Add support for internal hyperlinks

5 files changed, 84 insertions, 13 deletions

diff --git a/lib/document-to-html.js b/lib/document-to-html.js
index c068fba..000fc5c 100644
--- a/lib/document-to-html.js
+++ b/lib/document-to-html.js
@@ -234,7 +234,9 @@ function DocumentConversion(options) {
             callback();
         },
         "hyperlink": function(element, html, messages, callback) {
-            html.open(htmlPaths.element("a", {href: element.href}));
+            var href = element.anchor ? "#" + element.anchor : element.href;
+
+            html.open(htmlPaths.element("a", {href: href}));
             convertElements(element.children, html, messages, function(err) {
                 if (err) {
                     callback(err);
@@ -280,6 +282,17 @@ function DocumentConversion(options) {
         "lineBreak": function(element, html, messages, callback) {
             html.selfClosing(htmlPaths.element("br"));
             callback();
+        },
+        "bookmarkStart": function(element, html, messages, callback) {
+            if (element.name === "_GoBack") {
+                return callback();
+            }
+
+            html.open(htmlPaths.element("span", {
+                id: element.name
+            }), {alwaysWrite: true});
+            html.close();
+            callback();
         }
     };
     return {
diff --git a/lib/documents.js b/lib/documents.js
index 78bea1a..10d8646 100644
--- a/lib/documents.js
+++ b/lib/documents.js
@@ -13,7 +13,8 @@ var types = {
     table: "table",
     tableRow: "tableRow",
     tableCell: "tableCell",
-    lineBreak: "lineBreak"
+    lineBreak: "lineBreak",
+    bookmarkStart: "bookmarkStart"
 };
 
 function Document(children, options) {
@@ -74,7 +75,8 @@ function Hyperlink(children, options) {
     return {
         type: types.hyperlink,
         children: children,
-        href: options.href
+        href: options.href,
+        anchor: options.anchor
     };
 }
 
@@ -149,6 +151,13 @@ function LineBreak() {
     };
 }
 
+function BookmarkStart(options) {
+    return {
+        type: types.bookmarkStart,
+        name: options.name
+    };
+}
+
 exports.Document = Document;
 exports.Paragraph = Paragraph;
 exports.Run = Run;
@@ -163,5 +172,6 @@ exports.Table = Table;
 exports.TableRow = TableRow;
 exports.TableCell = TableCell;
 exports.LineBreak = LineBreak;
+exports.BookmarkStart = BookmarkStart;
 
 exports.verticalAlignment = verticalAlignment;
diff --git a/lib/docx/document-xml-reader.js b/lib/docx/document-xml-reader.js
index e290792..f25e7a3 100644
--- a/lib/docx/document-xml-reader.js
+++ b/lib/docx/document-xml-reader.js
@@ -167,10 +167,13 @@ function DocumentXmlReader(options) {
         },
         "w:hyperlink": function(element) {
             var relationshipId = element.attributes["r:id"];
+            var anchor         = element.attributes["w:anchor"];
             return readXmlElements(element.children).map(function(children) {
                 if (relationshipId) {
                     var href = relationships[relationshipId].target;
                     return new documents.Hyperlink(children, {href: href});
+                } else if (anchor) {
+                    return new documents.Hyperlink(children, {anchor: anchor});
                 } else {
                     return children;
                 }
@@ -195,7 +198,13 @@ function DocumentXmlReader(options) {
                 return new Result(new documents.LineBreak());
             }
         },
-        
+        "w:bookmarkStart": function(element){
+            var name = element.attributes["w:name"];
+            return readXmlElements(element.children).map(function(){
+                return new documents.BookmarkStart({name: name});
+            });
+        },
+
         "w:ins": readChildElements,
         "w:smartTag": readChildElements,
         "w:drawing": readChildElements,
@@ -245,7 +254,6 @@ var supportedImageTypes = {
 };
 
