feat: new requirements of package link 'media-type' attribute

rdeltour · rdeltour · commit 3199b8146b9b · 2022-11-27T21:20:16.000+01:00
This commit implements the new requirements of the `media-type` attribute of the package document `link` element. - new `OPF-093` error is reported when the `media-type` attribute is missing on linked resources located inside the EPUB container - new `OPF-094` error is reported when the `media-type` attribute is missing on a link where the `rel` keyword requires it (this was previously checked with Schematron) - new `OPF-095` error is reported when the media type of a "voicing" linked resource is not an audio type (this was previously checked with Schematron) These new checks are implemented in Java, and the old schematron logic is removed. Fixes #1307
diff --git a/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java b/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java
@@ -283,6 +283,9 @@ private void initialize()
     severities.put(MessageId.OPF_090, Severity.USAGE);
     severities.put(MessageId.OPF_091, Severity.ERROR);
     severities.put(MessageId.OPF_092, Severity.ERROR);
+    severities.put(MessageId.OPF_093, Severity.ERROR);
+    severities.put(MessageId.OPF_094, Severity.ERROR);
+    severities.put(MessageId.OPF_095, Severity.ERROR);
 
     // PKG
     severities.put(MessageId.PKG_001, Severity.WARNING);
diff --git a/src/main/java/com/adobe/epubcheck/messages/MessageId.java b/src/main/java/com/adobe/epubcheck/messages/MessageId.java
@@ -277,6 +277,9 @@ public enum MessageId implements Comparable<MessageId>
   OPF_090("OPF-090"),
   OPF_091("OPF-091"),
   OPF_092("OPF-092"),
+  OPF_093("OPF-093"),
+  OPF_094("OPF-094"),
+  OPF_095("OPF-095"),
 
   // Messages relating to the entire package
   PKG_001("PKG-001"),
diff --git a/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java b/src/main/java/com/adobe/epubcheck/opf/OPFHandler30.java
@@ -336,7 +336,7 @@ else if (EpubConstants.DCElements.equals(e.getNamespace()))
    * Document. Must be called after the parsing.
    * 
    * @return the metadata for the Rendition represented by the current Package
-   *         Document
+   *           Document
    */
   public MetadataSet getMetadata()
   {
@@ -350,7 +350,7 @@ public MetadataSet getMetadata()
    * called after the parsing.
    * 
    * @return the linked resources for the Rendition represented by the current
-   *         Package Document
+   *           Package Document
    */
   public LinkedResources getLinkedResources()
   {
@@ -363,7 +363,7 @@ public LinkedResources getLinkedResources()
    * the parsing.
    * 
    * @return the linked resources for the Rendition represented by the current
-   *         Package Document
+   *           Package Document
    */
   public ResourceCollections getCollections()
   {
@@ -446,28 +446,67 @@ private void processLink()
 
     String href = e.getAttribute("href");
     if (href != null)
-    { // check by schema
+    { // href presence is checked by schema
 
+      // check the 'href' URL
       URL url = checkURL(href);
-      if (context.isRemote(url)) {
+      if (context.isRemote(url))
+      {
         report.info(path, FeatureEnum.REFERENCE, href);
       }
-
       if (context.xrefChecker.isPresent())
       {
         context.xrefChecker.get().registerReference(url, Type.LINK, location());
       }
 
+      // check the 'rel' attribute
+      String rel = e.getAttribute("rel");
+      Set<Property> relSet = processLinkRel(rel);
+      Set<LINKREL_PROPERTIES> relEnum = Property.filter(relSet, LINKREL_PROPERTIES.class);
+
+      // check the 'media-type' attribute
+      String mediatype = e.getAttribute("media-type");
+      if (mediatype == null)
+      {
+        // media-type is required for in-container URLs
+        // NOTE: as legacy EPUB 3.2 collections made heavy use
+        // of local links with no media type, we only check this
+        // for metadata links, which may be a violation of EPUB.
+        if (!context.isRemote(url) && !metadataBuilders.isEmpty())
+        {
+          if (linkedResourcesBuilders.size() == 1)
+            report.message(MessageId.OPF_093, location());
+        }
+        // media-type is required by some keywords
+        else if (relEnum.stream().anyMatch(
+            keyword -> keyword == LINKREL_PROPERTIES.RECORD
+                || keyword == LINKREL_PROPERTIES.VOICING))
+        {
+          report.message(MessageId.OPF_094, location(), rel);
+        }
+      }
+      else
+      {
+        // 'voicing' links require an audio media type
+        if (relEnum.contains(LINKREL_PROPERTIES.VOICING) && !OPFChecker30.isAudioType(mediatype))
+        {
+          report.message(MessageId.OPF_095, location(), mediatype);
+        }
+      }
+
+      // check the 'properties' attribute
+      processLinkProperties(e.getAttribute("properties"));
+
+      // build the data model
       if (!linkedResourcesBuilders.isEmpty())
       {
-        processLinkProperties(e.getAttribute("properties"));
         LinkedResource resource = new LinkedResource.Builder(url).id(e.getAttribute("id"))
-            .rel(processLinkRel(e.getAttribute("rel"))).mimetype(e.getAttribute("media-type"))
-            .refines(e.getAttribute("refines")).build();
+            .rel(relSet).mimetype(mediatype).refines(e.getAttribute("refines")).build();
         linkedResourcesBuilders.peekFirst().add(resource);
       }
     }
 
