Handle internal hyperlinks created with complex fields

LLLLLLLUO · Aug 1, 2021 · 67142b6 · 67142b6
1 parent 2a3c8d6
commit 67142b6
Show file tree

Hide file tree

Showing 3 changed files with 54 additions and 14 deletions.
diff --git a/NEWS b/NEWS
@@ -1,3 +1,7 @@
+# 1.4.19
+
+* Handle internal hyperlinks created with complex fields.
+
 # 1.4.18
 
 * When extracting raw text, convert tab elements to tab characters.

diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
@@ -139,28 +139,33 @@ function BodyReader(options) {
         } else if (type === "end") {
             complexFieldStack.pop();
         } else if (type === "separate") {
-            var href = parseHyperlinkFieldCode(currentInstrText.join(''));
-            var complexField = href === null ? unknownComplexField : {type: "hyperlink", href: href};
+            var hyperlinkOptions = parseHyperlinkFieldCode(currentInstrText.join(''));
+            var complexField = hyperlinkOptions === null ? unknownComplexField : {type: "hyperlink", options: hyperlinkOptions};
             complexFieldStack.pop();
             complexFieldStack.push(complexField);
         }
         return emptyResult();
     }
 
-    function currentHyperlinkHref() {
+    function currentHyperlinkOptions() {
         var topHyperlink = _.last(complexFieldStack.filter(function(complexField) {
             return complexField.type === "hyperlink";
         }));
-        return topHyperlink ? topHyperlink.href : null;
+        return topHyperlink ? topHyperlink.options : null;
     }
 
     function parseHyperlinkFieldCode(code) {
-        var result = /\s*HYPERLINK "(.*)"/.exec(code);
-        if (result) {
-            return result[1];
-        } else {
-            return null;
+        var externalLinkResult = /\s*HYPERLINK "(.*)"/.exec(code);
+        if (externalLinkResult) {
+            return {href: externalLinkResult[1]};
+        }
+
+        var internalLinkResult = /\s*HYPERLINK\s+\\l\s+"(.*)"/.exec(code);
+        if (internalLinkResult) {
+            return {anchor: internalLinkResult[1]};
         }
+
+        return null;
     }
 
     function readInstrText(element) {
@@ -236,9 +241,9 @@ function BodyReader(options) {
                     var properties = _.find(children, isRunProperties);
                     children = children.filter(negate(isRunProperties));
 
-                    var hyperlinkHref = currentHyperlinkHref();
-                    if (hyperlinkHref !== null) {
-                        children = [new documents.Hyperlink(children, {href: hyperlinkHref})];
+                    var hyperlinkOptions = currentHyperlinkOptions();
+                    if (hyperlinkOptions !== null) {
+                        children = [new documents.Hyperlink(children, hyperlinkOptions)];
                     }
 
                     return new documents.Run(children, properties);

diff --git a/test/docx/body-reader.tests.js b/test/docx/body-reader.tests.js
@@ -246,7 +246,7 @@ test("complex fields", (function() {
     function isHyperlinkedRun(hyperlinkProperties) {
         return isRun({
             children: contains(
-                isHyperlink(_.extend({href: uri}, hyperlinkProperties))
+                isHyperlink(hyperlinkProperties)
             )
         });
     }
@@ -257,7 +257,7 @@ test("complex fields", (function() {
             assert.deepEqual(instrText, []);
         },
 
-        "runs in a complex field for hyperlinks are read as hyperlinks": function() {
+        "runs in a complex field for hyperlink without switch are read as external hyperlinks": function() {
             var hyperlinkRunXml = runOfText("this is a hyperlink");
             var paragraphXml = new XmlElement("w:p", {}, [
                 beginXml,
@@ -272,6 +272,33 @@ test("complex fields", (function() {
                 isEmptyRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
+                    href: uri,
+                    children: contains(
+                        isText("this is a hyperlink")
+                    )
+                }),
+                isEmptyRun
+            ));
+        },
+
+        "runs in a complex field for hyperlink with l switch 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")
                     )
@@ -324,6 +351,7 @@ test("complex fields", (function() {
                 isEmptyRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
+                    href: uri,
                     children: contains(
                         isText("this is a hyperlink")
                     )
@@ -356,6 +384,7 @@ test("complex fields", (function() {
                 isEmptyHyperlinkedRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
+                    href: uri,
                     children: contains(
                         isText("this is a hyperlink")
                     )
@@ -387,6 +416,7 @@ test("complex fields", (function() {
                 isEmptyHyperlinkedRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
+                    href: uri,
                     children: contains(
                         isText("John Doe")
                     )
@@ -415,6 +445,7 @@ test("complex fields", (function() {
                 isEmptyHyperlinkedRun,
                 isEmptyHyperlinkedRun,
                 isHyperlinkedRun({
+                    href: uri,
                     children: contains(
                         isText("this is a hyperlink")
                     )