README.Rmd

---
title: "SticsOnR"
output:
  github_document:
   html_preview: true
params:
  javastics_dist: 'JavaSTICS-1.5.2-STICS-10.1.0'
  #javastics_dist: 'JavaSTICS-1.41-stics-9.2'
  #java_cmd: '/usr/lib/jvm/java-1.11.0-openjdk-amd64/bin/java'
  java_cmd: 'java'
  root_path: '/home/plecharpent/Téléchargements'
  #root_path: 'F:/TESTS/test_SticsOnR'
  work_dir: 'example'
  windows: !r Sys.info()[['sysname']]=='Windows'
---


```{r setup, echo = FALSE, warning=FALSE}
suppressPackageStartupMessages(library(SticsRFiles))
suppressPackageStartupMessages(library(SticsOnR))
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-"
)
options(tibble.print_min = 5, tibble.print_max = 5)

javastics_dist <- params$javastics_dist
root_path <- normalizePath(params$root_path, winslash = "/")
javastics_path <- file.path(root_path, params$javastics_dist)
workspace_path <- file.path(javastics_path, params$work_dir)
output_path <- file.path(root_path, "gen_usms_xml2txt")


windows <- params$windows
linux <- !windows

java_cmd <- "java"
if (linux) {
  java_cmd <- normalizePath(params$java_cmd, winslash = "/")
}
```

```{r error=FALSE, echo=FALSE, results='hide'}
# For testing consistency betwen JavaSTICS and java version

# Adding a test to see if javastics is compatible with java version
res <- try(
  run_javastics(javastics_path,
    workspace_path,
    usm = "wheat", verbose = FALSE,
    java_cmd = java_cmd
  ),
  silent = TRUE
)


try_err <- attr(res, "class")


if (!is.null(try_err) && attr(try_err, "class") == "try-error") {
  ver <- system2(
    command = "java", args = "-version",
    stdout = TRUE, stderr = TRUE
  )
  stop(
    javastics_dist, ": javastics not compatible with java\n",
    ver[1],
    paste0(
      "\nA java path must be provided for a ",
      "compatible java version (11) to run_javastics"
    )
  )
}
```

The R package for the [STICS](https://eng-stics.paca.hub.inrae.fr/) model <img src="man/figures/logo.png" alt="logo" width="150" align="right" />

<!-- badges: start -->
[![Project Status: Active – The project has reached a stable, usable state and is being actively developed.](https://www.repostatus.org/badges/latest/active.svg)](https://www.repostatus.org/#active)
[![R build status](https://github.com/SticsRPacks/SticsOnR/workflows/R-CMD-check/badge.svg)](https://github.com/SticsRPacks/SticsOnR/actions)
[![Codecov test coverage](https://codecov.io/gh/SticsRPacks/SticsOnR/branch/main/graph/badge.svg)](https://app.codecov.io/gh/SticsRPacks/SticsOnR?branch=main)
[![DOI](https://zenodo.org/badge/166790380.svg)](https://zenodo.org/badge/latestdoi/166790380)
<!-- badges: end -->

The goal of SticsOnR is to perform simulations of the Stics model, downloadable with its graphical user interface from https://eng-stics.paca.hub.inrae.fr/Download.

If you want to be notified when a new release of this package is made, you can tick the Releases box in the "Watch / Unwatch => Custom" menu at the top right of [this page](https://github.com/SticsRPacks/SticsOnR).


## Prerequisites and technical tips

### JavaStics software

JavaStics must be installed and the minimal version is version 1.41.

The latest distribution version for Stics is downloadable [here](https://eng-stics.paca.hub.inrae.fr/Download).

The installation process only consists of unzipping the JavaStics archive, except for MacOS system  (see MacOS installation [specificities](#under-macos-systems))

### Under the Windows operating system

Be aware that the java virtual machine does not need to be installed to use the
JavaSTICS software, neither the graphical interface (`JavaStics.exe`) nor the command line interface (`JavaSticsCmd.exe`). Because, a Java machine is embedded in the JavaStics archive.

### Under linux operating systems

#### Java version

For using the JavaStics software (GUI and command line interface) under a `linux` operating system, the java version must be consistent with the JavaStics version

* for JavaStics 1.41, the java version must be at most a Java 8 version
* for JavaStics 1.5.0/1.5.1 versions, the java version must be at least a Java 11 version
* for JavaStics 1.5.2 version, the java version must be at least a Java 17 version 

So, for adapting the java version to the JavaStics version some manipulations must be done either by switching between versions through system commands [see here](https://sticsrpacks.github.io/SticsOnR/articles/Changing_java_version_linux.html), using a specific java executable path ([see running JavaStics](#javastics-command-line-interface)).


