aboutsummaryrefslogtreecommitdiff
path: root/lib/document-xml-reader.js
diff options
context:
space:
mode:
authorMichael Williamson <mike@zwobble.org>2013-04-08 20:16:05 +0100
committerMichael Williamson <mike@zwobble.org>2013-04-08 20:16:05 +0100
commit2f72e903f5f8bb7ac00c69778be4cb199017617d (patch)
tree0e24cac9cd0f6f3bddc72d6c32f4a2d9192f53dc /lib/document-xml-reader.js
parent2ea1d69fe5821a1e4f28e7c411e5bf41a498b2fa (diff)
Move DocumentReader to DocumentXmlReader in its own module
Diffstat (limited to 'lib/document-xml-reader.js')
-rw-r--r--lib/document-xml-reader.js158
1 files changed, 158 insertions, 0 deletions
diff --git a/lib/document-xml-reader.js b/lib/document-xml-reader.js
new file mode 100644
index 0000000..ea15204
--- /dev/null
+++ b/lib/document-xml-reader.js
@@ -0,0 +1,158 @@
+exports.DocumentXmlReader = DocumentXmlReader;
+
+var q = require("q");
+var _ = require("underscore");
+var path = require("path");
+
+var documents = require("./documents");
+var Result = require("./results").Result;
+var warning = require("./results").warning;
+
+
+function DocumentXmlReader(relationships, docxFile) {
+ 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;
+ }
+
+ 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)]
+ );
+ }
+ }
+ 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});
+ });
+ },
+
+ "w:drawing": function(element) {
+ return readXmlElement(element.first("wp:inline"));
+ },
+ "wp:inline": function(element) {
+ var relationshipId = element
+ .first("a:graphic")
+ .first("a:graphicData")
+ .first("pic:pic")
+ .first("pic:blipFill")
+ .first("a:blip")
+ .attributes["r:embed"];
+ var imagePath = relationships[relationshipId].target;
+ var readImage = function() {
+ return docxFile.read(path.join("word", imagePath));
+ };
+ var altText = element.first("wp:docPr").attributes.descr;
+
+ var image = documents.Image(readImage, altText)
+ return new Result(image);
+ }
+ };
+ return {
+ convertXmlToDocument: convertXmlToDocument,
+ readXmlElement: readXmlElement
+ };
+}
+
+var ignoreElements = {
+ "w:bookmarkStart": true,
+ "w:bookmarkEnd": true,
+ "w:sectPr": true,
+ "w:proofErr": true,
+ "w:lastRenderedPageBreak": true
+};
+
+function notNull(value) {
+ return value !== null;
+}
+
+function isParagraphProperties(element) {
+ return element.type === "paragraphProperties";
+}
+
+function isRunProperties(element) {
+ return element.type === "runProperties";
+}
+
+function negate(predicate) {
+ return function(value) {
+ return !predicate(value);
+ };
+}