Elasticsearch Intro Queries

queries.json

# Indexando un documento.
POST /newyork/article
{
  "title": "'Doctor Who' Versus 'Sherlock'? Keep Dreaming, Comic-Con",
  "keywords": ["doctor-who", "sherlock-holmes", "comic-con"],
  "pub_date": "2015-07-09T21:56:18Z",
  "print_page": 7,
  "section_name": {
    "display_name": "U.S.",
    "content": "us",
    "url": "http://www.nytimes.com/pages/national/index.html"
  },
  "web_url": "http://www.nytimes.com/aponline/2015/07/09/us/ap-us-comic-con-doctor-who-sherlock.html",
  "content": "..."
}

# Indexando un documento, cambiando el tipo de un campo.
POST /newyork/article/
{
  "title": "'Doctor Who' Versus 'Sherlock'? Keep Dreaming, Comic-Con",
  "pub_date": "nueve de julio de dos mil quince",
  "print_page": 7,
  "content": "..."
}

# Obtenemos el mapping de un tipo.
GET /newyork/article/_mapping

# Creamos un nuevo tipo con un mapping específico.
PUT /newyork/article2/_mapping
# Salida de lo de arriba, bajo el campo "article".
{
}

# Indexamos el documento con el nuevo tipo, ahora funciona.
POST /newyork/article2/
{
  "title": "'Doctor Who' Versus 'Sherlock'? Keep Dreaming, Comic-Con",
  "pub_date": "nueve de julio de dos mil quince",
  "print_page": 7,
  "content": "..."
}

# Hacemos una query por todos los documentos.
GET /newyork/article/_search
{
  "query": {
    "match_all": {}
  }
}

# Sólo documentos con la palabra "sherlock".
GET /nyt/article/_search
{
  "query": {
    "match": {
      "headline.main": "sherlock"
    }
  }
}

# Misma query, pero matcheando dos palabras simultáneamente.
GET /nyt/article/_search
{
  "query": {
    "match_phrase": {
      "headline.main": "doctor who"
    }
  }
}

# Dos queries, juntadas por una bool query.
GET /nyt/article/_search
{
  "query": {
    "bool": {
      "should": [
        {"match_phrase": {"headline.main": "doctor who"}},
        {"match": {"headline.main": "sherlock"}}
      ]
    }
  }
}

# Phrase match tiene que mantener posiciones relativas.
# minimum_number_should_match especifica cuántas cláusulas should deberían matchear.
GET /nyt/article/_search
{
  "query": {
    "bool": {
      "should": [
        {"match_phrase": {"headline.main": "who doctor"}},
        {"match": {"headline.main": "sherlock"}}
      ],
      "minimum_number_should_match": 2
    }
  }
}

# Para usar un filtro, usamos una filtered query.
# Filtramos acorde al valor del campo document_type.
GET /nyt/article/_search
{
  "query": {
    "filtered": {
      "query": {
        "bool": {
          "should": [
            {"match_phrase": {"headline.main": "doctor who"}},
            {"match": {"headline.main": "sherlock"}}
          ],
          "minimum_number_should_match": 1
        }
      },
      "filter": {
        "term": {"document_type": "article"}}
      }
    }
  }
}

# Se pueden especificar múltiples filters, agrupándolos con un and.
GET /nyt/article/_search
{
  "query": {
    "filtered": {
      "query": {
        "bool": {
          "should": [
            {"match_phrase": {"headline.main": "doctor who"}},
            {"match": {"headline.main": "sherlock"}}
          ],
          "minimum_number_should_match": 1
        }
      },
      "filter": {
        "and": [
          {"range": {"pub_date": {"gte": "now-2M"}}},
          {"term": {"document_type": "article"}}
        ]
      }
    }
  }
}

# Ejemplo de agregación.
GET /nyt/article/_search?search_type=count
{
  "query": {
    "match": {
      "headline.main": "Uruguay"
    }
  },
  "aggs": {
    "over_time": {
      "date_histogram": {
        "field": "pub_date",
        "format": "yyyy",
        "interval": "1y"
      }
    }
  }
}