Handle possibly invalid JSON (fixes #41)

TIBHannover · Mar 20, 2018 · cae151d · cae151d
1 parent 949a675
commit cae151d
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 0 deletions.
diff --git a/R/retrieve_data.R b/R/retrieve_data.R
@@ -91,10 +91,17 @@ retrieve_data <- function(searchTerm,
 #'
 #' @return A serialised JSON string.
 download <- function(URL, userpwd = paste(get_credentials(), collapse = ":")) {
+  gsub(
+    pattern = "[[:space:]]+",
+    replacement = " ",
+    perl = TRUE,
+    # Prevent "lexical error: invalid character inside string."
+    # https://github.com/jeroen/jsonlite/issues/47
   RCurl::getURL(
     URL,
     userpwd = userpwd,
     httpauth = 1L
+    )
   )
 }
 

diff --git a/tests/testthat/test-retrieve_data.R b/tests/testthat/test-retrieve_data.R
@@ -64,3 +64,15 @@ test_that("extracting a single field from a taxon-wide search works", {
                  )
                )))
 })
+
+test_that("normalising invalid JSON newlines works", {
+  is.list(retrieve_data(searchTerm = "Bacillus cytotoxicus",
+                        force_taxon_download = TRUE))
+  # https://bacdive.dsmz.de/api/bacdive/bacdive_id/1323/?format=json
+  # contains "sample_type": "Vegetable puree,severe food poisoning\noutbreak in France"
+
+  is.list(retrieve_data(searchTerm = "Bacillus halotolerans",
+                        force_taxon_download = TRUE))
+  # https://bacdive.dsmz.de/api/bacdive/bacdive_id/1847/?format=json
+  # contains "medium_composition": "Name: ISP 2 / Yeast Malt Agar (5265); 5265\r\nComposition
+})