{ "analyzer": { "keyword": { "filter": [ "truncate_keyword" ], "tokenizer": "no_splitting", "type": "custom" }, "lowercase_keyword": { "filter": [ "truncate_keyword", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "no_splitting", "type": "custom" }, "near_match": { "char_filter": [ "near_space_flattener" ], "filter": [ "truncate_keyword", "icu_normalizer" ], "tokenizer": "no_splitting", "type": "custom" }, "near_match_asciifolding": { "char_filter": [ "near_space_flattener" ], "filter": [ "truncate_keyword", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "no_splitting", "type": "custom" }, "plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icutokrep_no_camel_split", "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icutokrep_no_camel_split", "icu_normalizer" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "prefix": { "char_filter": [ "near_space_flattener" ], "filter": [ "icu_normalizer" ], "tokenizer": "prefix", "type": "custom" }, "prefix_asciifolding": { "char_filter": [ "near_space_flattener" ], "filter": [ "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "prefix", "type": "custom" }, "short_text": { "filter": [ "icu_normalizer", "aggressive_splitting", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "whitespace", "type": "custom" }, "short_text_search": { "filter": [ "icu_normalizer", "aggressive_splitting" ], "tokenizer": "whitespace", "type": "custom" }, "source_text_plain": { "char_filter": [ "word_break_helper_source_text" ], "filter": [ "icutokrep_no_camel_split", "icu_normalizer" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "source_text_plain_search": { "char_filter": [ "word_break_helper_source_text" ], "filter": [ "icutokrep_no_camel_split", "icu_normalizer" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "suggest": { "filter": [ "icutokrep_no_camel_split", "icu_normalizer", "suggest_shingle" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "suggest_reverse": { "filter": [ "icutokrep_no_camel_split", "icu_normalizer", "suggest_shingle", "reverse" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "text": { "char_filter": [ "globo_norm", "acronym_fixer", "camelCase_splitter", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_token_repair", "homoglyph_norm", "icu_normalizer", "polish_stop", "polish_stem", "stempel_pattern_filter", "icu_folding", "remove_empty", "stempel_stop" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "text_search": { "char_filter": [ "globo_norm", "acronym_fixer", "camelCase_splitter", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_token_repair", "homoglyph_norm", "icu_normalizer", "polish_stop", "polish_stem", "stempel_pattern_filter", "icu_folding", "remove_empty", "stempel_stop" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" }, "token_reverse": { "filter": [ "reverse" ], "tokenizer": "no_splitting", "type": "custom" }, "trigram": { "filter": [ "icu_normalizer" ], "tokenizer": "trigram", "type": "custom" }, "trigram_anchored": { "char_filter": [ "add_regex_start_end_anchors" ], "filter": [ "icu_normalizer" ], "tokenizer": "trigram", "type": "custom" }, "word_prefix": { "filter": [ "icutokrep_no_camel_split", "icu_normalizer", "prefix_ngram_filter" ], "tokenizer": "textify_icu_tokenizer", "type": "custom" } }, "char_filter": { "arabic_extended_norm": { "type": "limited_mapping", "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" ] }, "dotted_I_fix": { "type": "limited_mapping", "mappings": [ "\u0130=>I" ] }, "globo_norm": { "type": "mapping", "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=>" ] }, "icu_normalizer": { "type": "icu_normalizer", "name": "nfkc_cf" }, "near_space_flattener": { "type": "limited_mapping", "mappings": [ "'=>\\u0020", "\\u2019=>\\u0020", "\\u02BC=>\\u0020", "_=>\\u0020", "-=>\\u0020" ] }, "nnbsp_norm": { "type": "limited_mapping", "mappings": [ "\\u202F=>\\u0020" ] }, "regex_acronym_fixer": { "type": "pattern_replace", "pattern": "(?