analyzer not being applied to search time

[clintongormley]

Hiya

It may be something else, but it looks like the analyzer defined in the mapping is being applied at index time, but not at search time:

I'm trying to add a custom analyzer which will lowercase terms, but preserve them as the whole term, so eg ABC123 becomes the single term abc123

However, if I search for ABC123 I would expect it to be converted to a search for abc123, and this isn't happening. If I search for abc123 then I get the correct results:

curl -XPUT 'http://127.0.0.2:9200/foo/'  -d '
{
   "index" : {
      "analysis" : {
         "analyzer" : {
            "lc" : {
               "filter" : [
                  "lowercase"
               ],
               "tokenizer" : "keyword"
            }
         }
      }
   }
}
'
# {
#    "ok" : true,
#    "acknowledged" : true
# }


curl -XPUT 'http://127.0.0.2:9200/foo/bar/_mapping'  -d '
{
   "properties" : {
      "myfield" : {
         "type" : "string",
         "analyzer" : "lc"
      }
   }
}
'
# {
#    "ok" : true,
#    "acknowledged" : true
# }


curl -XPOST 'http://127.0.0.2:9200/foo/bar'  -d '
{
   "myfield" : "ABC123"
}
'
# {
#    "ok" : true,
#    "_index" : "foo",
#    "_id" : "f413800e-69bf-4ca6-a9bd-484e08a371cf",
#    "_type" : "bar"
# }


curl -XGET 'http://127.0.0.2:9200/foo/bar/_search'  -d '
{
   "query" : {
      "term" : {
         "myfield" : "ABC123"
      }
   }
}
'
# {
#    "hits" : {
#       "hits" : [],
#       "total" : 0
#    },
#    "_shards" : {
#       "failed" : 0,
#       "successful" : 5,
#       "total" : 5
#    }
# }


curl -XGET 'http://127.0.0.2:9200/foo/bar/_search'  -d '
{
   "query" : {
      "term" : {
         "myfield" : "abc123"
      }
   }
}
'
# {
#    "hits" : {
#       "hits" : [
#          {
#             "_source" : {
#                "myfield" : "ABC123"
#             },
#             "_index" : "foo",
#             "_id" : "f413800e-69bf-4ca6-a9bd-484e08a371cf",
#             "_type" : "bar"
#          }
#       ],
#       "total" : 1
#    },
#    "_shards" : {
#       "failed" : 0,
#       "successful" : 5,
#       "total" : 5
#    }
# }

[clintongormley]

Doh of course - it would only be analyzed if passed in as a query string, not as a term.

closing

[kimchy]

Yea, but note that there is the field query as well, which make things simpler than using the query_string query.