CODEWITHSUNDEEP
javascriptrethinkdbreql
totoro
totoro

Reputation: 195

RethinkDB GetNearest and Filter for Multi-Index

Is it possible to query nearby point that match with one or more tags? This is the data that I have:

r.db("test").table("Member"):

{ "POSTALCODE":  "526494" ,
"lat":  "1.3640" , "lon":  "103.9531" ,
"memberId":  "22b99a77-5f73-4331-a5da-dcdc46a4b4e7" ,
"membertags": ["tag1" ,"tag2" , "tag3" ] ,
"point": { "$reql_type$":  "GEOMETRY" , "coordinates": [103.9531 ,1.364],"type":  "Point"}
} {
"POSTALCODE":  "525842" ,
"lat":  "1.3516" , "lon":  "103.9384" ,
"memberId":  "79844285-7781-4bd4-819e-db54f90b8c8d" ,
"membertags": ["tag2" ,"tag3" ,"tag4" ,"tag5"] ,
"point": {"$reql_type$":  "GEOMETRY" ,"coordinates": [103.9384 ,1.3516] ,"type":  "Point" }
}

As highlighted in How do I run a filter on a getNearest() query in RethinkDB?, I could filter getNearest using eq 

var point = r.point(103.9531,1.364);  
r.db('test')
.table('Member')
.getNearest(point, {index: 'point', unit :'km'})
.filter(function(user) {
   return user('doc')('POSTALCODE').eq('525842') //1 RESULT
   //return user('doc')('membertags').eq('tag2') //EMPTY RESULT
})

but I could not get the result for multi-indexes (e.g. membertags) as suggested in rethinkdb with filter and getNearest commands (Option #1) - using .filter({membertags : 'tag2'}).

The secondary index was created using r.db("test").table("Member").index_create("membertags", multi=True). I'm aware that filter does not support secondary index, but could not find alternative solution that uses getNearest.

Upvotes: 2

Views: 144

Answers (1)

kureikain
kureikain

Reputation: 2314

If you can sacrifice performance to use filter instead of index, we can do this:

var point = r.point(103.9531,1.364);  
r.db('test')
.table('Member')
.getNearest(point, {index: 'point', unit :'km'})
.filter(function(user) {
  return user('doc')('membertags').contains('tag2')
})

Instead of using eq we now use contains to filter the result in which membertags field contaisn the tag.

Downside is that filter doesn't use index so performance maybe not that great.

Upvotes: 1

Related Questions