From bd8b2786247308c7bd82dc3cda627a23b4898958 Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Mon, 1 Apr 2013 17:35:15 +0100 Subject: Implement reading of hyperlink elements --- lib/docx-reader.js | 203 +++++++++++++-------- tests/docx-reader.tests.js | 30 +-- .../hyperlinks/word/_rels/document.xml.rels | 10 + tests/test-data/hyperlinks/word/document.xml | 18 ++ tests/testing.js | 9 +- 5 files changed, 177 insertions(+), 93 deletions(-) create mode 100644 tests/test-data/hyperlinks/word/_rels/document.xml.rels create mode 100644 tests/test-data/hyperlinks/word/document.xml diff --git a/lib/docx-reader.js b/lib/docx-reader.js index 3e4054f..71707ff 100644 --- a/lib/docx-reader.js +++ b/lib/docx-reader.js @@ -1,5 +1,5 @@ exports.read = read; -exports.readXmlElement = readXmlElement; +exports.DocumentReader = DocumentReader; var q = require("q"); var _ = require("underscore"); @@ -18,101 +18,144 @@ function read(docxFile) { .then(function(value) { return value.toString(); }); + }, + exists: function(path) { + return zipFile.names.indexOf(path) !== -1; } }; } - return docxFile.read("word/document.xml") - .then(xmlreader.read) - .then(convertXmlToDocument); -} - -function convertXmlToDocument(documentXml) { - var body = documentXml.root.first("w:body"); - - var result = readXmlElements(body.children) - .map(function(children) { - return new documents.Document(children); - }); - result.document = result.value; - return result; + return q.all([ + readXmlFile(docxFile, "word/_rels/document.xml.rels"), + readXmlFile(docxFile, "word/document.xml") + ]).spread(function(relationshipsXml, documentXml) { + var relationships = relationshipsXml ? readRelationships(relationshipsXml) : {}; + var reader = new DocumentReader(relationships); + return reader.convertXmlToDocument(documentXml); + }); } -function readXmlElements(elements) { - var results = elements.map(readXmlElement); - var values = _.pluck(results, "value").filter(notNull); - var messages = _.flatten(_.pluck(results, "messages"), true); - return new Result(values, messages); +function readXmlFile(docxFile, path) { + if (docxFile.exists(path)) { + return docxFile.read(path) + .then(xmlreader.read); + } else { + return null; + } } -function readXmlElement(element) { - if (element.type === "element") { - var handler = xmlElementReaders[element.name]; - if (handler) { - return handler(element); - } else if (!Object.prototype.hasOwnProperty.call(ignoreElements, element.name)) { - return new Result( - null, - [warning("An unrecognised element was ignored: " + element.name)] - ); +function readRelationships(relationshipsXml) { + var relationships = {}; + relationshipsXml.root.children.forEach(function(child) { + if (child.name === "Relationship") { + relationships[child.attributes.Id] = { + target: child.attributes.Target + }; } - } - return new Result(null); + }); + return relationships; } -var xmlElementReaders = { - "w:p": function(element) { - return readXmlElements(element.children) - .map(function(children) { - var properties = _.find(children, isParagraphProperties); - - return new documents.Paragraph( - children.filter(negate(isParagraphProperties)), - properties - ); - }); - }, - "w:pPr": function(element) { - var properties = { - type: "paragraphProperties" - }; - - var styleElement = element.first("w:pStyle"); - if (styleElement) { - properties.styleName = styleElement.attributes["w:val"]; - } +function DocumentReader(relationships) { + function convertXmlToDocument(documentXml) { + var body = documentXml.root.first("w:body"); - return new Result(properties); - }, - "w:r": function(element) { - return readXmlElements(element.children) + var result = readXmlElements(body.children) .map(function(children) { - var properties = _.find(children, isRunProperties); - - return new documents.Run( - children.filter(negate(isRunProperties)), - properties - ); + return new documents.Document(children); }); - }, - "w:rPr": function(element) { - var