Problems with geo_shape support

Elastic Stack Elasticsearch
1

I've been trying to get geo spatial search to work with Elasticsearch.
The geo_shape variety seems to do what I want and I'm interested in
indexing and retrieving points inside polygons. Simple radius search or
bounding box search is not good enough.

So, I indexed about 10 points in a simple test index and then ran the
following query. Each point has a location field with type geo_shape and a
location field:
{"id":"id0.1","type":"poi","name":{"en":["0.1"]},"location":{"type":"point",
"coordinates":[52.1,13.1]}}

{"id":"id0.2","type":"poi","name":{"en":["0.2"]},"location":{"type":"point",
"coordinates":[52.2,13.2]}}

{"id":"id0.3","type":"poi","name":{"en":["0.3"]},"location":{"type":"point",
"coordinates":[52.3,13.3]}}

{"id":"id0.4","type":"poi","name":{"en":["0.4"]},"location":{"type":"point",
"coordinates":[52.4,13.4]}}

{"id":"id0.5","type":"poi","name":{"en":["0.5"]},"location":{"type":"point",
"coordinates":[52.5,13.5]}}

{"id":"id0.6","type":"poi","name":{"en":["0.6"]},"location":{"type":"point",
"coordinates":[52.6,13.6]}}

{"id":"id0.7","type":"poi","name":{"en":["0.7"]},"location":{"type":"point",
"coordinates":[52.7,13.7]}}

{"id":"id0.8","type":"poi","name":{"en":["0.8"]},"location":{"type":"point",
"coordinates":[52.8,13.8]}}

{"id":"id0.9","type":"poi","name":{"en":["0.9"]},"location":{"type":"point",
"coordinates":[52.9,13.9]}}

{"id":"id1.0","type":"poi","name":{"en":["1.0"]},"location":{"type":"point",
"coordinates":[53.0,14.0]}}

I'm trying to run the following query:

{

"query" : {
    "geo_shape" : {
        "location" : {
            "shape" : {
                "type" : "envelope",
                "coordinates" : [[52.0, 53.0],[13.0, 14.0]]
            },
            "relation" : "intersects"
        }
    }
}

}

I have several issues with this query and variants of it:

  • what I really wanted is polygonal search, not an envelope search. Thius
    doesn't work at all. I get an npe if I use a polygon. For reference, here's
    the polygon I've been testing with, roughly corresponding to a circular
    area around Berlin

[

[

52.51795646971664,

13.7939833902653

],

[

52.61972051694795,

13.720499817631334

],

[

52.67575539585586,

13.56279280917535

],

[

52.66465768725302,

13.38110108186347

],

[

52.59066633862847,

13.244824700054153

],

[

52.48204353028336,

13.2060166097347

],

[

52.38027948305205,

13.279500182368666

],

[

52.32424460414414,

13.43720719082465

],

[

52.33534231274698,

13.61889891813653

],

[

52.40933366137153,

13.755175299945847

]

]

  • what I really wanted is relation: contains. This doesn't work either. I
    get a QueryParsingException[[myindex] Unknown shape operation [contains ]]
    if I try that
  • if I run the above query, it fails with a TooManyClauses[maxClauseCount
    is set to 1024]

Obviously, I'm doing something very wrong. But what?

I.e. how do I actually search for points contained in a polygon of
arbitrary size?

BTW. using Elasticsearch 0.20.2

--

2

It seems google groups messed up the formatting big time. Apologies. Here's
the plain text version:

I've been trying to get geo spatial search to work with elastic search.
The geo_shape variety seems to do what I want and I'm interested in
indexing and retrieving points inside polygons. Simple radius search or
bounding box search is not good enough.

So, I indexed about a few points in a simple test index and then ran the
following query. Each point has a location field with type geo_shape and a
location field:
{"id":"id0.1","type":"poi","name":{"en":["0.1"]},"location":{"type":"point",
"coordinates":[52.1,13.1]}}
{"id":"id0.2","type":"poi","name":{"en":["0.2"]},"location":{"type":"point",
"coordinates":[52.2,13.2]}}
{"id":"id0.3","type":"poi","name":{"en":["0.3"]},"location":{"type":"point",
"coordinates":[52.3,13.3]}}

etc.

I'm trying to run the following query:

{
"query" : {
"geo_shape" : {
"location" : {
"shape" : {
"type" : "envelope",
"coordinates" : [[52.0, 53.0],[13.0, 14.0]]
},
"relation" : "intersects"
}
}
}
}

I have several issues with this query and variants of it:

  • what I really wanted is polygonal search, not an envelope search. Thius
    doesn't work at all. I get an npe if I use a polygon. For reference, here's
    the polygon I've been testing with, roughly corresponding to a circular
    area around Berlin

[ [ 52.51795646971664, 13.7939833902653 ], [ 52.61972051694795,
13.720499817631334 ], [ 52.67575539585586, 13.56279280917535 ], [
52.66465768725302, 13.38110108186347 ], [ 52.59066633862847,
13.244824700054153 ], [ 52.48204353028336, 13.2060166097347 ], [
52.38027948305205, 13.279500182368666 ], [ 52.32424460414414,
13.43720719082465 ], [ 52.33534231274698, 13.61889891813653 ], [
52.40933366137153, 13.755175299945847 ] ]

  • what I really wanted is relation: contains. This doesn't work either. I
    get a QueryParsingException[[myindex] Unknown shape operation [contains ]]
    if I try that

  • if I run the above query, it fails with a TooManyClauses[maxClauseCount
    is set to 1024]

Obviously, I'm doing something very wrong. But what?

I.e. how do I actually search for points contained in a polygon of
arbitrary size?

--

3

I figured it out finally. If anyone else is struggling with this:

  1. relation:contains should be relation:within, apparently this changed in
    the implementation. I created a pull request for the documentation fix
  2. geojson polygons are 3d arrays and not 2d arrays. Passing in a 2d array
    causes the parser to throw an npe.
  3. the polygon should end with the same coordinate as it begins with

So, this query is correct:

{

"query" : {

    "filtered" : {

    "query" : {

        "match_all" : {}

    },

    "filter" : {

     "geo_shape" : {

         "location" : {

             "shape" : {

                 "type" : "Polygon",

"coordinates" :
[[[52.51795646971664,13.7939833902653],[52.61972051694795,13.720499817631334],[52.67575539585586,13.56279280917535],[52.66465768725302,13.38110108186347],[52.59066633862847,13.244824700054153],[52.48204353028336,13.2060166097347],[52.38027948305205,13.279500182368666],[52.32424460414414,13.43720719082465],[52.33534231274698,13.61889891813653],[52.40933366137153,13.755175299945847],[52.51795646971664,13.7939833902653]]]

},

             "relation" : "within"

         }

     }

 }

}

}

}

--