Move R script to vignette, add read_hdf5 function

Author: mrustl

DESCRIPTION

@@ -15,13 +15,17 @@ License: MIT + file LICENSE
 URL: https://github.com/KWB-R/kwb.raindrop
 BugReports: https://github.com/KWB-R/kwb.raindrop/issues
 Imports: 
+    dplyr,
     fs,
     hdf5r,
     kwb.utils,
     magrittr,
-    stringr
+    purrr, 
+    stringr,
+    tibble
 Suggests: 
     covr,
+    ggplot2,
     knitr,
     rmarkdown
 VignetteBuilder: 

NAMESPACE

@@ -1,7 +1,10 @@
 # Generated by roxygen2: do not edit by hand
 
 export("%>%")
+export(read_hdf5_timeseries)
 export(run_model)
+importFrom(dplyr,filter)
+importFrom(dplyr,pull)
 importFrom(fs,dir_create)
 importFrom(fs,file_copy)
 importFrom(fs,file_exists)
@@ -10,4 +13,7 @@ importFrom(hdf5r,H5File)
 importFrom(kwb.utils,catAndRun)
 importFrom(kwb.utils,resolve)
 importFrom(magrittr,"%>%")
+importFrom(purrr,map_dfr)
 importFrom(stringr,str_c)
+importFrom(tibble,as_tibble)
+importFrom(tibble,tibble)

R/.raindrop.R

@@ -1,6 +1,4 @@
-################################################################################
-### 1) Datei öffnen (lesen & schreiben) + Überblick
-################################################################################
+library(kwb.raindrop)
 
 path_list <- list(
   root_path = "C:/kwb/projects/raindrop/01102025_Raindrop_Daten",
@@ -24,78 +22,44 @@ path_list <- list(
   path_results_txt_multilayer = "<dir_target_output>/<file_results_txt_multilayer>", 
   path_target_input = "<dir_input>/<file_target>",
   path_target_output = "<dir_output>/<file_target>"
-  )
+)
 
 for(i in 1:100) {
-
-paths <- kwb.utils::resolve(path_list, dir_target = sprintf("s%05d", i))
-
-fs::dir_create(paths$dir_input)
-fs::dir_create(paths$dir_output)
-fs::dir_create(paths$dir_target_output)
-
-fs::file_copy(path = paths$path_base, 
-              new_path = paths$path_target_input, 
-              overwrite = TRUE)
-
-h5 <- hdf5r::H5File$new(paths$path_target_input, mode = "a")
-
-h5[["/Berechnungsparameter/Ergebnispfad"]]$read()
-
-new_path <- stringr::str_c(normalizePath(fs::path_abs(paths$dir_target_output)), 
-                           "\\")
-
-h5[["/Berechnungsparameter/Ergebnispfad"]][] <- new_path
-
-h5[["/Berechnungsparameter/Ergebnispfad"]]$read()
-
-# "a" = read/write (legt an, falls nicht da); alternativ "r+" = read/write, aber nicht neu anlegen
-
-h5$flush()
-## 2) Alle Handles schließen, sonst blockiert Windows das Umbenennen
-gc()
-h5$close_all()
-
-run_model(path_exe = paths$path_exe, 
-          path_input = paths$path_target_input)
+  
+  paths <- kwb.utils::resolve(path_list, dir_target = sprintf("s%05d", i))
+  
+  fs::dir_create(paths$dir_input)
+  fs::dir_create(paths$dir_output)
+  fs::dir_create(paths$dir_target_output)
+  
+  fs::file_copy(path = paths$path_base, 
+                new_path = paths$path_target_input, 
+                overwrite = TRUE)
+  
+  # "a" = read/write (legt an, falls nicht da); alternativ "r+" = read/write, aber nicht neu anlegen
+  h5 <- hdf5r::H5File$new(paths$path_target_input, mode = "a")
+  
+  h5[["/Berechnungsparameter/Ergebnispfad"]]$read()
+  
+  new_path <- stringr::str_c(normalizePath(fs::path_abs(paths$dir_target_output)), 
+                             "\\")
+  
+  h5[["/Berechnungsparameter/Ergebnispfad"]][] <- new_path
+  
+  h5[["/Berechnungsparameter/Ergebnispfad"]]$read()
+  
+  h5$flush()
+  
+  ## 2) Alle Handles schließen, sonst blockiert Windows das Umbenennen
+  gc()
+  h5$close_all()
+  
+  kwb.raindrop::run_model(path_exe = paths$path_exe,
+                          path_input = paths$path_target_input)
 }
 
