Merge pull request #1 from jdlehman/hyperlink-edge-cases

Hyperlink edge cases

2 files changed, 179 insertions, 9 deletions

diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
index e92f8b4..11aeec5 100644
--- a/lib/docx/body-reader.js
+++ b/lib/docx/body-reader.js
@@ -20,7 +20,8 @@ function createBodyReader(options) {
 }
 
 function BodyReader(options) {
-    var fieldHyperlink = null;
+    var complexFields = [];
+    var currentInstrText = [];
     var relationships = options.relationships;
     var contentTypes = options.contentTypes;
     var docxFile = options.docxFile;
@@ -147,8 +148,9 @@ function BodyReader(options) {
                     var properties = _.find(children, isRunProperties);
                     children = children.filter(negate(isRunProperties));
 
-                    if (fieldHyperlink !== null) {
-                        children = [new documents.Hyperlink(children, {href: fieldHyperlink})];
+                    var scopedHyperlink = scopedComplexFieldOfType(complexFields, "hyperlink");
+                    if (scopedHyperlink !== undefined) {
+                        children = [new documents.Hyperlink(children, {href: scopedHyperlink.href})];
                     }
 
                     return new documents.Run(children, properties);
@@ -158,15 +160,20 @@ function BodyReader(options) {
         "w:fldChar": function(element) {
             var type = element.attributes["w:fldCharType"];
             if (type === "end") {
-                fieldHyperlink = null;
+                complexFields.pop();
+            } else if (type === "separate") {
+                var href = parseHyperlinkFieldCode(currentInstrText.join(''));
+                if (href !== null) {
+                    complexFields.push({type: "hyperlink", href: href});
+                } else {
+                    complexFields.push(null);
+                }
+                currentInstrText = [];
             }
             return emptyResult();
         },
         "w:instrText": function(element) {
-            var hyperlink = parseHyperlinkFieldCode(element.text());
-            if (hyperlink !== null) {
-                fieldHyperlink = hyperlink;
-            }
+            currentInstrText.push(element.text());
             return emptyResult();
         },
         "w:t": function(element) {
@@ -530,3 +537,13 @@ function parseHyperlinkFieldCode(code) {
         return null;
     }
 }
+
+function complexFieldsOfType(complexFields, type) {
+    return complexFields.filter(function(complexField) {
+        return complexField.type === type;
+    });
+}
+
+function scopedComplexFieldOfType(complexFields, type) {
+    return _.last(complexFieldsOfType(complexFields, type));
+}
diff --git a/test/docx/body-reader.tests.js b/test/docx/body-reader.tests.js
index ab4995c..a670202 100644
--- a/test/docx/body-reader.tests.js
+++ b/test/docx/body-reader.tests.js
@@ -148,9 +148,13 @@ test("complex fields", (function() {
     var endXml = new XmlElement("w:r", {}, [
         new XmlElement("w:fldChar", {"w:fldCharType": "end"})
     ]);
+    var separateXml = new XmlElement("w:r", {}, [
+        new XmlElement("w:fldChar", {"w:fldCharType": "separate"})
+    ]);
     var hyperlinkInstrText = new XmlElement("w:instrText", {}, [
         xml.text(' HYPERLINK "' + uri + '"')
     ]);
+    var hyperlinkRunXml = runOfText("this is a hyperlink");
     
     return {
         "stores instrText returns empty result": function() {
@@ -160,10 +164,10 @@ test("complex fields", (function() {
         
         "runs in a complex field for hyperlinks are read as hyperlinks": function() {
             var hyperlinkRunXml = runOfText("this is a hyperlink");
-            var afterEndXml = runOfText("this will not be a hyperlink");
             var paragraphXml = new XmlElement("w:p", {}, [
                 beginXml,
                 hyperlinkInstrText,
+                separateXml,
                 hyperlinkRunXml,
                 endXml
             ]);
@@ -175,6 +179,14 @@ test("complex fields", (function() {
                     children: contains(
                         isHyperlink({
                             href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
                             children: contains(
                                 isText("this is a hyperlink")
                             )
@@ -190,6 +202,7 @@ test("complex fields", (function() {
             var paragraphXml = new XmlElement("w:p", {}, [
                 beginXml,
                 hyperlinkInstrText,
+                separateXml,
                 endXml,
                 afterEndXml
             ]);
@@ -197,6 +210,14 @@ test("complex fields", (function() {
             
             assertThat(paragraph.children, contains(
                 isEmptyRun,
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
                 isEmptyRun,
                 isRun({
                     children: contains(
@@ -204,6 +225,138 @@ test("complex fields", (function() {
                     )
                 })
             ));
+        },
+
+        "can handle split instrText elements": function() {
+            var hyperlinkInstrTextPart1 = new XmlElement("w:instrText", {}, [
+                xml.text(" HYPE")
+            ]);
+            var hyperlinkInstrTextPart2 = new XmlElement("w:instrText", {}, [
+                xml.text('RLINK "' + uri + '"')
+            ]);
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                hyperlinkInstrTextPart1,
+                hyperlinkInstrTextPart2,
+                separateXml,
+                hyperlinkRunXml,
+                endXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: contains(
+                                isText("this is a hyperlink")
+                            )
+                        })
+                    )
+                }),
+                isEmptyRun
+            ));
+        },
+
+        "hyperlink with a nested complex field uses the outer hyperlink": function() {
+            var authorInstrText = new XmlElement("w:instrText", {}, [
+                xml.text(' AUTHOR "John Doe"')
+            ]);
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                beginXml,
+                authorInstrText,
+                endXml,
+                hyperlinkInstrText,
+                separateXml,
+                hyperlinkRunXml,
+                endXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isEmptyRun,
+                isEmptyRun,
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: contains(
+                                isText("this is a hyperlink")
+                            )
+                        })
+                    )
+                }),
+                isEmptyRun
+          ));
+        },
+
+        "complex field nested within a hyperlink complex field is wrapped with the hyperlink": function() {
+            var authorInstrText = new XmlElement("w:instrText", {}, [
+                xml.text(' AUTHOR "John Doe"')
+            ]);
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                hyperlinkInstrText,
+                separateXml,
+                hyperlinkRunXml,
+                beginXml,
+                authorInstrText,
+                endXml,
+                endXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: contains(
+                                isText("this is a hyperlink")
+                            )
+                        })
+                    )
+                }),
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isEmptyRun,
+                isEmptyRun
+          ));
         }
     };
 })());