get_ecosite_history_from_NASIS_db: allow custom filtering on ES class…

…ifier field; thanks to @natearoe

R/get_ecosite_history_from_NASIS_db.R

@@ -4,13 +4,14 @@
 #' 
 #' @param best Should the "best" ecological site correlation be chosen? Creates field called `es_selection_method` with `"most recent"` or `"least missing data"` for resolving many:1 relationships in site history.
 #' @param SS Use selected set? Default: `TRUE`
+#' @param es_classifier Optional: character. Vector of classifier names (and corresponding records) to retain in final result. 
 #' @param dsn Path to SQLite data source, or a `DBIConnection` to database with NASIS schema.
 #'
 #' @seealso [get_extended_data_from_NASIS_db()]
 #' 
 #' @return a `data.frame`, or `NULL` on error
 #' @export
-get_ecosite_history_from_NASIS_db <- function(best = TRUE, SS = TRUE, dsn = NULL) {
+get_ecosite_history_from_NASIS_db <- function(best = TRUE, SS = TRUE, es_classifier = NULL, dsn = NULL) {
 
   .SD <- NULL
 
@@ -39,5 +40,5 @@ get_ecosite_history_from_NASIS_db <- function(best = TRUE, SS = TRUE, dsn = NULL
   }
 
   # load "best" siteecositehistory records: creates column 'es_selection_method' w/ "most recent" or "least missing data"
-  as.data.frame(data.table::as.data.table(ecositehistory)[, .pickBestEcosite(.SD), by = list(siteiid = ecositehistory$siteiid)])
+  as.data.frame(data.table::as.data.table(ecositehistory)[, .pickBestEcosite(.SD, es_classifier = es_classifier), by = list(siteiid = ecositehistory$siteiid)])
 }

R/utils.R

@@ -103,13 +103,17 @@
 
 ## TODO: this may need some review
 ## try and pick the best possible ecosite record
-.pickBestEcosite <- function(d) {
+.pickBestEcosite <- function(d, es_classifier = NULL) {
 
+  if (!is.null(es_classifier)) {
+    d <- d[which(d$es_classifier %in% es_classifier),]
+  }
+
 	# add a method field
 	d$es_selection_method <- NA_character_
 
 	# try to get the most recent:
-	d.order <- order(d$ecositecorrdate, decreasing=TRUE)
+	d.order <- order(d$ecositecorrdate, decreasing = TRUE)
 
 	# if there are multiple (unique) dates, return the most recent
 	if (length(unique(d$ecositecorrdate)) > 1) {