Merge pull request #110 from jdlehman/instrText-hyperlinks

Enable hyperlinks that are not defined with w:hyperlink

6 files changed, 347 insertions, 38 deletions

diff --git a/lib/docx/body-reader.js b/lib/docx/body-reader.js
index 529d6f2..a19ecdb 100644
--- a/lib/docx/body-reader.js
+++ b/lib/docx/body-reader.js
@@ -1,4 +1,5 @@
-exports.BodyReader = BodyReader;
+exports.createBodyReader = createBodyReader;
+exports._readNumberingProperties = readNumberingProperties;
 
 var _ = require("underscore");
 
@@ -7,8 +8,20 @@ var Result = require("../results").Result;
 var warning = require("../results").warning;
 var uriToZipEntryName = require("./uris").uriToZipEntryName;
 
+function createBodyReader(options) {
+    return {
+        readXmlElement: function(element) {
+            return new BodyReader(options).readXmlElement(element);
+        },
+        readXmlElements: function(elements) {
+            return new BodyReader(options).readXmlElements(elements);
+        }
+    };
+}
 
 function BodyReader(options) {
+    var complexFields = [];
+    var currentInstrText = [];
     var relationships = options.relationships;
     var contentTypes = options.contentTypes;
     var docxFile = options.docxFile;
@@ -125,7 +138,7 @@ function BodyReader(options) {
                     styleId: style.styleId,
                     styleName: style.name,
                     alignment: element.firstOrEmpty("w:jc").attributes["w:val"],
-                    numbering: readNumberingProperties(element.firstOrEmpty("w:numPr"))
+                    numbering: readNumberingProperties(element.firstOrEmpty("w:numPr"), numbering)
                 };
             });
         },
@@ -133,14 +146,37 @@ function BodyReader(options) {
             return readXmlElements(element.children)
                 .map(function(children) {
                     var properties = _.find(children, isRunProperties);
-                    
-                    return new documents.Run(
-                        children.filter(negate(isRunProperties)),
-                        properties
-                    );
+                    children = children.filter(negate(isRunProperties));
+
+                    var scopedHyperlink = scopedComplexFieldOfType(complexFields, "hyperlink");
+                    if (scopedHyperlink !== undefined) {
+                        children = [new documents.Hyperlink(children, {href: scopedHyperlink.href})];
+                    }
+
+                    return new documents.Run(children, properties);
                 });
         },
         "w:rPr": readRunProperties,
+        "w:fldChar": function(element) {
+            var type = element.attributes["w:fldCharType"];
+            if (type === "begin") {
+                currentInstrText = [];
+            } else if (type === "end") {
+                complexFields.pop();
+            } else if (type === "separate") {
+                var href = parseHyperlinkFieldCode(currentInstrText.join(''));
+                if (href !== null) {
+                    complexFields.push({type: "hyperlink", href: href});
+                } else {
+                    complexFields.push(null);
+                }
+            }
+            return emptyResult();
+        },
+        "w:instrText": function(element) {
+            currentInstrText.push(element.text());
+            return emptyResult();
+        },
         "w:t": function(element) {
             return elementResult(new documents.Text(element.text()));
         },
@@ -210,21 +246,12 @@ function BodyReader(options) {
         "wp:anchor": readDrawingElement,
         "v:imagedata": readImageData
     };
+    
     return {
         readXmlElement: readXmlElement,
-        readXmlElements: readXmlElements,
-        _readNumberingProperties: readNumberingProperties
+        readXmlElements: readXmlElements
     };
 
-    function readNumberingProperties(element) {
-        var level = element.firstOrEmpty("w:ilvl").attributes["w:val"];
-        var numId = element.firstOrEmpty("w:numId").attributes["w:val"];
-        if (level === undefined || numId === undefined) {
-            return null;
-        } else {
-            return numbering.findLevel(numId, level);
-        }
-    }
     
     function readTable(element) {
         return readXmlElements(element.children)
@@ -379,6 +406,17 @@ function BodyReader(options) {
             type + " style with ID " + styleId + " was referenced but not defined in the document");
     }
 }
+
+
+function readNumberingProperties(element, numbering) {
+    var level = element.firstOrEmpty("w:ilvl").attributes["w:val"];
+    var numId = element.firstOrEmpty("w:numId").attributes["w:val"];
+    if (level === undefined || numId === undefined) {
+        return null;
+    } else {
+        return numbering.findLevel(numId, level);
+    }
+}
     
