/// <summary>
/// Returns the text value of a node
/// </summary>
/// <param name="node">HtmlAgilityPack html node</param>
private string GetNodeValue(HtmlNode node)
{
return(UfHelpers.HtmlToText(node, false));
}
private void ParseUfElementValue(HtmlNode baseNode, UfElementDescriber ufElement, UfDataNode ufData)
{
// Create a single data node for whatever data insertion is needed.
UfDataNode ufd = new UfDataNode();
if (ufElement.CompoundName != string.Empty)
{
ufd.ParentNodeNames = ufData.ParentNodeNames + ufElement.CompoundName + " ";
}
else
{
ufd.ParentNodeNames = ufData.ParentNodeNames + ufElement.Name + " ";
}
ufd.ElementId = GetAttributeValue(baseNode, "id");
// A parent node in the data schema
if (ufElement.Elements.Count > 0)
{
if (ufElement.CompoundName == string.Empty)
{
// Add a emtpy structural node
ufd.Name = ufElement.Name;
}
else
{
// This is for compound structures, ie reviewer in hreview is a hcard
// Need to find a second attribute value to do this
HtmlAttribute att = baseNode.Attributes[ufElement.CompoundAttribute];
if (att != null)
{
if (UfHelpers.FindAttributeValue(att.Value.ToLower(), ufElement.CompoundName))
{
// Add a emtpy structural node using compound name
ufd.Name = ufElement.CompoundName;
}
}
}
// Recursion through the dom structure
foreach (UfElementDescriber ufChildElement in ufElement.Elements)
{
ParseUfElement(baseNode, ufChildElement, ufd, false);
}
}
// A value needs to be found
if (ufElement.Type != UfElementDescriber.PropertyTypes.None)
{
// Find child nodes with "value" or "value-title" classes
HtmlNodeCollection valueNodes = null;
HtmlNodeCollection valueTitleNodes = null;
// The value pattern
if (ufElement.Elements["value"] == null && ufElement.Name != "value")
{
valueNodes = baseNode.SelectNodes(".//*[contains(concat(' ', @class, ' '),' value ')]");
}
// The value-title pattern is only allow for some property types ie dates
// or name properties ie type, duration, geo, latitude and longitude
if (ufElement.Type == UfElementDescriber.PropertyTypes.Date ||
ufElement.Name == "type" ||
ufElement.Name == "duration" ||
ufElement.Name == "geo" ||
ufElement.Name == "latitude" ||
ufElement.Name == "longitude")
{
valueTitleNodes = baseNode.SelectNodes(".//*[contains(concat(' ', @class, ' '),' value-title ')]");
}
if (ufElement.Type == UfElementDescriber.PropertyTypes.UrlTextAttribute || ufElement.Type == UfElementDescriber.PropertyTypes.UrlTextTag || ufElement.Type == UfElementDescriber.PropertyTypes.UrlText)
{
string text = UfHelpers.HtmlToText(baseNode, false);
string link = UfHelpers.GetAbsoluteUrl(GetAttributeValue(baseNode, "href"), this.baseUrl, url);
string att = GetAttributeValue(baseNode, ufElement.Attribute);
ufd.Name = ufElement.Name;
UfDataNode ufd1 = new UfDataNode();
UfDataNode ufd2 = new UfDataNode();
UfDataNode ufd3 = new UfDataNode();
ufd1.Name = "text";
ufd1.Value = text;
ufd.Nodes.Add(ufd1);
ufd2.Name = "link";
ufd2.Value = link;
ufd.Nodes.Add(ufd2);
// Add the attribute value used for XFN like structures
if (ufElement.Type == UfElementDescriber.PropertyTypes.UrlTextAttribute)
{
ufd3.Name = ufElement.Attribute;
ufd3.Value = att;
ufd.Nodes.Add(ufd3);
}
// Add the tag element of the url
if (ufElement.Type == UfElementDescriber.PropertyTypes.UrlTextTag)
{
ufd3.Name = "tag";
ufd3.Value = UfHelpers.GetTagFromUrl(link);
ufd.Nodes.Add(ufd3);
}
if (ufElement.CompoundName == string.Empty)
{
ufData.Nodes.Add(ufd);
}
else
{
HtmlAttribute att1 = baseNode.Attributes[ufElement.CompoundAttribute];
if (att1 != null)
{
if (UfHelpers.FindAttributeValue(att1.Value.ToLower(), ufElement.CompoundName))
{
ufd.Name = ufElement.CompoundName;
ufData.Nodes.Add(ufd);
}
}
}
}
// The value excerpting pattern
else if (valueNodes != null)
{
string text = string.Empty;
foreach (HtmlNode node in valueNodes)
{
if (node.Name == "img" || node.Name == "area")
{
if (ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
text += GetAttributeValue(node, "title").Replace(" ", "") + " ";
}
else
{
text += GetAttributeValue(node, "title");
}
}
else if (node.Name == "abbr")
{
if (ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
text += GetAttributeValue(node, "title").Replace(" ", "") + " ";
}
else
{
text += GetAttributeValue(node, "title");
}
}
else
{
if (ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
text += UfHelpers.HtmlToText(node, false).Replace(" ", "") + " ";
}
else
{
text += UfHelpers.HtmlToText(node, false) + " ";
}
}
}
if (ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
// Take the fagmented bits and create a true ISODateTime string
ISODateTime isoDateTime = new ISODateTime();
text = isoDateTime.ParseUFFragmented(text);
}
ufd.Name = ufElement.Name;
ufd.Value = text.Trim();
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// The value-title excerpting pattern
else if (valueTitleNodes != null)
{
string text = GetAttributeValue(valueTitleNodes[0], "title");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Url from "a" or "link"
else if ((baseNode.Name == "a" || baseNode.Name == "link") && GetAttributeValue(baseNode, "href") != string.Empty && ufElement.Type == UfElementDescriber.PropertyTypes.Url)
{
string link = UfHelpers.GetAbsoluteUrl(GetAttributeValue(baseNode, "href"), this.baseUrl, url);
ufd.Name = ufElement.Name;
ufd.Value = link;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Url from "img"
else if ((baseNode.Name == "img" || baseNode.Name == "area") && GetAttributeValue(baseNode, "src") != string.Empty && ufElement.Type == UfElementDescriber.PropertyTypes.Url)
