Elastic Search multi match query can't ignore special characters - node.js

Question

I have a name field value as "abc_name" so when I search "abc_" I am getting proper results but when I search "abc_##£&-#&" still I am getting same results. I want my query to ignore this special characters that doesn't matches with my query.
My query has:
Multi_match
type as cross_fields
operator AND
I am using search_analyzer standard for my Fields
And I want this structure as it is otherwise it will affect my other Search behaviour
"name": {
"type": "text",
"fields": {
"keyword": {
"type": "keyword",
"ignore_above": 256
}
},
"analyzer": "autocomplete",
"search_analyzer": "standard"
}

Answer 1

Please see the below sample which would fit your use case where I've created a custom analyzer which would fit your use case:
Sample Mapping:
PUT some_test_index
{
"settings": {
"analysis": {
"analyzer": {
"my_custom_analyzer": {
"type": "custom",
"tokenizer": "custom_tokenizer",
"filter": ["lowercase", "3_5_edge_ngram"]
}
},
"tokenizer": {
"custom_tokenizer": {
"type": "pattern",
"pattern": "\\w+_+[^a-zA-Z\\d\\s_]+|\\s+". <---- Note this pattern
}
},
"filter": {
"3_5_edge_ngram": {
"type": "edge_ngram",
"min_gram": 3,
"max_gram": 5
}
}
}
},
"mappings": {
"properties": {
"my_field":{
"type": "text",
"analyzer": "my_custom_analyzer"
}
}
}
}
The above mentioned pattern would simply ignore the tokens with the format like abc_$%^^##. As a result this token would not be indexed.
Note that the way the analyzer works is:
First executes tokenizer
Then applies the edge_ngram filter on the tokens generated.
You can verify by simply removing the edge_ngram filter in the above mapping to first understand what tokens are getting generated via Analyze API which would be as below:
POST some_test_index/_analyze
{
"analyzer": "my_custom_analyzer",
"text": "abc_name asda efg_!##!## 1213_adav"
}
Tokens generated:
{
"tokens" : [
{
"token" : "abc_name",
"start_offset" : 0,
"end_offset" : 8,
"type" : "word",
"position" : 0
},
{
"token" : "asda",
"start_offset" : 9,
"end_offset" : 13,
"type" : "word",
"position" : 1
},
{
"token" : "1213_adav",
"start_offset" : 25,
"end_offset" : 34,
"type" : "word",
"position" : 2
}
]
}
Note that the token efg_!##!## has been removed.
I've added edge_ngram fitler as you would want the search to be successful if you search with abc_ if your tokens generated via tokenizer is abc_name.
Sample Document:
POST some_test_index/_doc/1
{
"my_field": "abc_name asda efg_!##!## 1213_adav"
}
Query Request:
Use-case 1:
POST some_test_index/_search
{
"query": {
"match": {
"my_field": "abc_"
}
}
}
Use-case-2:
POST some_test_index/_search
{
"query": {
"match": {
"my_field": "efg_!##!##"
}
}
}
Responses:
Response for use-case-1:
{
"took" : 0,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 1,
"relation" : "eq"
},
"max_score" : 0.47992462,
"hits" : [
{
"_index" : "some_test_index",
"_type" : "_doc",
"_id" : "1",
"_score" : 0.47992462,
"_source" : {
"my_field" : "abc_name asda efg_!##!## 1213_adav"
}
}
]
}
}
Response for use-case-2:
{
"took" : 0,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 0,
"relation" : "eq"
},
"max_score" : null,
"hits" : [ ]
}
}
Updated Answer:
Create your mapping as follows based on the index I've created and let me know if that works:
PUT some_test_index
{
"settings": {
"analysis": {
"analyzer": {
"my_custom_analyzer": {
"type": "custom",
"tokenizer": "punctuation",
"filter": ["lowercase"]
}
},
"tokenizer": {
"punctuation": {
"type": "pattern",
"pattern": "\\w+_+[^a-zA-Z\\d\\s_]+|\\s+"
}
}
}
},
"mappings": {
"properties": {
"my_field":{
"type": "text",
"analyzer": "autocompete", <----- Assuming you have already this in setting
"search_analyzer": "my_custom_analyzer". <----- Note this
}
}
}
}
Please try and let me know if this works for all your use-cases.