Nested Search Exercise

• Log-in into your ElasticSearch sandbox
• Make sure elastic search is running:

  sudo service elasticsearch restart
  

• Populate sample data using _bulk api:

  curl -XPOST http://localhost:9200/_bulk -d '{"index":{"_index":"nested-data","_type":"groups", "_id":"fans"}}
  {"group":"fans","members":[{"first":"John","last":"Smith"},{"first":"Alice","last":"White"}]}
  {"index":{"_index":"nested-data","_type":"groups", "_id":"supporters"}}
  {"group":"supporters","members":[{"first":"Alice","last":"Smith"},{"first":"John","last":"White"}]}
  '
  

• Confirm data has been populated:

  curl 'http://localhost:9200/nested-data/groups/_search?pretty=true'
  

• Expected result:

  {
  "took" : 2,
  "timed_out" : false,
  "_shards" : {
    "total" : 5,
    "successful" : 5,
    "failed" : 0
  },
  "hits" : {
    "total" : 2,
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "nested-data",
        "_type" : "groups",
        "_id" : "supporters",
        "_score" : 1.0,
        "_source" : {
          "group" : "supporters",
          "members" : [
            {
              "first" : "Alice",
              "last" : "Smith"
            },
            {
              "first" : "John",
              "last" : "White"
            }
          ]
        }
      },
      {
        "_index" : "nested-data",
        "_type" : "groups",
        "_id" : "fans",
        "_score" : 1.0,
        "_source" : {
          "group" : "fans",
          "members" : [
            {
              "first" : "John",
              "last" : "Smith"
            },
            {
              "first" : "Alice",
              "last" : "White"
            }
          ]
        }
      }
    ]
  }
  }
  

• Let's find group with John Smith in it:

  curl -XPOST 'http://localhost:9200/nested-data/groups/_search?pretty=true' -d '
  {
  "query" : {
      "bool" : {
        "must" : [
          {
              "match" : {"members.first" : "John"}
          },
          {
              "match" : {"members.last" : "Smith"}
          }
        ]
      }
    }
  }
  }'
  

• Expected result from Lucene's perspective:

  {
  "took" : 3,
  "timed_out" : false,
  "_shards" : {
    "total" : 5,
    "successful" : 5,
    "failed" : 0
  },
  "hits" : {
    "total" : 2,
    "max_score" : 0.51623213,
    "hits" : [
      {
        "_index" : "nested-data",
        "_type" : "groups",
        "_id" : "supporters",
        "_score" : 0.51623213,
        "_source" : {
          "group" : "supporters",
          "members" : [
            {
              "first" : "Alice",
              "last" : "Smith"
            },
            {
              "first" : "John",
              "last" : "White"
            }
          ]
        }
      },
      {
        "_index" : "nested-data",
        "_type" : "groups",
        "_id" : "fans",
        "_score" : 0.51623213,
        "_source" : {
          "group" : "fans",
          "members" : [
            {
              "first" : "John",
              "last" : "Smith"
            },
            {
              "first" : "Alice",
              "last" : "White"
            }
          ]
        }
      }
    ]
  }
  }
  

• Is that the expected result from the human perspective?
• To address the imperfection in the results we will recreate the index with nested mapping
• Delete the index:

  curl -XDELETE 'localhost:9200/nested-data'
  

• Re-create the index with mapping for nested type:

  curl -XPUT 'localhost:9200/nested-data' -d '
  {
  "mappings": {
    "groups": {
      "properties": {
        "members": {
          "type": "nested"
        }
      }
    } 
  }
  }'
  

• Re-populate sample data:

  curl -XPOST http://localhost:9200/_bulk -d '{"index":{"_index":"nested-data","_type":"groups", "_id":"fans"}}
  {"group":"fans","members":[{"first":"John","last":"Smith"},{"first":"Alice","last":"White"}]}
  {"index":{"_index":"nested-data","_type":"groups", "_id":"supporters"}}
  {"group":"supporters","members":[{"first":"Alice","last":"Smith"},{"first":"John","last":"White"}]}
  '
  

• Re-run the query:

  curl -XPOST 'http://localhost:9200/nested-data/groups/_search?pretty=true' -d '
  {
  "query" : {
      "bool" : {
        "must" : [
          {
              "match" : {"members.first" : "John"}
          },
          {
              "match" : {"members.last" : "Smith"}
          }
        ]
      }
    }
  }
  }'
  

• What results did you get?
• The reason for no results is that nested query uses different syntax:

  curl -XPOST 'http://localhost:9200/nested-data/groups/_search?pretty=true' -d '
  {
  "query": {
    "nested": {
      "path": "members",
      "query": {
        "bool": {
          "must": [
            { "match": { "members.first": "John" }},
            { "match": { "members.last":  "Smith" }} 
          ]
        }
      }
    }
  }
  }'
  

• Expected Results:

  {
  "took" : 6,
  "timed_out" : false,
  "_shards" : {
    "total" : 5,
    "successful" : 5,
    "failed" : 0
  },
  "hits" : {
    "total" : 1,
    "max_score" : 1.3862944,
    "hits" : [
      {
        "_index" : "nested-data",
        "_type" : "groups",
        "_id" : "fans",
        "_score" : 1.3862944,
        "_source" : {
          "group" : "fans",
          "members" : [
            {
              "first" : "John",
              "last" : "Smith"
            },
            {
              "first" : "Alice",
              "last" : "White"
            }
          ]
        }
      }
    ]
  }
  }
  

• Is the nested document mapping worth the hassle?