In ropensci/elasticdsl: Elasticsearch DSL

knitr::opts_chunk$set(
  comment = "#>",
  warning = FALSE,
  message = FALSE
)

elastic DSL

You can define any query you can do against the Elasticsearch HTTP API with the Search() function in this package. However, writing queries like

'{
  "aggs": {
      "stats" : {
          "terms" : {
              "field" : "text_entry"
          }
      }
  }
}'

or as a nested list

list(aggs = list(stats = list(terms = list(field = "text_entry"))))

Is quite painful. Enter the elastic DSL, which is inspired by the Python library elasticsearch-dsl, but steals the general non-standard evaluation workflow from R's dplyr.

Here's what the workflow looks like:

index("shakespeare") %>%
   bool(must_not = list(term=list(speaker="KING HENRY IV"))) %>%
   range( speech_number == 5, line_id > 3 ) %>%
   Search()

Or, if you want to modify something about the search you can add on Search() like

index("shakespeare") %>%
   bool(must_not = list(term=list(speaker="KING HENRY IV"))) %>%
   range( speech_number == 5, line_id > 3 ) %>%
   Search(explain = TRUE)

Load the library

library("elastic")

Define object to query on

The index() function defines the index you want to query. You can optionally define a type within an index. The function has a nice print method too.

index("shakespeare")

index("shakespeare", "scene")

Generate queries

bool query

bool(must_not = list(term=list(speaker="KING HENRY IV")))

range query

range( speech_number == 5, line_id > 3 )

Put it all together

index("shakespeare") %>%
  bool(must_not = list(term=list(speaker="KING HENRY IV"))) %>%
  range( speech_number == 5, line_id > 3 )

ropensci/elasticdsl documentation built on May 18, 2022, 9:53 a.m.