 var ignoreElements = {
-    "w:bookmarkStart": true,
     "w:bookmarkEnd": true,
     "w:sectPr": true,
     "w:proofErr": true,
diff --git a/tests/document-to-html.tests.js b/tests/document-to-html.tests.js
index 3f50cbd..4863cf1 100644
--- a/tests/document-to-html.tests.js
+++ b/tests/document-to-html.tests.js
@@ -239,7 +239,18 @@ describe('DocumentConverter', function() {
             assert.equal(result.value, '<a href="http://www.example.com">Hello.</a>');
         });
     });
-    
+
+    test('docx hyperlink with anchor is converted to <a>', function() {
+        var hyperlink = new documents.Hyperlink(
+            [runOfText("Hello.")],
+            {anchor: "_Peter"}
+        );
+        var converter = new DocumentConverter();
+        return converter.convertToHtml(hyperlink).then(function(result) {
+            assert.equal(result.value, '<a href="#_Peter">Hello.</a>');
+        });
+    });
+
     test('docx tab is converted to tab in HTML', function() {
         var tab = new documents.Tab();
         var converter = new DocumentConverter();
@@ -405,6 +416,27 @@ describe('DocumentConverter', function() {
             assert.equal(result.value.indexOf("<p>Hello.</p>"), 0);
         });
     });
+
+    test('referenced bookmarks are coverted to anchors', function() {
+        var bookmarkStart = new documents.BookmarkStart({name: "_Peter"});
+        var hyperlink = new documents.Hyperlink(
+            [runOfText("Hello.")],
+            {anchor: "_Peter"}
+        );
+        var converter = new DocumentConverter();
+        var document = new documents.Document([bookmarkStart, hyperlink]);
+        return converter.convertToHtml(document).then(function(result) {
+            assert.equal(result.value, '<span id="_Peter"></span><a href="#_Peter">Hello.</a>');
+        });
+    });
+
+    test('unreferenced bookmarks are not converted', function() {
+        var bookmarkStart = new documents.BookmarkStart({name: "_Unreferenced"});
+        var converter = new DocumentConverter();
+        return converter.convertToHtml(bookmarkStart).then(function(result) {
+            assert.equal(result.value, '');
+        });
+    });
 });
 
 function paragraphOfText(text, styleId, styleName) {
diff --git a/tests/docx/document-xml-reader.tests.js b/tests/docx/document-xml-reader.tests.js
index 0b5a75f..35c806b 100644
--- a/tests/docx/document-xml-reader.tests.js
+++ b/tests/docx/document-xml-reader.tests.js
@@ -217,12 +217,12 @@ describe("readXmlElement: ", function() {
             }]
         );
     });
-    
-    test("w:bookmarkStart is ignored without warning", function() {
-        var ignoredElement = new XmlElement("w:bookmarkStart");
-        var result = readXmlElement(ignoredElement);
-        assert.deepEqual(result.messages, []);
-        assert.deepEqual([], result.value);
+
+    test("w:bookmarkStart is read as a bookmarkStart", function() {
+        var bookmarkStart = new XmlElement("w:bookmarkStart", {"w:name": "_Peter", "w:id": "42"});
+        var result = readXmlElement(bookmarkStart);
+        assert.deepEqual(result.value.name, "_Peter");
+        assert.deepEqual(result.value.type, "bookmarkStart");
     });
     
     test("can read inline pictures", function() {
@@ -354,7 +354,15 @@ describe("readXmlElement: ", function() {
         assert.deepEqual(result.value.href, "http://example.com");
         assert.deepEqual(result.value.children[0].type, "run");
     });
-    
+
+    test("w:hyperlink is read as document hyperlink if it has an anchor", function() {
+        var runXml = new XmlElement("w:r", {}, []);
+        var hyperlinkXml = new XmlElement("w:hyperlink", {"w:anchor": "_Peter"}, [runXml]);
+        var result = readXmlElement(hyperlinkXml);
+        assert.deepEqual(result.value.anchor, "_Peter");
+        assert.deepEqual(result.value.children[0].type, "run");
+    });
+
     test("w:hyperlink is ignored if it does not have a relationship ID", function() {
         var runXml = new XmlElement("w:r", {}, []);
         var hyperlinkXml = new XmlElement("w:hyperlink", {}, [runXml]);