Merge pull request #905 from openpreserve/karenhanson-update-to-epub-…

…5-0-1

FIX: Bump ePub -> 1.3 in test audit.

jhove-bbt/scripts/create-1.29-target.sh

@@ -55,3 +55,6 @@ echo "TEST BASELINE: Creating baseline"
 # Simply copy baseline for now we're not making any changes
 echo " - copying ${baselineRoot} baseline to ${targetRoot}"
 cp -R "${baselineRoot}" "${targetRoot}"
+
+# Update release details for ePub module
+find "${targetRoot}" -type f -name "audit.jhove.xml" -exec sed -i 's/^   <module release="1.2">EPUB-ptc<\/module>$/   <module release="1.3">EPUB-ptc<\/module>/' {} \;

jhove-ext-modules/pom.xml

@@ -13,7 +13,7 @@
 
   <properties>
     <jwat.version>1.0.3</jwat.version>
-    <epubcheck.version>4.2.6</epubcheck.version>
+    <epubcheck.version>5.1.0</epubcheck.version>
   </properties>
 
   <build>

jhove-ext-modules/src/main/java/org/ithaka/portico/jhove/module/EpubModule.java

@@ -82,9 +82,9 @@ public class EpubModule extends ModuleBase {
     private static final String FORMATNAME = "EPUB";
 
     private static final String NAME = "EPUB-ptc";
-    private static final String RELEASE = "1.2";
-    private static final int[] DATE = { 2023, 03, 16 };
-    private static final String RIGHTS_YEAR = "2019";
+    private static final String RELEASE = "1.3";
+    private static final int[] DATE = { 2023, 06, 12 };
+    private static final String RIGHTS_YEAR = "2023";
     private static final String[] FORMAT = { FORMATNAME };
     private static final String COVERAGE = FORMATNAME;
     private static final String[] MIMETYPE = { EPUB_MEDIATYPE };
@@ -94,18 +94,18 @@ public class EpubModule extends ModuleBase {
     private static final String NOTE = "This module uses EPUBCheck for testing of EPUB files.";
 
     // EPUB agent information
-    private static final String EPUB_AGENTNAME = "International Digital Publishing Forum";
+    private static final String EPUB_AGENTNAME = "International Digital Publishing Forum and World Wide Web Consortium";
     private static final AgentType EPUB_AGENTTYPE = AgentType.STANDARD;
-    private static final String EPUB_AGENTADDRESS = "International Digital Publishing Forum (IDPF), "
-            + "113 Cherry Street, Suite 70-719, Seattle, WA 98104";
-    private static final String EPUB_AGENTWEBSITE = "http://idpf.org";
-    private static final String EPUB_AGENTEMAIL = "membership@idpf.org";
-    private static final String EPUB_AGENTPHONE = "+1-206-451-7250";
+    private static final String EPUB_AGENTADDRESS = "W3C Inc, "
+            + "401 Edgewater Place, Suite 600, Wakefield, MA 01880, USA";
+    private static final String EPUB_AGENTWEBSITE = "https://www.w3.org/";
+    private static final String EPUB_AGENTEMAIL = "membership@w3.org";
+    private static final String EPUB_AGENTPHONE = "+1-339-273-2711";
 
     // EPUB format doc information
     private static final String EPUB_FORMATDOCTITLE = FORMATNAME;
-    private static final String EPUB_FORMATDOCDATE = "2019-05-15";
-    private static final String EPUB_FORMATDOCURL = "http://www.idpf.org/epub/dir/";
+    private static final String EPUB_FORMATDOCDATE = "2023-05-25";
+    private static final String EPUB_FORMATDOCURL = "https://www.w3.org/TR/epub/";
 
     // Signatures
     private static final String EPUB_EXTENSION = ".epub";

jhove-ext-modules/src/main/java/org/ithaka/portico/jhove/module/epub/JhoveRepInfoReport.java

@@ -72,6 +72,8 @@ public class JhoveRepInfoReport extends MasterReport {
     protected static final String ISO_DATE_PATTERN = "yyyy-MM-dd'T'HH:mm:ss'Z'";
     protected static final String FALLBACK_FORMAT = "application/octet-stream";
 
