to double ss
  12     - "[^[:Letter:] [:Number:] [:Space:]] >"
  13     - "[:Lm:] >"
  14     - ":: [[:Number:]] Latin ()"
  15     - ":: [[:Number:]] Ascii ();"
  16     - ":: [[:Number:]] NFD ();"
  17     - "[[:Nonspacing Mark:] [:Cf:]] >;"
  18     - "[:Space:]+ > ' '"
  19 transliteration:
  20     - ":: Latin ()"
  21     - !include icu-rules/extended-unicode-to-asccii.yaml
  22     - ":: Ascii ()"
  23     - ":: NFD ()"
  24     - "[^[:Ascii:]] >"
  25     - ":: lower ()"
  26     - ":: NFC ()"
  27 sanitizers:
  28     - step: split-name-list
  29     - step: strip-brace-terms
  30 token-analysis:
  31     - variants:
  32           - !include icu-rules/variants-bg.yaml
  33           - !include icu-rules/variants-ca.yaml
  34           - !include icu-rules/variants-cs.yaml
  35           - !include icu-rules/variants-da.yaml
  36           - !include icu-rules/variants-de.yaml
  37           - !include icu-rules/variants-el.yaml
  38           - !include icu-rules/variants-en.yaml
  39           - !include icu-rules/variants-es.yaml
  40           - !include icu-rules/variants-et.yaml
  41           - !include icu-rules/variants-eu.yaml
  42           - !include icu-rules/variants-fi.yaml
  43           - !include icu-rules/variants-fr.yaml
  44           - !include icu-rules/variants-gl.yaml
  45           - !include icu-rules/variants-hu.yaml
  46           - !include icu-rules/variants-it.yaml
  47           - !include icu-rules/variants-ja.yaml
  48           - !include icu-rules/variants-mg.yaml
  49           - !include icu-rules/variants-ms.yaml
  50           - !include icu-rules/variants-nl.yaml
  51           - !include icu-rules/variants-no.yaml
  52           - !include icu-rules/variants-pl.yaml
  53           - !include icu-rules/variants-pt.yaml
  54           - !include icu-rules/variants-ro.yaml
  55           - !include icu-rules/variants-ru.yaml
  56           - !include icu-rules/variants-sk.yaml
  57           - !include icu-rules/variants-sl.yaml
  58           - !include icu-rules/variants-sv.yaml
  59           - !include icu-rules/variants-tr.yaml
  60           - !include icu-rules/variants-uk.yaml
  61           - !include icu-rules/variants-vi.yaml