Use getElementsByTagName for safer image parsing

3 files changed, 41 insertions, 16 deletions

diff --git a/lib/document-xml-reader.js b/lib/document-xml-reader.js
index a391a21..1724b5b 100644
--- a/lib/document-xml-reader.js
+++ b/lib/document-xml-reader.js
@@ -23,9 +23,7 @@ function DocumentXmlReader(relationships, docxFile) {
 
     function readXmlElements(elements) {
         var results = elements.map(readXmlElement);
-        var values = _.flatten(_.pluck(results, "value"));
-        var messages = _.flatten(_.pluck(results, "messages"), true);
-        return new Result(values, messages);
+        return Result.combine(results);
     }
 
     function readXmlElement(element) {
@@ -108,19 +106,24 @@ function DocumentXmlReader(relationships, docxFile) {
             return readXmlElements(element.children);
         },
         "wp:inline": function(element) {
-            var relationshipId = element
-                .first("a:graphic")
-                .first("a:graphicData")
-                .first("pic:pic")
-                .first("pic:blipFill")
-                .first("a:blip")
-                .attributes["r:embed"];
-            var imagePath = relationships[relationshipId].target;
-            var readImage = docxFile.read.bind(docxFile, path.join("word", imagePath));
-            var altText = element.first("wp:docPr").attributes.descr;
+            var blips = element
+                .getElementsByTagName("a:graphic")
+                .getElementsByTagName("a:graphicData")
+                .getElementsByTagName("pic:pic")
+                .getElementsByTagName("pic:blipFill")
+                .getElementsByTagName("a:blip");
             
-            var image = documents.Image(readImage, altText)
-            return new Result(image);
+            var results = blips.map(function(blip) {
+                var relationshipId = blip.attributes["r:embed"];
+                var imagePath = relationships[relationshipId].target;
+                var readImage = docxFile.read.bind(docxFile, path.join("word", imagePath));
+                var altText = element.first("wp:docPr").attributes.descr;
+                
+                var image = documents.Image(readImage, altText)
+                return new Result(image);
+            });
+            
+            return Result.combine(results);
         }
     };
     return {
diff --git a/lib/results.js b/lib/results.js
index d5dd7af..c12d194 100644
--- a/lib/results.js
+++ b/lib/results.js
@@ -1,3 +1,6 @@
+var _ = require("underscore");
+
+
 exports.Result = Result;
 exports.warning = warning;
 
@@ -18,6 +21,12 @@ Result.prototype.flatMapThen = function(func) {
     });
 };
 
+Result.combine = function(results) {
+    var values = _.flatten(_.pluck(results, "value"));
+    var messages = _.flatten(_.pluck(results, "messages"), true);
+    return new Result(values, messages);
+};
+
 function warning(message) {
     return {
         type: "warning",
diff --git a/lib/xmlreader.js b/lib/xmlreader.js
index 08347c4..3b14a93 100644
--- a/lib/xmlreader.js
+++ b/lib/xmlreader.js
@@ -99,7 +99,20 @@ Element.prototype.first = function(name) {
 }
 
 Element.prototype.getElementsByTagName = function(name) {
-    return _.filter(this.children, function(child) {
+    var elements = _.filter(this.children, function(child) {
         return child.name == name;
     });
+    return toElementList(elements);
 };
+
+var elementListPrototype = {
+    getElementsByTagName: function(name) {
+        return toElementList(_.flatten(this.map(function(element) {
+            return element.getElementsByTagName(name);
+        }, true)));
+    }
+};
+
+function toElementList(array) {
+    return _.extend(array, elementListPrototype);
+}