feat: check that overlays audio file URLs have no fragments

rdeltour · rdeltour · commit fdabe6646a74 · 2022-11-27T21:20:15.000+01:00
This is reported as `MED-014` (repurposed). Fixes #1251
diff --git a/src/main/java/com/adobe/epubcheck/overlay/OverlayHandler.java b/src/main/java/com/adobe/epubcheck/overlay/OverlayHandler.java
@@ -11,7 +11,6 @@
 import com.adobe.epubcheck.messages.MessageId;
 import com.adobe.epubcheck.opf.OPFChecker30;
 import com.adobe.epubcheck.opf.ValidationContext;
-import com.adobe.epubcheck.opf.XRefChecker;
 import com.adobe.epubcheck.opf.XRefChecker.Type;
 import com.adobe.epubcheck.util.EpubConstants;
 import com.adobe.epubcheck.vocab.AggregateVocab;
@@ -163,8 +162,17 @@ private void processAudioSrc()
   {
 
     URL url = checkURL(currentElement().getAttribute("src"));
+
+    // check that the URL has no fragment
+    if (url.fragment() != null)
+    {
+      report.message(MessageId.MED_014, location(), url.fragment());
+      url = URLUtils.docURL(url);
+    }
+
     if (url != null && context.xrefChecker.isPresent())
     {
+
       // check that the audio type is a core media type resource
       String mimeType = context.xrefChecker.get().getMimeType(url);
       if (mimeType != null && !OPFChecker30.isBlessedAudioType(mimeType))
diff --git a/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties b/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
@@ -142,7 +142,7 @@ MED_010=EPUB Content Documents referenced from a Media Overlay must specify the
 MED_011=EPUB Content Document referenced from multiple Media Overlay Documents.
 MED_012=The "media-overlay" attribute does not match the ID of the Media Overlay that refers to this document.
 MED_013=Media Overlay Document referenced from the "media-overlay" attribute does not contain a reference to this Content Document.
-MED_014=A non-empty fragment identifier is required.
+MED_014=Media overlay audio file URLs must not have a fragment, but found '#%1$s'. 
 MED_015=Media overlay text references must be in reading order. Text target "%1$s" is before the previous link target in %2$s order.
 MED_016=Media Overlays total duration should be the sum of the durations of all Media Overlays documents.
 MED_017=URL fragment should indicate an element ID, but found '#%1$s'.
diff --git a/src/test/resources/epub3/09-media-overlays/files/audio-src-fragment-error.smil b/src/test/resources/epub3/09-media-overlays/files/audio-src-fragment-error.smil
@@ -0,0 +1,13 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<smil xmlns="https://www.w3.org/ns/SMIL" xmlns:epub="https://www.idpf.org/2007/ops" version="3.0">
+    <body>
+        <par id="par1">
+            <text src="chapter1.xhtml#text1"/>
+            <audio src="chapter1.mp3" clipBegin="0s" clipEnd="10s"/>
+        </par>
+        <par id="par2">
+            <text src="chapter1.xhtml#text2"/>
+            <audio src="chapter1.mp3#fragment" clipBegin="10s" clipEnd="20s"/>
+        </par>
+    </body>
+</smil>
diff --git a/src/test/resources/epub3/09-media-overlays/media-overlays.feature b/src/test/resources/epub3/09-media-overlays/media-overlays.feature
@@ -92,6 +92,12 @@ Feature: EPUB 3 — Media Overlays
   
   #### 9.2.2.8 The `audio` element
 
+  @spec @xref:sec-smil-audio-elem
+  Scenario: Report an audio file URL with a fragment
+    When checking document 'audio-src-fragment-error.smil'
+    Then error MED-014 is reported
+    And no other errors or warnings are reported
+
   @spec @xref:sec-smil-audio-elem
   Scenario: Allow clock values with the full clock syntax (`hh:mm:ss.milli`)
     When checking document 'clock-value-full-syntax-valid.smil'
