public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
string sizeAttr = atts ["size"].Value;
if (sizeAttr != null)
{
Matcher 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.IsEmpty())
{
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.Add(0, size);
}
else
{
instance.fontSizeStack.Add(0, null);
}
}
else
{
instance.fontSizeStack.Add(0, null);
}
return(false);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
if (--instance.inAnchor == 0)
{
if (instance.inIgnorableElement == 0)
{
instance.AddWhitespaceIfNecessary();
instance.tokenBuilder.Append(NBoilerpipeContentHandler.ANCHOR_TEXT_END);
instance.tokenBuilder.Append(' ');
instance.sbLastWasWhitespace = true;
}
}
return(false);
}
/// <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);
}
/// <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
this.End(instance, localName);
}
if (instance.inIgnorableElement == 0)
{
instance.AddWhitespaceIfNecessary();
instance.tokenBuilder.Append(NBoilerpipeContentHandler.ANCHOR_TEXT_START);
instance.tokenBuilder.Append(' ');
instance.sbLastWasWhitespace = true;
}
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);
}
if (instance.inIgnorableElement == 0)
{
instance.AddWhitespaceIfNecessary();
instance.tokenBuilder.Append(NBoilerpipeContentHandler.ANCHOR_TEXT_START);
instance.tokenBuilder.Append(' ');
instance.sbLastWasWhitespace = true;
}
return(false);
}
/// <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;
}
public NBoilerpipeHtmlParser(NBoilerpipeContentHandler contentHandler)
{
this.contentHandler = contentHandler;
}
public bool Start(NBoilerpipeContentHandler instance, string localName,HtmlAttributeCollection atts)
{
instance.inIgnorableElement++;
return true;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddLabelAction(action);
return true;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddWhitespaceIfNecessary();
instance.AddLabelAction(action);
return false;
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
instance.fontSizeStack.RemoveAt(0);
return false;
}
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 End(NBoilerpipeContentHandler instance, string localName)
{
instance.FlushBlock();
instance.inBody--;
return false;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddLabelAction(action);
return(true);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
instance.AddWhitespaceIfNecessary();
return(false);
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.AddWhitespaceIfNecessary();
instance.AddLabelAction(action);
return(false);
}
/// <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 End(NBoilerpipeContentHandler instance, string localName)
{
instance.fontSizeStack.RemoveFirst();
return(false);
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
return(false);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
instance.FlushBlock();
instance.inBody--;
return(false);
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.FlushBlock();
instance.inBody++;
return false;
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
return(true);
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
return false;
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool End(NBoilerpipeContentHandler instance, string localName)
{
return(t1.End(instance, localName) | t2.End(instance, localName));
}
/// <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)
{
instance.inIgnorableElement++;
return(true);
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool End(NBoilerpipeContentHandler instance, string localName)
{
return t1.End(instance, localName) | t2.End(instance, localName);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
instance.inIgnorableElement--;
return(true);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
instance.AddWhitespaceIfNecessary();
return false;
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool End(NBoilerpipeContentHandler instance, string localName)
{
labelStack.RemoveLast();
return(isBlockLevel);
}
public bool End(NBoilerpipeContentHandler instance, string localName)
{
return true;
}
/// <exception cref="Sharpen.SAXException"></exception>
public bool End(NBoilerpipeContentHandler instance, string localName)
{
labelStack.RemoveLast();
return isBlockLevel;
}
public bool End(NBoilerpipeContentHandler instance, string localName )
{
instance.inIgnorableElement--;
return true;
}
public bool End (NBoilerpipeContentHandler instance, string localName)
{
if (--instance.inAnchor == 0)
{
if (instance.inIgnorableElement == 0)
{
instance.AddWhitespaceIfNecessary();
instance.tokenBuilder.Append(NBoilerpipeContentHandler.ANCHOR_TEXT_END);
instance.tokenBuilder.Append(' ');
instance.sbLastWasWhitespace = true;
}
}
else
instance.inIgnorableElement--;
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;
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
instance.FlushBlock();
instance.inBody++;
return(false);
}
public NBoilerpipeHtmlParser(NBoilerpipeContentHandler contentHandler)
{
this.contentHandler = contentHandler;
}
/// <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;
}
