From d2d1135260df4e5b648c0e03646705a3d05a1db1 Mon Sep 17 00:00:00 2001
From: Alex Osborne <aosborne@nla.gov.au>
Date: Wed, 6 Dec 2023 15:02:43 +0900
Subject: [PATCH] CdxRequestEncoder: Catch IllegalArgumentException when
 request has invalid content-type header

---
 .../jwarc/cdx/CdxRequestEncoder.java            |  7 ++++++-
 .../jwarc/cdx/CdxRequestEncoderTest.java        | 17 ++++++++++++++---
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/src/org/netpreserve/jwarc/cdx/CdxRequestEncoder.java b/src/org/netpreserve/jwarc/cdx/CdxRequestEncoder.java
index db57870..5d0c735 100644
--- a/src/org/netpreserve/jwarc/cdx/CdxRequestEncoder.java
+++ b/src/org/netpreserve/jwarc/cdx/CdxRequestEncoder.java
@@ -28,7 +28,12 @@ public static String encode(HttpRequest httpRequest) throws IOException {
         out.append("__wb_method=");
         out.append(httpRequest.method());
         int maxLength = out.length() + 1 + QUERY_STRING_LIMIT;
-        MediaType baseContentType = httpRequest.contentType().base();
+        MediaType baseContentType;
+        try {
+            baseContentType = httpRequest.contentType().base();
+        } catch (IllegalArgumentException e) {
+            baseContentType = MediaType.OCTET_STREAM;
+        }
         InputStream stream = new BufferedInputStream(httpRequest.body().stream(), BUFFER_SIZE);
         if (baseContentType.equals(MediaType.WWW_FORM_URLENCODED)) {
             encodeFormBody(stream, out);
diff --git a/test/org/netpreserve/jwarc/cdx/CdxRequestEncoderTest.java b/test/org/netpreserve/jwarc/cdx/CdxRequestEncoderTest.java
index a427bb5..6972aad 100644
--- a/test/org/netpreserve/jwarc/cdx/CdxRequestEncoderTest.java
+++ b/test/org/netpreserve/jwarc/cdx/CdxRequestEncoderTest.java
@@ -57,6 +57,8 @@ public class CdxRequestEncoderTest {
                     String.join(" ", Collections.nCopies(1000, "this is very long")) + "\"}", "x=1"),
             new Case("__wb_method=POST&x=%2",
                     MediaType.WWW_FORM_URLENCODED, "x=%2"),
+            new Case("__wb_method=POST&__wb_post_data=eD0lMg==",
+                    "application/x-www-form-urlencoded, text/plain", "x=%2"),
             generateLargeCase()
     };
 
@@ -74,6 +76,7 @@ private static Case generateLargeCase() {
     public static class Case {
         final String expected;
         final MediaType requestType;
+        String requestTypeString;
         final byte[] requestBody;
         final String queryString;
 
@@ -96,14 +99,22 @@ public Case(String expected, MediaType requestType, byte[] requestBody, String q
             this.queryString = queryString;
         }
 
+        public Case(String expected, String requestType, String requestBody) {
+            this(expected, MediaType.parse("null/null"), requestBody);
+            requestTypeString = requestType;
+        }
+
         public HttpRequest request() {
             String target = "/foo";
             if (queryString != null) {
                 target += "?" + queryString;
             }
-            return new HttpRequest.Builder("POST", target)
-                    .body(requestType, requestBody)
-                    .build();
+            HttpRequest.Builder builder = new HttpRequest.Builder("POST", target)
+                    .body(requestType, requestBody);
+            if (requestTypeString != null) {
+                builder.setHeader("Content-Type", requestTypeString);
+            }
+            return builder.build();
         }
     }