+    // check hreflang attribute
     String hreflang = e.getAttribute("hreflang");
     if (hreflang != null && !hreflang.isEmpty())
     {
diff --git a/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties b/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
@@ -289,6 +289,9 @@ OPF_089=The "alternate" link rel keyword cannot be paired with other keywords.
 OPF_090=It is encouraged to use MIME media type "%1$s" instead of "%2$s".
 OPF_091=The item href URL must not have a fragment identifier.
 OPF_092=Language tag "%1$s" is not well-formed: %2$s
+OPF_093=The "media-type" attribute is required for linked resources located in the EPUB container
+OPF_094=The "media-type" attribute is required for "%1$s" links.
+OPF_095=The "media-type" attribute of "voicing" links must be an audio MIME type, but found "%1$s". 
 
 #Package
 PKG_001=Validating the EPUB against version %1$s but detected version %2$s.
diff --git a/src/main/resources/com/adobe/epubcheck/schema/30/package-30.sch b/src/main/resources/com/adobe/epubcheck/schema/30/package-30.sch
@@ -34,16 +34,15 @@
     
     <pattern id="opf.link.record">
         <rule context="opf:link[tokenize(@rel,'\s+')='record']">
-            <assert test="exists(@media-type)">The type of "record" references must be identifiable
-                from the link element’s "media-type" attribute.</assert>
+            <!--<assert test="exists(@media-type)">**checked in java**</assert>-->
             <assert test="empty(@refines)">"record" links only applies to the Publication (must not
                 have a "refines" attribute).</assert>
         </rule>
     </pattern>
     
