public bool Process(TextDocument doc) {
bool changes = false;
List<TextBlock> blocks = doc.TextBlocks;
var blocksNew = new List<TextBlock>();
foreach (TextBlock tb in blocks) {
string text = tb.Text;
string[] paragraphs = Regex.Split(text, "[\n\r]+");
if (paragraphs.Length < 2) {
blocksNew.Add(tb);
continue;
}
bool isContent = tb.IsContent;
List<string> labels = (tb.Labels ?? Enumerable.Empty<string>()).ToList();
foreach (String p in paragraphs) {
var tbP = new TextBlock(p) { IsContent = isContent };
tbP.AddLabels(labels);
blocksNew.Add(tbP);
changes = true;
}
}
if (changes) {
blocks.Clear();
blocks.AddRange(blocksNew);
}
return changes;
}
private static bool Classify(TextBlock prev, TextBlock curr, TextBlock next) {
bool isContent = (curr.LinkDensity > 0 && next.NumWords > 11)
|| (curr.NumWords > 19
|| (next.NumWords > 6 && next.LinkDensity == 0 && prev.LinkDensity == 0
&& (curr.NumWords > 6 || prev.NumWords > 7 || next.NumWords > 19)));
if (curr.IsContent != isContent) {
curr.IsContent = isContent;
return true;
}
return false;
}
protected bool Classify(TextBlock prev, TextBlock curr, TextBlock next) {
bool isContent;
if (curr.LinkDensity <= 0.333333) {
if (prev.LinkDensity <= 0.555556) {
if (curr.TextDensity <= 9) {
if (next.TextDensity <= 10) {
if (prev.TextDensity <= 4) {
isContent = false;
} else {
isContent = true;
}
} else {
isContent = true;
}
} else {
if (next.TextDensity == 0) {
isContent = false;
} else {
isContent = true;
}
}
} else {
if (next.TextDensity <= 11) {
isContent = false;
} else {
isContent = true;
}
}
} else {
isContent = false;
}
if (curr.IsContent != isContent) {
curr.IsContent = isContent;
return true;
}
return false;
}
protected bool Classify(TextBlock prev, TextBlock curr, TextBlock next) {
bool isContent;
if (curr.LinkDensity <= 0.333333) {
if (prev.LinkDensity <= 0.555556) {
if (curr.NumWords <= 16) {
if (next.NumWords <= 15) {
if (prev.NumWords <= 4) {
isContent = false;
} else {
isContent = true;
}
} else {
isContent = true;
}
} else {
isContent = true;
}
} else {
if (curr.NumWords <= 40) {
if (next.NumWords <= 17) {
isContent = false;
} else {
isContent = true;
}
} else {
isContent = true;
}
}
} else {
isContent = false;
}
if (curr.IsContent != isContent) {
curr.IsContent = isContent;
return true;
}
return false;
}
protected static int GetNumFullTextWords(TextBlock tb) {
return GetNumFullTextWords(tb, 9);
}
protected static int GetNumFullTextWords(TextBlock tb, float minTextDensity) {
if (tb.TextDensity >= minTextDensity) {
return tb.NumWords;
}
return 0;
}
/// <summary>
/// Merges specified <see cref="TextBlock"/> with this <see cref="TextBlock"/>.
/// </summary>
/// <param name="other">Then <see cref="TextBlock"/> to merge with.</param>
public void MergeNext(TextBlock other) {
StringBuilder sb = _text;
sb.Append('\n');
sb.Append(other._text);
_numWords += other._numWords;
_numWordsInAnchorText += other._numWordsInAnchorText;
_numWordsInWrappedLines += other._numWordsInWrappedLines;
_numWrappedLines += other._numWrappedLines;
_offsetBlocksStart = Math.Min(_offsetBlocksStart, other._offsetBlocksStart);
_offsetBlocksEnd = Math.Max(_offsetBlocksEnd, other._offsetBlocksEnd);
InitDensities();
IsContent |= other.IsContent;
if (_containedTextElements == null) {
_containedTextElements = (BitArray)other._containedTextElements.Clone();
} else {
_containedTextElements.Or(other._containedTextElements);
}
_numFullTextWords += other._numFullTextWords;
if (other._labels != null) {
if (_labels == null) {
_labels = new List<String>(other._labels);
} else {
_labels.AddRange(other._labels);
}
}
_tagLevel = Math.Min(_tagLevel, other._tagLevel);
}
/// <summary>
/// Adds the labels to the <see cref="TextBlock" /> if the condition is met.
/// </summary>
/// <param name="block">The <see cref="TextBlock" /> to add the labels to.</param>
public override void AddTo(TextBlock block) {
if (_condition.MeetsCondition(block)) {
AddLabelsTo(block);
}
}
protected virtual void AddLabelsTo(TextBlock tb) {
tb.AddLabels(Labels);
}
public virtual void AddTo(TextBlock block) {
AddLabelsTo(block);
}