+    protected static final String FRAGMENT_START = "#";
+
     public JhoveRepInfoReport(String ePubName) {
         this.setEpubFileName(PathUtil.removeWorkingDirectory(ePubName));
 
@@ -146,7 +148,8 @@ public void info(String resource, FeatureEnum feature, String value) {
             this.references.add(value);
             break;
         case RESOURCE:
-            this.resources.add(value);
+            String no_fragment = value.split(FRAGMENT_START)[0];
+            this.resources.add(no_fragment);
             break;
         case DC_LANGUAGE:
             this.language = value;

jhove-ext-modules/src/test/java/org/ithaka/portico/jhove/module/EpubModuleTest.java

@@ -74,7 +74,7 @@ public class EpubModuleTest {
     private static final String EPUB3_TITLE_ENCODING = "src/test/resources/epub/epub3-multiple-renditions.epub";
 
     private static final String EXPECTED_MEDIATYPE = "application/epub+zip";
-    private static final String EXPECTED_VERSION_3_2 = "3.2";
+    private static final String EXPECTED_VERSION_3_3 = "3.3";
     private static final String PNG_MIMETYPE = "image/png";
     private static final String XHTML_MIMETYPE = "application/xhtml+xml";
     private static final String NCX_MIMETYPE = "application/x-dtbncx+xml";
@@ -99,9 +99,11 @@ public void parseValidEpub3PropertiesTest() throws Exception {
         assertEquals(0, info.getMessage().size()); // no errors
         assertEquals("EPUB", info.getFormat());
         assertEquals(EXPECTED_MEDIATYPE, info.getMimeType());
-        assertEquals(EXPECTED_VERSION_3_2, info.getVersion());
+        assertEquals(EXPECTED_VERSION_3_3, info.getVersion());
         // these may change, so just check they aren't null
-        assertNotNull(info.getCreated());
+        // note: this property is missing in 5.1.0 of epubcheck but 
+        // due for fix in next maintence release, uncomment when fixed.
+        //assertNotNull(info.getCreated());
         assertNotNull(info.getLastModified());
 
         Property metadata = info.getProperty(EPUBMETADATA_KEY);
@@ -248,7 +250,9 @@ public void parseValidEpub2PropertiesTest() throws Exception {
         assertEquals(EXPECTED_MEDIATYPE, info.getMimeType());
         assertEquals("2.0.1", info.getVersion());
         // may change, so just check it isn't null
-        assertNotNull(info.getCreated());
+        // note: this property is missing in 5.1.0 of epubcheck but 
+        // due for fix in next maintence release, uncomment when fixed.
+        //assertNotNull(info.getCreated());
 
         Property metadata = info.getProperty(EPUBMETADATA_KEY);
         Map<String, Object> props = toMap(metadata);
@@ -397,7 +401,7 @@ public void parseImproperlyCompressedEpubTest() throws Exception {
         File epubFile = new File(ZIPPED_EPUB_FILEPATH);
         RepInfo info = parseAndCheckValidity(epubFile, RepInfo.FALSE, RepInfo.FALSE);
         assertEquals(EXPECTED_MEDIATYPE, info.getMimeType());
-        assertEquals(EXPECTED_VERSION_3_2, info.getVersion());
+        assertEquals(EXPECTED_VERSION_3_3, info.getVersion());
         assertEquals(1, info.getMessage().size());
         assertEquals("PKG-006", info.getMessage().get(0).getId());
     }
@@ -425,7 +429,7 @@ public void parseNonEpubTest() throws Exception {
         File epubFile = new File(WRONG_EXT_NOT_AN_EPUB_FILEPATH);
         RepInfo info = parseAndCheckValidity(epubFile, RepInfo.FALSE, RepInfo.FALSE);
         List<Message> msgs = info.getMessage();
-        final int expectedNumMessages = 3;
+        final int expectedNumMessages = 2;
         assertEquals(expectedNumMessages, msgs.size());
     }
 
