Handle hyperlink complex fields with unquoted hrefs

3 files changed, 66 insertions, 8 deletions

diff --git a/NEWS b/NEWS
index d9b8255..02dd689 100644
--- a/NEWS
+++ b/NEWS
@@ -2,6 +2,8 @@
 
 * Handle hyperlinked wp:anchor and wp:inline elements.
 
+* Handle hyperlink complex fields with unquoted hrefs.
+
 # 1.11.0
 
 * Ignore style definitions using a style ID that has already been used.
diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
index febbbc2..74add0a 100644
--- a/lib/docx/body-reader.js
+++ b/lib/docx/body-reader.js
@@ -203,14 +203,16 @@ function BodyReader(options) {
     }
 
     function parseInstrText(instrText, fldChar) {
-        var externalLinkResult = /\s*HYPERLINK "(.*)"/.exec(instrText);
+        var externalLinkResult = /^\s*HYPERLINK\s+(?:"(.*)"|([^\\]\S*))/.exec(instrText);
         if (externalLinkResult) {
-            return {type: "hyperlink", options: {href: externalLinkResult[1]}};
+            var href = externalLinkResult[1] === undefined ? externalLinkResult[2] : externalLinkResult[1];
+            return {type: "hyperlink", options: {href: href}};
         }
 
-        var internalLinkResult = /\s*HYPERLINK\s+\\l\s+"(.*)"/.exec(instrText);
+        var internalLinkResult = /^\s*HYPERLINK\s+\\l\s+(?:"(.*)"|([^\\]\S*))/.exec(instrText);
         if (internalLinkResult) {
-            return {type: "hyperlink", options: {anchor: internalLinkResult[1]}};
+            var anchor = internalLinkResult[1] === undefined ? internalLinkResult[2] : internalLinkResult[1];
+            return {type: "hyperlink", options: {anchor: anchor}};
         }
 
         var checkboxResult = /\s*FORMCHECKBOX\s*/.exec(instrText);
diff --git a/test/docx/body-reader.tests.js b/test/docx/body-reader.tests.js
index dee38a5..26c57e2 100644
--- a/test/docx/body-reader.tests.js
+++ b/test/docx/body-reader.tests.js
@@ -332,11 +332,13 @@ test("complex fields", (function() {
             assert.deepEqual(instrText, []);
         },
 
-        "runs in a complex field for hyperlink without switch are read as external hyperlinks": function() {
+        "runs in a complex field for hyperlink without switch with quoted target are read as external hyperlinks": function() {
             var hyperlinkRunXml = runOfText("this is a hyperlink");
             var paragraphXml = new XmlElement("w:p", {}, [
                 beginXml,
-                hyperlinkInstrText,
+                new XmlElement("w:instrText", {}, [
+                    xml.text(' HYPERLINK "http://example.com"')
+                ]),
                 separateXml,
                 hyperlinkRunXml,
                 endXml
@@ -347,7 +349,7 @@ test("complex fields", (function() {
                 isEmptyRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
-                    href: uri,
+                    href: "http://example.com",
                     children: contains(
                         isText("this is a hyperlink")
                     )
@@ -356,7 +358,33 @@ test("complex fields", (function() {
             ));
         },
 
-        "runs in a complex field for hyperlink with l switch are read as internal hyperlinks": function() {
+        "runs in a complex field for hyperlink without switch with unquoted target are read as external hyperlinks": function() {
+            var hyperlinkRunXml = runOfText("this is a hyperlink");
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                new XmlElement("w:instrText", {}, [
+                    xml.text(' HYPERLINK http://example.com')
+                ]),
+                separateXml,
+                hyperlinkRunXml,
+                endXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isEmptyHyperlinkedRun,
+                isHyperlinkedRun({
+                    href: "http://example.com",
+                    children: contains(
+                        isText("this is a hyperlink")
+                    )
+                }),
+                isEmptyRun
+            ));
+        },
+
+        "runs in a complex field for hyperlink with l switch with quoted target are read as internal hyperlinks": function() {
             var hyperlinkRunXml = runOfText("this is a hyperlink");
             var paragraphXml = new XmlElement("w:p", {}, [
                 beginXml,
@@ -382,6 +410,32 @@ test("complex fields", (function() {
             ));
         },
 
+        "runs in a complex field for hyperlink with l switch with unquoted target are read as internal hyperlinks": function() {
+            var hyperlinkRunXml = runOfText("this is a hyperlink");
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                new XmlElement("w:instrText", {}, [
+                    xml.text(' HYPERLINK \\l InternalLink')
+                ]),
+                separateXml,
+                hyperlinkRunXml,
+                endXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isEmptyHyperlinkedRun,
+                isHyperlinkedRun({
+                    anchor: "InternalLink",
+                    children: contains(
+                        isText("this is a hyperlink")
+                    )
+                }),
+                isEmptyRun
+            ));
+        },
+
         "runs after a complex field for hyperlinks are not read as hyperlinks": function() {
             var afterEndXml = runOfText("this will not be a hyperlink");
             var paragraphXml = new XmlElement("w:p", {}, [