aboutsummaryrefslogtreecommitdiff
path: root/lib
diff options
context:
space:
mode:
authorMichael Williamson <mike@zwobble.org>2013-04-01 17:35:15 +0100
committerMichael Williamson <mike@zwobble.org>2013-04-01 17:35:15 +0100
commitbd8b2786247308c7bd82dc3cda627a23b4898958 (patch)
treee30fc62c22bbb0dc358155c5c0caebb126688091 /lib
parent25109ca7e318e188229cac92552ba0dcd7e897a1 (diff)
Implement reading of hyperlink elements
Diffstat (limited to 'lib')
-rw-r--r--lib/docx-reader.js203
1 files changed, 123 insertions, 80 deletions
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,