 
-#### Read results
-
-### HDF-5
-
-i <- 1 
-paths <- kwb.utils::resolve(path_list, dir_target = sprintf("s%05d", i))
-
-
-read_hdf5_timeseries <- function(ts_groupvariable) {
-
-# 1) Nur Datasets (keine Subgruppen) auflisten
-ds_ts_groupvariable <- ts_groupvariable$ls() %>%
-  tibble::as_tibble() %>%
-  dplyr::filter(obj_type == "H5I_DATASET") %>%
-  dplyr::pull(name)
-
-ds_ts_groupvariable
-
-
-# als benannte Liste mit Matrizen (2 x N)
-states_list <- setNames(lapply(ds_ts_groupvariable, function(nm) ts_groupvariable[[nm]]$read()), ds_ts_groupvariable)
-
-# oder direkt als langes tibble (time/value angenommen: 1. Zeile = Zeit, 2. Zeile = Wert)
-states_long <- purrr::map_dfr(ds_ts_groupvariable, function(nm) {
-  m <- ts_groupvariable[[nm]]$read()
-  tibble::tibble(
-    variable = nm,
-    time  = as.numeric(m[1, ]),
-    value = as.numeric(m[2, ])
-  )
-})
-
-states_long
-}
-
+### Read results for first run
 
 simulation_names <- basename(fs::dir_ls(paths$dir_output))
 
@@ -105,34 +69,40 @@ errors_df <- lapply(simulation_names, function(s_name) {
   paths <- kwb.utils::resolve(path_list, dir_target = s_name)
 
   if(fs::file_exists(paths$path_errors_hdf5)) {
-  error_hdf <- hdf5r::H5File$new(paths$path_errors_hdf5, mode = "r")
-  
-  tibble::tibble(id = i, 
-                 path = paths$path_errors_hdf5,
-                 number_of_errors = error_hdf[["AnzahlFehler"]]$read()
-                 )
+    error_hdf <- hdf5r::H5File$new(paths$path_errors_hdf5, mode = "r")
+    
+    tibble::tibble(id = i, 
+                   path = paths$path_errors_hdf5,
+                   number_of_errors = error_hdf[["AnzahlFehler"]]$read()
+    )
   }
 }) %>% 
   dplyr::bind_rows()
 
+
+i <- 1 
+paths <- kwb.utils::resolve(path_list, dir_target = sprintf("s%05d", i))
+
+# "a" = read/write (legt an, falls nicht da); alternativ "r+" = read/write, aber nicht neu anlegen
 res_hdf <- hdf5r::H5File$new(paths$path_results_hdf5, mode = "r")
 
 hdf5_results <- list(
-  rates = read_hdf5_timeseries(ts_groupvariable = res_hdf5[["Raten"]]),
-  additional_evapotranspiration = read_hdf5_timeseries(ts_groupvariable = res_hdf5[["Zusaetzliche Variablen Evapotranspiration"]]),
-  additional_infiltration = read_hdf5_timeseries(ts_groupvariable = res_hdf5[["Zusaetzliche Variablen Infiltration"]]),
-  states = read_hdf5_timeseries(ts_groupvariable = res_hdf5[["Zustandsvariablen"]])
+  rates = kwb.raindrop::read_hdf5_timeseries(res_hdf5[["Raten"]]),
+  additional_evapotranspiration = kwb.raindrop::read_hdf5_timeseries(res_hdf5[["Zusaetzliche Variablen Evapotranspiration"]]),
+  additional_infiltration = kwb.raindrop::read_hdf5_timeseries(res_hdf5[["Zusaetzliche Variablen Infiltration"]]),
+  states = kwb.raindrop::read_hdf5_timeseries(res_hdf5[["Zustandsvariablen"]])
 )
 
