An htmlwidget for jQuery QueryBuilder
jQuery QueryBuilder is a jQuery plugin offering an simple interface to create complex queries written by @mistic100. The javascript library can be found at https://github.com/mistic100/jQuery-QueryBuilder with documentation and examples at http://querybuilder.js.org/.
This htmlwidget works in shiny apps and uses jQuery QueryBuilder to build a query using data frame columns and dplyr to filter the data frame from the query.
library(devtools)
devtools::install_github("harveyl888/queryBuilder")
Minimally, queryBuilder can be used with a data frame and list of filters as follows
queryBuilder(data, filters)
queryBuilder can take a number of arguments:
data
: A data frame containing the data to be filtered.filters
: A list of lists containing filter information (see below).autoassign
: Boolean. If true then the filter information is automatically assigned according to column class and the filter
argument is ignored (default = false).default_condition
: Default condition for rules (can be 'AND' or 'OR', default = 'AND').allow_empty
: Boolean. If set to true then no validation error is thrown when the builder is entirely empty (default = false).display_errors
: Boolean. If set to true then an icon and tooltip explaining the error will be displayed (default = true).display_empty_filter
: Boolean. If true then an empty option will be included for each rule. If false then the first filter will be selected when creating the rule (default = true).chosen
: Boolean. Use jquery chosen (https://github.com/harvesthq/chosen) to select filter for rulesjQuery QueryBuilder can create powerful queries from a series of inputs. The filters
option of this widget can be used to specify specific columns of a data frame and determine how they behave in jQuery QueryBuilder. It is constructed as a list of lists containing a number of named attributes.
For example, using the mtcars
data, filters for mpg and cyl can be constructed as follows:
filters = list(list(name = 'mpg', type = 'double', min = min(mtcars$mpg), max = max(mtcars$mpg), step = 0.1),
list(name = 'cyl', type = 'double'))
Here both mpg and cyl are defined with type double. In addition, mpg has an allowable range between its minimum and maximum values.
Filter types include integer, double, string, date and boolean. In addition, select, radio and the selectize plugin are recognized.
If autoassign is set to true then all the columns from the data frame will be used as potential filters and they will be assigned according to the column class as follows:
numeric
: filter type = doubleinteger
: filter type = integercharacter
: filter type = stringfactor
: filter type = string, input = selectizeDate
: filter type = datelogical
: filter type = boolean, input = radioThe queryBuilder widget returns a number of variables back to Shiny apps, each of which are prefixed by the element id and outlined below for an element created using queryBuilderOutput('querybuilder')
:
querybuilder_filters
: stringified JSON representation of input filters. querybuilder_out
: result of jQuery QueryBuilder's getRules method. This output is used to subsequently filter the data frame.querybuilder_validate
: result of jQuery QueryBuilder's validate method.In addition to constructing the htmlwidget, an additional function called filterTable is defined:
filterTable(input$querybuilder_out, data, 'table')
returns a filtered data frame for element querybuilder
on data frame data
.
filterTable(input$querybuilder_out, data, 'text')
returns the input to dplyr filter_ for element querybuilder
on data frame data
.
A Shiny example can be found under inst/examples/example01. In this case part of the mtcars data are used with some additonal columns:
This example can also be found at http://162.243.57.47:3838/home/harvey/queryBuilder/
jQuery queryBuilder is a powerful tool and there are some features which are somewhat experimental.
A trending function has been added. This will filter based on a series of values that increase or decrease. The filter takes the form of
list(name = 'Trend', type = 'string', input = 'function_0')
This instructs the builder to create a rule with the desired operators and a selectize, sortable output to hold the values.
If filters are added with an input that starts with group
then they'll be treated in two different ways depending on the operator. If the operator belongs to an optgroup called Group
then the value can be chosen from a list of other inputs from the same group. This allows for group comparison such as Group_1 > Group_2
. If the operator belongs to a different optgroup then the value is treated as a scalar such as Group_1 > 12
. This allows for complex rules to be built with minimal setup.
An example of each of these features can be found under inst/examples/example02.