Ramping Up on ElasticSearch Query DSL

gistfile1.json

// 1) Lets start with the simplest query that you can run in
//    the head plugin for ElasticSearch located at the url:
//    http://localhost:9200/_plugin/head/
{
  "query": {
    "match_all": {}
  }
}

// 2) Before jumping into forming queries with the ES Query DSL
//    syntax alone, lets try out a few queries written within
//    the comfort zone of those who are already familiar with
//    Lucene and/or Solr
{
  "query": {
    "query_string": {
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  }
}

// 3) Lets try it again but with slightly different syntax, it gives
//    exactly the same number of total hits as the previous query.
{
  "query": {
    "query_string": {
      "query": "camera AND laptop",
      "use_dis_max": true
    }
  }
}

// 4) What if we want to specify the exact fields or terms, which should
// be searched for the query? Well, here's how:
{
  "query": {
    "query_string": {
      "fields": [
        "name",
        "shortDescription"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  }
}

// 5) Someone who is not yet well-versed with JSON, may ask: what if I just want
// to search inside one field exactly for a match and not multiple ones?
// Simple, don't specify multiple fields:
{
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  }
}

// 6) That's all good but now we want to limit the fields coming back
//    in the response. It would be nice to specify exactly the fields
//    that we want to retrieve:
{
  "fields": [
    "name",
    "shortDescription",
    "longDescription"
  ],
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  }
}

// 7) Now lets try something different like a range query to try and
//    group our results so far into price buckets.
//
// WARNING: Make sure to POST to the exact index:
//              http://localhost:9200/my_index/_search/
//          And not against any & all indices
//              http://localhost:9200/_search/
//          Otherwise you may run into an exception like the following:
//              org.elasticsearch.search.facet.FacetPhaseExecutionException:
//              Facet [range1]: No mapping found for key_field [regularPrice]
//          from other indices that don't share the same fields:
{
  "fields": [
    "name",
    "upc",
    "salePrice",
    "regularPrice"
  ],
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "range1": {
      "range": {
        "regularPrice": [
          {
            "to": 100
          },
          {
            "from": 100,
            "to": 200
          },
          {
            "from": 200,
            "to": 300
          },
          {
            "from": 300
          }
        ]
      }
    }
  }
}

// 8) The total # of results is simple to find out but what about getting 
//    all the results back? Well both getting the results back and doing so
//    in an orderly and paginated manner is easy to do like so:
// 
//    8.1) Get the results starting from offset 0 and get a total of 2 entries
{
  "from": 0,
  "size": 2,
  "fields": [
    "name"
  ],
  "query": {
    "match_all": {}
  }
}
//    8.2) Get the results starting from offset 2 and get a total of 2 entries
{
  "from": 2,
  "size": 2,
  "fields": [
    "name"
  ],
  "query": {
    "match_all": {}
  }
}

// 9) If your index doesn't change often, then neither will the position of results
//    being returned by the query. These results are positioned by the score allotted
//    to each one of them in terms of how likely Lucene/ES thinks they are to be
//    the closest-fit / best-answer to your query.
//
//    Lets say you want random results, what then? How can you ask ES to change the
//    score and return different results for every query? Here's how:
{
  "from": 0,
  "size": 2,
  "fields": [
    "name",
    "upc"
  ],
  "query": {
    "custom_score": {
      "query": {
        "match_all": {}
      },
      "script": "random()"
    }
  }
}

// In this example by separating out the key and value facet fields,
// the range buckets are filled based on the key field ... while the
// averages etc. are calculated based on the the value field ... the
// assumption here (I suppose) is that the documents which have the
// key fields present, will also have the value field present (optimistic?)
{
  "fields": [
    "name",
    "upc",
    "salePrice",
    "regularPrice"
  ],
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "range1": {
      "range": {
        "key_field": "regularPrice",
        "value_field": "salePrice",
        "ranges": [
          {
            "to": 100
          },
          {
            "from": 100,
            "to": 200
          },
          {
            "from": 200,
            "to": 300
          },
          {
            "from": 300
          }
        ]
      }
    }
  }
}

// Good example of using histogram facets if the ranges that you specify
// happen to be at regular intervals anyway.
{
  "fields": [
    "name",
    "upc",
    "salePrice",
    "regularPrice"
  ],
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "histo1": {
      "histogram": {
        "field": "regularPrice",
        "interval": 100
      }
    }
  }
}

// The following is supposed to be the same but seems to give more metrics
{
  "fields": [
    "name",
    "upc",
    "salePrice",
    "regularPrice"
  ],
  "query": {
    "query_string": {
      "fields": [
        "name"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "histo1": {
      "histogram": {
        "key_field": "regularPrice",
        "value_field": "salePrice",
        "interval": 100
      }
    }
  }
}

// This is a really exciting example where we take our query results and
// bucket them by making a date histogram facet, therefore, shirking all
// responsibility of figuring out which items fall in which year by startDate
// All work is done by ES! Can you say $profit$ :)
{
  "fields": [
    "name",
    "upc",
    "salePrice",
    "regularPrice"
  ],
  "query": {
    "query_string": {
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "histo1": {
      "date_histogram": {
        "key_field": "startDate",
        "value_field": "startDate",
        "interval": "year"
      }
    }
  }
}

// What if you have a scenario where out want to figure out if there are
// really good sales going on in the area that you are interested in?
// Then statistical facet queries should get your motor started.
{
  "query": {
    "query_string": {
      "fields": [
        "name",
        "description"
      ],
      "query": "+camera +laptop",
      "use_dis_max": true
    }
  },
  "facets": {
    "stat1": {
      "statistical": {
        "field": "regularPrice"
      }
    },
    "stat2": {
      "statistical": {
        "field": "salePrice"
      }
    }
  }
}