{ "analyzer": { "bg_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "bg_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "bg_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "bg_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "bulgarian_stop", "bulgarian_stemmer", "bg_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "bg_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "bulgarian_stop", "bulgarian_stemmer", "bg_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "bo_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bo_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bo_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bo_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bug_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bug_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bug_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "bug_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "cdo_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "cdo_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "cdo_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "cdo_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "cjk_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "cjk_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "cjk_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "cjk_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "cs_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "cs_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "cs_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "czech_stop", "czech_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "cs_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "czech_stop", "czech_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "de_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper", "german_charfilter" ], "filter": [ "de_icu_normalizer", "preserve_original_recorder", "de_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "de_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper", "german_charfilter" ], "filter": [ "de_icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "de_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "german_charfilter" ], "filter": [ "de_icu_normalizer", "german_stop", "german_normalization", "german_stemmer", "de_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "de_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "german_charfilter" ], "filter": [ "de_icu_normalizer", "german_stop", "german_normalization", "german_stemmer", "de_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "en_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "en_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "en_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "possessive_english", "icu_normalizer", "stop", "icu_folding", "remove_empty", "kstem", "custom_stem" ], "tokenizer": "standard", "type": "custom" }, "en_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "possessive_english", "icu_normalizer", "stop", "icu_folding", "remove_empty", "kstem", "custom_stem" ], "tokenizer": "standard", "type": "custom" }, "fr_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "fr_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "fr_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "french_charfilter" ], "filter": [ "french_elision", "icu_normalizer", "french_stop", "french_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "fr_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "french_charfilter" ], "filter": [ "french_elision", "icu_normalizer", "french_stop", "french_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "ja_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "ja_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "ja_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "ja_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "cjk_charfilter" ], "filter": [ "cjk_width", "icu_normalizer", "cjk_bigram", "cjk_stop", "ja_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "ja_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "cjk_charfilter" ], "filter": [ "cjk_width", "icu_normalizer", "cjk_bigram", "cjk_stop", "ja_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "jv_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "jv_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "jv_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "jv_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "sv_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "sv_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sv_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "sv_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "swedish_stop", "swedish_stemmer", "sv_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sv_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "swedish_stop", "swedish_stemmer", "sv_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "th_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "th_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "th_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "th_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "thai_repl_pat", "thai_charfilter" ], "filter": [ "icu_normalizer", "thai_stop", "th_icu_folding", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "th_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "thai_repl_pat", "thai_charfilter" ], "filter": [ "icu_normalizer", "thai_stop", "th_icu_folding", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "wuu_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "wuu_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "wuu_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "wuu_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "dotted_I_fix", "arabic_extended_norm" ], "filter": [ "decimal_digit", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "zh_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "smartcn_stop", "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "zh_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "smartcn_stop", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "zh_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "dotted_I_fix", "arabic_extended_norm", "invis_cleanup", "stconvertfix", "tsconvert" ], "filter": [ "smartcn_stop", "icu_normalizer", "remove_empty" ], "tokenizer": "smartcn_tokenizer", "type": "custom" }, "zh_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "dotted_I_fix", "arabic_extended_norm", "invis_cleanup", "stconvertfix", "tsconvert" ], "filter": [ "smartcn_stop", "icu_normalizer", "remove_empty" ], "tokenizer": "smartcn_tokenizer", "type": "custom" } }, "char_filter": { "arabic_extended_norm": { "mappings": [ "\\uFB8E=>\\u0643", "\\uFB8F=>\\u0643", "\\uFB90=>\\u0643", "\\uFB91=>\\u0643", "\\u06A9=>\\u0643", "\\u06AA=>\\u0643", "\\uFEDB=>\\u0643", "\\uFEDC=>\\u0643", "\\uFED9=>\\u0643", "\\uFEDA=>\\u0643", "\\uFBFC=>\\u064A", "\\uFBFD=>\\u064A", "\\uFBFE=>\\u064A", "\\uFBFF=>\\u064A", "\\u06CC=>\\u064A", "\\uFBE8=>\\u064A", "\\uFBE9=>\\u064A", "\\uFEEF=>\\u064A", "\\uFEF0=>\\u064A", "\\u0649=>\\u064A", "\\u06CD=>\\u064A", "\\uFBE4=>\\u064A", "\\uFBE5=>\\u064A", "\\uFBE6=>\\u064A", "\\uFBE7=>\\u064A", "\\u06D0=>\\u064A", "\\uFBA6=>\\u0647", "\\uFBA7=>\\u0647", "\\uFBA8=>\\u0647", "\\uFBA9=>\\u0647", "\\u06C1=>\\u0647", "\\u06C0=>\\u0647", "\\uFBA4=>\\u0647", "\\uFBA5=>\\u0647", "\\u06D5=>\\u0647" ], "type": "mapping" }, "cjk_charfilter": { "mappings": [ "\u309b=>\\u3099", "\u309c=>\\u309a" ], "type": "mapping" }, "dotted_I_fix": { "mappings": [ "\u0130=>I" ], "type": "mapping" }, "french_charfilter": { "mappings": [ "\\u02BC=>\\u0027" ], "type": "mapping" }, "german_charfilter": { "mappings": [ "\u1e9e=>\u00df" ], "type": "mapping" }, "globo_norm": { "mappings": [ "`=>'", "\u00b4=>'", "\u02b9=>'", "\u02bb=>'", "\u02bc=>'", "\u02bd=>'", "\u02be=>'", "\u02bf=>'", "\u02cb=>'", "\u055a=>'", "\\u05F3=>'", "\u2018=>'", "\u2019=>'", "\u201b=>'", "\u2032=>'", "\u2035=>'", "\ua78c=>'", "\uff07=>'", "\uff40=>'", "\\u202F=>\\u0020", "\u02c8=>", "\u02cc=>", "\\u0640=>", "\u066c=>,", "\u060c=>,", "\u055b=>", "\u055c=>", "\u055e=>", "\u00b5=>\u03bc", "\\u05F0=>\\u05D5\\u05D5", "\\u05F1=>\\u05D5\\u05D9", "\\u05F2=>\\u05D9\\u05D9", "\\uFB1F=>\\u05D9\\u05D9\\u05B7", "\\u05D9\\u05B7\\u05D9=>\\u05D9\\u05D9\\u05B7", "\u0d23\u0d4d\\u200D=>\u0d7a", "\u0d28\u0d4d\\u200D=>\u0d7b", "\u0d30\u0d4d\\u200D=>\u0d7c", "\u0d32\u0d4d\\u200D=>\u0d7d", "\u0d33\u0d4d\\u200D=>\u0d7e", "\u09a4\u09cd\\u200D=>\u09ce", "\\u061C=>" ], "type": "mapping" }, "invis_cleanup": { "mappings": [ "\\u200B=>\\u0020", "\\u00AD=>", "\\u200C=>", "\\u200D=>", "\\u2060=>", "\\uFEFF=>", "\\u200E=>", "\\u200F=>", "\\u202A=>", "\\u202B=>", "\\u202C=>", "\\u202D=>", "\\u202E=>", "\\u2066=>", "\\u2067=>", "\\u2068=>", "\\u2069=>", "\\u2061=>", "\\u2062=>", "\\u2063=>", "\\u2064=>", "\ufe00=>", "\ufe01=>", "\ufe02=>", "\ufe03=>", "\ufe04=>", "\ufe05=>", "\ufe06=>", "\ufe07=>", "\ufe08=>", "\ufe09=>", "\ufe0a=>", "\ufe0b=>", "\ufe0c=>", "\ufe0d=>", "\ufe0e=>", "\ufe0f=>", "\udb40\udd00=>", "\udb40\udd01=>", "\udb40\udd02=>", "\udb40\udd03=>", "\udb40\udd04=>", "\udb40\udd05=>", "\udb40\udd06=>", "\udb40\udd07=>", "\udb40\udd08=>", "\udb40\udd09=>", "\udb40\udd0a=>", "\udb40\udd0b=>", "\udb40\udd0c=>", "\udb40\udd0d=>", "\udb40\udd0e=>", "\udb40\udd0f=>", "\udb40\udd10=>", "\udb40\udd11=>", "\udb40\udd12=>", "\udb40\udd13=>", "\udb40\udd14=>", "\udb40\udd15=>", "\udb40\udd16=>", "\udb40\udd17=>", "\udb40\udd18=>", "\udb40\udd19=>", "\udb40\udd1a=>", "\udb40\udd1b=>", "\udb40\udd1c=>", "\udb40\udd1d=>", "\udb40\udd1e=>", "\udb40\udd1f=>", "\udb40\udd20=>", "\udb40\udd21=>", "\udb40\udd22=>", "\udb40\udd23=>", "\udb40\udd24=>", "\udb40\udd25=>", "\udb40\udd26=>", "\udb40\udd27=>", "\udb40\udd28=>", "\udb40\udd29=>", "\udb40\udd2a=>", "\udb40\udd2b=>", "\udb40\udd2c=>", "\udb40\udd2d=>", "\udb40\udd2e=>", "\udb40\udd2f=>", "\udb40\udd30=>", "\udb40\udd31=>", "\udb40\udd32=>", "\udb40