#### System libraries
For the linux operating system, the SticsOnR package may require to install an `xslt` library.

If the SticsOnR installation fails, and the `xslt` library is missing, the error message indicates what is the name of the xslt library to be installed (according to the common linux distributions). For example, for the Ubuntu or Debian OS `libxslt1-dev` must be installed.

### Under MacOS systems
The STICS executable must be compiled and imported in the JavaStics directory. The procedure is fully described
in the JavaStics documentation (see JavaStics_documentation.html under the doc folder), in the prerequisites sub-section under the Software section. Java version installation management is also described in it.

### Files/directories paths syntax
Under all systems, file paths must not contain any special character or space.

Under unix like systems, using the `~` in files or directories paths may cause errors in SticsOnR functions. So, it is safer for the moment to use absolute paths. This will be fixed in the future versions.

#### Remote installation tools

For installing packages from the Github site an additional package must be installed. One can use either `devtools` or [`remotes`](https://github.com/r-lib/remotes#readme)

For `devtools`, installation requires system dependent tools. They must be installed first.

* For Windows
RTools must be installed using the latest installer from  [here](https://cran.r-project.org/bin/windows/Rtools/)
* For linux
Development tools must be installed first, like the `build-essentials` package for a Debian like distribution for example.

Then the `devtools` package can be installed using:

```{r eval=FALSE}
install.packages("devtools")
```

For `remotes`, it can be directly installed using:

```{r eval=FALSE}
install.packages("remotes")
```


## Installation

### Recommended installation: `SticsRPacks`

The best way to install the packages from `SticsRPacks`, from which `SticsOnR` is part of, is by installing the `[SticsRPacks]` package:

```{r eval=FALSE}
devtools::install_github("SticsRPacks/SticsRPacks")
```

- With `remotes`

```{r eval=FALSE}
remotes::install_github("SticsRPacks/SticsRPacks")
```

The package will install the packages for you at the latest release version.

### Other way: install each package independently

#### SticsOnR

The package installation can be remotely done directly from [GitHub](https://github.com/) using either `devtools` or the lightweight `remote` one package

The latest release version can be installed using:

- With `devtools`

```{r eval=FALSE}
devtools::install_github("SticsRPacks/SticsOnR@*release")
```

- With `remotes`

```{r eval=FALSE}
remotes::install_github("SticsRPacks/SticsOnR@*release")
```

Normally, all the package dependencies will be installed for CRAN packages.

#### SticsRFiles

`SticsRFiles` must be installed manually using the above syntax, just replacing  **SticsOnR** with **SticsRFiles**.

## Loading the packages library
```{r lib_load}
library(SticsOnR)
library(SticsRFiles)
```


## Running the model

Here are basic examples which show you how to run the model either from a R model interface or a JavaStics (command line) one. More complete examples will be detailed in a specific documentation later.

### JavaStics command line interface

The JavaStics installation folder (for example, `r javastics_dist`) contains an `example` workspace folder with a set of runnable usms.


For running simulations from it, we can use the `run_javastics()` function.


* Specifying the JavaStics folder   
<pre>javastics_path <- `r paste0('/path/to/',javastics_dist)`</pre>

* Specifying a workspace
  * as a sub-folder of JavaStics    
<pre>workspace_path <- "example"</pre>
  * or an absolute path to an external folder    
<pre>workspace_path <- "/path/to/javastics/workspace"</pre>    
  
```{r eval = windows, echo = TRUE, include = TRUE}
########## For Windows or linux with a compatible java version ################
# Running specific usms from the workspace
run_javastics(javastics_path, workspace_path,
  usm = c("banana", "wheat"),
  verbose = FALSE
)

# Running all usms contained in the workspace
run_javastics(javastics_path, workspace_path,
  verbose = FALSE
)

# Getting information about execution:
runs_info <- run_javastics(javastics_path, workspace_path,
  usm = c("banana", "wheat"), verbose = FALSE
)

runs_info
```
* For linux systems, if the java version is not compatible
with the JavaStics version a compatible java executable path must be used
<pre>java_cmd <- "/path/to/java/exe"</pre>    

```{r eval = linux, echo = TRUE, include = TRUE}
################ Only for linux with a specific java executable################

# Running specific usms from the workspace
run_javastics(javastics_path, workspace_path,
  usm = c("banana", "wheat"),
  verbose = FALSE, java_cmd = java_cmd
)

# Running all usms contained in the workspace
run_javastics(javastics_path, workspace_path,
  verbose = FALSE, java_cmd = java_cmd
)

# Getting information about execution:
runs_info <- run_javastics(javastics_path, workspace_path,
  usm = c("banana", "wheat"),
  verbose = FALSE, java_cmd = java_cmd
)

runs_info
```

In the returned information, the error field name gives a list of messages from the JavaStics command line interface. If any `Error` key word appears in a message, the corresponding simulation failed. But, at the moment it is impossible to identify what is the error's origin. Things must be checked manually in the workspace, after running again the faulty usm (because the model input files are overwritten at each usms simulation).