     <pattern id="opf.link.voicing">
         <rule context="opf:link[tokenize(@rel,'\s+')='voicing']">
-            <assert test="starts-with(normalize-space(@media-type),'audio/')">"voicing" links must have a "media-type" attribute identifying an audio MIME type.</assert>
+            <!--<assert test="starts-with(normalize-space(@media-type),'audio/')">**checked in java**</assert>-->
             <assert test="exists(@refines)">"voicing" links must have a "refines" attribute.</assert>
         </rule>
     </pattern>
diff --git a/src/test/resources/epub3/03-resources/files/file-url-in-package-document-error.opf b/src/test/resources/epub3/03-resources/files/file-url-in-package-document-error.opf
@@ -5,7 +5,7 @@
   <dc:language>en</dc:language>
   <dc:identifier id="q">NOID</dc:identifier>
   <meta property="dcterms:modified">2017-06-14T00:00:01Z</meta>
-  <link href="file:example" rel="acquire"/>
+  <link href="file:example" rel="acquire" media-type="text/html"/>
 </metadata>
 <manifest>
   <item id="content_001"  href="content_001.xhtml" media-type="application/xhtml+xml"/>
diff --git a/src/test/resources/epub3/05-package-document/files/link-to-publication-resource-error.opf b/src/test/resources/epub3/05-package-document/files/link-to-publication-resource-error.opf
@@ -8,7 +8,7 @@
         <dc:identifier id="uid">NOID</dc:identifier>
         <meta property="dcterms:modified">2019-01-01T12:00:00Z</meta>
         <!--Linked resources MUST NOT be listed in the manifest.-->
-        <link rel="example:property" href="contents.xhtml#ch1"/>
+        <link rel="example:property" href="contents.xhtml#ch1" media-type="application/xhtml+xml"/>
     </metadata>
     <manifest>
         <item id="t001" href="contents.xhtml" properties="nav" media-type="application/xhtml+xml"/>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/book.xml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/book.xml
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/content_001.xhtml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/content_001.xhtml
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/nav.xhtml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/nav.xhtml
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/package.opf b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/EPUB/package.opf
@@ -0,0 +1,18 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<package xmlns="https://www.idpf.org/2007/opf" version="3.0" xml:lang="en" unique-identifier="q"
+  prefix="test: https://example.org">
+  <metadata xmlns:dc="https://purl.org/dc/elements/1.1/">
+    <dc:title id="title">Minimal EPUB 3.0</dc:title>
+    <dc:language>en</dc:language>
+    <dc:identifier id="q">NOID</dc:identifier>
+    <meta property="dcterms:modified">2017-06-14T00:00:01Z</meta>
+    <link rel="test:link" href="book.xml"/>
+  </metadata>
+  <manifest>
+    <item id="content_001" href="content_001.xhtml" media-type="application/xhtml+xml"/>
+    <item id="nav" href="nav.xhtml" media-type="application/xhtml+xml" properties="nav"/>
+  </manifest>
+  <spine>
+    <itemref idref="content_001"/>
+  </spine>
+</package>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/META-INF/container.xml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/META-INF/container.xml
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/mimetype b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-local-error/mimetype
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/content_001.xhtml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/content_001.xhtml
@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html xmlns="https://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
+	<head>
+		<meta charset="utf-8" />
+		<title>Minimal EPUB</title>
+	</head>
+	<body>
+		<h1>Loomings</h1>
+		<p>Call me Ishmael.</p>
+	</body>
+</html>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/nav.xhtml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/nav.xhtml
@@ -0,0 +1,14 @@
+<!DOCTYPE html>
+<html xmlns="https://www.w3.org/1999/xhtml" xmlns:epub="https://www.idpf.org/2007/ops" xml:lang="en" lang="en">
+	<head>
+		<meta charset="utf-8"/>
+		<title>Minimal Nav</title>
+	</head>
+	<body>
+		<nav epub:type="toc">
+			<ol>
+				<li><a href="content_001.xhtml">content 001</a></li>
+			</ol>
+		</nav>
+	</body>
+</html>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/package.opf b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/EPUB/package.opf
@@ -0,0 +1,18 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<package xmlns="https://www.idpf.org/2007/opf" version="3.0" xml:lang="en" unique-identifier="q"
+  prefix="test: https://example.org">
+  <metadata xmlns:dc="https://purl.org/dc/elements/1.1/">
+    <dc:title id="title">Minimal EPUB 3.0</dc:title>
+    <dc:language>en</dc:language>
+    <dc:identifier id="q">NOID</dc:identifier>
+    <meta property="dcterms:modified">2017-06-14T00:00:01Z</meta>
+    <link rel="test:link" href="https://example.org/"/>
+  </metadata>
+  <manifest>
+    <item id="content_001" href="content_001.xhtml" media-type="application/xhtml+xml"/>
+    <item id="nav" href="nav.xhtml" media-type="application/xhtml+xml" properties="nav"/>
+  </manifest>
+  <spine>
+    <itemref idref="content_001"/>
+  </spine>
+</package>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/META-INF/container.xml b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/META-INF/container.xml
@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<container xmlns="urn:oasis:names:tc:opendocument:xmlns:container" version="1.0">
+	<rootfiles>
+		<rootfile full-path="EPUB/package.opf" media-type="application/oebps-package+xml"/>
+	</rootfiles>
+</container>
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/mimetype b/src/test/resources/epub3/05-package-document/files/package-link-media-type-missing-remote-valid/mimetype
@@ -0,0 +1 @@
+application/epub+zip
diff --git a/src/test/resources/epub3/05-package-document/files/package-link-missing-media-type-error/EPUB/package.opf b/src/test/resources/epub3/05-package-document/files/package-link-missing-media-type-error/EPUB/package.opf
diff --git a/src/test/resources/epub3/05-package-document/package-document.feature b/src/test/resources/epub3/05-package-document/package-document.feature
@@ -334,12 +334,15 @@ Feature: EPUB 3 — Package document
     And no other errors or warnings are reported
 
   @spec @xref:sec-link-elem
-  Scenario: A link to a local resource must declare a media type  
-    When checking file 'package-link-missing-media-type-error'
-    # Then error RSC-005 is reported
-    # And the message contains 'missing required attribute "media-type"'
-    # And no other errors or warnings are reported
-    Then no other errors or warnings are reported
+  Scenario: Report a missing 'media-type' attribute on links to container resources  
+    When checking file 'package-link-media-type-missing-local-error'
+    Then error OPF-093 is reported
+    And no other errors or warnings are reported
+
+  @spec @xref:sec-link-elem
+  Scenario: Allow a missing 'media-type' attribute on links to remote resources  
+    When checking file 'package-link-media-type-missing-remote-valid'
+    Then no errors or warnings are reported
 
