public HttpScraper(string sourceUri, string xpath)
{
web = new FormProcessorWeb(true);
web.UserAgent = "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0)";
processor = new FormProcessor(web);
form = processor.GetForm(sourceUri, xpath, FormQueryModeEnum.Nested);
}
public void SetForm(Form _form)
{
form = _form;
}
/// <summary>
/// Posts the form back to the server with a payload
/// assembled from the specified Form object's HtmlNodeCollection.
/// </summary>
/// <param name="form">The form.</param>
/// <returns></returns>
public HtmlDocument SubmitForm(Form form)
{
return SubmitForm(form.elements, form.action);
}
public void SetForm(HtmlDocument doc, string url, string xpath)
{
form = processor.GetForm(doc, url, xpath, FormQueryModeEnum.Nested);
}
/// <summary>
/// Attempts to extract, parse, and return an HTMLFormElement object
/// from the content contained in the doc object, using the specified
/// XPath statement and queryMode.
/// </summary>
/// <param name="doc">The doc.</param>
/// <param name="url">The URL.</param>
/// <param name="xpath">The xpath.</param>
/// <param name="queryMode">The query mode.</param>
/// <returns></returns>
public Form GetForm(HtmlDocument doc, string url, string xpath, // # FIX THIS FUNCTION
FormQueryModeEnum queryMode)
{
HtmlNode formNode
= doc.DocumentNode.SelectSingleNode(xpath);
if (formNode == null)
return null;
HtmlNodeCollection formNodes = null;
Form form = null;
// Absolutize the form's action attribute
AbsolutizeForm(formNode, url);
if (formNode != null) // TODO : fix this
{
// If queryMode is nested then just apply the path from
// the point of the containing element
if (queryMode == FormQueryModeEnum.Nested)
{
formNodes = formNode.SelectNodes(NodeSelectPath);
}
else if (queryMode == FormQueryModeEnum.Adjacent)
{
// Otherwise, the form is not properly the parent of
// all its nodes, so grab all child nodes of the entire document
// *TODO* make this take into account the node position
formNodes = doc.DocumentNode.SelectNodes(NodeSelectPath);
// Add these nodes to the form element
}
if (formNodes != null)
{
form = new Form(formNode, formNodes);
}
}
return form; // could be null!
}