 var supportedImageTypes = {
     "image/png": true,
@@ -491,3 +529,22 @@ function joinElements(first, second) {
 function identity(value) {
     return value;
 }
+
+function parseHyperlinkFieldCode(code) {
+    var result = /\s*HYPERLINK "(.*)"/.exec(code);
+    if (result) {
+        return result[1];
+    } else {
+        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/lib/docx/docx-reader.js b/lib/docx/docx-reader.js
index 0c7d57f..5aa95ca 100644
--- a/lib/docx/docx-reader.js
+++ b/lib/docx/docx-reader.js
@@ -7,7 +7,7 @@ var documents = require("../documents");
 var Result = require("../results").Result;
 
 var readXmlFromZipFile = require("./office-xml-reader").readXmlFromZipFile;
-var BodyReader = require("./body-reader").BodyReader;
+var createBodyReader = require("./body-reader").createBodyReader;
 var DocumentXmlReader = require("./document-xml-reader").DocumentXmlReader;
 var relationshipsReader = require("./relationships-reader");
 var contentTypesReader = require("./content-types-reader");
@@ -95,7 +95,7 @@ function readXmlFileWithBody(name, options, func) {
     });
     
     return readRelationshipsFromZipFile(options.docxFile).then(function(relationships) {
-        var bodyReader = new BodyReader({
+        var bodyReader = new createBodyReader({
             relationships: relationships,
             contentTypes: options.contentTypes,
             docxFile: options.docxFile,
diff --git a/test/docx/body-reader.tests.js b/test/docx/body-reader.tests.js
index f605c2d..a670202 100644
--- a/test/docx/body-reader.tests.js
+++ b/test/docx/body-reader.tests.js
@@ -3,6 +3,7 @@ var path = require("path");
 
 var _ = require("underscore");
 var hamjest = require("hamjest");
+var assertThat = hamjest.assertThat;
 var promiseThat = hamjest.promiseThat;
 var allOf = hamjest.allOf;
 var contains = hamjest.contains;
@@ -10,7 +11,14 @@ var hasProperties = hamjest.hasProperties;
 var willBe = hamjest.willBe;
 var FeatureMatcher = hamjest.FeatureMatcher;
 
-var BodyReader = require("../../lib/docx/body-reader").BodyReader;
+var documentMatchers = require("./document-matchers");
+var isEmptyRun = documentMatchers.isEmptyRun;
+var isHyperlink = documentMatchers.isHyperlink;
+var isRun = documentMatchers.isRun;
+var isText = documentMatchers.isText;
+
+var createBodyReader = require("../../lib/docx/body-reader").createBodyReader;
+var _readNumberingProperties = require("../../lib/docx/body-reader")._readNumberingProperties;
 var documents = require("../../lib/documents");
 var xml = require("../../lib/xml");
 var XmlElement = xml.Element;
@@ -25,7 +33,7 @@ var createFakeDocxFile = testing.createFakeDocxFile;
 function readXmlElement(element, options) {
     options = Object.create(options || {});
     options.styles = options.styles || new Styles({}, {});
-    return new BodyReader(options).readXmlElement(element);
+    return createBodyReader(options).readXmlElement(element);
 }
 
 function readXmlElementValue(element, options) {
@@ -106,10 +114,7 @@ test("numbering properties are converted to numbering at specified level", funct
     
     var numbering = new Numbering({"42": {"1": {isOrdered: true, level: "1"}}});
     
-    var reader = new BodyReader({
-        numbering: numbering
-    });
-    var numberingLevel = reader._readNumberingProperties(numberingPropertiesXml);
+    var numberingLevel = _readNumberingProperties(numberingPropertiesXml, numbering);
     assert.deepEqual(numberingLevel, {level: "1", isOrdered: true});
 });
 
@@ -120,10 +125,7 @@ test("numbering properties are ignored if w:ilvl is missing", function() {
     
     var numbering = new Numbering({"42": {"1": {isOrdered: true, level: "1"}}});
     
-    var reader = new BodyReader({
-        numbering: numbering
-    });
-    var numberingLevel = reader._readNumberingProperties(numberingPropertiesXml);
+    var numberingLevel = _readNumberingProperties(numberingPropertiesXml, numbering);
     assert.equal(numberingLevel, null);
 });
 
@@ -134,13 +136,231 @@ test("numbering properties are ignored if w:numId is missing", function() {
     
     var numbering = new Numbering({"42": {"1": {isOrdered: true, level: "1"}}});
     
-    var reader = new BodyReader({
-        numbering: numbering
-    });
-    var numberingLevel = reader._readNumberingProperties(numberingPropertiesXml);
+    var numberingLevel = _readNumberingProperties(numberingPropertiesXml, numbering);
     assert.equal(numberingLevel, null);
 });
 
