From 8eb849250bae7b595bbd0a2de95ef46ab39efb0c Mon Sep 17 00:00:00 2001
From: Romain Deltour <rdeltour@gmail.com>
Date: Mon, 21 Nov 2022 00:17:21 +0100
Subject: [PATCH] feat: no longer check collection `role`

This commit suppresses and removes the code of the following checks:
- `OPF-068`: checked that a role token were one the defined roles in
  the collection role registry
  (see https://idpf.github.io/epub-registries/roles/)
- `OPF-069`: checked that a custom role URL did not contain the string
  "idpf.org".

These restrictions were removed in EPUB 3.3.

The attribute value is still tokenized, and any URL-looking string
  (in the shape `scheme:something`) is checked to be a valid URL string.

Fix #1350.
---
 .../epubcheck/messages/DefaultSeverities.java |  4 +--
 .../com/adobe/epubcheck/opf/OPFHandler30.java | 26 +++----------------
 .../messages/MessageBundle.properties         |  6 ++---
 .../files/collection-role-url-valid.opf       |  8 ++++++
 4 files changed, 16 insertions(+), 28 deletions(-)

diff --git a/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java b/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java
index 613b8c5b4..aaa61f901 100644
--- a/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java
+++ b/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java
@@ -257,8 +257,8 @@ private void initialize()
     severities.put(MessageId.OPF_065, Severity.ERROR);
     severities.put(MessageId.OPF_066, Severity.ERROR);
     severities.put(MessageId.OPF_067, Severity.ERROR);
-    severities.put(MessageId.OPF_068, Severity.ERROR);
-    severities.put(MessageId.OPF_069, Severity.ERROR);
+    severities.put(MessageId.OPF_068, Severity.SUPPRESSED);
+    severities.put(MessageId.OPF_069, Severity.SUPPRESSED);
     severities.put(MessageId.OPF_070, Severity.WARNING);
     severities.put(MessageId.OPF_071, Severity.ERROR);
     severities.put(MessageId.OPF_072, Severity.USAGE);
diff --git a/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java b/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java
index a6055fcd4..d5592ff44 100644
--- a/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java
+++ b/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java
@@ -407,36 +407,16 @@ private List<String> processCollectionRole(String roleAtt)
       if (URLUtils.isAbsoluteURLString(role))
       {
         // Role is an absolute IRI
-        // check that the host component doesn't contain 'idpf.org'
         try
         {
-          URL url = URL.parse(role);
-          if (url.authority() != null && url.authority().contains("idpf.org"))
-          {
-            report.message(MessageId.OPF_069, location(), role);
-          }
-          else
-          {
-            rolesBuilder.add(role);
-          }
+          URL.parse(role);
         } catch (GalimatiasParseException e)
         {
           report.message(MessageId.OPF_070, location(), role);
+          break;
         }
       }
-      else
-      {
-        // Role is a NMTOKEN
-        // Check that it's in the reserved role list
-        if (ResourceCollection.Roles.fromString(role).isPresent())
-        {
-          rolesBuilder.add(role);
-        }
-        else
-        {
-          report.message(MessageId.OPF_068, location(), role);
-        }
-      }
+      rolesBuilder.add(role);
     }
     return rolesBuilder.build();
   }
diff --git a/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties b/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
index 462b671af..638f73a0a 100644
--- a/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
+++ b/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
@@ -247,9 +247,9 @@ OPF_064=OPF declares type "%1$s", validating using profile "%2$s".
 OPF_065=Invalid metadata declaration, probably due to a cycle in "refines" metadata.
 OPF_066=Missing "dc:source" or "source-of" pagination metadata. The pagination source must be identified using the "dc:source" and "source-of" properties when the content includes page break markers.
 OPF_067=The resource "%1$s" must not be listed both as  a "link" element in the package metadata and as a manifest item.
-OPF_068=Unknown collection role "%1$s".
-OPF_069=Custom collection role URI "%1$s" must not include the string "idpf.org" in its host component.
-OPF_070=Custom collection role "%1$s" is an invalid URI.
+OPF_068=
+OPF_069=
+OPF_070=Custom collection role "%1$s" is an invalid URL.
 OPF_071=Index collections must only contain resources pointing to XHTML Content Documents.
 OPF_072=Metadata element "%1$s" is empty.
 OPF_073=External identifiers must not appear in the document type declaration.
diff --git a/src/test/resources/epub3/05-package-document/files/collection-role-url-valid.opf b/src/test/resources/epub3/05-package-document/files/collection-role-url-valid.opf
index ad7dad5eb..9d57e9079 100644
--- a/src/test/resources/epub3/05-package-document/files/collection-role-url-valid.opf
+++ b/src/test/resources/epub3/05-package-document/files/collection-role-url-valid.opf
@@ -16,4 +16,12 @@
     <collection role="http://example.org/">
         <link href="http://example.org/resource"/>
     </collection>
+    <collection role="http://idpf.org.example.org/">
+        <!-- using 'idpf.org' in the URL authority was forbidden in EPUB 3.2 -->
+        <link href="http://example.org/resource"/>
+    </collection>
+    <collection role="unknown-token">
+        <!-- using roles not in the role vocab was forbidden in EPUB 3.2 -->
+        <link href="http://example.org/resource"/>
+    </collection>
 </package>