public static List <string> GetTextLinesSurroundedByRectangle(List <CharBox> cbs, RectangleF r, TextAutoInsertSpace textAutoInsertSpace, CharFilter charFilter)
{
cbs = GetCharBoxsSurroundedByRectangle(cbs, r);
return(Page.GetLines(cbs, textAutoInsertSpace, charFilter).Select(a => a.GetString()).ToList());
}
//public static string GetTextSurroundedByRectangle(List<CharBox> cbs, RectangleF r, TextAutoInsertSpace textAutoInsertSpace, Template.CharFilter charFilter)
//{
// return string.Join("\r\n", GetTextLinesSurroundedByRectangle(cbs, r, textAutoInsertSpace, charFilter));
//}
public static List <Page.Line <CharBox> > GetLinesSurroundedByRectangle(List <CharBox> cbs, RectangleF r, TextAutoInsertSpace textAutoInsertSpace, CharFilter charFilter)
{
cbs = GetCharBoxsSurroundedByRectangle(cbs, r);
return(Page.GetLines(cbs, textAutoInsertSpace, charFilter));
}
public static string GetText(IEnumerable <CharBox> cbs, TextAutoInsertSpace textAutoInsertSpace, CharFilter charFilter)
{
return(string.Join("\r\n", GetTextLines(cbs, textAutoInsertSpace, charFilter)));
}
public static List <Line <CharBoxT> > GetLines <CharBoxT>(IEnumerable <CharBoxT> cbs, TextAutoInsertSpace textAutoInsertSpace, CharFilter charFilter) where CharBoxT : CharBox, new()
{
if (textAutoInsertSpace?.IgnoreSourceSpaces == true)
{
cbs = cbs.Where(a => a.Char != " ");
}
if (charFilter != null)//to filter out wrong OCR chars like borders etc which brakes lines
{
//SizeF s=new SizeF(ignoreCharsBiggerThan.Width*Settings.Constants.Pdf2ImageResolutionRatio)
float maxWidth = charFilter.MaxWidth <= 0 ? float.MaxValue : charFilter.MaxWidth;
float maxHeight = charFilter.MaxHeight <= 0 ? float.MaxValue : charFilter.MaxHeight;
cbs = cbs.Where(a => a.R.Width >= charFilter.MinWidth && a.R.Width <= maxWidth && a.R.Height >= charFilter.MinHeight && a.R.Height <= maxHeight);
}
List <Line <CharBoxT> > lines = new List <Line <CharBoxT> >();
foreach (CharBoxT cb in cbs)
{
for (int i = 0; i < lines.Count; i++)
{
float mY = cb.R.Bottom - cb.R.Height / 2;
if (mY < lines[i].Top)
{
Line <CharBoxT> l = new Line <CharBoxT> {
Top = cb.R.Top, Bottom = cb.R.Bottom
};
l.CharBoxs.Add(cb);
lines.Insert(i, l);
goto NEXT_CHAR;
}
if (mY <= lines[i].Bottom)//the char's center is in the line
{
lines[i].CharBoxs.Add(cb);
if (lines[i].Top > cb.R.Top)
{
lines[i].Top = cb.R.Top;
}
if (lines[i].Bottom < cb.R.Bottom)
{
lines[i].Bottom = cb.R.Bottom;
}
goto NEXT_CHAR;
}
}
{
Line <CharBoxT> l = new Line <CharBoxT> {
Top = cb.R.Top, Bottom = cb.R.Bottom
};
l.CharBoxs.Add(cb);
lines.Add(l);
}
NEXT_CHAR :;
}
for (int i = 1; i < lines.Count; i++)
{
float intersectionH2 = (lines[i - 1].Bottom - lines[i].Top) * 2;
if (intersectionH2 > lines[i - 1].Height || intersectionH2 > lines[i].Height)
{
lines[i - 1].CharBoxs.AddRange(lines[i].CharBoxs);
if (lines[i - 1].Top > lines[i].Top)
{
lines[i - 1].Top = lines[i].Top;
}
if (lines[i - 1].Bottom < lines[i].Bottom)
{
lines[i - 1].Bottom = lines[i].Bottom;
}
lines.RemoveAt(i);
i--;
}
}
lines.ForEach(a => a.CharBoxs = a.CharBoxs.OrderBy(b => b.R.X).ToList());
if (textAutoInsertSpace?.Threshold > 0)
{
foreach (Line <CharBoxT> l in lines)
{
for (int i = 1; i < l.CharBoxs.Count; i++)
{
CharBox cb0 = l.CharBoxs[i - 1];
CharBox cb = l.CharBoxs[i];
if (/*cb0.Char != " " && */ cb.R.Left - cb0.R.Right > (/*cb0.R.Width*/ 0.8 / cb0.R.Height + /*cb.R.Width*/ 0.8 / cb.R.Height) * textAutoInsertSpace.Threshold)
{
float spaceWidth = (cb0.R.Width + cb.R.Width) / 2;
int spaceNumber = (int)Math.Ceiling((cb.R.Left - cb0.R.Right) / spaceWidth);
for (int j = 0; j < spaceNumber; j++)
{
l.CharBoxs.Insert(i, new CharBoxT {
Char = textAutoInsertSpace.Representative, R = new RectangleF(cb0.R.Right + spaceWidth * j, cb0.R.Y, spaceWidth, cb.R.Height)
});
}
i += spaceNumber;
}
}
}
}
return(lines);
}
public static List <string> GetTextLines <CharBoxT>(IEnumerable <CharBoxT> cbs, TextAutoInsertSpace textAutoInsertSpace, CharFilter charFilter) where CharBoxT : CharBox, new()
{
return(GetLines(cbs, textAutoInsertSpace, charFilter).Select(a => a.GetString()).ToList());
}