localeFilter:
  filterType: locale
  includeChildren: false
  # Note: en_GB doesn't noticeably increase the data size, so let's include it too.
  includelist:
    - en_US
    - en_GB
featureFilters:
  # Dictionaries are needed only for non-latin and non-cyrillic scripts.
  brkitr_dictionaries: exclude
  brkitr_rules:
    filterType: regex
    excludelist:
      - ^.*_cj$  # Chinese
  coll_tree:
    filterType: language
    # Include top 23 languages with Latin, Cyrillic or Greek script
    # (selected by the number of speakers and file size).
    includelist:
      - be
      - cs
      - de
      - el
      - en
      - es
      - ff
      - fr
      - ha
      - hu
      - ig
      - it
      - kk
      - ms
      - nl
      - pl
      - pt
      - ro
      - ru
      - tr
      - uk
      - xh
      - zu
  # Conversion mappings are needed only for legacy character encodings.
  conversion_mappings: exclude
  stringprep:
    # Exclude profiles that no one is likely to need.
    excludelist:
      - rfc3722      # iSCSI names
      - rfc3920node  # XMPP
      - rfc3920res   # XMPP
resourceFilters:
  - categories: [coll_tree]
    rules:
      - -/collations/emoji
      - -/collations/private-unihan
  - categories: [translit]
    rules:
      - -/
      - +/%%Parent
      - +/%%ALIAS
      - +/*/ASCII-Latin
      - +/*/Latin-ASCII
      - +/*/Accents-Any
      - +/*/Any-Accents
      - +/*/Publishing-Any
      - +/*/Any-Publishing
      - +/*/Cyrillic-Latin
      - +/*/Cyrl-Latn
      - +/*/Latin-Cyrillic
      - +/*/Latn-Cyrl
      - +/*/Greek-Latin
      - +/*/Grek-Latn
      - +/*/Greek-Latin/BGN
      - +/*/el-el_Latn/BGN
      - +/*/Latin-Greek
      - +/*/Latn-Grek
      - +/*/Russian-Latin/BGN
      - +/*/ru-ru_Latn/BGN
      - +/*/Latin-Russian/BGN
      - +/*/ru_Latn-ru/BGN
      - +/*/Ukrainian-Latin/BGN
      - +/*/uk-uk_Latn/BGN
      - +/*/de-ASCII
      - +/*/de-t-de-d0-ascii
# Use smaller UCAData
# https://unicode-org.github.io/icu/userguide/icu_data/buildtool.html#collation-ucadata
collationUCAData: implicithan