### Running the model using the stics executable directly

We need for that a JavaStics folder and a directory with text input files for Stics, or a folder containing individual sub-directories for usms.

These directories can be generated using the `[SticsRFiles::gen_usms_xml2txt()]` function from the **SticsRFiles** package, by converting automatically XML files to Stics input text files. See the documentation [here](https://sticsrpacks.github.io/SticsRFiles/articles/Generating_Stics_text_files.html).

Example of use:
```{r eval = windows, echo = TRUE}
# For Windows
# Generating files for all the usms contained in the workspace
SticsRFiles::gen_usms_xml2txt(javastics_path,
  workspace = workspace_path,
  out_dir = output_path, verbose = FALSE
)
```

```{r eval = linux, echo = TRUE}
# For linux, using a specific java executable
# Generating files for all the usms contained in the workspace
SticsRFiles::gen_usms_xml2txt(javastics_path,
  workspace = workspace_path,
  out_dir = output_path, verbose = FALSE,
  java_cmd = java_cmd
)
```

```{r, echo=F, results='hide', eval = TRUE}
unlink(x = file.path(output_path, "intercrop_pea_barley"), recursive = TRUE)
```

The `run_stics()` function can be used as follows with one folder or multiple sub-folders.

The Stics executable path is set according to each operating system:

* for windows   
<pre>stics_path <- file.path(javastics_path,"bin","stics_modulo.exe")</pre>
* for linux    
<pre>stics_path <- file.path(javastics_path,"bin","stics_modulo")</pre>
* for Mac
<pre>stics_path <- file.path(javastics_path,"bin","stics_modulo_mac")</pre>


```{r run_stics_int, echo = FALSE, results='hide'}
# Specifying the Stics executable file path
stics_path <- file.path(javastics_path, "bin", "stics_modulo.exe")
# for linux
if (is_unix()) {
  stics_path <- file.path(javastics_path, "bin", "stics_modulo")
}
```


```{r run_stics}
# Specifying a directory containing Stics input files
# For example reusing a generated sub-directory in the previous section
# of the document
# Running one usm
files_dir_path <- file.path(output_path, "banana")
run_stics(stics_path, files_dir_path)

# Specifying a root directory containing usms individual directories
# For example reusing a generated directory in the previous section
# of the document
# Running two usms
run_stics(stics_path, output_path, usm = c("banana", "wheat"))

# Running all the usms defined in the sub-directories of output_path
run_stics(stics_path, output_path, usm = "all")

# Getting returned information about stics runs
runs_info <- run_stics(stics_path, output_path, usm = c("banana", "wheat"))

runs_info
```


### Advanced simulations parameterization

A specific function `stics_wrapper()` is dedicated to manage simulations with a higher level
of parameterization than what `run_stics()` offers.

This `stics_wrapper()` function allows:

- Forcing the values of a set of parameters (common or specific values per USM)
- Returning simulated daily outputs for each usm with possible dates and variables filtering
- Parallelizing simulations, and displaying execution time
- Run Usms in successive mode

As the `run_stics()` function, the `stics_wrapper()` operates on directories
containing text stics input files.

#### Defining simulations options

Simulation options can be fixed using the `stics_wrapper_options()` function.
Both of them are mandatory: the model executable path and the directory path containing usms sub-directories with text input files.

A template is returned by the function when called with no arguments:

```{r}
stics_wrapper_options()
```

For the example, we will use the default stics model version shipping with JavaStics and the directory where individual usms input directories have been generated:

```{r eval=FALSE}
sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  workspace = output_path, verbose = FALSE
)
```

By default, `stics_wrapper_options()` checks that `javastics`, `stics_exe` and `workspace` exists.

There are different solutions if you need to use a custom version of stics:

1. if it is already listed in the preference (e.g. added in JavaStics), simply provide its name (ID):

```{r, eval=FALSE}
sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  stics_exe = "stics_custom",
  workspace = output_path, verbose = FALSE
)
```

2. if it is located in the bin directory of the JavaStics installation directory, provide the executable name:

```{r, eval=FALSE}
sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  stics_exe = "stics_custom.exe",
  workspace = output_path, verbose = FALSE
)
```

3. if it is located in any other folder, provide the full path to the executable name, and no need to use `javastics_path`:

```{r, eval=FALSE}
sim_options <- stics_wrapper_options(
  stics_exe = "path/to/stics_custom.exe",
  workspace = output_path, verbose = FALSE
)
```

#### Simple simulations cases


- Without filtering usms or outputs

```{r, eval=FALSE}
results <- stics_wrapper(model_options = sim_options)
```

- Filtering on usms list

```{r, eval=FALSE}
usm <- c("wheat", "maize")

results <- stics_wrapper(model_options = sim_options, situation = usm)
```

- Filtering outputs on variables

```{r, eval=FALSE}
usm <- c("wheat", "maize")

stics_wrapper(
  model_options = sim_options, situation = usm,
  var = c("masec_n", "mafruit")
)
```

- Filtering outputs on variables and dates for several USMs

The argument `sit_var_dates_mask` must contain a named list (named by usms names) containing
data.frames, as the sim_list element of the list returned by stics\_wrapper (see here-after) or as observations data.