properties = { - type: "runProperties" - }; - - var styleElement = element.first("w:rStyle"); - if (styleElement) { - properties.styleName = styleElement.attributes["w:val"]; + result.document = result.value; + return result; + } + + function readXmlElements(elements) { + var results = elements.map(readXmlElement); + var values = _.pluck(results, "value").filter(notNull); + var messages = _.flatten(_.pluck(results, "messages"), true); + return new Result(values, messages); + } + + function readXmlElement(element) { + if (element.type === "element") { + var handler = xmlElementReaders[element.name]; + if (handler) { + return handler(element); + } else if (!Object.prototype.hasOwnProperty.call(ignoreElements, element.name)) { + return new Result( + null, + [warning("An unrecognised element was ignored: " + element.name)] + ); + } } - properties.isBold = !!element.first("w:b"); - properties.isItalic = !!element.first("w:i"); - - return new Result(properties); - }, - "w:t": function(element) { - return new Result(new documents.Text(element.children[0].value)); + return new Result(null); } -}; + + var xmlElementReaders = { + "w:p": function(element) { + return readXmlElements(element.children) + .map(function(children) { + var properties = _.find(children, isParagraphProperties); + + return new documents.Paragraph( + children.filter(negate(isParagraphProperties)), + properties + ); + }); + }, + "w:pPr": function(element) { + var properties = { + type: "paragraphProperties" + }; + + var styleElement = element.first("w:pStyle"); + if (styleElement) { + properties.styleName = styleElement.attributes["w:val"]; + } + + return new Result(properties); + }, + "w:r": function(element) { + return readXmlElements(element.children) + .map(function(children) { + var properties = _.find(children, isRunProperties); + + return new documents.Run( + children.filter(negate(isRunProperties)), + properties + ); + }); + }, + "w:rPr": function(element) { + var properties = { + type: "runProperties" + }; + + var styleElement = element.first("w:rStyle"); + if (styleElement) { + properties.styleName = styleElement.attributes["w:val"]; + } + properties.isBold = !!element.first("w:b"); + properties.isItalic = !!element.first("w:i"); + + return new Result(properties); + }, + "w:t": function(element) { + return new Result(new documents.Text(element.children[0].value)); + }, + "w:hyperlink": function(element) { + var relationshipId = element.attributes["r:id"]; + var href = relationships[relationshipId].target; + return readXmlElements(element.children) + .map(function(children) { + return new documents.Hyperlink(children, {href: href}); + }); + } + }; + return { + convertXmlToDocument: convertXmlToDocument, + readXmlElement: readXmlElement + }; +} var ignoreElements = { "w:bookmarkStart": true, diff --git a/tests/docx-reader.tests.js b/tests/docx-reader.tests.js index ef880f8..68efb0a 100644 --- a/tests/docx-reader.tests.js +++ b/tests/docx-reader.tests.js @@ -37,48 +37,56 @@ describe("docx-reader", function() { }); }); - test("paragraph properties are not included as child of paragraph", function() { + test("hyperlink hrefs are read from relationships file", function() { var docxFile = createFakeDocxFile({ - "word/document.xml": testData("paragraphStyles/word/document.xml") + "word/document.xml": testData("hyperlinks/word/document.xml"), + "word/_rels/document.xml.rels": testData("hyperlinks/word/_rels/document.xml.rels") }); return docxReader.read(docxFile).then(function(result) { var paragraph = result.document.children[0]; assert.equal(1, paragraph.children.length); + var hyperlink = paragraph.children[0]; + assert.equal(hyperlink.href, "http://www.example.com"); + assert.equal(hyperlink.children.length, 1); }); }); }); +function readXmlElement(element) { + return new docxReader.DocumentReader({}).readXmlElement(element); +} + describe("readElement", function() { test("reads styles from