Ukázka ICU Folding. Předpokládá Elasticsearch 0.90.0 a nainstalovaný ICU plugin 1.9.0

gistfile1.sh

#!/bin/sh

curl -X DELETE 'localhost:9200/i/'

curl -X POST 'localhost:9200/i/' -d '{
  "settings" : {
    "number_of_shards" : 1,
    "number_of_replicas" : 0,
    "analysis" : {
      "analyzer" : {
        "icu_folding" : {
          "type" : "custom",
          "tokenizer" : "whitespace",
          "filter" : ["icu_folding"]
        },
        "ascii_folding" : {
          "type" : "custom",
          "tokenizer" : "whitespace",
          "filter" : ["asciifolding","lowercase"]
        }
      }
    }
  }
}'

# exit; and test manually...

# ascii folding and icu folding work the same way (except the lowercasing which has to be added into ascii_filding)
curl 'localhost:9200/i/_analyze?analyzer=icu_folding&pretty=true' -d 'Běloučký kůň úpěl ódy!'
curl 'localhost:9200/i/_analyze?analyzer=ascii_folding&pretty=true' -d 'Běloučký kůň úpěl ódy!'

# Ascii folding works in some cases...
curl 'localhost:9200/i/_analyze?analyzer=icu_folding&pretty=true' -d 'dž ¼ № ℃ ™ Æ Ȣ ffi '
curl 'localhost:9200/i/_analyze?analyzer=ascii_folding&pretty=true' -d 'dž ¼ № ℃ ™ Æ Ȣ ffi '

# Ascii folding noop here... ICU folding rocks!
curl 'localhost:9200/i/_analyze?analyzer=icu_folding&pretty=true' -d 'º o ª a ℹ i ℇ e'
curl 'localhost:9200/i/_analyze?analyzer=ascii_folding&pretty=true' -d 'º o ª a ℹ i ℇ e'

for those who are lazy or simply cant try this right away, the differences are:
icu_folding: dž -> dz, ¼ -> 1/4, № -> no ...
ascii_folding: dž -> dz, ¼ -> ¼, № -> № ...
icu_folding: º -> o, o -> o, ª -> a, a -> a, ℹ -> i, i -> i, ℇ -> e, e -> e
ascii_folding: º -> º, o -> o, ª -> ª, a -> a, ℹ -> ℹ, i -> i, ℇ -> ℇ, e -> e