fix: fail gracefully when running on a non-EPUB file (#1134)

rdeltour · web-flow · commit 2083f05ddd8e · 2020-05-01T16:22:36.000+02:00
Running EPUBCheck on a non-EPUB package file (for instance an Office Open XML document) caused a a `NullPointerException`. It now fails gracefully with a fatal error (RSC-002) potentially followed by side-effect errors (typically RSC-001, PKG-006). Fixes #1050
diff --git a/src/main/java/com/adobe/epubcheck/api/EpubCheck.java b/src/main/java/com/adobe/epubcheck/api/EpubCheck.java
@@ -268,10 +268,12 @@ void checkExtension(OCFPackage ocf, String extension)
         else
         {
           List<String> opfPaths = ocf.getOcfData().getEntries(OPFData.OPF_MIME_TYPE);
-          if(ocf.getOpfData().get(opfPaths.get(0)).getVersion() == EPUBVersion.VERSION_3) {
-            report.message(MessageId.PKG_024, EPUBLocation.create(epubFile.getName(), extension));
-          } else {
-            report.message(MessageId.PKG_017, EPUBLocation.create(epubFile.getName(), extension));
+          if (!opfPaths.isEmpty()) {
+            if(ocf.getOpfData().get(opfPaths.get(0)).getVersion() == EPUBVersion.VERSION_3) {
+              report.message(MessageId.PKG_024, EPUBLocation.create(epubFile.getName(), extension));
+            } else {
+              report.message(MessageId.PKG_017, EPUBLocation.create(epubFile.getName(), extension));
+            }
           }
         }
       }
diff --git a/src/main/java/com/adobe/epubcheck/xml/XMLParser.java b/src/main/java/com/adobe/epubcheck/xml/XMLParser.java
@@ -196,6 +196,9 @@ public void process()
       try
       {
         InputStream in = closer.register(context.resourceProvider.getInputStream(path));
+        if (in == null) {
+          return; // Abort processing. Missing required files are reported elsewhere.
+        }
         // System.err.println("DEBUG XMLParser#process on" + resource);
         if (!in.markSupported())
         {
diff --git a/src/test/java/com/adobe/epubcheck/api/Epub30CheckTest.java b/src/test/java/com/adobe/epubcheck/api/Epub30CheckTest.java
@@ -367,4 +367,12 @@ public void testEdupubRenditions_Invalid_NoRenditionDCType()
     Collections.addAll(expectedErrors, MessageId.RSC_005);
     testValidateDocument("invalid/edupub-multiple-renditions-nodctype-rendition.epub");
   }
+
+  @Test
+  public void testNotAnEPUB()
+  {
+    Collections.addAll(expectedFatals, MessageId.RSC_002);
+    Collections.addAll(expectedErrors, MessageId.PKG_006, MessageId.RSC_001);
+    testValidateDocument("invalid/not-an-epub.docx");
+  }
 }
diff --git a/src/test/resources/30/epub/invalid/not-an-epub.docx b/src/test/resources/30/epub/invalid/not-an-epub.docx

Original file line number	Diff line number	Diff line change
`@@ -268,10 +268,12 @@ void checkExtension(OCFPackage ocf, String extension)`
`268`	`268`	`else`
`269`	`269`	`{`
`270`	`270`	`List<String> opfPaths = ocf.getOcfData().getEntries(OPFData.OPF_MIME_TYPE);`
`271`		`- if(ocf.getOpfData().get(opfPaths.get(0)).getVersion() == EPUBVersion.VERSION_3) {`
`272`		`- report.message(MessageId.PKG_024, EPUBLocation.create(epubFile.getName(), extension));`
`273`		`- } else {`
`274`		`- report.message(MessageId.PKG_017, EPUBLocation.create(epubFile.getName(), extension));`
	`271`	`+ if (!opfPaths.isEmpty()) {`
	`272`	`+ if(ocf.getOpfData().get(opfPaths.get(0)).getVersion() == EPUBVersion.VERSION_3) {`
	`273`	`+ report.message(MessageId.PKG_024, EPUBLocation.create(epubFile.getName(), extension));`
	`274`	`+ } else {`
	`275`	`+ report.message(MessageId.PKG_017, EPUBLocation.create(epubFile.getName(), extension));`
	`276`	`+ }`
`275`	`277`	`}`
`276`	`278`	`}`
`277`	`279`	`}`
Original file line number	Diff line number	Diff line change
`@@ -196,6 +196,9 @@ public void process()`
`196`	`196`	`try`
`197`	`197`	`{`
`198`	`198`	`InputStream in = closer.register(context.resourceProvider.getInputStream(path));`
	`199`	`+ if (in == null) {`
	`200`	`+ return; // Abort processing. Missing required files are reported elsewhere.`
	`201`	`+ }`
`199`	`202`	`// System.err.println("DEBUG XMLParser#process on" + resource);`
`200`	`203`	`if (!in.markSupported())`
`201`	`204`	`{`
Original file line number	Diff line number	Diff line change
`@@ -367,4 +367,12 @@ public void testEdupubRenditions_Invalid_NoRenditionDCType()`
`367`	`367`	`Collections.addAll(expectedErrors, MessageId.RSC_005);`
`368`	`368`	`testValidateDocument("invalid/edupub-multiple-renditions-nodctype-rendition.epub");`
`369`	`369`	`}`
	`370`	`+`
	`371`	`+ @Test`
	`372`	`+ public void testNotAnEPUB()`
	`373`	`+ {`
	`374`	`+ Collections.addAll(expectedFatals, MessageId.RSC_002);`
	`375`	`+ Collections.addAll(expectedErrors, MessageId.PKG_006, MessageId.RSC_001);`
	`376`	`+ testValidateDocument("invalid/not-an-epub.docx");`
	`377`	`+ }`
`370`	`378`	`}`