elasticsearch ngram backslash escape - elasticsearch

Question

I am using a ngram analysier on my elastic search index. This is needed for the search capablity I require. I am searching for a document with a name called "l/test_V0001". When I search using "l/test" i am only getting results for "l" the / is working as a escape character and not as a text. I have searched and found this is a common issue and expected but can find no work around.
When i search the API for "l/test_V0001" I can find the result I am after. However when doing the same search via the java API I still only get results for "l".
here is the API search:
{
"query": {
"multi_match": {
"query": "l/test_V0001",
"fields": ["name", "name.partial", "name.text"]
}
}
}
and the mapping for the index:
{
"settings": {
"index": {
"max_ngram_diff": 20,
"search.idle.after": "10m"
},
"analysis": {
"analyzer": {
"ngram3_analyzer": {
"tokenizer": "ngram3_tokenizer",
"filter": [
"lowercase"
]
}
},
"tokenizer": {
"ngram3_tokenizer": {
"type": "ngram",
"min_gram": 3,
"max_gram": 20
}
}
}
},
"mappings": {
"dynamic": "strict",
"properties": {
"name": {
"type": "keyword",
"fields": {
"partial": {
"type": "text",
"analyzer": "ngram3_analyzer",
"search_analyzer": "keyword"
},
"text": {
"type": "text"
}
}
},
"value": {
"type": "integer"
}
}
}
}
any help on this or a work around would be great!

Answer 1

so after a bit of digging I found the answer using custom token chars. This is has added to the index mapping:
"tokenizer": {
"ngram3_tokenizer": {
"type": "ngram",
"min_gram": 3,
"max_gram": 20,
"token_chars": [
"letter",
"digit",
"symbol",
"custom"
],
"custom_token_chars": "/"
}
so my full index now looks like:
{
"settings": {
"index": {
"max_ngram_diff": 20,
"search.idle.after": "10m"
},
"analysis": {
"analyzer": {
"ngram3_analyzer": {
"tokenizer": "ngram3_tokenizer",
"filter": [
"lowercase"
]
}
},
"tokenizer": {
"ngram3_tokenizer": {
"type": "ngram",
"min_gram": 3,
"max_gram": 20,
"token_chars": [
"letter",
"digit",
"symbol",
"custom"
],
"custom_token_chars": "/"
}
}
}
},
"mappings": {
"dynamic": "strict",
"properties": {
"name": {
"type": "keyword",
"fields": {
"partial": {
"type": "text",
"analyzer": "ngram3_analyzer",
"search_analyzer": "keyword"
},
"text": {
"type": "text"
}
}
},
"value": {
"type": "integer"
}
}
}
}
this works for both rest client and java API