Latenise token on query time - elasticsearch

Question

I need to latenise the query tokens that I use when querying (or filtering). I can do this on application level, but I was wondering if elasticsearch provides an out of the box solution.
I'm using ES 1.7.5 (as a service)

Answer 1

By default elasticsearch will use the same analyzer at index time and query time but it is possible to specify a search_analyzer which will only be used at query time.
Let's take a look at the following example:
# First we define an analyzer which will fold non ascii characters called `latinize`.
PUT books
{
"settings": {
"analysis": {
"analyzer": {
"latinize": {
"tokenizer": "standard",
"filter": ["asciifolding"]
}
}
}
},
"mappings": {
"book": {
"properties": {
"name": {
"type": "string",
"analyzer": "standard", # We use the standard analyzer at index time.
"search_analyzer": "latinize" # But we use the latinize analyzer at query time.
}
}
}
}
}
# Now let's create a document and search for it with a non latinized string.
POST books/book
{
"name": "aaoaao"
}
POST books/_search
{
"query": {
"match": {
"name": "ääöääö"
}
}
}
And bam! There is our document.
{
"took": 3,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"failed": 0
},
"hits": {
"total": 1,
"max_score": 0.30685282,
"hits": [
{
"_index": "books",
"_type": "book",
"_id": "AVkIXdNyDpmDHTvI6Cp1",
"_score": 0.30685282,
"_source": {
"name": "aaoaao"
}
}
]
}
}