@@ -451,7 +455,7 @@ public void parseNonEpubWithEpubExtensionTest() throws Exception {
         RepInfo info = parseAndCheckValidity(epubFile, RepInfo.FALSE, RepInfo.FALSE);
         assertEquals(OCTET_MIMETYPE, info.getMimeType());
         List<Message> msgs = info.getMessage();
-        final int expectedNumMessages = 3;
+        final int expectedNumMessages = 2;
         assertEquals(expectedNumMessages, msgs.size());
     }
 
@@ -508,7 +512,7 @@ public void parseEpubWithMissingFontsTest() throws Exception {
         Map<String, Object> fontinfo = new HashMap<String, Object>();
         font.forEach(f -> fontinfo.put(f.getName(), f.getValue()));
 
-        // only one font in this file, listed but missing.
+        // a single font file is listed - it is supposed to be embedded but is missing.
         assertEquals("Courier", fontinfo.get(PROPNAME_FONTNAME));
         assertEquals(true, fontinfo.get(PROPNAME_FONTFILE));
 
@@ -579,18 +583,12 @@ public void parseEpubMissingOpfTest() throws Exception {
         File epubFile = new File(EPUB2_MISSING_OPF_FILEPATH);
         RepInfo info = parseAndCheckValidity(epubFile, RepInfo.FALSE, RepInfo.FALSE);
 
-        assertEquals(OCTET_MIMETYPE, info.getMimeType());
-
         Set<String> msgCodes = new HashSet<String>();
-        assertEquals(2, info.getMessage().size());
-        Message msg1 = info.getMessage().get(0);
-        Message msg2 = info.getMessage().get(1);
-        assertTrue(msg1 instanceof ErrorMessage);
-        msgCodes.add(msg1.getId());
-        assertTrue(msg2 instanceof ErrorMessage);
-        msgCodes.add(msg2.getId());
+        assertEquals(1, info.getMessage().size());
+        Message msg = info.getMessage().get(0);
+        assertTrue(msg instanceof ErrorMessage);
+        msgCodes.add(msg.getId());
         assertTrue(msgCodes.contains("OPF-002"));
-        assertTrue(msgCodes.contains("RSC-001"));
     }
 
     /**
@@ -665,8 +663,8 @@ public void checkSignaturesEpub2WithEncryptionTest() throws Exception {
     public void parseEpub3TitleEncodingTest() throws Exception {
         File epubFile = new File(EPUB3_TITLE_ENCODING);
         String expectedTitle = "महाभारत";
-        // well formed and valid
-        RepInfo info = parseAndCheckValidity(epubFile, RepInfo.TRUE, RepInfo.TRUE);
+        // well formed but not valid (this is inconsequential to the test, we're just checking title)
+        RepInfo info = parseAndCheckValidity(epubFile, RepInfo.TRUE, RepInfo.FALSE);
 
         Property metadata = info.getProperty(PROPNAME_EPUB_METADATA);
         Map<String, Object> props = toMap(metadata);

jhove-ext-modules/src/test/java/org/ithaka/portico/jhove/module/epub/JhoveRepInfoReportTest.java

@@ -4,6 +4,7 @@
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertTrue;
 
+import java.io.File;
 import java.text.SimpleDateFormat;
 import java.util.Arrays;
 import java.util.Date;
@@ -45,9 +46,10 @@ public class JhoveRepInfoReportTest {
     private static final MessageId WARN_MSG_ID = MessageId.CHK_001;
     private static final String WARN_MSG = "Consider yourself warned";
     private static final String WARN_MSG_SUGGEST = "Don't do it again!";
-
-    private EPUBLocation messageLoc = EPUBLocation.create("epub.opf");
-    private EPUBLocation messageLoc2 = EPUBLocation.create("content.xhtml");
+
+    private EPUBLocation messageLoc = EPUBLocation.of(new File("epub.opf"));
+    private EPUBLocation messageLoc2 = EPUBLocation.of(new File("content.xhtml"));
+
     private String messageArg = "fakearg";