outputs: Simulation outputs

Description

Accessor functions for the outputs slots in a simList object.

If a module saves a file to disk during events, it can be useful to keep track of the files that are saved e.g., for saveSimList() so that all files can be added to the archive. In addition to setting outputs at the simInit stage, a module developer can also put this in a using any saving mechanism that is relevant (e.g., qs::qsave, saveRDS etc.). When a module event does this it can be useful to register that saved file. registerOutputs offers an additional mechanism to do this. See examples.

Usage

outputs(sim)
# S4 method for simList
outputs(sim)
outputs(sim) <- value
# S4 method for simList
outputs(sim) <- value
registerOutputs(filename, sim, ...)
outputArgs(sim)
# S4 method for simList
outputArgs(sim)
outputArgs(sim) <- value
# S4 method for simList
outputArgs(sim) <- value

Value

A simList which will be the sim passed in with a new object registered in the outputs(sim)

Arguments

sim: A simList. If missing, then the function will search in the call stack, so it will find it if it is in a SpaDES module.
value: The object to be stored at the slot. See Details.
filename: The filename to register in the outputs(sim) data.frame. If missing, an attempt will be made to search for either a file or filename argument in the call itself. This means that this function can be used with the pipe, as long as the returned return from the upstream pipe function is a filename or if it is NULL (e.g., saveRDS), then it will find the file argument and use that.
...: Not used.

outputs function or argument in <code>simInit</code>

outputs accepts a data.frame similar to the inputs data.frame, but with up to 6 columns.

`objectName`	required, character string indicating the name of the object in the `simList` that will be saved to disk (without the `sim$` prefix).
`file`	optional, a character string indicating the file path to save to. The default is to concatenate `objectName` with the model timeunit and `saveTime`, separated by underscore, '`_`'. So a default filename would be `"Fires_year1.rds"`.
`fun`	optional, a character string indicating the function to use to save that file. The default is `saveRDS()`
`package`	optional character string indicating the package in which to find the `fun`);
`saveTime`	optional numeric, indicating when in simulation time the file should be saved. The default is the lowest priority at `end(sim)`, i.e., at the very end.
`arguments`	is a list of lists of named arguments, one list for each `fun`. For example, if `fun = "write.csv"`, `arguments = list(row.names = TRUE)` will pass the argument `row.names = TRUE` to `write.csv` If there is only one list, then it is assumed to apply to all files and will be recycled as per normal R rules of recycling for each `fun`.

See the modules vignette for more details (browseVignettes("SpaDES.core")).

Details

These functions are one of three mechanisms to add information about which output files to save.

As arguments to a simInit call. Specifically, inputs or outputs. See ?simInit.
With the outputs(simList) function call.
By adding a function called .inputObjects inside a module, which will be executed during the simInit call. This last way is the most "modular" way to create default data sets for your model.

See below for more details.

Note using registerOutputs: a user can pass any other arguments to registerOutputs that are in the outputs(sim) data.frame, such as objectName, fun, package, though these will not be used to save the files as this function is only about registering an output that has already been saved.

Examples

Run this code

#######################
# outputs
#######################

tmpdir <- file.path(tempdir(), "outputs") |> checkPath(create = TRUE)
tmpFile <- file.path(tmpdir, "temp.rds")
tempObj <- 1:10

# Can add data.frame of outputs directly into simInit call
sim <- simInit(objects = c("tempObj"),
               outputs = data.frame(objectName = "tempObj"),
               paths = list(outputPath = tmpdir))
outputs(sim) # To see what will be saved, when, what filename
sim <- spades(sim)
outputs(sim) # To see that it was saved, when, what filename

# Also can add using assignment after a simList object has been made
sim <- simInit(objects = c("tempObj"), paths = list(outputPath = tmpdir))
outputs(sim) <- data.frame(objectName = "tempObj", saveTime = 1:10)
sim <- spades(sim)
outputs(sim) # To see that it was saved, when, what filename.

# can do highly variable saving
tempObj2 <- paste("val", 1:10)
df1 <- data.frame(col1 = tempObj, col2 = tempObj2)
sim <- simInit(objects = c("tempObj", "tempObj2", "df1"),
  paths = list(outputPath = tmpdir))
outputs(sim) <- data.frame(
     objectName = c(rep("tempObj", 2), rep("tempObj2", 3), "df1"),
     saveTime = c(c(1, 4), c(2, 6, 7), end(sim)),
     fun = c(rep("saveRDS", 5), "write.csv"),
     package = c(rep("base", 5), "utils"),
     stringsAsFactors = FALSE)
# since write.csv has a default of adding a column, x, with rownames, must add additional
#   argument for 6th row in data.frame (corresponding to the write.csv function)
outputArgs(sim)[[6]] <- list(row.names = FALSE)
sim <- spades(sim)
outputs(sim)

# read one back in just to test it all worked as planned
newObj <- read.csv(dir(tmpdir, pattern = "year10.csv", full.name = TRUE))
newObj

# using saving with SpaDES-aware methods
# To see current ones SpaDES can do
.saveFileExtensions()

library(terra)
ras <- rast(ncol = 4, nrow = 5)
ras[] <- 1:20

sim <- simInit(objects = c("ras"), paths = list(outputPath = tmpdir))
outputs(sim) <- data.frame(
  file = "test",
  fun = "writeRaster",
  package = "terra",
  objectName = "ras",
  stringsAsFactors = FALSE)

simOut <- spades(sim)
outputs(simOut)
newRas <- rast(dir(tmpdir, full.name = TRUE, pattern = ".tif")[1])
all.equal(newRas, ras) # Should be TRUE
# Clean up after
unlink(tmpdir, recursive = TRUE)
# For `registerOutputs`
sim <- simInit()
# This would normally be a save call, e.g., `writeRaster`
tf <- reproducible::tempfile2(fileext = ".tif")
sim <- registerOutputs(sim, filename = tf)

# Using a pipe
tf <- reproducible::tempfile2(fileext = ".rds")
sim$a <- 1
sim <- saveRDS(sim$a, tf) |> registerOutputs()
# confirm:
outputs(sim) # has object --> saved = TRUE

Run the code above in your browser using DataLab