Include log feature for the get_polis_api_data function

R/get_polis_api_data.R

@@ -34,6 +34,8 @@
 #'                        Default is retrieved from the environment variable
 #'                        'POLIS_API_KEY'. An explicit API key can be provided
 #'                        if required.
+#' @param log_results Logical indicating whether to log the results of the API
+#' @param log_file_path Path to the directory where the log file will be saved
 #' @param updated_dates Logical indicating whether to use the 'LastUpdateDate'
 #' @return A data frame containing the requested data aggregated from all pages
 #'         of the API response. Each row represents a record, and columns
@@ -45,13 +47,15 @@
 #' }
 #' @export
 
-get_polis_api_data <- function(min_date,
+get_polis_api_data <- function(min_date = "2021-01-01",
                                max_date = Sys.Date(),
                                data_type = "cases",
                                region = "AFRO",
                                select_vars = NULL,
-                               updated_dates = TRUE,
-                               polis_api_key) {
+                               updated_dates = FALSE,
+                               polis_api_key,
+                               log_results = FALSE,
+                               log_file_path) {
 
   # API Endpoint and URL Construction
   api_endpoint <- "https://extranet.who.int/polis/api/v2/"
@@ -85,5 +89,41 @@ get_polis_api_data <- function(min_date,
   # process API response
   full_data <- process_api_response(response)
 
+  # log results
+  if (log_results) {
+
+    # Check if log file name is provided
+    if (is.null(log_file_path)) {
+      warning("No log file name provided. Logging is disabled.")
+      return(invisible(NULL))
+    }
+
+    # set up log file name
+    log_file_name <- paste0(log_file_path, "/", "polis_data_update_log.rds")
+
+    # Construct the log message
+    log_message <- data.frame(
+      Region = tools::toTitleCase(region),
+      QueryStartDate = as.Date(min_date, format = "%Y-%m-%d"),
+      QueryEndDate = as.Date(max_date, format = "%Y-%m-%d"),
+      DataStartDate = min(as.Date(full_data[[date_field]])),
+      DataEndDate = max(as.Date(full_data[[date_field]])),
+      PolisDataType = as.character(endpoint_suffix),
+      NumberOfVariables = ncol(full_data),
+      NumberOfRows = format(nrow(full_data), big.mark = ",")
+    )
+
+    if (file.exists(log_file_name)) {
+      log_data <- epiCleanr::import(log_file_name)
+      log_data <- rbind(log_data, log_message)
+    } else {
+      log_data <- log_message
+    }
+
+    # Save log file
+    epiCleanr::export(log_data, log_file_name)
+
+  }
+
   return(full_data)
 }

R/save_polis_data.R

@@ -37,7 +37,16 @@ save_polis_data <- function(polis_data, polis_path,
   cli::cli_process_done(  )
 
   # Check existing datasets and keep only the 5 most recent
-  existing_files <- list.files(polis_path, full.names = TRUE)
+  existing_files <- list.files(
+    polis_path, pattern = "\\.rds$", full.names = TRUE)
+
+  # Check existing RDS datasets and keep only the 5 most recent
+  existing_files <- list.files(
+    polis_path, pattern = "\\.rds$", full.names = TRUE)
+
+  # Exclude files that contain 'polis_data_update_log' in the name
+  existing_files <- grep(
+    "polis_data_update_log", existing_files, value = TRUE, invert = TRUE)
 
   if (length(existing_files) > 5) {
     # Sort files by date, assuming the naming convention holds the date info