public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
HtmlNode thisContent = thisPart.ParsedContent;
string pattern = Settings["pattern"];
string attribute = Settings["attribute"];
string value = Settings["value"];
if (thisContent.SelectNodes(pattern) == null)
{
//No such pattern. Return the part with no changes.
return thisPart;
}
foreach (HtmlNode thisNode in thisContent.SelectNodes(pattern))
{
counter++;
if (thisNode.Attributes[attribute] != null)
{
thisNode.Attributes[attribute].Value = value;
}
else
{
thisNode.Attributes.Add(attribute, value);
}
}
thisPart.ParsedContent = thisContent;
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Pattern: {0}; Attribute: {1}; Value: {2} Set attribute on (3) element(s)", pattern, attribute, value, counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
List<string> domainsToRemove = Settings["removeDomains"].Split(new Char[] {','}).ToList<string>();
if (thisPart.ParsedContent.SelectNodes(".//a") == null)
{
return thisPart;
}
foreach (HtmlNode link in thisPart.ParsedContent.SelectNodes(".//a"))
{
if (link.Attributes["href"] == null)
{
continue;
}
foreach (string domainToRemove in domainsToRemove)
{
if (link.Attributes["href"].Value.Contains(domainToRemove))
{
counter++;
link.Attributes["href"].Value = link.Attributes["href"].Value.Replace(String.Concat("http://", domainToRemove), "");
link.Attributes["href"].Value = link.Attributes["href"].Value.Replace(String.Concat("https://", domainToRemove), "");
}
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Removed {0} domain(s)", counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string pattern = Settings["pattern"];
pattern = pattern.Replace("&", "&");
if (thisPart.ParsedContent.SelectNodes(pattern) != null)
{
foreach (HtmlNode nodeToRemove in thisPart.ParsedContent.SelectNodes(pattern))
{
// Only remove nodes with nothing but whitespace in them
if (nodeToRemove.InnerHtml.Trim().Length == 0)
{
counter++;
nodeToRemove.ParentNode.RemoveChild(nodeToRemove);
}
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Pattern: {0}; Removed {1} element(s).", pattern, counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string pattern = Settings["pattern"];
string attribute = Settings["attribute"];
string value = null;
if (Settings.ContainsKey("value"))
{
value = Settings["value"];
}
if (thisPart.ParsedContent.SelectNodes(pattern) != null)
{
foreach (HtmlNode thisNode in thisPart.ParsedContent.SelectNodes(pattern))
{
if (thisNode.Attributes[attribute] != null)
{
//If we have a value, and this attribute doesn't match that value, skip it
if (value != null && value.ToLowerInvariant() != thisNode.Attributes[attribute].Value.ToLowerInvariant())
{
continue;
}
counter++;
thisNode.Attributes[attribute].Remove();
}
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format(" Pattern: {0}; Attribute: {1}; Removed {2} attribute(s).", pattern, attribute, counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
if (!thisPart.Content.Contains(Settings["character"]))
{
thisPart.AddMessage(this, StandardMessageTypes.Notice, String.Format("Character \"{0}\" not found.", Settings["character"]));
}
int lengthBefore = thisPart.Content.Length;
var parts = thisPart.Content.Split(Settings["character"].ToCharArray());
thisPart.Content = String.Join(String.Empty, parts.Take(parts.Length - 1).ToArray());
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Length before: {0}; Length after: {1}", lengthBefore.ToString(), thisPart.Content.Length.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int lengthBeforeStripping = thisPart.Content.Length;
string content = thisPart.Content;
thisPart.Content = content.Trim();
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Length before: {0}; Length after: {1}", lengthBeforeStripping.ToString(), thisPart.Content.Length.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string pattern = "img";
string directory = String.Empty;
if (Settings.Exists("directory"))
{
directory = Settings["directory"];
}
else
{
directory = ArtifactManager.GetArtifactPath("downloaded-assets");
}
//Have any images?