\udd33=>", "\udb40\udd34=>", "\udb40\udd35=>", "\udb40\udd36=>", "\udb40\udd37=>", "\udb40\udd38=>", "\udb40\udd39=>", "\udb40\udd3a=>", "\udb40\udd3b=>", "\udb40\udd3c=>", "\udb40\udd3d=>", "\udb40\udd3e=>", "\udb40\udd3f=>", "\udb40\udd40=>", "\udb40\udd41=>", "\udb40\udd42=>", "\udb40\udd43=>", "\udb40\udd44=>", "\udb40\udd45=>", "\udb40\udd46=>", "\udb40\udd47=>", "\udb40\udd48=>", "\udb40\udd49=>", "\udb40\udd4a=>", "\udb40\udd4b=>", "\udb40\udd4c=>", "\udb40\udd4d=>", "\udb40\udd4e=>", "\udb40\udd4f=>", "\udb40\udd50=>", "\udb40\udd51=>", "\udb40\udd52=>", "\udb40\udd53=>", "\udb40\udd54=>", "\udb40\udd55=>", "\udb40\udd56=>", "\udb40\udd57=>", "\udb40\udd58=>", "\udb40\udd59=>", "\udb40\udd5a=>", "\udb40\udd5b=>", "\udb40\udd5c=>", "\udb40\udd5d=>", "\udb40\udd5e=>", "\udb40\udd5f=>", "\udb40\udd60=>", "\udb40\udd61=>", "\udb40\udd62=>", "\udb40\udd63=>", "\udb40\udd64=>", "\udb40\udd65=>", "\udb40\udd66=>", "\udb40\udd67=>", "\udb40\udd68=>", "\udb40\udd69=>", "\udb40\udd6a=>", "\udb40\udd6b=>", "\udb40\udd6c=>", "\udb40\udd6d=>", "\udb40\udd6e=>", "\udb40\udd6f=>", "\udb40\udd70=>", "\udb40\udd71=>", "\udb40\udd72=>", "\udb40\udd73=>", "\udb40\udd74=>", "\udb40\udd75=>", "\udb40\udd76=>", "\udb40\udd77=>", "\udb40\udd78=>", "\udb40\udd79=>", "\udb40\udd7a=>", "\udb40\udd7b=>", "\udb40\udd7c=>", "\udb40\udd7d=>", "\udb40\udd7e=>", "\udb40\udd7f=>", "\udb40\udd80=>", "\udb40\udd81=>", "\udb40\udd82=>", "\udb40\udd83=>", "\udb40\udd84=>", "\udb40\udd85=>", "\udb40\udd86=>", "\udb40\udd87=>", "\udb40\udd88=>", "\udb40\udd89=>", "\udb40\udd8a=>", "\udb40\udd8b=>", "\udb40\udd8c=>", "\udb40\udd8d=>", "\udb40\udd8e=>", "\udb40\udd8f=>", "\udb40\udd90=>", "\udb40\udd91=>", "\udb40\udd92=>", "\udb40\udd93=>", "\udb40\udd94=>", "\udb40\udd95=>", "\udb40\udd96=>", "\udb40\udd97=>", "\udb40\udd98=>", "\udb40\udd99=>", "\udb40\udd9a=>", "\udb40\udd9b=>", "\udb40\udd9c=>", "\udb40\udd9d=>", "\udb40\udd9e=>", "\udb40\udd9f=>", "\udb40\udda0=>", "\udb40\udda1=>", "\udb40\udda2=>", "\udb40\udda3=>", "\udb40\udda4=>", "\udb40\udda5=>", "\udb40\udda6=>", "\udb40\udda7=>", "\udb40\udda8=>", "\udb40\udda9=>", "\udb40\uddaa=>", "\udb40\uddab=>", "\udb40\uddac=>", "\udb40\uddad=>", "\udb40\uddae=>", "\udb40\uddaf=>", "\udb40\uddb0=>", "\udb40\uddb1=>", "\udb40\uddb2=>", "\udb40\uddb3=>", "\udb40\uddb4=>", "\udb40\uddb5=>", "\udb40\uddb6=>", "\udb40\uddb7=>", "\udb40\uddb8=>", "\udb40\uddb9=>", "\udb40\uddba=>", "\udb40\uddbb=>", "\udb40\uddbc=>", "\udb40\uddbd=>", "\udb40\uddbe=>", "\udb40\uddbf=>", "\udb40\uddc0=>", "\udb40\uddc1=>", "\udb40\uddc2=>", "\udb40\uddc3=>", "\udb40\uddc4=>", "\udb40\uddc5=>", "\udb40\uddc6=>", "\udb40\uddc7=>", "\udb40\uddc8=>", "\udb40\uddc9=>", "\udb40\uddca=>", "\udb40\uddcb=>", "\udb40\uddcc=>", "\udb40\uddcd=>", "\udb40\uddce=>", "\udb40\uddcf=>", "\udb40\uddd0=>", "\udb40\uddd1=>", "\udb40\uddd2=>", "\udb40\uddd3=>", "\udb40\uddd4=>", "\udb40\uddd5=>", "\udb40\uddd6=>", "\udb40\uddd7=>", "\udb40\uddd8=>", "\udb40\uddd9=>", "\udb40\uddda=>", "\udb40\udddb=>", "\udb40\udddc=>", "\udb40\udddd=>", "\udb40\uddde=>", "\udb40\udddf=>", "\udb40\udde0=>", "\udb40\udde1=>", "\udb40\udde2=>", "\udb40\udde3=>", "\udb40\udde4=>", "\udb40\udde5=>", "\udb40\udde6=>", "\udb40\udde7=>", "\udb40\udde8=>", "\udb40\udde9=>", "\udb40\uddea=>", "\udb40\uddeb=>", "\udb40\uddec=>", "\udb40\udded=>", "\udb40\uddee=>", "\udb40\uddef=>" ], "type": "mapping" }, "nnbsp_norm": { "mappings": [ "\\u202F=>\\u0020" ], "type": "mapping" }, "regex_acronym_fixer": { "pattern": "(?