Clean give_hints_authors()

Author: mrustl

R/give_hints.R

@@ -16,6 +16,24 @@ give_hints_project_names <- function(project_names, dbg = TRUE) {
 }
 
 
+get_indices_multiple_authors <- function(author_names) {
+  which(unlist(lapply(author_names, function(x)
+    length(stringr::str_split(x, pattern = ",")[[1]]))) > 2)
+}
+
+get_indices_firstname_lastname <-  function(author_names) {
+  which(
+  stringr::str_detect(author_names, "^\\s?\\w+\\.?\\s+\\w+") &
+    !stringr::str_detect(author_names, ",")
+)}
+
+
+get_indices_ends_with_semicolon <- function(author_names) {
+  which(author_names %>%
+  stringr::str_trim() %>%
+  stringr::str_detect(".*,$"))
+}
+
 #' Helper Function: Give Hints For Author Names
 #'
 #' @param author_names vector with author names to check
@@ -30,26 +48,24 @@ give_hints_author_names <- function(author_names, dbg = TRUE) {
     "Generate hints for 'Author Names'",
     dbg = dbg,
     expr = {
-      multiple_authors_idx <- which(unlist(lapply(author_names, function(x)
-        length(stringr::str_split(x, pattern = ",")[[1]]))) > 2)
-
-      firstname_lastname_idx <- which(
-        stringr::str_detect(author_names, "^\\s?\\w+\\.?\\s+\\w+") &
-          !stringr::str_detect(author_names, ",")
-      )
-
-      if (length(multiple_authors_idx) > 0) {
-        author_names[multiple_authors_idx] <- "fix_multiple_authors_per_line"
-      }
+      multiple_authors_idx <- get_indices_multiple_authors(author_names)
+      firstname_lastname_idx <- get_indices_firstname_lastname(author_names)
+      ends_with_semicolon_idx <- get_indices_ends_with_semicolon(author_names)
 
-      if (length(firstname_lastname_idx) > 0) {
-        author_names[firstname_lastname_idx] <- paste0(
-          "fix_firstname_lastname_with_",
-          "lastname_semicolon_firstname"
-        )
-      }
 
-      replace_na_with_value(author_names, "add_author_lastname_semicolon_firstname")
+      author_fullname$value %>%
+      replace_indices_with_value(multiple_authors_idx,
+                                 "fix_multiple_authors_per_line",
+                                 dbg) %>%
+      replace_indices_with_value(firstname_lastname_idx,
+                                 paste0("fix_firstname_lastname_with_",
+                                 "lastname_semicolon_firstname"),
+                                 dbg) %>%
+      replace_indices_with_value(ends_with_semicolon_idx,
+                                 "fix_author_name_by_deleting_last_semicolon",
+                                 dbg) %>%
+      replace_na_with_value("add_author_lastname_semicolon_firstname",
+                            dbg)
     }
   )
 }

R/helper.R

@@ -143,9 +143,30 @@ tidy_df <- function(df, exclude_cols = "rec_number") {
 }
 
 
-replace_na_with_value <- function(vector, value) {
-  if (any(is.na(vector))) {
-    vector[is.na(vector)] <- value
+replace_na_with_value <- function(vector, value, dbg = TRUE) {
+
+  na_indices <- which(is.na(vector))
+
+  if (any(na_indices)) {
+    msg <- sprintf("Replacing %d NA values with %s",
+                   length(na_indices),
+                   value)
+    kwb.utils::catAndRun(msg, expr = {vector[na_indices] <- value}, dbg = dbg)
+  }
+
+  vector
+}
+
+
+replace_indices_with_value <- function(vector, indices, value, dbg = TRUE) {
+
+  n_indices <- length(indices)
+
+  if (n_indices > 0) {
+    msg <- sprintf("Replacing %d values with %s",
+                   n_indices,
+                   value)
+    kwb.utils::catAndRun(msg, expr = {vector[indices] <- value}, dbg = dbg)
   }
 
   vector