It defines a mask: stics\_wrapper will return a result for each USM, variable and date that contains at least a value (*i.e.* different from NA) in the mask.

The stics\_wrapper function returns a list that contains two elements:

- error, a boolean indicating if an error occurs during the simulations,
- sim\_list, a named list of data.frames containing the simulated values for the requested USMs, variables and dates.


```{r}
obs_list <- get_obs(
  workspace = workspace_path, usm = c("wheat", "maize"),
  verbose = FALSE
)

# Observations table for wheat
obs_list$wheat

sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  workspace = output_path, verbose = TRUE
)

results <- stics_wrapper(
  model_options = sim_options,
  sit_var_dates_mask = obs_list
)
head(results)
```

Some warnings may occur, as in this case (that is why `results$error` is TRUE), indicating that observed variables and/or observations dates are missing in simulated data.
Concerning the dates, this may be due to the USMs simulation period that may not include observed dates.
For the variables, this may be due to an incorrect spelling of the variables in obs\_list.


#### Simulations with forcing parameters

- Applying a single parameter values vector for all the selected usms

Parameters values are prescribed using the `param_values` argument. It can be a named vector containing the values and names of the parameters to force.
In this case, the same values will be applied for all the simulated usms.

```{r, eval=FALSE}
param_values <- c(0.002, 50)
names(param_values) <- c("dlaimax", "durvieF")

results <- stics_wrapper(
  model_options = sim_options, situation = usm,
  param_values = param_values
)
```

- Defining different parameters values depending on the usms

`param_values` can also be a data.frame or a tibble having one named column per parameter and an optional column named `situation` containing the name of the situations (USMs for Stics) that allows to define different values of the parameters for different situations.

```{r}
# Let's run usm wheat with c(dlaimax=0.001, durvieF=50),
# usm pea with c(dlaimax=0.001, durvieF=60),
# and usm maize with c(dlaimax=0.001, durvieF=70)
param_values <- data.frame(
  situation = c("wheat", "pea", "maize"),
  dlaimax = c(0.001, 0.001, 0.001),
  durvieF = c(50, 60, 70)
)

# Let's display it
param_values

results <- stics_wrapper(
  model_options = sim_options,
  param_values = param_values,
  situation = c("wheat", "maize")
)
```

#### Simulations in successive mode

USMs can be run in successive mode, if they are adequately defined (*i.e.* if the beginning and end of simulations are consistent), using the option `successive_usms`.

`successive_usms` is a list of vectors containing the names of the UMSs to consider as successive (e.g. `list(c("usm1.1","usm1.2"),c("usm2.1","usm2.2"))` defines 2 successions usm1.1 -> usm1.2 and usm2.1 -> usm2.2).

#### Other Optional arguments

- Displaying execution time

```{r eval=FALSE}
sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  workspace = output_path,
  time_display = TRUE,
  verbose = FALSE
)

results <- stics_wrapper(model_options = sim_options)
```

- Activating parallel execution

On may specify the number of cores to use with the cores argument.

```{r eval=FALSE}
sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  workspace = output_path,
  parallel = TRUE, time_display = TRUE,
  cores = 2, verbose = FALSE
)

results <- stics_wrapper(model_options = sim_options)
```

If cores is not given, parallel execution is performed over machine total cores number minus 1.

```{r eval=FALSE}
library(parallel)

# Used cores number
detectCores() - 1

sim_options <- stics_wrapper_options(
  javastics = javastics_path,
  workspace = output_path, parallel = TRUE,
  time_display = TRUE, verbose = FALSE
)

results <- stics_wrapper(model_options = sim_options)
```

## Getting help

If you have any question or suggestion or if you want to report a bug, please do it via the GitHub [issues](https://github.com/SticsRPacks/SticsOnR/issues).

Thanks for that, this would greatly help us to improve this package.

## Citation

If you have used this package for a study that led to a publication or report, please cite us. You can either use the citation tool from Github if you used the last version, or use `citation("SticsOnR")` from R otherwise.

## Code of conduct

Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.

The package is under intensive development, so you can fill an issue or request us a feature [here](https://github.com/SticsRPacks/SticsOnR/issues) at any time.