Create an EdgeNGram analyzer supporting both sides in Azure Search

问题

When defining a custom analyzer for Azure Search there is an option of defining a token filter from this list. I am trying to support search of both prefix and infix. For example: if a field contains the name: 123 456, I want the searchable terms to contain:

1
12
123
23
3
4
45
456
56
6

When using the EdgeNGramTokenFilterV2 which seems to do the trick, there is an option of defining a "side" property, but only "front" and "back" are supported, not both. the "front" (default) value generates this list:

1
12
123
4
45
456

and back generates:

123
23
3
456
56
6

I tried using two token two EdgeNGramTokenFilterV2s, but this creates terms from combining the two filters such as: "2" or "5":

1
12
123
23
3
4
45
456
56
6
2 // Unwanted
5 // Unwanted

I also tried using a "reverse" token, but this reverses everything and the results are still wrong.

I am using only one search field ("Name") and would prefer it to stay like this. (Thought of the option of using a different field named "name_reverse" with a different analyzer, but this is very inefficient and will cause a lot of headache when connecting the search engine to the data source.

For easier reference, this is the current index creation request:

{
 "name": "testindexboth",  
 "fields": [
  {"name": "id", "type": "Edm.String", "key": true },
  {"name": "Name", "type": "Edm.String", "searchable": true, "analyzer": "myAnalyzer"}
 ],
 "myAnalyzer": [
  {
   "name": "myAnalyzer",
   "@odata.type": "#Microsoft.Azure.Search.CustomAnalyzer",
   "tokenizer": "standard_v2",
    "tokenFilters":["front_filter", "back_filter"]
  }],

    "tokenFilters":[
            {
               "name":"front_filter",
               "@odata.type":"#Microsoft.Azure.Search.EdgeNGramTokenFilterV2",
               "maxGram":15,
               "side": "front"
            },
                        {
               "name":"back_filter",
               "@odata.type":"#Microsoft.Azure.Search.EdgeNGramTokenFilterV2",
               "maxGram":15,
               "side": "back"
            }
        ]
}

Is there an option of combining both, without getting them scramble up the results?

回答1:

Add two fields to your index, with two different custom analyzers: one for prefix, one for suffix. When querying, query against both fields.

来源：https://stackoverflow.com/questions/46756336/create-an-edgengram-analyzer-supporting-both-sides-in-azure-search