+test("complex fields", (function() {
+    var uri = "http://example.com";
+    var beginXml = new XmlElement("w:r", {}, [
+        new XmlElement("w:fldChar", {"w:fldCharType": "begin"})
+    ]);
+    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() {
+            var instrText = readXmlElementValue(hyperlinkInstrText);
+            assert.deepEqual(instrText, []);
+        },
+        
+        "runs in a complex field for hyperlinks are read as hyperlinks": function() {
+            var hyperlinkRunXml = runOfText("this is a hyperlink");
+            var paragraphXml = new XmlElement("w:p", {}, [
+                beginXml,
+                hyperlinkInstrText,
+                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
+            ));
+        },
+        
+        "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", {}, [
+                beginXml,
+                hyperlinkInstrText,
+                separateXml,
+                endXml,
+                afterEndXml
+            ]);
+            var paragraph = readXmlElementValue(paragraphXml);
+            
+            assertThat(paragraph.children, contains(
+                isEmptyRun,
+                isRun({
+                    children: contains(
+                        isHyperlink({
+                            href: uri,
+                            children: []
+                        })
+                    )
+                }),
+                isEmptyRun,
+                isRun({
+                    children: contains(
+                        isText("this will not be a hyperlink")
+                    )
+                })
+            ));
+        },
+
+        "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
+          ));
+        }
+    };
+})());
+
 test("run has no style if it has no properties", function() {
     var runXml = runWithProperties([]);
     var run = readXmlElementValue(runXml);
@@ -861,6 +1081,11 @@ function createLinkedBlip(relationshipId) {
     return new XmlElement("a:blip", {"r:link": relationshipId});
 }
 
+function runOfText(text) {
+    var textXml = new XmlElement("w:t", {}, [xml.text(text)]);
+    return new XmlElement("w:r", {}, [textXml]);
+}
+
 function assertImageBuffer(element, expectedImageBuffer) {
     return element.read()
         .then(function(readValue) {
diff --git a/test/docx/comments-reader.tests.js b/test/docx/comments-reader.tests.js
index 4bb0e9b..0b3c0ce 100644
--- a/test/docx/comments-reader.tests.js
+++ b/test/docx/comments-reader.tests.js
@@ -1,14 +1,14 @@
 var assert = require("assert");
 
 var createCommentsReader = require("../../lib/docx/comments-reader").createCommentsReader;
-var BodyReader = require("../../lib/docx/body-reader").BodyReader;
+var createBodyReader = require("../../lib/docx/body-reader").createBodyReader;
 var documents = require("../../lib/documents");
 var xml = require("../../lib/xml");
 var test = require("../test")(module);
 
 
 function readComment(element) {
-    var bodyReader = new BodyReader({});
+    var bodyReader = createBodyReader({});
     var commentsReader = createCommentsReader(bodyReader);
     var comments = commentsReader(element);
     assert.equal(comments.value.length, 1);
diff --git a/test/docx/document-matchers.js b/test/docx/document-matchers.js
new file mode 100644
index 0000000..b750bed
--- /dev/null
+++ b/test/docx/document-matchers.js
@@ -0,0 +1,27 @@
+var hamjest = require("hamjest");
+var _ = require("underscore");
+
+var documents = require("../../lib/documents");
+
+
+exports.isEmptyRun = isRun({children: []});
+exports.isRun = isRun;
+exports.isText = isText;
+exports.isHyperlink = isHyperlink;
+
+
+function isRun(properties) {
+    return isDocumentElement(documents.types.run, properties);
+}
+
+function isText(text) {
+    return isDocumentElement(documents.types.text, {value: text});
+}
+
+function isHyperlink(properties) {
+    return isDocumentElement(documents.types.hyperlink, properties);
+}
+
+function isDocumentElement(type, properties) {
+    return hamjest.hasProperties(_.extend({type: hamjest.equalTo(type)}, properties));
+}
diff --git a/test/docx/notes-reader.tests.js b/test/docx/notes-reader.tests.js
index 1d4cf9c..71bb794 100644
--- a/test/docx/notes-reader.tests.js
+++ b/test/docx/notes-reader.tests.js
@@ -1,14 +1,14 @@
 var assert = require("assert");
 
 var createFootnotesReader = require("../../lib/docx/notes-reader").createFootnotesReader;
-var BodyReader = require("../../lib/docx/body-reader").BodyReader;
+var createBodyReader = require("../../lib/docx/body-reader").createBodyReader;
 var documents = require("../../lib/documents");
 var XmlElement = require("../../lib/xml").Element;
 var test = require("../test")(module);
 
 
 test('ID and body of footnote are read', function() {
-    var bodyReader = new BodyReader({});
+    var bodyReader = new createBodyReader({});
     var footnoteBody = [new XmlElement("w:p", {}, [])];
     var footnotes = createFootnotesReader(bodyReader)(
         new XmlElement("w:footnotes", {}, [