Add ignoreEmptyParagraphs option

5 files changed, 28 insertions, 2 deletions

diff --git a/README.md b/README.md
index 04fcfc6..238685f 100644
--- a/README.md
+++ b/README.md
@@ -301,6 +301,9 @@ Converts the source document to HTML.
     
   * `convertUnderline`: by default, the underlining of any text is ignored.
     Set this option to [`mammoth.underline.element(name)`](#underline) to override the default behaviour.
+  
+  * `ignoreEmptyParagraphs`: by default, empty paragraphs are ignored.
+    Set this option to `false` to preserve empty paragraphs in the output.
 
 * Returns a promise containing a result.
   This result has the following properties:
diff --git a/lib/document-to-html.js b/lib/document-to-html.js
index e9d35b9..220ae2b 100644
--- a/lib/document-to-html.js
+++ b/lib/document-to-html.js
@@ -1,6 +1,7 @@
-var promises = require("./promises");
 var async = require("async");
+var _ = require("underscore");
 
+var promises = require("./promises");
 var documents = require("./documents");
 var htmlPaths = require("./html-paths");
 var HtmlGenerator = require("./html-generator").HtmlGenerator;
@@ -32,7 +33,7 @@ function DocumentConversion(options) {
     
     var noteReferences = [];
     
-    options = options || {};
+    options = _.extend({ignoreEmptyParagraphs: true}, options);
     var idPrefix = options.idPrefix === undefined ?
         Math.floor(Math.random() * 1000000000000000) :
         options.idPrefix;
@@ -76,6 +77,11 @@ function DocumentConversion(options) {
     
     function convertParagraph(element, html, messages, callback) {
         html.satisfyPath(styleForParagraph(element, messages));
+        
+        if (!options.ignoreEmptyParagraphs) {
+            html.writeAll();
+        }
+        
         setImmediate(function() {
             convertElements(element.children, html, messages, callback);
         });
diff --git a/lib/html-generator.js b/lib/html-generator.js
index be1ebd2..a071a5e 100644
--- a/lib/html-generator.js
+++ b/lib/html-generator.js
@@ -90,6 +90,7 @@ function HtmlGenerator(options) {
         open: open,
         close: close,
         closeAll: closeAll,
+        writeAll: writeAll,
         asString: asString,
         append: append,
         selfClosing: selfClosing
diff --git a/tests/mammoth.tests.js b/tests/mammoth.tests.js
index 247ef17..6395afd 100644
--- a/tests/mammoth.tests.js
+++ b/tests/mammoth.tests.js
@@ -40,6 +40,22 @@ describe('mammoth', function() {
         });
     });
     
+    test('empty paragraphs are ignored by default', function() {
+        var docxPath = path.join(__dirname, "test-data/empty.docx");
+        return mammoth.convertToHtml({path: docxPath}).then(function(result) {
+            assert.equal(result.value, "");
+            assert.deepEqual(result.messages, []);
+        });
+    });
+    
+    test('empty paragraphs are preserved if ignoreEmptyParagraphs is false', function() {
+        var docxPath = path.join(__dirname, "test-data/empty.docx");
+        return mammoth.convertToHtml({path: docxPath}, {ignoreEmptyParagraphs: false}).then(function(result) {
+            assert.equal(result.value, "<p></p>");
+            assert.deepEqual(result.messages, []);
+        });
+    });
+    
     test('style map can be expressed as string', function() {
         var docxFile = createFakeDocxFile({
             "word/document.xml": testData("simple/word/document.xml")
diff --git a/tests/test-data/empty.docx b/tests/test-data/empty.docx
new file mode 100644
index 0000000..e10b869
--- /dev/null
+++ b/tests/test-data/empty.docx