+### Plot results
 
 for(name in names(hdf5_results)) {
-
-gg <- hdf5_results[[name]] %>% 
-  ggplot2::ggplot(ggplot2::aes(x = time, y = value)) + 
-  ggplot2::facet_wrap(~ variable, ncol = 1, scales = "free_y") +
-  ggplot2::geom_line() +
-  ggplot2::labs(title = name) +
-  ggplot2::theme_bw() 
-
-plot(gg)
+  
+  gg <- hdf5_results[[name]] %>% 
+    ggplot2::ggplot(ggplot2::aes(x = time, y = value)) + 
+    ggplot2::facet_wrap(~ variable, ncol = 1, scales = "free_y") +
+    ggplot2::geom_line() +
+    ggplot2::labs(title = name) +
+    ggplot2::theme_bw() 
+  
+  plot(gg)
 }

R/read_hdf5_timeseries.R

@@ -0,0 +1,74 @@
+#' Read HDF5 Results Time Series from HDF5 Group into a Long Tibble
+#'
+#' @description
+#' Extracts all **datasets** (no subgroups) from a given HDF5 group and
+#' converts each 2×N matrix into a tidy long table. It assumes the first row
+#' holds the time/index vector and the second row the values.
+#'
+#' @param ts_groupvariable `hdf5r::H5Group`
+#'   An open HDF5 group whose children are time-series datasets stored as 2×N
+#'   numeric matrices (`[1, ] = time/index`, `[2, ] = value`). Typically a group
+#'   like `"/Zustandsvariablen"` or similar in your model output file.
+#'
+#' @return A `tibble` with columns:
+#' \itemize{
+#'   \item `variable` (`character`): dataset name within the group.
+#'   \item `time` (`numeric`): time or index taken from the first row.
+#'   \item `value` (`numeric`): values taken from the second row.
+#' }
+#'
+#' @details
+#' The function lists all child objects of `ts_groupvariable`, filters for
+#' datasets (`H5I_DATASET`), reads each dataset into memory, and stacks them
+#' into one long tibble. Datasets are expected to be **2×N**; if your storage
+#' differs (e.g., `N×2` or 1D), adapt the reading logic accordingly.
+#'
+#' @section Assumptions:
+#' - Each dataset under `ts_groupvariable` is a numeric matrix of shape 2×N.
+#' - Row 1 is the time/index vector; row 2 contains the values.
+#'
+#' @examples
+#' \dontrun{
+#' library(hdf5r)
+#' f <- H5File$new("Optimierung_MuldenRigole.h5", mode = "r")
+#' grp <- f[["Zustandsvariablen"]]
+#' ts_long <- read_hdf5_timeseries(grp)
+#' head(ts_long)
+#' f$close_all()
+#' }
+#'
+#' @seealso
+#'   \code{\link[hdf5r]{H5File}}, \code{\link[hdf5r]{H5Group}}
+#'
+#' @importFrom tibble tibble as_tibble
+#' @importFrom dplyr filter pull
+#' @importFrom purrr map_dfr
+#' @export
+
+read_hdf5_timeseries <- function(ts_groupvariable) {
+  
+  # 1) Nur Datasets (keine Subgruppen) auflisten
+  ds_ts_groupvariable <- ts_groupvariable$ls() %>%
+    tibble::as_tibble() %>%
+    dplyr::filter(obj_type == "H5I_DATASET") %>%
+    dplyr::pull(name)
+  
+  ds_ts_groupvariable
+  
+  
+  # als benannte Liste mit Matrizen (2 x N)
+  states_list <- setNames(lapply(ds_ts_groupvariable, function(nm) {
+    ts_groupvariable[[nm]]$read()}), nm = ds_ts_groupvariable)
+  
+  # oder direkt als langes tibble (time/value angenommen: 1. Zeile = Zeit, 2. Zeile = Wert)
+  states_long <- purrr::map_dfr(ds_ts_groupvariable, function(nm) {
+    m <- ts_groupvariable[[nm]]$read()
+    tibble::tibble(
+      variable = nm,
+      time  = as.numeric(m[1, ]),
+      value = as.numeric(m[2, ])
+    )
+  })
+  
+  states_long
+}