This _should_ fix #117

R/process_csv.R

@@ -8,16 +8,10 @@
 #' @noRd
 
 .process_csv <- function(x, isd_history) {
-
   # Import data from the website for individual stations or tempdir() for all --
-  DT <-
-    fread(x,
-      colClasses = c("STATION" = "character"),
-      strip.white = TRUE
-    )
+  DT <- fread(x, strip.white = TRUE, keepLeadingZeros = TRUE)
 
   # Replace 99.99 et al. with NA
-  set(DT, j = "PRCP", value = as.character(DT[["PRCP"]]))
   set(DT,
     i = which(DT[["PRCP"]] == "99.99"),
     j = "PRCP",
@@ -32,7 +26,6 @@
     "GUST",
     "SNDP"
   )]) {
-    set(DT, j = col, value = as.character(DT[[col]]))
     set(DT,
       i = which(DT[[col]] == "999.9"),
       j = col,
@@ -49,7 +42,6 @@
     "MAX",
     "MIN"
   )]) {
-    set(DT, j = col, value = as.character(DT[[col]]))
     set(DT,
       i = which(DT[[col]] == "9999.9"),
       j = col,
@@ -73,6 +65,13 @@
   # Add STNID col --------------------------------------------------------------
   DT[, STNID := gsub("^(.{6})(.*)$", "\\1-\\2", DT$STATION)]
 
+  # Correct STP values ---------------------------------------------------------
+  # The NCEI supplied CSV files are broken, they lop off the "1" in values >1000
+  # See https://github.com/ropensci/GSODR/issues/117
+  DT[, STP := fifelse(substr(STP, 1, 1) == 0,
+                      sprintf("%s%s", 1, DT$STP[[5]]),
+                      STP, na = NA)]
+
   # Add and convert date related columns ---------------------------------------
   DT[, YEARMODA := as.Date(DATE, format = "%Y-%m-%d")]
   DT[, YEAR := as.integer(substr(DATE, 1, 4))]