check exisiting files for historical means

ecohealthalliance · Oct 13, 2023 · a5b6c4a · a5b6c4a
1 parent 4ef3b4d
commit a5b6c4a
Show file tree

Hide file tree

Showing 4 changed files with 1,020 additions and 800 deletions.
diff --git a/R/calculate_weather_anomalies.R b/R/calculate_weather_anomalies.R
@@ -27,7 +27,6 @@ calculate_weather_anomalies <-
     save_filename <- glue::glue("weather_anomaly_{date_selected}.gz.parquet")
     message(paste0("Calculating weather anomalies for ", date_selected))
 
-
     # Check if file already exists
     existing_files <- list.files(weather_anomalies_directory)
     if(save_filename %in% existing_files & !overwrite) {

diff --git a/R/calculate_weather_historical_means.R b/R/calculate_weather_historical_means.R
@@ -12,20 +12,26 @@
 calculate_weather_historical_means <- function(nasa_weather_transformed,
                                                nasa_weather_directory_transformed,
                                                weather_historical_means_directory,
-                                               days_of_year) {
+                                               days_of_year,
+                                               overwrite = FALSE) {
 
   # Set filename
   doy <- days_of_year
   doy_frmt <- str_pad(doy,width = 3, side = "left", pad = "0")
   save_filename <- glue::glue("historical_weather_mean_doy_{doy_frmt}.gz.parquet")
-
   message(paste("calculating historical weather means and standard deviations for doy", doy_frmt))
 
+  # Check if file already exists
+  existing_files <- list.files(weather_historical_means_directory)
+  if(save_filename %in% existing_files & !overwrite) {
+    message("file already exists, skipping download")
+    return(file.path(weather_historical_means_directory, save_filename))
+  }
   # Open dataset to transformed data
-  weather_dataset <- open_dataset(nasa_weather_directory_transformed)
+  weather_transformed_dataset <- open_dataset(nasa_weather_directory_transformed)
 
   # Filter for day of year and calculate historical means and standard deviations
-  historical_means <- weather_dataset |> 
+  historical_means <- weather_transformed_dataset |> 
     filter(day_of_year == doy) |> 
     group_by(x, y, day_of_year) |> 
     summarize(historical_relative_humidity_mean = mean(relative_humidity),

diff --git a/_targets.R b/_targets.R
@@ -305,7 +305,8 @@ data_targets <- tar_plan(
   tar_target(weather_historical_means, calculate_weather_historical_means(nasa_weather_transformed, # enforce dependency
                                                                           nasa_weather_directory_transformed,
                                                                           weather_historical_means_directory,
-                                                                          days_of_year),
+                                                                          days_of_year,
+                                                                          overwrite = FALSE),
              pattern = days_of_year,
              format = "file", 
              repository = "local"),  

diff --git a/_targets/meta/meta b/_targets/meta/meta