add record existence checks in read_all

R/parse.R

@@ -3,16 +3,27 @@
 #' Creates a list of tibbles containing relevant information from all records
 #' stored in the input directory.
 #'
-#' @param path character
+#' @param path path to records directory
+#' @param pattern search pattern to find records in directory
 #'
 #' @export
-read_all <- function(path="inst/records/") {
+read_all <- function(path="inst/records/", pattern="-data\\.xml") {
+    # append file separator if needed
+    path <- make_directory_path(path)
+    
     cat("Reading all records from", path, "\n")
-    available_protocols <- list.files(path)
-    res <- pblapply(available_protocols, read_one, path=path)
 
+    # list all files in directory and filter by search pattern
+    fs <- list.files(path)
+    available_protocols <- fs[str_detect(fs, pattern)]
     if (length(available_protocols) == 0)
-        stop("The given directory is empty or does not exist.")
+        stop(paste0("The given directory does not exist or does not contain files matching \"",
+                    pattern,
+                    "\"."))
+
+    # parse records one by one and remove null entries
+    res <- compact %$% pblapply(available_protocols, read_one, path=path)
+    if (length(res) == 0) stop("No valid records found. Did you fetch successfully?")
 
     lapply(res, `[[`, "speaker") %>%
         bind_rows() %>%
@@ -55,7 +66,8 @@ read_all <- function(path="inst/records/") {
 # this reads all currently parseable data from one xml
 read_one <- function(name, path) {
     x <- tryCatch(read_xml(paste0(path, name)),
-                  error = function(c) NULL)
+                  error = function(c) NULL,
+                  warning = function(c) NULL)
     if (is.null(x)) return(NULL)
     # extract date of session
     date <- xml_attr(x, "sitzung-datum")