run properties", function() { var runStyleXml = new XmlElement("w:rStyle", {"w:val": "Emphasis"}); var runPropertiesXml = createRunPropertiesXml([runStyleXml]); - var result = docxReader.readXmlElement(runPropertiesXml); + var result = readXmlElement(runPropertiesXml); assert.equal(result.value.styleName, "Emphasis"); }); test("isBold is false if bold element is not present", function() { var runPropertiesXml = createRunPropertiesXml([]); - var result = docxReader.readXmlElement(runPropertiesXml); + var result = readXmlElement(runPropertiesXml); assert.equal(result.value.isBold, false); }); test("isBold is true if bold element is present", function() { var boldXml = new XmlElement("w:b"); var runPropertiesXml = createRunPropertiesXml([boldXml]); - var result = docxReader.readXmlElement(runPropertiesXml); + var result = readXmlElement(runPropertiesXml); assert.equal(result.value.isBold, true); }); test("isItalic is false if bold element is not present", function() { var runPropertiesXml = createRunPropertiesXml([]); - var result = docxReader.readXmlElement(runPropertiesXml); + var result = readXmlElement(runPropertiesXml); assert.equal(result.value.isItalic, false); }); test("isItalic is true if bold element is present", function() { var italicXml = new XmlElement("w:i"); var runPropertiesXml = createRunPropertiesXml([italicXml]); - var result = docxReader.readXmlElement(runPropertiesXml); + var result = readXmlElement(runPropertiesXml); assert.equal(result.value.isItalic, true); }); @@ -86,7 +94,7 @@ describe("readElement", function() { var runStyleXml = new XmlElement("w:rStyle", {"w:val": "Emphasis"}); var runPropertiesXml = new XmlElement("w:rPr", {}, [runStyleXml]); var runXml = new XmlElement("w:r", {}, [runPropertiesXml]); - var result = docxReader.readXmlElement(runXml); + var result = readXmlElement(runXml); assert.equal(result.value.properties.styleName, "Emphasis"); }); @@ -94,13 +102,13 @@ describe("readElement", function() { var runStyleXml = new XmlElement("w:rStyle", {"w:val": "Emphasis"}); var runPropertiesXml = new XmlElement("w:rPr", {}, [runStyleXml]); var runXml = new XmlElement("w:r", {}, [runPropertiesXml]); - var result = docxReader.readXmlElement(runXml); + var result = readXmlElement(runXml); assert.deepEqual(result.value.children, []); }); test("emits warning on unrecognised element", function() { var unrecognisedElement = new XmlElement("w:not-an-element"); - var result = docxReader.readXmlElement(unrecognisedElement); + var result = readXmlElement(unrecognisedElement); assert.deepEqual( result.messages, [{ @@ -112,7 +120,7 @@ describe("readElement", function() { test("w:bookmarkStart is ignored without warning", function() { var ignoredElement = new XmlElement("w:bookmarkStart"); - var result = docxReader.readXmlElement(ignoredElement); + var result = readXmlElement(ignoredElement); assert.deepEqual(result.messages, []); assert.equal(null, result.value); }); diff --git a/tests/test-data/hyperlinks/word/_rels/document.xml.rels b/tests/test-data/hyperlinks/word/_rels/document.xml.rels new file mode 100644 index 0000000..baa4102 --- /dev/null +++ b/tests/test-data/hyperlinks/word/_rels/document.xml.rels @@ -0,0 +1,10 @@ + + + + + + + + + + \ No newline at end of file diff --git a/tests/test-data/hyperlinks/word/document.xml b/tests/test-data/hyperlinks/word/document.xml new file mode 100644 index 0000000..f1ce575 --- /dev/null +++ b/tests/test-data/hyperlinks/word/document.xml @@ -0,0 +1,18 @@ + + + + + + + coconuts + + + + + + + + + + + diff --git a/tests/testing.js b/tests/testing.js index f2cf55a..af8ec94 100644 --- a/tests/testing.js +++ b/tests/testing.js @@ -23,10 +23,15 @@ function testData(testDataPath) { function createFakeDocxFile(files) { function read(path) { - return files[path]; + return q.when(files[path]); + } + + function exists(path) { + return !!files[path]; } return { - read: read + read: read, + exists: exists }; } -- cgit v1.2.3