<=(?:^|\\P{L})\\p{L})[.\uff0e](\\p{L})(?=\\P{L}|$)", "replacement": "$1", "type": "pattern_replace" }, "regex_camelCase": { "pattern": "(\\p{Ll}[\\p{M}\\p{Cf}]*)([\\p{Lu}\\p{Lt}])", "replacement": "$1 $2", "type": "pattern_replace" }, "stconvertfix": { "mappings": [ "\\u606d\\u5f18=>\\u606d \\u5f18", "\\u5138=>\\u3469" ], "type": "mapping" }, "thai_charfilter": { "mappings": [ "_=>\\u0020", ";=>\\u0020", ":=>\\u0020", "\u00b7=>\\u0020", "\u2027=>\\u0020", "\u0e03=>\u0e02", "\u0e05=>\u0e04", "\\u0e4d\\u0e32=>\\u0e33", "\\u0e4d\\u0e48\\u0e32=>\\u0e48\\u0e33", "\\u0e4d\\u0e49\\u0e32=>\\u0e49\\u0e33", "\\u0e33\\u0e48=>\\u0e48\\u0e33", "\\u0e33\\u0e49=>\\u0e49\\u0e33", "\\u0E34\\u0E4D=>\\u0E36", "\\u0E4D\\u0E34=>\\u0E36" ], "type": "mapping" }, "thai_repl_pat": { "pattern": "(?<=\\p{Nd})([\u0e01-\u0e4f])|(?<=[\u0e01-\u0e4f])(\\p{Nd})|(?<=[\u0e01-\u0e4f])\\.([\u0e01-\u0e4f])", "replacement": " $1$2$3", "type": "pattern_replace" }, "tsconvert": { "convert_type": "t2s", "delimiter": "#", "keep_both": false, "type": "stconvert" }, "word_break_helper": { "mappings": [ "_=>\\u0020", ":=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", "\uff0e=>\\u0020", "\uff3f=>\\u0020", "\uff1a=>\\u0020", "\u00b7=>\\u0020" ], "type": "mapping" } }, "filter": { "bg_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0419\u0439]" }, "bulgarian_stemmer": { "language": "bulgarian", "type": "stemmer" }, "bulgarian_stop": { "stopwords": "_bulgarian_", "type": "stop" }, "cjk_stop": { "stopwords": "_english_", "type": "stop" }, "custom_stem": { "rules": "guidelines => guideline", "type": "stemmer_override" }, "czech_stemmer": { "language": "czech", "type": "stemmer" }, "czech_stop": { "stopwords": "_czech_", "type": "stop" }, "de_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u00c4\u00e4\u00d6\u00f6\u00dc\u00fc\u1e9e\u00df]" }, "de_icu_normalizer": { "name": "nfkc_cf", "type": "icu_normalizer", "unicodeSetFilter": "[^\u1e9e\u00df]" }, "french_elision": { "articles": [ "l", "m", "t", "qu", "n", "s", "j", "d", "c", "jusqu", "quoiqu", "lorsqu", "puisqu" ], "articles_case": true, "type": "elision" }, "french_stemmer": { "language": "light_french", "type": "stemmer" }, "french_stop": { "stopwords": "_french_", "type": "stop" }, "german_stemmer": { "language": "light_german", "type": "stemmer" }, "german_stop": { "stopwords": "_german_", "type": "stop" }, "icu_folding": { "type": "icu_folding" }, "icu_normalizer": { "name": "nfkc_cf", "type": "icu_normalizer" }, "ja_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u304c-\u30fe]" }, "possessive_english": { "language": "possessive_english", "type": "stemmer" }, "remove_empty": { "min": 1, "type": "length" }, "smartcn_stop": { "stopwords": [ "," ], "type": "stop" }, "sv_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u00c5\u00e5\u00c4\u00e4\u00d6\u00f6]" }, "swedish_stemmer": { "language": "swedish", "type": "stemmer" }, "swedish_stop": { "stopwords": "_swedish_", "type": "stop" }, "th_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\\u0E47-\\u0E4E]" }, "thai_stop": { "stopwords": "_thai_", "type": "stop" } }, "tokenizer": [] }