Throw more informative error when word/document.xml cannot be found

4 files changed, 26 insertions, 8 deletions

diff --git a/NEWS b/NEWS
index 07e8ee6..3c3a715 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,7 @@
+# 1.2.0
+
+* Throw more informative error when word/document.xml cannot be found.
+
 # 1.1.0
 
 * Add support for comments.
diff --git a/lib/docx/docx-reader.js b/lib/docx/docx-reader.js
index 57830d7..0c7d57f 100644
--- a/lib/docx/docx-reader.js
+++ b/lib/docx/docx-reader.js
@@ -60,16 +60,20 @@ function read(docxFile, input) {
         };
     }).then(function(result) {
         return readXmlFileWithBody("document", result, function(bodyReader, xml) {
-            return result.notes.flatMap(function(notes) {
-                return result.comments.flatMap(function(comments) {
-                    var reader = new DocumentXmlReader({
-                        bodyReader: bodyReader,
-                        notes: notes,
-                        comments: comments
+            if (xml) {
+                return result.notes.flatMap(function(notes) {
+                    return result.comments.flatMap(function(comments) {
+                        var reader = new DocumentXmlReader({
+                            bodyReader: bodyReader,
+                            notes: notes,
+                            comments: comments
+                        });
+                        return reader.convertXmlToDocument(xml);
                     });
-                    return reader.convertXmlToDocument(xml);
                 });
-            });
+            } else {
+                throw new Error("Could not find word/document.xml in ZIP file. Are you sure this is a valid .docx file?");
+            }
         });
     });
 }
diff --git a/test/mammoth.tests.js b/test/mammoth.tests.js
index f4d445b..bf3d2fd 100644
--- a/test/mammoth.tests.js
+++ b/test/mammoth.tests.js
@@ -368,3 +368,13 @@ test('extractRawText can use .docx files represented by a Buffer', function() {
             assert.deepEqual(result.messages, []);
         });
 });
+
+
+test('should throw error if file is not a valid docx document', function() {
+    var docxPath = path.join(__dirname, "test-data/empty.zip");
+    return mammoth.convertToHtml({path: docxPath}).then(function(result) {
+        assert.ok(false, "Expected error");
+    }, function(error) {
+        assert.equal(error.message, "Could not find word/document.xml in ZIP file. Are you sure this is a valid .docx file?");
+    });
+});
diff --git a/test/test-data/empty.zip b/test/test-data/empty.zip
new file mode 100644
index 0000000..15cb0ec
--- /dev/null
+++ b/test/test-data/empty.zip