   @spec @xref:sec-link-elem
   Scenario: the link 'rel' attribute can have multiple properties
diff --git a/src/test/resources/epub3/D-vocabularies/files/link-rel-record-mediatype-missing-error.opf b/src/test/resources/epub3/D-vocabularies/files/link-rel-record-mediatype-missing-error.opf
@@ -6,7 +6,7 @@
         <dc:language>en</dc:language>
         <dc:identifier id="uid">NOID</dc:identifier>
         <meta property="dcterms:modified">2019-01-01T12:00:00Z</meta>
-        <link rel="record" href="record.atom"/>
+        <link rel="record" href="https://example.org/record.atom"/>
     </metadata>
     <manifest>
         <item id="t001" href="contents.xhtml" properties="nav" media-type="application/xhtml+xml"/>
diff --git a/src/test/resources/epub3/D-vocabularies/files/link-rel-voicing-mediatype-missing-error.opf b/src/test/resources/epub3/D-vocabularies/files/link-rel-voicing-mediatype-missing-error.opf
@@ -6,7 +6,7 @@
         <dc:language>en</dc:language>
         <dc:identifier id="uid">NOID</dc:identifier>
         <meta property="dcterms:modified">2019-01-01T12:00:00Z</meta>
-        <link rel="voicing" refines="#title" href="title.mp3"/>
+        <link rel="voicing" refines="#title" href="https://example.org/title.mp3"/>
     </metadata>
     <manifest>
         <item id="t001" href="contents.xhtml" properties="nav" media-type="application/xhtml+xml"/>
diff --git a/src/test/resources/epub3/D-vocabularies/metadata-link.feature b/src/test/resources/epub3/D-vocabularies/metadata-link.feature
@@ -33,7 +33,7 @@ Feature: EPUB 3 — Vocabularies — Metadata link vocabulary
     When checking file 'link-rel-alternate-with-other-keyword-error.opf'
     Then error OPF-089 is reported
     And no other errors or warnings are reported
-
+  
 
   #### D.4.1.3, D.4.1.4, D.4.1.5, D.4.1.9 *-record
   
@@ -44,6 +44,8 @@ Feature: EPUB 3 — Vocabularies — Metadata link vocabulary
       | OPF-086 | "mods-record" is deprecated      |
       | OPF-086 | "onix-record" is deprecated      |
       | OPF-086 | "xmp-record" is deprecated       |
+    And error OPF-093 is reported 4 times
+      # note: 'media-type' is now required, even on deprecated properties
     And no other errors or warnings are reported
 
     
@@ -62,10 +64,9 @@ Feature: EPUB 3 — Vocabularies — Metadata link vocabulary
     Then no errors or warnings are reported
   
   @spec @xref:sec-record
-  Scenario: a 'record' link must have a 'media-type' attribute 
+  Scenario: a 'record' link must have a 'media-type' attribute even when remote 
     When checking file 'link-rel-record-mediatype-missing-error.opf'
-    Then error RSC-005 is reported
-    And the message contains "media-type"
+    Then error OPF-094 is reported
     And no other errors or warnings are reported
 
   Scenario: a 'record' link type can be further identified with a 'properties' attribute
@@ -95,17 +96,15 @@ Feature: EPUB 3 — Vocabularies — Metadata link vocabulary
     And no other errors or warnings are reported
 
   @spec @xref:sec-voicing
-  Scenario: a 'voicing' link must have a 'media-type' attribute
+  Scenario: a 'voicing' link must have a 'media-type' attribute even when remote
     When checking file 'link-rel-voicing-mediatype-missing-error.opf'
-    Then error RSC-005 is reported
-    And the message contains 'must have a "media-type" attribute'
+    Then error OPF-094 is reported
     And no other errors or warnings are reported
 
   @spec @xref:sec-voicing
   Scenario: a 'voicing' link resource must have an audio media type
     When checking file 'link-rel-voicing-mediatype-not-audio-error.opf'
-    Then error RSC-005 is reported
-    And the message contains 'must have a "media-type" attribute identifying an audio MIME type'
+    Then error OPF-095 is reported
     And no other errors or warnings are reported
 
 
@@ -115,4 +114,6 @@ Feature: EPUB 3 — Vocabularies — Metadata link vocabulary
     When checking file 'link-rel-xml-signature-deprecated-warning.opf'
     Then warning OPF-086 is reported
     And the message contains '"xml-signature" is deprecated'
+    And error OPF-093 is reported
+      # note: 'media-type' is now required, even on deprecated properties
     And no other errors or warnings are reported