<=(?:^|\\P{L})\\p{L})[.\uff0e](\\p{L})(?=\\P{L}|$)", "replacement": "$1" }, "regex_camelCase": { "type": "pattern_replace", "pattern": "(\\p{Ll}[\\p{M}\\p{Cf}]*)([\\p{Lu}\\p{Lt}])", "replacement": "$1 $2" }, "word_break_helper": { "type": "limited_mapping", "mappings": [ "_=>\\u0020", ":=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", "\uff0e=>\\u0020", "\uff3f=>\\u0020", "\uff1a=>\\u0020", "\u00b7=>\\u0020" ] }, "word_break_helper_source_text": { "type": "limited_mapping", "mappings": [ "_=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", ":=>\\u0020" ] } }, "filter": { "aggressive_splitting": { "type": "word_delimiter_graph", "stem_english_possessive": false, "preserve_original": false }, "asciifolding": { "type": "asciifolding", "preserve_original": false }, "asciifolding_preserve": { "type": "asciifolding", "preserve_original": true }, "icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0104\u0105\u0106\u0107\u0118\u0119\u0141\u0142\u0143\u0144\u00d3\u00f3\u015a\u015b\u0179\u017a\u017b\u017c]" }, "icu_nfkc_normalization": { "type": "icu_normalizer", "name": "nfkc" }, "icu_normalizer": { "type": "icu_normalizer", "name": "nfkc_cf" }, "icutokrep_no_camel_split": { "type": "icu_token_repair", "keep_camel_split": false }, "lowercase": { "type": "lowercase" }, "polish_stop": { "type": "stop", "stopwords": [ "vol", "o.o.", "o.o", "mgr", "godz", "z\u0142", "www", "pl", "ul", "tel", "hab", "prof", "in\u017c", "dr", "i", "u", "aby", "albo", "ale", "ani", "a\u017c", "bardzo", "bez", "bo", "bowiem", "by", "byli", "bym", "by\u0142", "by\u0142a", "by\u0142o", "by\u0142y", "by\u0107", "b\u0119dzie", "b\u0119d\u0105", "chce", "cho\u0107", "co", "coraz", "co\u015b", "czy", "czyli", "cz\u0119sto", "dla", "do", "gdy", "gdyby", "gdy\u017c", "gdzie", "go", "ich", "im", "inne", "i\u017c", "ja", "jak", "jakie", "jako", "je", "jednak", "jednym", "jedynie", "jego", "jej", "jest", "jeszcze", "je\u015bli", "je\u017celi", "ju\u017c", "j\u0105", "kiedy", "kilku", "kto", "kt\u00f3ra", "kt\u00f3re", "kt\u00f3rego", "kt\u00f3rej", "kt\u00f3ry", "kt\u00f3rych", "kt\u00f3rym", "kt\u00f3rzy", "lat", "lecz", "lub", "ma", "maj\u0105", "mamy", "mi", "mia\u0142", "mimo", "mnie", "mog\u0105", "mo\u017ce", "mo\u017cna", "mu", "musi", "na", "nad", "nam", "nas", "nawet", "nic", "nich", "nie", "niej", "nim", "ni\u017c", "no", "nowe", "np", "nr", "o", "od", "ok", "on", "one", "oraz", "pan", "po", "pod", "ponad", "poniewa\u017c", "poza", "przed", "przede", "przez", "przy", "raz", "razie", "roku", "r\u00f3wnie\u017c", "si\u0119", "sobie", "spos\u00f3b", "swoje", "s\u0105", "ta", "tak", "takich", "takie", "tak\u017ce", "tam", "te", "tego", "tej", "temu", "ten", "teraz", "te\u017c", "to", "trzeba", "tu", "tych", "tylko", "tym", "tys", "tzw", "t\u0119", "w", "we", "wie", "wi\u0119c", "wszystko", "w\u015br\u00f3d", "w\u0142a\u015bnie", "z", "za", "za\u015b", "ze", "\u017ce", "\u017ceby", "ii", "iii", "iv", "vi", "vii", "viii", "ix", "xi", "xii", "xiii", "xiv", "xv" ] }, "prefix_ngram_filter": { "type": "edgeNGram", "max_gram": 255 }, "remove_empty": { "type": "length", "min": 1 }, "stempel_pattern_filter": { "type": "pattern_replace", "pattern": "^([a-z\u0105\u0142]?[a-z\u0107\u0144]|..\u0107|\\d.*\u0107)$", "replacement": "" }, "stempel_stop": { "type": "stop", "stopwords": [ "owa\u0107", "iwa\u0107", "obi\u0107", "sn\u0105\u0107", "ywa\u0107", "ium", "my", "um" ] }, "suggest_shingle": { "type": "shingle", "min_shingle_size": 2, "max_shingle_size": 3, "output_unigrams": true }, "truncate_keyword": { "type": "truncate_norm", "length": 5000 } }, "normalizer": { "keyword": { "type": "custom", "filter": [ "truncate_keyword" ] }, "lowercase_keyword": { "type": "custom", "filter": [ "truncate_keyword", "icu_normalizer" ] } }, "tokenizer": { "no_splitting": { "type": "keyword" }, "prefix": { "type": "edgeNGram", "max_gram": 255 }, "trigram": { "type": "nGram", "min_gram": 3, "max_gram": 3 } } }