private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute current in (IEnumerable <HtmlAttribute>)collection)
{
current.Value = HtmlEntity.Entitize(current.Value);
}
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute attribute in (IEnumerable <HtmlAttribute>)collection)
{
attribute.Value = Entitize(attribute.Value);
}
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute at in collection)
{
at.Value = Entitize(at.Value);
}
}
private bool IsHidden(HtmlAttributeCollection atts)
{
if (atts.Contains("class"))
{
return atts["class"].Value.Split(' ').Any(str => str == "hidden" || str.StartsWith("promo") || str.Contains("comment"));
}
else
return false;
}
//private static IWebElement CreateNMock2IWebElement(this HtmlNode htmlNode)
//{
// var mockIWebElement = new Mockery().NewMock<IWebElement>();
// var attributes = htmlNode.Attributes;
// if (attributes["id"] == null)
// Stub.On(mockIWebElement).Method("GetAttribute").With("id").Will(Return.Value(""));
// foreach (var att in attributes)
// Stub.On(mockIWebElement).Method("GetAttribute").With(att.Name).Will(Return.Value(att.Value));
// Stub.On(mockIWebElement).GetProperty("TagName").Will(Return.Value(htmlNode.Name));
// var text = htmlNode.SelectSingleNode("text()");
// if (text != null)
// Stub.On(mockIWebElement).GetProperty("Text").Will(Return.Value(text.InnerText));
// else
// Stub.On(mockIWebElement).GetProperty("Text").Will(Return.Value(""));
// return mockIWebElement;
//}
//public static IWebElement ToNMock2IWebElement(this HtmlNode htmlNode)
//{
// var mockIWebElement = CreateNMock2IWebElement(htmlNode);
// if (htmlNode.HasChildNodes)
// {
// foreach (var child in htmlNode.ChildNodes)
// {
// var tempElement = new Mockery().NewMock<IWebElement>();
// tempElement = CreateNMock2IWebElement(child);
// var byID = By.Id(child.Id);
// Stub.On(mockIWebElement).Method("FindElement").With(byID).Will(Return.Value(tempElement));
// }
// }
// return mockIWebElement;
//}
private static string GetAttributeValue(HtmlAttributeCollection attributes, string attributeName)
{
var attribute = attributes[attributeName];
if (attribute != null)
{
return(attribute.Value);
}
return("");
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute at in collection)
{
if (at.Value == null)
{
continue;
}
at.Value = Entitize(at.Value);
}
}
public static bool EqualsAttributes(this HtmlAttributeCollection collection, HtmlAttributeCollection second)
{
if(collection.Count != second.Count)
return false;
bool all = true;
for (int i = 0; i < collection.Count; i++)
{
if (collection[i] != second[i])
{
all = false;
}
}
return all;
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
IList<string> labels = new AList<string> (5);
labels.AddItem (DefaultLabels.MARKUP_PREFIX + localName);
string classVal = atts ["class"].Value;
if (classVal != null && classVal.Length > 0) {
classVal = PAT_NUM.Matcher (classVal).ReplaceAll ("#");
classVal = classVal.Trim ();
string[] vals = classVal.Split ("[ ]+");
labels.AddItem (DefaultLabels.MARKUP_PREFIX + "." + classVal.Replace (' ', '.'));
if (vals.Length > 1) {
foreach (string s in vals) {
labels.AddItem (DefaultLabels.MARKUP_PREFIX + "." + s);
}
}
}
var att = atts["id"];
var id = ( atts !=null) ? att.Name : "";
if (id != null && id.Length > 0) {
id = PAT_NUM.Matcher (id).ReplaceAll ("#");
labels.AddItem (DefaultLabels.MARKUP_PREFIX + "#" + id);
}
ICollection<string> ancestors = GetAncestorLabels ();
IList<string> labelsWithAncestors = new AList<string> ((ancestors.Count + 1) * labels
.Count);
foreach (string l in labels) {
foreach (string an in ancestors) {
labelsWithAncestors.AddItem (an);
labelsWithAncestors.AddItem (an + " " + l);
}
labelsWithAncestors.AddItem (l);
}
instance.AddLabelAction (new LabelAction (Sharpen.Collections.ToArray (labelsWithAncestors
, new string[labelsWithAncestors.Count])));
labelStack.AddItem (labels);
return isBlockLevel;
}
static Item GetItemFromAttr(HtmlAttributeCollection itemAttr)
{
string hold = "";
//Data quality 6 is normal, 1 is genuine
Item tradeItem = new Item();
tradeItem.Name = Types[Int32.Parse(itemAttr["data-quality"].Value)] +
itemAttr["data-name"].Value;
if (itemAttr.Contains("data-particleeffectname"))
tradeItem.Name += " with " + itemAttr["data-particleeffectname"].Value;
if (itemAttr.Contains("data-customname"))
tradeItem.Name += " called " + itemAttr["data-customname"].Value;
if (itemAttr.Contains("data-customdescription"))
tradeItem.Name += " with description " + itemAttr["data-customdescription"].Value;
if (itemAttr.Contains("data-tint"))
{
if (!Tints.TryGetValue(Int32.Parse(itemAttr["data-tint"].Value), out hold))
tradeItem.Name += " painted a colour I haven't found yet. Contact me";
else
