private static HtmlFormDefinition PopulateForm(Match formMatch, string pageHtml)
{
var parsedForm = new HtmlFormDefinition {
PageHtml = pageHtml
};
foreach (var attribute in Utility.ParseAttributes(formMatch.Groups[RegexLibrary.ParseFormsAttributesGroup].Value))
{
parsedForm.attributes.Add(attribute.Key, attribute.Value);
}
// TODO: need to remove comments from form HTML before parsing out controls; commented out controls will be found!
// Populate controls
var controlMatches = RegexCache.Instance.Regex(RegexLibrary.ParseFormControls, RegexLibrary.ParseFormControlsOptions).Matches(formMatch.Groups[RegexLibrary.ParseFormsBodyGroup].Value);
foreach (Match controlMatch in controlMatches)
{
HtmlFormControl control;
if (controlMatch.Groups[RegexLibrary.ParseFormControlsInputGroup].Value.Length > 0)
{
var inputControl = new InputHtmlFormControl(controlMatch.Value);
if (inputControl.ControlType == InputHtmlFormControlType.Radio)
{
control = new InputRadioHtmlFormControl(controlMatch.Value);
}
else if (inputControl.ControlType == InputHtmlFormControlType.CheckBox)
{
control = new InputCheckBoxHtmlFormControl(controlMatch.Value);
}
else
{
// Generic control
control = inputControl;
}
}
else if (controlMatch.Groups[RegexLibrary.ParseFormControlsSelectGroup].Value.Length > 0)
{
control = new SelectHtmlFormControl(controlMatch.Value);
}
else if (controlMatch.Groups[RegexLibrary.ParseFormControlsTextAreaGroup].Value.Length > 0)
{
control = new TextAreaHtmlFormControl(controlMatch.Value);
}
else
{
throw new System.Net.WebException(string.Format(CultureInfo.CurrentCulture, NScrapeResources.UnsupportedHtmlControl, controlMatch.Value));
}
if (control.Name != null)
{
parsedForm.controls.Add(control);
}
}
return(parsedForm);
}
/// <summary>
/// Builds the request data to be used to submit an ASPX form.
/// </summary>
/// <param name="eventTargetValue">Contains the value for <b>__EVENTTARGET</b>, the control doing the submission.</param>
/// <param name="eventArgumentValue">Contains the value for <b>__EVENTARGUMENT</b>, any additional information.</param>
/// <returns>The request data in <b>application/x-www-form-urlencoded</b> format.</returns>
/// <remarks>
/// See <see href="http://www.evagoras.com/2011/02/10/how-postback-works-in-asp-net/">How postback works in ASP.NET</see> for a good overview on the topic.
/// </remarks>
protected string BuildAspxPostBackRequest(string eventTargetValue, string eventArgumentValue)
{
var eventTarget = Controls.SingleOrDefault(c => c.Name == EventTargetName) as InputHtmlFormControl;
if (eventTarget == null)
{
// If __EVENTTARGET is not in the HTML (injected via JavaScript?), add it manually.
eventTarget = new InputHtmlFormControl("<input type=\"hidden\" name=\"__EVENTTARGET\" id=\"__EVENTTARGET\" value=\"\" />");
Controls.Add(eventTarget);
}
var eventArgument = Controls.SingleOrDefault(c => c.Name == EventArgumentName) as InputHtmlFormControl;
if (eventArgument == null)
{
// If __EVENTARGUMENT is not in the HTML (injected via JavaScript?), add it manually.
eventArgument = new InputHtmlFormControl("<input type=\"hidden\" name=\"__EVENTARGUMENT\" id=\"__EVENTARGUMENT\" value=\"\" />");
Controls.Add(eventArgument);
}
// Most __doPostBack() examples found via Googling unescaped the event target parameter,
// but some did not, so play it safe and see which kind we have
if (RegexCache.Instance.Regex(RegexLibrary.MatchDoPostBack, RegexLibrary.MatchDoPostBackOptions).IsMatch(Html))
{
eventTarget.Value = eventTargetValue.Replace('$', ':');
}
else
{
eventTarget.Value = eventTargetValue;
}
eventArgument.Value = eventArgumentValue;
// Build the request using an empty string so that none of the buttons in the form are included.
var request = BuildRequest(string.Empty);
// Do not persist the values
eventTarget.Value = string.Empty;
eventArgument.Value = string.Empty;
return(request);
}
/// <summary>
/// Builds the request data to be used to submit an ASPX form.
