fix: improve reporting of invalid URL host parts

- fix #1034: make the message more genric ("Couldn't parse host…")
- fix #1079: don't report underscores used in the URL host part
- expand test `testValidateXHTMLUrlChecksInvalid`

src/main/java/com/adobe/epubcheck/ops/OPSHandler.java

@@ -22,8 +22,10 @@
 
 package com.adobe.epubcheck.ops;
 
+import java.net.MalformedURLException;
 import java.net.URI;
 import java.net.URISyntaxException;
+import java.net.URL;
 import java.util.Locale;
 import java.util.Stack;
 
@@ -226,15 +228,22 @@ else if (".".equals(href))
     {
       report.info(path, FeatureEnum.REFERENCE, href);
 
-      /*
-       * #708 report invalid HTTP/HTTPS URLs
-       * uri.scheme may be correct, but missing a : or a / from the //
-       * leads to uri.getHost() == null
-       */
+      // Report if the host part couldn't be parsed correctly
+      // (either due to missing slashes (issue #708) or invalid characters (issue #1034)
       if (uri.getHost() == null)
       {
-        int missingSlashes = uri.getSchemeSpecificPart().startsWith("/") ? 1 : 2;
-        report.message(MessageId.RSC_023, parser.getLocation(), uri, missingSlashes, uri.getScheme());
+        try
+        {
+          // if the URL contains underscore characters, try reparsing it without them,
+          // as underscores are accepted by browsers in the host part (even if it's disallowed)
+          // see issue #1079 
+          if (!href.contains("_") || new URI(href.replace('_', 'x')).getHost() == null) {
+            report.message(MessageId.RSC_023, parser.getLocation(), uri);
+          }
+        } catch (URISyntaxException ignored)
+        {
+          // ignored (well-formedness errors are caught earlier)
+        }
       }
     }
 

src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties

@@ -326,7 +326,7 @@ RSC_019=EPUBs with Multiple Renditions should contain a META-INF/metadata.xml fi
 RSC_020='%1$s' is not a valid URI.
 RSC_021=A Search Key Map Document must point to Content Documents ('%1$s' was not found in the spine).
 RSC_022=Cannot check image details (requires Java version 7 or higher).
-RSC_023=The URL '%1$s' is missing %2$d slash(es) '/' after the protocol '%3$s:'
+RSC_023=Couldn't parse host of URL '%1$s' (probably due to disallowed characters or missing slashes after the protocol)
 
 #Scripting
 SCP_001=Use of Javascript eval() function in EPUB scripts is a security risk.

src/test/java/com/adobe/epubcheck/ops/OPSCheckerTest.java

@@ -230,7 +230,7 @@ public void testValidateXHTMLLINKInvalid()
   public void testValidateXHTMLUrlChecksInvalid()
   {
     Collections.addAll(expectedErrors, MessageId.RSC_020);
-    Collections.addAll(expectedWarnings, MessageId.HTM_025, MessageId.RSC_023, MessageId.RSC_023);
+    Collections.addAll(expectedWarnings, MessageId.HTM_025, MessageId.RSC_023, MessageId.RSC_023, MessageId.RSC_023);
     testValidateDocument("xhtml/invalid/url-checks_issue-708.xhtml", "application/xhtml+xml",
         EPUBVersion.VERSION_3);
   }

src/test/resources/30/single/xhtml/invalid/url-checks_issue-708.xhtml

@@ -11,7 +11,9 @@
 			<a href="httpf://www.youtube.com/watch?v=xxxxxxxxxxx">Unsupported URI scheme (HTM-025)</a>
 			<a href="https:/www.youtube.com/watch?v=xxxxxxxxxxx">URL is missing slashes after protocol (RSC-023)</a>
 			<a href="https:www.youtube.com/watch?v=xxxxxxxxxxx">URL is missing slashes after protocol (RSC-023)</a>
-
+			<a href="https://w,w.example.com/watch?v=xxxxxxxxxxx">Host contains an invalid character (RSC-023)</a>
+
+			<a href="https://w_w.example.com">Underscore in hosts are accepted in most browsers</a>
 			<a href="https://www.youtube.com/watch?v=xxxxxxxxxxx">Valid URI</a>
 			<a href="https://youtube.com/watch?v=xxxxxxxxxxx">Valid URI</a>
 			<a href="https://youtube.com/watch?v=xxxxxx%20xxxx">Valid URI</a>