Elasticsearch Analyzer忽略连字符

Question

We are in the process of implementing Elasticsearch and using .Nest in a .NET solution. We've created and loaded an index having several fields. We wish to define and analyzer that will produce the same search results for an itemNumber field when the user enters UNV-1234 or UNV1234 . The itemNumber field is limited to item number only with no additional words. However the item number field may have a number separated by a space.

Based on my searches, it seems that the keyword analyzer would provide the desired results. This is not working for us.

Can someone provide information on how best to accomplish this?

"itemNumber": {
    "type": "string",
    "index": "not_analyzed",
    "fields": {
        "_english": {
        "type": "string",
        "analyzer": "english"
        },
        "_keyword": {
        "type": "string",
        "analyzer": "keyword"
        },
        "_standard": {
        "type": "string",
        "analyzer": "standard"
        }
    }
}

Answer 1

You need to define a custom analyzer with "keyword" tokenizer and a pattern replace token filter to remove any special character and use it for your field. The analyzer can be defined as below

    "analysis" : { 
        "filter" : { 
            "cleanspecial": { 
                "type": "pattern_replace", 
                "pattern": "[^a-zA-Z0-9]", 
                "replacement": "" 
            } 
        }, 
        "analyzer" : { 
            "cleanspecialanalyzer": { 
                "filter": ["cleanspecial"], 
                "type": "custom", 
                "tokenizer": "keyword" 
             }
         }
    }

Note: Please verify the pattern, I haven't tested it.

You can change the mapping as below

"itemNumber": {
    "type": "string",
    "index": "not_analyzed",
    "fields": {
        "_english": {
        "type": "string",
        "analyzer": "english"
        },
        "_keyword": {
        "type": "string",
        "analyzer": "cleanspecialanalyzer"
        },
        "_standard": {
        "type": "string",
        "analyzer": "standard"
        }
    }
}

Search can happen on the field itemNumber._keyword