execute.handler: execute.handler

Description

Execute the handler

Usage

execute.handler(analysis.page, params, plot.file, file.params = list(), device = svg, annotate.plot = analysis.page$annotate.plot, max.annotated.regions = 5000, logger = create.logger(stderr(), log4r:::FATAL + 1))

Arguments

analysis.page

AnalysisPage object

params

Named list of parameters. These can include arguments to svg and arguments to the handler function. If there are any extra arguments then an error is thrown.

plot.file

Path to file to create. Should not exist already.

file.params

Named list of parameters (but defaults to empty list). These will be passed through as-is and should correspond to FILE uploads (being length-2 lists with $name and $tmp_name elements).

device

The plotting device function to use. Default: svg. You might specify png instead (you are passing the actual function here, not its name).

annotate.plot

Logical, indicating whether I should try to annotate the SVG plot. (If you aren't using the SVG device then this should be set to FALSE to not waste time trying to annotate the plot.) Default: analysis.page$annotate.plot

max.annotated.regions

Integer. If the handler returns more than this many regions then do not try to annotate them in the plot. Default: 5000

logger

log4r object. Default: no logging (FATAL + 1)

Value

AnnotatedDataFrame, but throws error if the handler is not making a plot, or is returning invalid data.

Details

execute.handler executes the plot function in the handler based on the parameter list, checks that the output is valid, adds the SVG attributes to the plot, and returns an AnnotatedDataFrame.

All of the parameters in the parameter list are JSON decoded. Even though this is really just extra work for the scalar parameters, we do it because otherwise it is confusing who needs to be de/encoded and who doesn't.

It is OK if your handler doesn't turn off the device when it's done. This wrapper will check if the current device hasn't changed. If so, it will call dev.off. This is useful because then you can use the same function in an interactive session, and also saves you one line of code. It's also OK if your handler *does* turn off the device. Then the current device will have decreased and the wrapper will known not to call dev.off again.

It is also OK if your handler returns a data.frame instead of an AnnotatedDataFrame. It just has to have x, y. An AnnotatedDataFrame will be built The interpretation of the fields in the AnnotatedDataFrame depend on your front end, but the guidelines are like this:

type: "text", "numeric" or "none", to set sorting and filtering options.

labelDescription

A display name for the column, instead of showing the actual name.

If $no.plot is true then the plotting device won't be opened or closed, and of course the plot won't be annotated.

If annotate.data.frame is set then your data.frame is converted to an AnnotatedDataFrame and your AnnotatedDataFrame is converted to an AnalysisPageDataNode of "table" type automatically.

Examples

Run this code

page <- new.analysis.page(AnalysisPageServer:::sine.handler)
plot.file <- tempfile(fileext = ".svg")
plist <- lapply(list(xmin=-2*pi, xmax=2*pi, n= 50), rjson::toJSON)
sine.data <- AnalysisPageServer:::execute.handler(page, plist, plot.file=plot.file)
# now sine.data is an AnnotatedDataFrame