+ normalizeXML: function(nativeNode) {
+ var doc = this,
+ prefixLength = 'dc:'.length;
+
+ $(nativeNode).find('metadata').each(function() {
+ var metadataNode = $(this),
+ owner = doc.createDocumentNode(metadataNode.parent()[0]),
+ metadata = owner.getMetadata();
+
+ metadataNode.children().each(function() {
+ metadata.add({key: (this.tagName).toLowerCase().substr(prefixLength), value: $(this).text()}, {undoable: false});
+ });
+ metadataNode.remove();
+ });
+ nativeNode.normalize();
+
+ $(nativeNode).find(':not(iframe)').addBack().contents()
+ .filter(function() {return this.nodeType === Node.TEXT_NODE;})
+ .each(function() {
+ var el = $(this),
+ text = {original: el.text(), trimmed: $.trim(el.text())},
+ elParent = el.parent(),
+ hasSpanParent = elParent.prop('tagName') === 'SPAN',
+ hasSpanBefore = el.prev().length && $(el.prev()).prop('tagName') === 'SPAN',
+ hasSpanAfter = el.next().length && $(el.next()).prop('tagName') === 'SPAN';
+
+
+ var addInfo = function(toAdd, where, transformed, original) {
+ var parentContents = elParent.contents(),
+ idx = parentContents.index(el[0]),
+ prev = idx > 0 ? parentContents[idx-1] : null,
+ next = idx < parentContents.length - 1 ? parentContents[idx+1] : null,
+ target, key;
+
+ if(where === 'above') {
+ target = prev ? $(prev) : elParent;
+ key = prev ? 'orig_after' : 'orig_begin';
+ } else if(where === 'below') {
+ target = next ? $(next) : elParent;
+ key = next ? 'orig_before' : 'orig_end';
+ } else { throw new Error();}
+
+ target.data(formatter_prefix + key, toAdd);
+ if(transformed !== undefined) {
+ target.data(formatter_prefix + key + '_transformed', transformed);
+ }
+ if(original !== undefined) {
+ target.data(formatter_prefix + key + '_original', original);
+ }
+ };
+
+ text.transformed = text.trimmed;
+
+ if(hasSpanParent || hasSpanBefore || hasSpanAfter) {
+ var startSpace = /\s/g.test(text.original.substr(0,1)),
+ endSpace = /\s/g.test(text.original.substr(-1)) && text.original.length > 1;
+ text.transformed = (startSpace && (hasSpanParent || hasSpanBefore) ? ' ' : '');
+ text.transformed += text.trimmed;
+ text.transformed += (endSpace && (hasSpanParent || hasSpanAfter) ? ' ' : '');
+ } else {
+ if(text.trimmed.length === 0 && text.original.length > 0 && elParent.contents().length === 1) {
+ text.transformed = ' ';
+ }
+ }