{
string link = UfHelpers.GetAbsoluteUrl(GetAttributeValue(baseNode, "src"), this.baseUrl, url);
ufd.Name = ufElement.Name;
ufd.Value = link;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Email from "a" or "link"
else if (baseNode.Name == "a" && GetAttributeValue(baseNode, "href") != string.Empty && ufElement.Type == UfElementDescriber.PropertyTypes.Email)
{
string address = UfHelpers.CleanEmailAddress(GetAttributeValue(baseNode, "href"));
ufd.Name = ufElement.Name;
ufd.Value = address;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Tel from "object"
else if (baseNode.Name == "object" && (GetAttributeValue(baseNode, "data") != "") && ufElement.Name == "tel")
{
UfHelpers.TelOptimization(ufd, GetAttributeValue(baseNode, "data"));
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Date from "time"
else if (baseNode.Name == "time" && GetAttributeValue(baseNode, "datetime") != "" && ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
string text = GetAttributeValue(baseNode, "datetime");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Date from "abbr"
else if (baseNode.Name == "abbr" && GetAttributeValue(baseNode, "title") != string.Empty && ufElement.Type == UfElementDescriber.PropertyTypes.Date)
{
string text = GetAttributeValue(baseNode, "title");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Text from "abbr"
else if (baseNode.Name == "abbr" || baseNode.Name == "acronym" && GetAttributeValue(baseNode, "title") != string.Empty)
{
string text = GetAttributeValue(baseNode, "title");
ufd.Name = ufElement.Name;
// This is for geo been used as a location in hcalandar
if (ufElement.CompoundName != string.Empty)
{
ufd.Name = ufElement.CompoundName;
}
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Text from "input"
else if (baseNode.Name == "input" && GetAttributeValue(baseNode, "value") != string.Empty)
{
string text = GetAttributeValue(baseNode, "value");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Tel from "area"
else if (baseNode.Name == "area" && (GetAttributeValue(baseNode, "href") != "") && ufElement.Name == "tel")
{
UfHelpers.TelOptimization(ufd, GetAttributeValue(baseNode, "href"));
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Text and url from "area"
else if (baseNode.Name == "area" && (GetAttributeValue(baseNode, "href") != string.Empty || GetAttributeValue(baseNode, "alt") != string.Empty))
{
if ((ufElement.Type == UfElementDescriber.PropertyTypes.Url || ufElement.Type == UfElementDescriber.PropertyTypes.Email) && GetAttributeValue(baseNode, "href") != string.Empty)
{
string text = GetAttributeValue(baseNode, "href");
if (ufElement.Type == UfElementDescriber.PropertyTypes.Email)
{
text = UfHelpers.CleanEmailAddress(text);
}
if (ufElement.Type == UfElementDescriber.PropertyTypes.Url)
{
text = UfHelpers.GetAbsoluteUrl(text, this.baseUrl, url);
}
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
else if (GetAttributeValue(baseNode, "alt") != string.Empty)
{
string text = GetAttributeValue(baseNode, "alt");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
}
// Url/Image from "object"
else if (baseNode.Name == "object" && GetAttributeValue(baseNode, "data") != string.Empty && (ufElement.Type == UfElementDescriber.PropertyTypes.Url || ufElement.Type == UfElementDescriber.PropertyTypes.Image))
{
string text = UfHelpers.GetAbsoluteUrl(GetAttributeValue(baseNode, "data"), this.baseUrl, url);
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Image from "img" or "area"
else if ((baseNode.Name == "img" || baseNode.Name == "area") && GetAttributeValue(baseNode, "src") != string.Empty && ufElement.Type == UfElementDescriber.PropertyTypes.Image)
{
string text = UfHelpers.GetAbsoluteUrl(GetAttributeValue(baseNode, "src"), this.baseUrl, url);
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Text from "img" longdesc attribute
else if (baseNode.Name == "img" && GetAttributeValue(baseNode, "longdesc") != string.Empty)
{
string text = GetAttributeValue(baseNode, "longdesc");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);;
}
// Text from "img" alt attribute
else if (baseNode.Name == "img" && GetAttributeValue(baseNode, "alt") != string.Empty)
{
string text = GetAttributeValue(baseNode, "alt");
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
// Text for type/value structures with no found children
else if (ufElement.NodeType == UfElementDescriber.StructureTypes.TypeValuePair)
{
// if no chidren nodes ie type/value are found use text
// the calls for a children node type and value are alway both thier parent
if (ufd.Nodes.Count == 0)
{
// Add text from node value
string text = UfHelpers.HtmlToText(baseNode, false);
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
else
{
// Add child type/value pair
ufd.Name = ufElement.Name;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
}
// Text from Html node collect
else if (ufElement.Type == UfElementDescriber.PropertyTypes.FormattedText)
{
string text = UfHelpers.HtmlToText(baseNode, true);
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
else
{
// Text from node value
//string text = FindValuePattern(baseNode, ufElement);
//if(text == string.Empty)
// text = HtmlToText(baseNode, false);
string text = UfHelpers.HtmlToText(baseNode, false);
ufd.Name = ufElement.Name;
ufd.Value = text;
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
}
else
{
AddNewDateNode(baseNode, ufData, ufd, ufElement);
}
}