if (thisPart.ParsedContent.SelectNodes(pattern) != null)
{
//Loop through them
foreach(HtmlNode image in thisPart.ParsedContent.SelectNodes(pattern))
{
//If they're remote, skip it
if (image.Attributes["src"].Value.StartsWith("http"))
{
continue;
}
counter++;
//Get the corrected path with the domain (to make the request)
string assetPathWithDomain = thisPart.Page.GetContextualUrl(image.Attributes["src"].Value, true);
//Get the corrected path without the domain (to save the file)
string assetPathWithoutDomain = thisPart.Page.GetContextualUrl(image.Attributes["src"].Value, false);
//Add the directory from the settings
string newImagePath = String.Concat(directory, "/", assetPathWithoutDomain);
//Get the file data
Byte[] fileData = Utils.MakeBinaryHttpRequest(assetPathWithDomain);
//Save the file
ArtifactManager.SaveFileData(fileData, newImagePath);
//Correct the reference to the image
image.Attributes["src"].Value = newImagePath;
thisPart.Meta.Add(String.Concat("Downloaded Asset ", counter.ToString()), assetPathWithoutDomain);
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Saved {0} image(s) locally.", counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int lengthBeforeStripping = thisPart.Content.Length;
string content = thisPart.Content;
thisPart.Content = Regex.Replace(content, @"<!--.*?-->", String.Empty, RegexOptions.Singleline);
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Length before: {0}; Length after: {1}; Comment characters removed: {2}", lengthBeforeStripping.ToString(), thisPart.Content.Length.ToString(), (lengthBeforeStripping - thisPart.Content.Length).ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
string content = thisPart.Content;
int lengthBefore = content.Length;
string newString = Settings.Exists("newString") ? Settings["newString"] : String.Empty;
string oldString = Settings["oldString"];
thisPart.Content = content.Replace(oldString, newString);
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Length before: {0}; Length after: {1}", lengthBefore.ToString(), thisPart.Content.Length.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string pattern = Settings.Exists("tag") ? Settings["tag"] : "*";
string className = Settings.Exists("class") ? Settings["class"] : "*";
if (thisPart.ParsedContent.SelectNodes(pattern) != null)
{
foreach (HtmlNode matchingNode in thisPart.ParsedContent.SelectNodes(pattern))
{
if (matchingNode.Attributes["class"] == null)
{
continue;
}
counter++;
if (className == "*")
{
matchingNode.Attributes["class"].Remove();
continue;
}
List<string> currentClasses = matchingNode.Attributes["class"].Value.Split(new Char[] { ' ' }).ToList<string>();
currentClasses.Remove(className);
//If we now have no classes, just remove the whole thing
if (currentClasses.Count == 0)
{
matchingNode.Attributes["class"].Remove();
}
else
{
//Otherwise, set it back to what's left
matchingNode.Attributes["class"].Value = String.Join(" ", currentClasses.ToArray());
}
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format(" Pattern: {0}; Class: {1}; Removed classes from {2} tag(s).", pattern, className, counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
string pattern = Settings["pattern"];
string result = String.Empty;
if (thisPart.ParsedContent.SelectSingleNode(pattern) != null)
{
thisPart.Content = thisPart.ParsedContent.SelectSingleNode(pattern).InnerHtml;
result = "Found and inflated";
}
else
{
result = "Not found";
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format(" Pattern: {0}; Result: {1}", pattern, result ));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
using (Document doc = new Document(thisPart.Content))
{
doc.ShowWarnings = false;
doc.Quiet = true;
doc.IndentBlockElements = AutoBool.Yes;
doc.OutputXhtml = true;
doc.OutputBodyOnly = AutoBool.Yes;
doc.WrapAt = 10000;
doc.CleanAndRepair();
string parsed = doc.Save();
thisPart.Content = parsed;
thisPart.AddMessage(this, StandardMessageTypes.Executed, "Processed by HTMLTidy");
}
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string oldTag = Settings["old"].TrimStart('/');
string newTag = Settings["new"].TrimStart('/');
if (thisPart.ParsedContent.SelectNodes("//" + oldTag) == null)
{
return thisPart;
}
foreach (HtmlNode matchingNode in thisPart.ParsedContent.SelectNodes("//" + oldTag))
{
counter++;
matchingNode.Name = newTag;
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Old tag: {0}; New tag: {1}; Swapped {2} tag(s).", oldTag, newTag, counter.ToString()));
return thisPart;
}
public override PagePart Execute(PagePart thisPart)
{
int counter = 0;
string pattern = Settings["pattern"];
pattern = pattern.Replace("&", "&");
bool preserveContents = Settings.Exists("preserveContents") ? Convert.ToBoolean(Settings["preserveContents"]) : false;
if (thisPart.ParsedContent.SelectNodes(pattern) != null)
{
foreach (HtmlNode nodeToRemove in thisPart.ParsedContent.SelectNodes(pattern))
{
counter++;
nodeToRemove.ParentNode.RemoveChild(nodeToRemove, preserveContents);
}
}
thisPart.AddMessage(this, StandardMessageTypes.Executed, String.Format("Pattern: {0}; Removed {1} element(s).", pattern, counter.ToString()));
return thisPart;
}
