adjust regex in get-key-values.R since there might be empty values or…

… new-line characters in OSM values

Fix #250

R/get-key-values.R

@@ -247,7 +247,7 @@ get_keys = function(text, values = FALSE, which_keys = NULL) {
     # 1. Follow ^" or ," (where ^ denotes the start of a line)
     # and
     # 2. Precede the character "=>" (i.e. the delimiter)
-    pattern = '(?<=^\\"|,\\").+?(?=\\"=>\\")',
+    pattern = '(?<=^\\"|\\",\\").+?(?=\\"=>\\")',
     text = text,
     perl = TRUE
   )
@@ -266,7 +266,7 @@ get_keys = function(text, values = FALSE, which_keys = NULL) {
   # 4. Otherwise, we need to extract the values. I will use a regex that is
   # analogous to the previous query (inverting the lookahead and lookbehind)
   regexp_values = gregexpr(
-    pattern = '(?<=(\\"=>\\")).+?(?=\\"$|\\",)',
+    pattern = '(?<=(\\"=>\\")).*?(?=\\"$|\\",\\"|,\\")',
     text = text,
     perl = TRUE
   )

tests/testthat/test-get-keys.R

@@ -56,6 +56,14 @@ test_that("get_keys (values): more complicated examples", {
     expected = list(C = c("D", "E")),
     ignore_attr = TRUE
   )
+
+  # there might be empty values or newlines, see
+  # https://github.com/ropensci/osmextract/issues/250
+  expect_identical(
+    object = unclass(get_keys('"A"=>"\n","B"=>"C","D"=>""', values = TRUE)),
+    expected = list(A = "", B = "C", D = ""),
+    ignore_attr = TRUE
+  )
 })
 
 test_that("oe_get_keys: simplest examples work", {