NOTE: An answer was found, see the end of this post and the comments for more information.
ElasticSearch (ver 6) is my kryptonite. I have been trying to add a sorting to a query for far too long and only getting errors in response:
- No mapping
- Fielddata is disabled on text fields by default. Set fielddata=true ...
- when I try to set "fielddata=true" ... I get more errors
- [field_sort] unknown field [fielddata], parser not found
I've tried ingore_unmapped and get the same "unknown field" error.
The ES documentation is so generic that it is unhelpful, and nothing here on SO has helped... I've looked at/read after running into various errors
- https://stackoverflow.com/a/17051944/10358406
- unknown field [dest], parser not found- error coming while reindexing
- Elasticsearch: Expected field name but got START_OBJECT
- how to set fielddata=true in kibana
So now...
Here is my current query that I use through Postman. I use Nest in my C# project to build this up. I know how to build queries if I can get the json structure right in Postman:
{
"from": 0,
"size": 50,
"aggs": {
"specs": {
"nested": {
"path": "specs"
},
"aggs": {
"names": {
"terms": {
"field": "specs.name",
"size": 10
},
"aggs": {
"specValues": {
"terms": {
"field": "specs.value",
"size": 10
}
}
}
}
}
}
},
"query": {
"constant_score": {
"filter": {
"bool": {
"must": [
{
"term": {
"Id": {
"value": 1
}
}
}
],
"must_not": [
{
"terms": {
"partId": []
}
}
],
"should": [
{
"terms": {
"picCode": [
"b02"
]
}
},
{
"terms": {
"partId": []
}
}
]
}
}
}
}
}
The results I get back are like this:
{
"took": 13,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"skipped": 0,
"failed": 0
},
"hits": {
"total": 281,
"max_score": 1,
"hits": [
{
"_index": "09812734",
"_type": "part",
"_id": "1234:1",
"_score": 1,
"_source": {
"id": "1234:1",
"partId": 1234,
"mcfCostCenterId": 1,
"oemCode": "ABC",
"oemDescription": "Blah blah blah",
"oemPartCode": "123456",
"picCode": "B02",
"description": "other blah blah",
"isServiceable": false,
"marketingDescription": "this thing does stuff"",
"salesVolume": 0,
"searchSortOrder": 0,
"catalogSortOrders": [],
"specs": [
{
"name": "Color",
"value": "NA"
},
{
"name": "Diameter",
"value": "7.0000"
},
{
"name": "OtherSpec",
"value": "Q"
},
{
"name": "LastSpec",
"value": "FuBar"
}
]
}
}
]
},
"aggregations": {
"specs": {
"doc_count": 18,
"names": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "Length",
"doc_count": 4,
"specValues": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "2",
"doc_count": 1
},
{
"key": "3",
"doc_count": 1
},
{
"key": "8",
"doc_count": 1
}
]
}
}
]
}
}
}
}
I need to sort on description or salesVolume in the results.
We have another query that does sorting, so I used that same code for this query, but get the above errors. That code builds up the json as follows:
{
"from": 0,
"size": 50,
"sort": [
{
"foo.bar": {
"missing": "_last",
"order": "desc",
"nested_filter": {
"term": {
"foo.bar": {
"value": "frankenstein"
}
}
},
"nested_path": "_source"
}
}
],
"aggs": {
"specs": {
"nested": {
"path": "specs"
...
As I try to edit that to make it relevant to this search, I just get errors and no results.
EDIT: So, getting "salesVolume" is pretty easy because it isn't a text field.
{
"from": 0,
"size": 50,
"sort": [
{
"salesVolume": {
"missing": "_last",
"order": "desc"
}
}
],
"aggs": {
...
This sets up a sort order no problem. So, the crux is text... I just can't figure out where to set the fielddata to true without erroring out.
EDIT 2:
In my DTO I have the field already set up with the keyword attribute.
public class MyClass
{
public string Id
[Text]
public string ThisCode
[Keyword]
public string MySortableTextField
... //and so forth
}
The search will still not perform and errors out.
ANSWER:
^^^ The Keyword attribute worked. I was not placing this in the right project. We have a project that builds out all our db indexes and that was still using the Text attribute. Once a indexing was redone then I was able to apply the sort order without errors.
