Rails Searchkick has_many indexing and searching - elasticsearch

Question

I am able to search by customer_id, name, lastname, and kids id, name, lastname and "birthdate"
The searching by id must be exact and it is. The searching by names or lastname has misspelling with distance 2 and it works. But I want to search also by kid_birthdate with match exact (mispelling, distance 0)
So far whenever I search by birthdate the results are returned like misspelling distance 2. I don't know how to search exact dates.
Rails 5.1.0.rc1
elasticsearch-5.0.3
searchkick-2.2.0
class Customer < ActiveRecord::Base
include Searchable
def search_data
attributes.merge avatar_url: avatar.url, kids: kids
end
has_many :kids
...
end
class Kid < ActiveRecord::Base
belongs_to :customer
def reindex_customer
customer.reindex async: true
end
...
end
module Searchable
extend ActiveSupport::Concern
included do
SEARCH_RESULTS_PER_PAGE = 10
def self.elastic_search(query, opts = { page: 1 })
# This regex accept string that contains digits or dates
regexp = /(\d+)|(^(0[1-9]|1\d|2\d|3[01])-(0[1-9]|1[0-2])-(19|20)\d{2}$)/
distance = query.match?(regexp) ? 0 : 2 #This is for calculate the distance for misspelling 0 for digits and dates and 2 for strings
options = { load: false,
match: :word_middle,
misspellings: { edit_distance: distance },
per_page: SEARCH_RESULTS_PER_PAGE,
page: opts[:page] }
search query, options
end
end
end
My index contains customer data with her/his kids data. Kids are nested under her/his parent customer.
How can I force the searching for exact matching for dates
For this query:
curl http://localhost:9200/customers_development/_search?pretty -d '{"query":{"dis_max":{"queries":[{"match":{"_all":{"query":"28388","boost":10,"operator":"and","analyzer":"searchkick_search"}}},{"match":{"_all":{"query":"28388","boost":10,"operator":"and","analyzer":"searchkick_search2"}}},{"match":{"_all":{"query":"28388","boost":1,"operator":"and","analyzer":"searchkick_search","fuzziness":0,"prefix_length":0,"max_expansions":3,"fuzzy_transpositions":true}}},{"match":{"_all":{"query":"28388","boost":1,"operator":"and","analyzer":"searchkick_search2","fuzziness":0,"prefix_length":0,"max_expansions":3,"fuzzy_transpositions":true}}}]}},"size":10,"from":0,"timeout":"11s"}'
This is how the index looks:
{
"took": 11,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"failed": 0
},
"hits": {
"total": 1,
"max_score": 97.29381,
"hits": [
{
"_index": "customers_development_20170913145033808",
"_type": "customer",
"_id": "28388",
"_score": 97.29381,
"_source": {
"id": 28388,
"created_at": "2017-07-10T19:49:43.856Z",
"updated_at": "2017-09-13T03:01:51.727Z",
"name": "Linda",
"lastname": "Schott",
"email": "linda.schott#web.de",
"avatar": null,
"phone": null,
"mobile": null,
"erster_kontakt": null,
"memo": null,
"brief_title": null,
"newsletter": null,
"avatar_url": "/no_customer.png",
"kids": [
{
"id": 34229,
"name": "Jakob",
"lastname": "Schott",
"birthdate": "2013-03-22",
"age": "4,5",
"avatar": {
"url": "/avatars/kid/34229/Jellyfish.png",
"thumb": {
"url": "/avatars/kid/34229/thumb_Jellyfish.png"
}
},
"created_at": "2017-07-10T19:50:16.058Z",
"updated_at": "2017-09-13T03:02:52.962Z",
"customer_id": 28388,
"member": null,
"year_certified": null,
"zahlart": null,
"tn_merge_markiert": null,
"family": null,
"medal": "black",
"score": 30,
"current_level": "swimmys"
},
{
"id": 34228,
"name": "Lilith",
"lastname": "Schott",
"birthdate": "2013-03-22",
"age": "4,5",
"avatar": {
"url": "/avatars/kid/34228/Penguins.png",
"thumb": {
"url": "/avatars/kid/34228/thumb_Penguins.png"
}
},
"created_at": "2017-07-10T19:50:16.058Z",
"updated_at": "2017-09-13T03:02:52.962Z",
"customer_id": 28388,
"member": null,
"year_certified": null,
"zahlart": null,
"tn_merge_markiert": null,
"family": null,
"medal": "green",
"score": 17,
"current_level": "beginner"
},
{
"id": 27718,
"name": "Johanna",
"lastname": "Plischke",
"birthdate": "2010-12-29",
"age": "6,8",
"avatar": {
"url": "/avatars/kid/27718/Koala.png",
"thumb": {
"url": "/avatars/kid/27718/thumb_Koala.png"
}
},
"created_at": "2017-07-10T19:50:16.034Z",
"updated_at": "2017-09-13T04:01:15.261Z",
"customer_id": 28388,
"member": null,
"year_certified": null,
"zahlart": null,
"tn_merge_markiert": null,
"family": null,
"medal": "red",
"score": 27,
"current_level": ""
}
]
}
}
]
}
}

Answer 1

Let's analyze the part of query:
"match":{
"_all":{
"query":"28388",
"boost":1,
"operator":"and",
"analyzer":"searchkick_search",
"fuzziness":0,
"prefix_length":0,
"max_expansions":3,
"fuzzy_transpositions":true
}
}
_all
You have said your kids is the nested field but you just search _all, so the first thing we should make it clear is whether kids is included in _all.
As the document says:
Sets the default include_in_all value for all the properties within the nested object. Nested documents do not have their own _all field. Instead, values are added to the _all field of the main “root” document.
So, first question is whether the index nested type has set include_in_all to false which makes nested fields can't be search by _all.
Nested Query
Or you can choose nested query to query nested object:
GET /_search
{
"query": {
"nested" : {
"path" : "kids",
"score_mode" : "avg",
"query" : {
"query_string": {
"fields": ["kids.birthdate"],
"query": "xxx"
}
}
}
}
}
Fuzzy
When it comes to misspelling, Elasticsearch recommend us to use fuzzy query:
GET /_search
{
"query": {
"fuzzy" : {
"name" : {
"value" : "xxx",
"boost" : 1.0,
"fuzziness" : 2,
"prefix_length" : 0,
"max_expansions": 100
}
}
}
}
Combine Query
And finally, we can combine them using bool query:
POST _search
{
"query": {
"bool" : {
"must" : [{
"nested" : {
"path" : "kids",
"query" : {
"query_string": {
"fields": ["kids.birthdate"],
"query": "xxx"
}
}
}
},
{ "fuzzy" : {
"name" : {
"value" : "xxx",
"boost" : 1.0,
"fuzziness" : 2,
"prefix_length" : 0,
"max_expansions": 100
}
}
}]
}
}
}
I am not familiar with Ruby, so that all I can help. Hope that helpful.