Query Elastic Search where word indexed has space - elasticsearch

Question

I recently tried to use Elastic search. However, I am struggling to query for the following scenario:
I have my index set up with this:
"analysis": {
"index_analyzer": {
"my_index_analyzer": {
"type": "custom",
"tokenizer": "standard",
"filter": ["standard", "lowercase", "nGram"],
"char-filter": ["my_pattern"]
}
},
"search_analyzer": {
"my_search_analyzer": {
"type": "custom",
"tokenizer": "standard",
"filter": ["standard", "lowercase", "nGram"],
"char-filter": ["my_pattern"]
}
},
"filter": {
"nGram": {
"type": "nGram",
"min_gram": 3,
"max_gram": 40
}
},
"char_filter" : {
"my_pattern":{
"type":"pattern_replace",
"pattern":"\u0020",
"replacement":""
}
}
And the documents that are indexed are :
{
name:'My self'
},
{
name:'Hell o'
}
If I search for Myself, I am expecting it to return the first JSON object, however this is not happening..
I am searching using this (where term is just the string being searched):
var query = {
match: {
location: term
}
};
client.search({
index: 'requests',
analyzer:'my_search_analyzer',
body: {
query:query
}
})
I would really appreciate some guidance on this!
Kind Regards
JB

Answer 1

You are almost there, your index definition only has some small issues and typos which we will fix:
You don't need index_analyzer and search_analyzer simply define my_index_analyzer and my_search_analyzer directly under the analyzer element.
char-filter should read char_filter (with underscore)
your space pattern needs an additional backslash
This is the corrected settings/mappings that I used:
{
"settings": {
"analysis": {
"analyzer": {
"my_index_analyzer": { <--- 1. directly under analyzer
"type": "custom",
"tokenizer": "standard",
"filter": [
"standard",
"lowercase",
"nGram"
],
"char_filter": [ <--- 2. underscore
"my_pattern"
]
},
"my_search_analyzer": { <--- 1. directly under analyzer
"type": "custom",
"tokenizer": "standard",
"filter": [
"standard",
"lowercase",
"nGram"
],
"char_filter": [ <--- 2. underscore
"my_pattern"
]
}
},
"filter": {
"nGram": {
"type": "nGram",
"min_gram": 3,
"max_gram": 40
}
},
"char_filter": {
"my_pattern": {
"type": "pattern_replace",
"pattern": "\\u0020", <--- 3. additional backslash
"replacement": ""
}
}
}
},
"mappings": {
"request": {
"properties": {
"location": {
"type": "string",
"index_analyzer": "my_index_analyzer"
}
}
}
}
}
Then you can index your two sample documents:
curl -XPUT localhost:9200/requests/request/1 -d '{"location":"My self"}'
curl -XPUT localhost:9200/requests/request/2 -d '{"location":"Hell o"}'
And you'll get what you expect:
curl -XPOST localhost:9200/requests/request/_search -d '{
"query": {
"match": {
"location": "Myself"
}
}
}'
will return the document with My self