Elastic phrase prefix working phrase isnt - elasticsearch

Question

I am trying to return all documents that contain a string in the userName & documentName.
Data:
{
"userName" : "johnwick",
"documentName": "john",
"office":{
"name":"my_office"
}
},
{
"userName" : "johnsnow",
"documentName": "snowy",
"office": {
"name":"Abraham deVilliers"
}
},
{
"userName" : "johnnybravo",
"documentName": "bravo",
"office": {
"name":"blabla"
}
},
{
"userName" : "moana",
"documentName": "disney",
"office": {
"name":"deVilliers"
}
},
{
"userName" : "stark",
"documentName": "marvel",
"office": {
"name":"blabla"
}
}
I can perform an exact string match with:
}
_source": [ "userName", "documentName"],
"query": {
"multi_match": {
"query": "johnsnow",
"fields": [ "userName", "documentName"]
}
}
}
This successfully returns:
{
"userName" : "johnsnow",
"documentName": "snowy",
"office": {
"name":"Abraham deVilliers"
}
}
If i use type: phrase_fix with john i also get returned successfully 3 results.
But then i try with:
{
"query": {
"multi_match": {
"query": "ohn", // <---- match all docs that contain 'ohn'
"type": "phrase_prefix"
"fields": [ "userName", "documentName"]
}
}
}
Zero results are returned.

Answer 1

What you are looking for is the infix search and you need to have ngram tokenizer with a search time analyzer to achieve that.
Complete example with your sample data
Index mapping and setting
{
"settings": {
"analysis": {
"filter": {
"autocomplete_filter": {
"type": "Ingram", --> note this
"min_gram": 1,
"max_gram": 10
}
},
"analyzer": {
"autocomplete": {
"type": "custom",
"tokenizer": "standard",
"filter": [
"lowercase",
"autocomplete_filter"
]
}
}
},
"index.max_ngram_diff" : 10 --> this you can reduce based on your requirement.
},
"mappings": {
"properties": {
"userName": {
"type": "text",
"analyzer": "autocomplete",
"search_analyzer": "standard"
},
"documentName": {
"type": "text",
"analyzer": "autocomplete",
"search_analyzer": "standard"
}
}
}
}
Sample your docs and then use the same search query, I indexed only first and last doc for brevity and it returned me first doc
"hits": [
{
"_index": "infix",
"_type": "_doc",
"_id": "1",
"_score": 5.7100673,
"_source": {
"userName": "johnwick",
"documentName": "john"
}
}
]