/// <exception cref="NBoilerpipe.BoilerpipeProcessingException"></exception>
public bool Process(TextDocument doc)
{
bool changes = false;
IList <TextBlock> blocks = doc.GetTextBlocks();
IList <TextBlock> blocksNew = new AList <TextBlock>();
foreach (TextBlock tb in blocks)
{
string text = tb.GetText();
string[] paragraphs = text.Split("[\n\r]+");
if (paragraphs.Length < 2)
{
blocksNew.AddItem(tb);
continue;
}
bool isContent = tb.IsContent();
ICollection <string> labels = tb.GetLabels();
foreach (string p in paragraphs)
{
TextBlock tbP = new TextBlock(p);
tbP.SetIsContent(isContent);
tbP.AddLabels(labels);
blocksNew.AddItem(tbP);
changes = true;
}
}
if (changes)
{
blocks.Clear();
Sharpen.Collections.AddAll(blocks, blocksNew);
}
return(changes);
}
protected void AddTextBlock(TextBlock tb)
{
foreach (int l in fontSizeStack)
{
tb.AddLabels("font-" + l);
break;
}
foreach (List <LabelAction> labels in labelStacks)
{
if (labels != null)
{
foreach (LabelAction label in labels)
{
label.AddTo(tb);
}
}
}
textBlocks.Add(tb);
}
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);
}
protected virtual void AddLabelsTo(TextBlock tb)
{
tb.AddLabels(Labels);
}
protected internal void AddLabelsTo(TextBlock tb)
{
tb.AddLabels(labels);
}