tradeItem.Name += " painted " + Tints[Int32.Parse(itemAttr["data-tint"].Value)];
}
tradeItem.ID = Int32.Parse(itemAttr["data-tf2itemid"].Value);
tradeItem.Level = Int32.Parse(itemAttr["data-level"].Value);
return tradeItem;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddWhitespaceIfNecessary();
instance.AddLabelAction(action);
return false;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
try
{
var alt = atts.Contains("alt") ? atts["alt"].Value : "";
if (alt.Length < 5)
{
alt = (atts.Contains("title") ? atts["title"].Value : alt);
}
int width = Math.Max(atts.Contains("width") ? int.Parse(atts["width"].Value.TrimEnd('p', 'x', ';')) : 0, 1);
int height = Math.Max(atts.Contains("height") ? int.Parse(atts["height"].Value.TrimEnd('p', 'x', ';')) : 0, 1);
var src = atts.Contains("src") ? atts["src"].Value : FindAlternateSrc(atts);
bool isWikimedia = false;
if (instance.inIgnorableElement <= 0 && !string.IsNullOrWhiteSpace(src) &&
(alt.Length > 5 || width > 400 || height > 320 || (isWikimedia = src.StartsWith("//upload.wikimedia.org"))))
{
var altWidthHeight = FindAlternateWidthHieght(src);
width = Math.Max(altWidthHeight.Item1, width);
height = Math.Max(altWidthHeight.Item2, height);
if (src.StartsWith("//"))
src = "http:" + src;
if (width > 400 || height > 320 || isWikimedia)
{
var tb = new Document.TextBlock("", new Sharpen.BitSet(), Math.Max((Math.Max(width, height) / 6), alt.Length), 0, 0, 0, 0, src);
tb.SetIsContent(true);
instance.textBlocks.Add(tb);
}
}
instance.inIgnorableElement++;
return true;
}
catch(Exception ex)
{
Debug.WriteLine("during boilerpipe parsing: " + ex.ToString());
}
instance.inIgnorableElement++;
return true;
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute htmlAttribute in (IEnumerable<HtmlAttribute>) collection)
htmlAttribute.Value = HtmlEntity.Entitize(htmlAttribute.Value);
}
private string FindAlternateSrc(HtmlAttributeCollection atts)
{
foreach (var att in atts)
{
if (att.Value.EndsWith(".jpg") || att.Value.EndsWith(".png"))
return att.Value;
}
return null;
}
public bool Start(NBoilerpipeContentHandler instance, string localName,HtmlAttributeCollection atts)
{
instance.inIgnorableElement++;
return true;
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute at in collection)
{
at.Value = Entitize(at.Value);
}
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
return t1.Start(instance, localName, atts) | t2.Start(instance, localName,atts);
}
public bool Start (NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
string sizeAttr = atts ["size"].Value;
if (sizeAttr != null)
{
var m = CommonTagActions.PAT_FONT_SIZE.Matcher(sizeAttr);
if (m.Matches())
{
string rel = m.Group(1);
int val = System.Convert.ToInt32(m.Group(2));
int size;
if (rel.Length == 0)
{
// absolute
size = val;
}
else
{
// relative
int? prevSize;
if (instance.fontSizeStack.Count == 0)
{
prevSize = 3;
}
else
{
prevSize = 3;
foreach (int? s in instance.fontSizeStack)
{
if (s != null)
{
prevSize = s;
break;
}
}
}
if (rel[0] == '+')
{
size = (int)prevSize + val;
}
else
{
size = (int)prevSize - val;
}
}
instance.fontSizeStack.Insert(0, size);
}
else
{
instance.fontSizeStack.Insert(0, null);
}
}
else
{
instance.fontSizeStack.Insert(0, null);
}
return false;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
return false;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.FlushBlock();
instance.inBody++;
return false;
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool Start (NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
if (instance.inAnchor++ > 0) {
// as nested A elements are not allowed per specification, we
// are probably reaching this branch due to a bug in the XML
// parser
//System.Console.Error.WriteLine ("Warning: SAX input contains nested A elements -- You have probably hit a bug in your HTML parser (e.g., NekoHTML bug #2909310). Please clean the HTML externally and feed it to boilerpipe again. Trying to recover somehow..."
//);
//this.End (instance, localName);
instance.inIgnorableElement++;
}
if (instance.inIgnorableElement == 0) {
instance.AddWhitespaceIfNecessary ();
instance.tokenBuilder.Append(NBoilerpipeContentHandler.ANCHOR_TEXT_START);
instance.tokenBuilder.Append(' ');
instance.sbLastWasWhitespace = true;
}
return false;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddLabelAction(action);
return true;
}
/// <summary>
/// Gets the value of an HTML attribute safely from an HTMLAttributeCollection
/// </summary>
/// <param name="p_hacCollection"></param>
/// <param name="p_strItem"></param>
/// <returns></returns>
private static string GetHtmlAttributeValue(HtmlAttributeCollection p_hacCollection, string p_strItem)
{
if (p_hacCollection.Contains(p_strItem)) return ((HtmlAttribute)p_hacCollection[p_strItem]).Value;
else return "";
}
