Evaluating R Code
Evaluating R code with evalR
The WebR.evalR()
method can be used to evaluate R code and retrieve the result of the computation. The method takes two arguments,
code
- The R code to evaluate.options
- Optional: AEvalROptions
object controlling how the code is evaluated.
A promise is returned resolving to a reference to the result of the computation, given as an RObject
proxy.
The options
argument provides advanced control over how the R code is evaluated and captured. For example, this can be used to enable R’s autoprint feature.
The env
property of the options
argument can be set when you wish to evaluate the R code within an R environment, rather than the default global environment. This could be used, for example, to sandbox an evaluation in a local environment to avoid <-
creating variables in the global environment. Providing an environment is also a convenient way to pass data from Javascript to the R process.
The normal stream output is written to the JavaScript console. The error stream is also written to the console except when it is emitted with structured R conditions and captureConditions
is true
(the default). In this case, R errors raised during evaluation are rethrown as JavaScript exceptions, and R warnings and messages are rethrown as JavaScript warnings.
RObject
references returned by evalR()
are subject to memory management and should be destroyed when no longer in use. The related Shelter.evalR()
method can be used to automatically manage returned R objects using a webR shelter.
Returning JavaScript values when evaluating R code
A selection of convenience methods return the result of evaluating R code as a promise resolving to a JavaScript object, rather than an RObject
reference. The benefit is that since the returned object is of a raw JavaScript type, it does not need to be memory managed like an RObject
result would be.
Method | Returned object type |
---|---|
WebR.evalRBoolean() |
boolean |
WebR.evalRNumber() |
number |
WebR.evalRString() |
string |
WebR.evalRVoid() |
No return value |
The methods in the table above are shortcuts to the more general WebR.evalRRaw()
method. This takes an additional outputType
argument, which determines the type of raw JavaScript object returned. Using the evalRRaw()
method additionally allows for the return of JavaScript Array
objects.
objectType argument value |
Returned object type |
---|---|
'boolean' |
boolean |
'number' |
number |
'string' |
string |
'boolean[]' |
Array of boolean values |
'number[]' |
Array of number values |
'string[]' |
Array of string values |
'void' |
No return value |
Evaluating R code and capturing output with captureR
The Shelter.captureR()
method is lower level and more flexible than evalR()
. It allows to capture any stream output or conditions raised during evaluation of R code, in addition to returning the result of the computation.
Unlike evalR()
which only returns one R object, captureR()
returns a variable number of objects when R conditions are captured. Since this makes memory management of individual objects unwieldy, captureR()
requires the shelter approach to memory management, where all the sheltered objects are destroyed at once.
The method takes two arguments,
code
- The R code to evaluateoptions
- Optional: AEvalROptions
object controlling how the code is evaluated.
A promise is returned resolving resolving to a JavaScript object with two properties,
result
- The result of the computation, as anRObject
proxy.output
- Output captured during execution, an array of objects with propertiestype
anddata
.
Stream output is returned as elements of output
with a type
property of stdout
or stderr
. The line of output is returned in the form of a JavaScript string in the data
property.
Any raised conditions are also included as elements of output
with a type
property of message
, warning
, or error
. A reference to the condition is returned in the form of an RObject
proxy in the data
property.