If decoding fails, fall back to using the original text. Addresses #1226

stanfordnlp · Dec 3, 2021 · 20fe1e9 · 20fe1e9
1 parent aac9202
commit 20fe1e9
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/src/edu/stanford/nlp/pipeline/StanfordCoreNLPServer.java b/src/edu/stanford/nlp/pipeline/StanfordCoreNLPServer.java
@@ -327,7 +327,11 @@ private Annotation getDocument(Properties props, HttpExchange httpExchange) thro
 
         String text = IOUtils.slurpReader(IOUtils.encodedInputStreamReader(httpExchange.getRequestBody(), encoding));
         if (contentType.equals(URL_ENCODED)) {
-          text = URLDecoder.decode(text, encoding);
+          try {
+            text = URLDecoder.decode(text, encoding);
+          } catch (IllegalArgumentException e) {
+            // ignore decoding errors so that libraries which don't specify a content type might not fail
+          }
         }
         // We use to trim. But now we don't. It seems like doing that is illegitimate. text = text.trim();