{ "job": { "content": [ { "reader": { "name": "hivereader", "parameter": { "connection": [ { "table": [ "org_space_tenant_preferences" ] } ], "column": [ { "name": "id", "type": "STRING" }, { "name": "business_code", "type": "STRING" }, { "name": "eid", "type": "STRING" }, { "name": "company_id", "type": "STRING" }, { "name": "field_code", "type": "STRING" }, { "name": "field_name", "type": "STRING" }, { "name": "field_value", "type": "STRING" }, { "name": "create_time", "type": "STRING" }, { "name": "update_time", "type": "STRING" } ] } }, "writer": { "name": "eswriter", "parameter": { "postAddAlias": [ "algo_space_tenant_preferences_sec" ], "index": "algo_space_tenant_preferences_${cyctime}", "indexStrategy": "RECREATE", "batchSize": 200, "timeout": 5000, "settings": { "index": { "refresh_interval": "1s", "number_of_shards": "3", "max_result_window": "50000", "number_of_replicas": "1", "analysis": { "normalizer": { "lowercase": { "type": "custom", "filter": [ "lowercase" ], "char_filter": [ "point_mapping" ] }, "uppercase": { "type": "custom", "filter": [ "uppercase" ] } }, "char_filter": { "punctuation_mapping": { "mappings": [ "? => \\u0020sentence1\\u0020 ", "? => \\u0020sentence2\\u0020 ", "。 => \\u0020sentence3\\u0020 ", ". => \\u0020sentence4\\u0020 ", "! => \\u0020sentence5\\u0020 ", "! => \\u0020sentence6\\u0020 ", ", => \\u0020punctuation0\\u0020 ", ", => \\u0020punctuation1\\u0020 ", "; => \\u0020punctuation2\\u0020 ", ": => \\u0020punctuation3\\u0020 ", ": => \\u0020punctuation4\\u0020 ", "“ => \\u0020punctuation5\\u0020 ", "\" => \\u0020punctuation6\\u0020 ", "” => \\u0020punctuation7\\u0020 ", "‘ => \\u0020punctuation8\\u0020 ", "' => \\u0020punctuation9\\u0020 ", "’ => \\u0020punctuation10\\u0020 ", "( => \\u0020punctuation11\\u0020 ", ") => \\u0020punctuation12\\u0020 ", "( => \\u0020punctuation13\\u0020 ", ") => \\u0020punctuation14\\u0020 ", "、 => \\u0020punctuation15\\u0020 ", "` => \\u0020punctuation16\\u0020 ", "+ => \\u0020punctuation17\\u0020 ", "- => \\u0020punctuation18\\u0020 ", "& => \\u0020punctuation19\\u0020 ", "| => \\u0020punctuation20\\u0020 ", "[ => \\u0020punctuation21\\u0020 ", "] => \\u0020punctuation22\\u0020 ", "^ => \\u0020punctuation23\\u0020 ", "@ => \\u0020punctuation25\\u0020 ", "# => \\u0020punctuation26\\u0020 ", "$ => \\u0020punctuation27\\u0020 ", "% => \\u0020punctuation28\\u0020 ", "* => \\u0020punctuation29\\u0020 ", "\\n => \\u0020punctuation30\\u0020 ", "℃ => \\u0020punctuation31\\u0020", "℉ => \\u0020punctuation32\\u0020", "± => \\u0020punctuation33\\u0020", "mg/l => \\u0020punctuation34\\u0020", "° => \\u0020punctuation35\\u0020", "™ => \\u0020punctuation36\\u0020", "® => \\u0020punctuation37\\u0020" ], "type": "mapping" }, "point_mapping": { "mappings": [ "? =>", "? => \\u0020sentence2\\u0020 ", "。 => \\u0020sentence3\\u0020 ", ". =>", "! => \\u0020sentence5\\u0020 ", "! => \\u0020sentence6\\u0020 ", ", => \\u0020punctuation0\\u0020 ", ", => \\u0020punctuation1\\u0020 ", "; => \\u0020punctuation2\\u0020 ", ": => \\u0020punctuation3\\u0020 ", ": => \\u0020punctuation4\\u0020 ", "“ => \\u0020punctuation5\\u0020 ", "\" => \\u0020punctuation6\\u0020 ", "” => \\u0020punctuation7\\u0020 ", "‘ => \\u0020punctuation8\\u0020 ", "' => \\u0020punctuation9\\u0020 ", "’ => \\u0020punctuation10\\u0020 ", "( => \\u0020punctuation11\\u0020 ", ") => \\u0020punctuation12\\u0020 ", "( => \\u0020punctuation13\\u0020 ", ") => \\u0020punctuation14\\u0020 ", "、 => \\u0020punctuation15\\u0020 ", "` => \\u0020punctuation16\\u0020 ", "+ => \\u0020punctuation17\\u0020 ", "- => \\u0020punctuation18\\u0020 ", "& => \\u0020punctuation19\\u0020 ", "| => \\u0020punctuation20\\u0020 ", "[ => \\u0020punctuation21\\u0020 ", "] => \\u0020punctuation22\\u0020 ", "^ => \\u0020punctuation23\\u0020 ", "@ => \\u0020punctuation25\\u0020 ", "# =>", "$ => \\u0020punctuation27\\u0020 ", "% => \\u0020punctuation28\\u0020 ", "* => \\u0020punctuation29\\u0020 ", "\\n => \\u0020punctuation30\\u0020 ", "℃ => \\u0020punctuation31\\u0020", "℉ => \\u0020punctuation32\\u0020", "± => \\u0020punctuation33\\u0020", "mg/l => \\u0020punctuation34\\u0020", "° => \\u0020punctuation35\\u0020", "™ => \\u0020punctuation36\\u0020", "® => \\u0020punctuation37\\u0020" ], "type": "mapping" } }, "analyzer": { "my_analyzer": { "filter": [ "lowercase" ], "char_filter": [ "punctuation_mapping" ], "type": "custom", "tokenizer": "standard" }, "my_analyzer1": { "filter": [ "lowercase" ], "char_filter": [ "point_mapping" ], "type": "custom", "tokenizer": "my_tokenizer1" }, "my_analyzer2": { "filter": [ "lowercase" ], "char_filter": [ "point_mapping" ], "type": "custom", "tokenizer": "my_tokenizer2" } }, "tokenizer": { "my_tokenizer2": { "token_chars": [ "letter", "digit" ], "min_gram": "2", "type": "ngram", "max_gram": "3" }, "my_tokenizer1": { "token_chars": [ "letter", "digit" ], "min_gram": "1", "type": "ngram", "max_gram": "1" } } } } }, "copy_to_fields": { "title_brief_claim": { "type": "text", "analyzer": "my_analyzer" }, "title_brief_claim2": { "type": "text", "analyzer": "my_analyzer" } }, "column": [ { "name": "id", "type": "keyword" }, { "name": "business_code", "type": "keyword" }, { "name": "eid", "type": "keyword" }, { "name": "company_id", "type": "keyword" }, { "name": "field_code", "type": "keyword" }, { "name": "field_name", "type": "keyword" }, { "name": "field_value", "type": "text" }, { "name": "create_time", "type": "keyword" }, { "name": "update_time", "type": "keyword" } ] } } } ], "setting": { "errorLimit": { "record": 1, "percentage": 0 }, "speed": { "readerChannel": 1, "writerChannel": 2 }, "platform": { "defaultParams": { "cyctime": "$[yyyyMMddHHmmss]" }, "jvm": { "perTaskMemory": 4 }, "enableDirty": true, "customQueues": {} } } } }