Torturing Word Delimiter TokenFilter in ElasticSearch

gistfile1.sh

#!/bin/sh

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

curl -X POST 'localhost:9200/i/' -d '{
  "settings" : {
    "analysis" : {
      "analyzer" : {
        "crazy" : {
          "type" : "custom",
          "tokenizer" : "keyword",
          "filter" : [ "replace-whitespaces", "strip_underscore" ]
        }
      },
      "filter" : {
        "replace-whitespaces" : {
          "type" : "pattern_replace",
          "pattern" : "\\s+",
          "replacement" : "_"
        },
        "strip_underscore" : {
          "type" : "word_delimiter",
          "split_on_numerics" : false,
          "split_on_case_change" : false,
          "generate_word_parts" : false,
          "generate_number_parts" : false,
          "catenate_all" : true
        }
      }
    }
  }
}'

curl 'localhost:9200/i/_analyze?analyzer=crazy&pretty=true' -d 'Hello-Hello 101 World'

# {"ok":true,"acknowledged":true}{"ok":true,"acknowledged":true}{
#   "tokens" : [ {
#     "token" : "HelloHello101World",
#     "start_offset" : 0,
#     "end_offset" : 21,
#     "type" : "word",
#     "position" : 1
#   } ]
# }