/// </summary>
/// <param name="eventTargetValue">Contains the value for <b>__EVENTTARGET</b>, the control doing the submission.</param>
/// <param name="eventArgumentValue">Contains the value for <b>__EVENTARGUMENT</b>, any additional information.</param>
/// <returns>The request data in <b>application/x-www-form-urlencoded</b> format.</returns>
/// <remarks>
/// See <see href="http://www.evagoras.com/2011/02/10/how-postback-works-in-asp-net/">How postback works in ASP.NET</see> for a good overview on the topic.
/// </remarks>
protected string BuildAspxPostBackRequest( string eventTargetValue, string eventArgumentValue )
{
var eventTarget = Controls.SingleOrDefault( c => c.Name == EventTargetName ) as InputHtmlFormControl;
if ( eventTarget == null ) {
// If __EVENTTARGET is not in the HTML (injected via JavaScript?), add it manually.
eventTarget = new InputHtmlFormControl( "<input type=\"hidden\" name=\"__EVENTTARGET\" id=\"__EVENTTARGET\" value=\"\" />" );
Controls.Add( eventTarget );
}
var eventArgument = Controls.SingleOrDefault( c => c.Name == EventArgumentName ) as InputHtmlFormControl;
if ( eventArgument == null ) {
// If __EVENTARGUMENT is not in the HTML (injected via JavaScript?), add it manually.
eventArgument = new InputHtmlFormControl( "<input type=\"hidden\" name=\"__EVENTARGUMENT\" id=\"__EVENTARGUMENT\" value=\"\" />" );
Controls.Add( eventArgument );
}
// Most __doPostBack() examples found via Googling unescaped the event target parameter,
// but some did not, so play it safe and see which kind we have
if ( RegexCache.Instance.Regex( RegexLibrary.MatchDoPostBack, RegexLibrary.MatchDoPostBackOptions ).IsMatch( Html ) ) {
eventTarget.Value = eventTargetValue.Replace( '$', ':' );
}
else {
eventTarget.Value = eventTargetValue;
}
eventArgument.Value = eventArgumentValue;
// Build the request using an empty string so that none of the buttons in the form are included.
var request = BuildRequest( string.Empty );
// Do not persist the values
eventTarget.Value = string.Empty;
eventArgument.Value = string.Empty;
return request;
}
private static HtmlFormDefinition PopulateForm( Match formMatch, string pageHtml )
{
var parsedForm = new HtmlFormDefinition {
PageHtml = pageHtml
};
foreach ( var attribute in Utility.ParseAttributes( formMatch.Groups[RegexLibrary.ParseFormsAttributesGroup].Value ) ) {
parsedForm.attributes.Add( attribute.Key, attribute.Value );
}
// TODO: need to remove comments from form HTML before parsing out controls; commented out controls will be found!
// Populate controls
var controlMatches = RegexCache.Instance.Regex( RegexLibrary.ParseFormControls, RegexLibrary.ParseFormControlsOptions ).Matches( formMatch.Groups[RegexLibrary.ParseFormsBodyGroup].Value );
foreach ( Match controlMatch in controlMatches ) {
HtmlFormControl control;
if ( controlMatch.Groups[RegexLibrary.ParseFormControlsInputGroup].Value.Length > 0 ) {
var inputControl = new InputHtmlFormControl( controlMatch.Value );
if ( inputControl.ControlType == InputHtmlFormControlType.Radio ) {
control = new InputRadioHtmlFormControl( controlMatch.Value );
}
else if ( inputControl.ControlType == InputHtmlFormControlType.CheckBox ) {
control = new InputCheckBoxHtmlFormControl( controlMatch.Value );
}
else {
// Generic control
control = inputControl;
}
}
else if ( controlMatch.Groups[RegexLibrary.ParseFormControlsSelectGroup].Value.Length > 0 ) {
control = new SelectHtmlFormControl( controlMatch.Value );
}
else if ( controlMatch.Groups[RegexLibrary.ParseFormControlsTextAreaGroup].Value.Length > 0 ) {
control = new TextAreaHtmlFormControl( controlMatch.Value );
}
else {
throw new System.Net.WebException( string.Format( CultureInfo.CurrentCulture, NScrapeResources.UnsupportedHtmlControl, controlMatch.Value ) );
}
if ( control.Name != null ) {
parsedForm.controls.Add( control );
}
}
return parsedForm;
}
