bool IsEndOfLine(IBlock block, BlockLine line)
{
float startOfBlock = block.GetX();
float endOfBlock = block.GetX() + block.GetWidth();
float endOfLine = line.GetX() + line.GetWidth();
if ((block.GetX() + block.GetWidth() - line.GetX()) < 0)
{
return(true);
}
else
{
return(false);
}
}
public BlockPage Process(BlockPage page)
{
GroupFontLineHelper groupFont = null;
BlockLine line = null;
IBlock last = null;
var result = new BlockPage();
foreach (var block in page.AllBlocks)
{
if (last != null)
{
if (IsEndOfWord(block) || IsEndOfLine(block, line))
{
last = null;
result.Add(line);
}
else
{
line.Text += block.GetText();
line.Width = block.GetX() + block.GetWidth() - line.GetX();
groupFont.MergeFont((Block)block);
}
}
if ((last == null))
{
var b = (Block)block;
string text = block.GetText();
line = new BlockLine()
{
Text = text,
X = block.GetX(),
H = block.GetH(),
Width = block.GetWidth(),
Height = block.GetHeight(),
HasBackColor = b.HasBackColor,
HasLargeSpace = false,
// might be inaccurate
FontFullName = b.FontFullName,
FontName = b.FontName,
FontSize = b.FontSize, // BE CAREFUL!
FontStyle = b.FontStyle
// now the settings are done in GroupFontLineHelper
};
// TODO: validar a entrada duas vezes
if (groupFont != null)
{
groupFont.Done();
}
groupFont = new GroupFontLineHelper(line, b);
if (line.Width <= 0 || line.Height <= 0)
{
PdfReaderException.AlwaysThrow("line.Width <= 0 || line.Height <= 0");
}
}
last = block;
}
if (groupFont != null)
{
groupFont.Done();
}
return(result);
}
public BlockPage Process(BlockPage page)
{
GroupFontLineHelper groupFont = null;
BlockLine line = null;
IBlock last = null;
string last_hidden = null;
var result = new BlockPage();
foreach (var block in page.AllBlocks)
{
if (block is BlockHidden)
{
var blockHidden = (BlockHidden)block;
if (last_hidden != null)
{
PdfReaderException.Warning("last_hidden != null: hidden text will be overwritten");
}
last_hidden = blockHidden.GetHiddenText();
continue;
}
if (last != null)
{
if (CheckSubfonts(line, (Block)block))
{
bool isBackspace = CheckBackspace(line, block);
float endofblock = block.GetX() + block.GetWidth();
float endofline = line.GetX() + line.GetWidth();
if (endofblock > endofline)
{
line.Width = block.GetX() + block.GetWidth() - line.GetX();
}
if (line.Width <= 0)
{
PdfReaderException.AlwaysThrow("line.Width <= 0");
}
// conside same line: update text and Width
// we dont add space character (should we?)
line.Text += block.GetText();
// gather statistics
statBackspace += (isBackspace) ? 1 : 0;
statSubfonts++;
// does not update 'last' variable!!
continue;
}
}
if ((last == null) || (CompareLine(block, last) != 0))
{
var b = (Block)block;
string text = block.GetText();
if (last_hidden != null)
{
text = last_hidden + text;
last_hidden = null;
}
line = new BlockLine()
{
Text = text,
X = block.GetX(),
H = block.GetH(),
Width = block.GetWidth(),
Height = block.GetHeight(),
HasBackColor = b.HasBackColor,
HasLargeSpace = false,
// might be inaccurate
FontFullName = b.FontFullName,
FontName = b.FontName,
FontSize = b.FontSize, // BE CAREFUL!
FontStyle = b.FontStyle
// now the settings are done in GroupFontLineHelper
};
// TODO: validar a entrada duas vezes
if (groupFont != null)
{
groupFont.Done();
}
groupFont = new GroupFontLineHelper(line, b);
if (line.Width <= 0 || line.Height <= 0)
{
PdfReaderException.AlwaysThrow("line.Width <= 0 || line.Height <= 0");
}
result.Add(line);
}
else
{
string separator = (ShouldAddSpace(last, block)) ? " " : "";
// same line: update text and Width
float startOfBlock = block.GetX();
float endOfBlock = block.GetX() + block.GetWidth();
float endOfLine = line.GetX() + line.GetWidth();
line.Text += separator + block.GetText();
line.Width = block.GetX() + block.GetWidth() - line.GetX();
if (line.Width <= 0)
{
PdfReaderException.AlwaysThrow("line.Width <= 0");
}
bool couldBeTable = ShouldAddLargeSpace(last, block);
if (couldBeTable)
{
line.HasLargeSpace = true;
}
// walking backwards
// very strict check: sometimes the start overlaps with the ending
//if (startOfBlock < endOfLine)
// throw new InvalidOperationException();
// soft check: end of block should never that low unless it is an overlap
if (endOfBlock < endOfLine)
{
PdfReaderException.AlwaysThrow("endOfBlock < endOfLine", new IBlock[] { last, block });
}
groupFont.MergeFont((Block)block);
}
last = block;
}
if (groupFont != null)
{
groupFont.Done();
}
return(result);
}