mapping parent/child with _id mongo river elasticsearch - mongodb

Question

I've 3 collections in my database under mongo: shows - venues - dropdowns
shows are mapped like below
"show": {
"properties" : {
"description": {
"type": "string"
},
"image": {
"type": "string"
},
"site": {
"type": "string"
},
"title" : {
"type" : "multi_field",
"fields" : {
"title" : {"type" : "string", "index" : "analyzed"},
"raw_title" : {"type" : "string", "index" : "not_analyzed", "store": "no"}
}
}
}
}
venues like this
"venues": {
"properties" : {
"name" : {
"type" : "multi_field",
"fields" : {
"name" : {"type" : "string", "index" : "analyzed"},
"raw_name" : {"type" : "string", "index" : "not_analyzed", "store": "no"}
}
},
"city" : {
"type" : "multi_field",
"fields" : {
"city" : {"type" : "string", "index" : "analyzed"},
"raw_city" : {"type" : "string", "index" : "not_analyzed", "store": "no"}
}
},
"region" : {
"type" : "multi_field",
"fields" : {
"region" : {"type" : "string", "index" : "analyzed"},
"raw_region" : {"type" : "string", "index" : "not_analyzed", "store": "no"}
}
},
"state" : {
"type": "boolean"
}
}
}
and I've this model in mongo for dropdowns:
{
created: {
type: Date,
default: Date.now
},
analytics: {
type: String,
default: '',
trim: true
},
state: {
type: Boolean,
default: false,
index: true
},
show: {
type: Schema.ObjectId,
ref: 'Show'
},
venues:[{
venue:{
type: Schema.ObjectId,
ref: 'Venue',
index: true
},
site: {
type: String,
trim: true,
index: true
}
}]
}
I'd map dropdowns with parent/child schema into my index, but I can't understand if is possibile with ObjectId because I've tried with this mapping:
"dropdown": {
"properties" : {
"state": {
"type": "boolean"
},
"analytics": {
"type": "string"
},
"_parent":{
"type" : "show"
},
"venues" : {
"properties" : {
"venue" : {
"_parent": {
"type" : "venues"
}
}
},
"site" : {"type" : "string"}
}
}
}
But I received this error:
MapperParsingException[No type specified for property [show]]
There is anyway to setting up correctly my index?

Answer 1

Issue is that you're specifying _parent incorrectly. You have to set it not in properties field, but next to it. Please see documentation and example from it:
PUT /company
{
"mappings": {
"branch": {},
"employee": {
"_parent": {
"type": "branch"
}
}
}
}
So following that logic, I've taken your mappings, simplified it a bit and made it work:
PUT /test
{
"mappings": {
"show": {
"properties": {
"description": {
"type": "string"
},
"image": {
"type": "string"
},
"site": {
"type": "string"
},
"title": {
"type": "multi_field",
"fields": {
"title": {
"type": "string",
"index": "analyzed"
},
"raw_title": {
"type": "string",
"index": "not_analyzed",
"store": "no"
}
}
}
}
},
"venues": {
"properties": {
"name": {
"type": "multi_field",
"fields": {
"name": {
"type": "string",
"index": "analyzed"
},
"raw_name": {
"type": "string",
"index": "not_analyzed",
"store": "no"
}
}
},
"city": {
"type": "multi_field",
"fields": {
"city": {
"type": "string",
"index": "analyzed"
},
"raw_city": {
"type": "string",
"index": "not_analyzed",
"store": "no"
}
}
},
"region": {
"type": "multi_field",
"fields": {
"region": {
"type": "string",
"index": "analyzed"
},
"raw_region": {
"type": "string",
"index": "not_analyzed",
"store": "no"
}
}
},
"state": {
"type": "boolean"
}
}
},
"dropdown": {
"_parent": {
"type": "show"
},
"properties": {
"state": {
"type": "boolean"
},
"analytics": {
"type": "string"
},
"venues": {
"type": "object",
"_parent": {
"type": "venues"
},
"site": {
"type": "string"
}
}
}
}
}
}
I've tried this by myself on Elasticsearch 1.7.1 and it worked fine.
However, I'm not sure if you can declare _parent relationship inside nested documents as you did for venues. My mapping query didn't throw an error and accepted it. However, looking on how it got parsed in head plugin - _parent was eliminated and only object part remained as seen in screenshot:
If I tried to index it without specifying type - this error is thrown:
"MapperParsingException[mapping [dropdown]]; nested:
MapperParsingException[No type specified for property [venues]];