Move get_all_forecast_dirs() to hewr

hewr/R/parse_dir_name.R → hewr/R/directory_utils.R

@@ -1,8 +1,13 @@
+#' Utilities for handling and parsing directory names
+#' based on pyrenew-hew pipeline conventions.
+
 disease_map_lower <- list(
   "covid-19" = "COVID-19",
   "influenza" = "Influenza"
 )
 
+#' Parse model batch directory name.
+#'
 #' Parse the name of a model batch directory
 #' (i.e. a directory representing a single
 #' report date and disease pair, but potentially
@@ -40,6 +45,8 @@ parse_model_batch_dir_name <- function(model_batch_dir_name) {
   ))
 }
 
+#' Parse model run directory path.
+#'
 #' Parse path to a model run directory
 #' (i.e. a directory representing a run for a
 #' particular location, disease, and reference
@@ -61,3 +68,39 @@ parse_model_run_dir_path <- function(model_run_dir_path) {
     parse_model_batch_dir(batch_dir)
   ))
 }
+
+
+#' Get forecast directories.
+#'
+#' Get all the subdirectories within a parent directory
+#' that match the pattern for a forecast run for a
+#' given disease and optionally a given report date.
+#'
+#' @param dir_of_forecast_dirs Directory in which to look for
+#' subdirectories representing individual forecast date / pathogen /
+#' dataset combinations.
+#' @param diseases Names of the diseases to match, as a vector of strings,
+#' or a single disease as a string.
+#' @return A vector of paths to the forecast subdirectories.
+get_all_forecast_dirs <- function(dir_of_forecast_dirs,
+                                  diseases) {
+  # disease names are lowercase by convention
+  match_patterns <- stringr::str_c(tolower(diseases),
+    "_r",
+    collapse = "|"
+  )
+
+  dirs <- tibble::tibble(
+    dir_path = fs::dir_ls(
+      dir_of_forecast_dirs,
+      type = "directory"
+    )
+  ) |>
+    dplyr::filter(str_starts(
+      fs::path_file(dir_path),
+      match_patterns
+    )) |>
+    dplyr::pull(dir_path)
+
+  return(dirs)
+}

pipelines/collate_score_tables.R

@@ -12,35 +12,6 @@ purrr::walk(script_packages, \(pkg) {
 })
 
 
-#' Get all the subdirectories within a parent directory
-#' that match the pattern for a forecast run for a
-#' given disease and optionally a given report date.
-#'
-#' @param parent_dir Directory in which to look for forecast subdirectories.
-#' @param diseases Names of the diseases to match, as a vector of strings,
-#' or a single disease as a string.
-#' @return A vector of paths to the forecast subdirectories.
-get_all_forecast_dirs <- function(dir_of_forecast_date_dirs,
-                                  diseases) {
-  # disease names are lowercase by convention
-  match_patterns <- str_c(tolower(diseases), "_r", collapse = "|")
-
-  dirs <- tibble::tibble(
-    dir_path = fs::dir_ls(
-      dir_of_forecast_date_dirs,
-      type = "directory"
-    )
-  ) |>
-    dplyr::filter(str_starts(
-      fs::path_file(dir_path),
-      match_patterns
-    )) |>
-    dplyr::pull(dir_path)
-
-  return(dirs)
-}
-
-
 process_loc_date_score_table <- function(model_run_dir) {
   table_path <- fs::path(model_run_dir,
     "score_table",
@@ -137,7 +108,7 @@ collate_scores_for_date <- function(model_run_dir,
 collate_all_score_tables <- function(model_base_dir,
                                      disease,
                                      score_file_save_path = NULL) {
-  date_dirs_to_process <- get_all_forecast_dirs(
+  date_dirs_to_process <- hewr::get_all_forecast_dirs(
     model_base_dir,
     diseases = disease
   )