private PdfTextElement BuildTextElement()
{
PdfTextElement textElem = new PdfTextElement();
textElem.Font = _font;
textElem.FontSize = _fontSize;
textElem.Matrix = _textMatrixCurrent.Multiply(_graphicsMatrix);
textElem.RawText = _sbText.ToString();
textElem.VisibleText = PdfString_ToUnicode(textElem.RawText, _font);
PdfCharElement lastChar = _listCharacters[_listCharacters.Count - 1];
textElem.VisibleWidth = (lastChar.Displacement + lastChar.Width) * textElem.Matrix.Matrix[0, 0];
textElem.VisibleHeight = (_font.Height * _fontSize) * textElem.Matrix.Matrix[1, 1];
textElem.Characters = new List <PdfCharElement>();
foreach (PdfCharElement c in _listCharacters)
{
textElem.Characters.Add(new PdfCharElement
{
Char = c.Char,
Displacement = (c.Displacement * textElem.Matrix.Matrix[0, 0]),
Width = (c.Width * textElem.Matrix.Matrix[0, 0]),
});
}
textElem.Childs = new List <PdfTextElement>();
return(textElem);
}
public string GetFieldAsString(string field, bool fuzzy = true)
{
PdfTextElement fieldTitle = FindElementByText(field, fuzzy);
if (fieldTitle == null)
{
return(null);
}
double titleX = fieldTitle.GetX();
var fieldData = new List <PdfTextElement>();
foreach (PdfTextElement elem in _textElements)
{
if (TextElementHorizontalIntersection(fieldTitle, elem) == false)
{
continue;
}
double elemX = elem.GetX();
if (elemX <= titleX)
{
continue;
}
fieldData.Add(elem);
}
if (fieldData.Count == 0)
{
return(null);
}
return(fieldData.OrderBy(elem => elem.GetX()).FirstOrDefault().VisibleText);
}
private bool TextElementVerticalIntersection(PdfTextElement elem1, double elem2X1, double elem2X2)
{
double elem1X1 = elem1.GetX();
double elem1X2 = elem1.GetX() + elem1.VisibleWidth;
return(elem1X2 >= elem2X1 && elem2X2 >= elem1X1);
}
private void SplitTextElements()
{
var textElementsSplitted = new List <PdfTextElement>();
while (_textElements.Count > 0)
{
PdfTextElement elem = _textElements[0];
_textElements.Remove(elem);
double maxWidth = elem.MaxWidth();
int prevBreak = 0;
for (int i = 1; i < elem.Characters.Count; i++)
{
double prevCharEnd = elem.Characters[i - 1].Displacement + elem.Characters[i - 1].Width;
double charSeparation = elem.Characters[i].Displacement - prevCharEnd;
if (charSeparation > maxWidth)
{
PdfTextElement partElem = elem.SubPart(prevBreak, i);
textElementsSplitted.Add(partElem);
prevBreak = i;
}
}
if (prevBreak == 0)
{
textElementsSplitted.Add(elem);
continue;
}
PdfTextElement lastElem = elem.SubPart(prevBreak, elem.Characters.Count);
textElementsSplitted.Add(lastElem);
}
_textElements = textElementsSplitted;
}
private void AddTextElement(PdfTextElement textElement)
{
if (string.IsNullOrEmpty(textElement.VisibleText.Trim()))
{
return;
}
_textElements.Add(textElement);
}
public PdfTextElementColumn(PdfTextElement head, IEnumerable <PdfTextElement> elements, double y, double x1, double x2)
{
HeadTextElement = head;
Elements = elements;
Y = y;
X1 = x1;
X2 = x2;
}
private bool TextElementHorizontalIntersection(PdfTextElement elem1, PdfTextElement elem2)
{
double elem1Y1 = elem1.GetY();
double elem1Y2 = elem1.GetY() + elem1.VisibleHeight;
double elem2Y1 = elem2.GetY();
double elem2Y2 = elem2.GetY() + elem2.VisibleHeight;
return(elem1Y2 >= elem2Y1 && elem2Y2 >= elem1Y1);
}
private static void DrawTextElement(PdfTextElement textElement, Graphics gc, Pen penTextElem, Pen penCharElem, int scale, int pageHeight, double pageXMin, double pageYMin, Brush brushText)
{
if (textElement == null)
{
return;
}
double textElementX = textElement.GetX() - pageXMin;
double textElementY = textElement.GetY() - pageYMin;
double textElementWidth = textElement.VisibleWidth;
double textElementHeight = textElement.VisibleHeight;
string textElementText = textElement.VisibleText;
string textElementFontName = (textElement.Font == null ? string.Empty : textElement.Font.Name);
if (textElementHeight < 0.0001)
{
return;
}
double textElementPageX = textElementX;
double textElementPageY = pageHeight - textElementY;
if (penTextElem != null)
{
DrawRoundedRectangle(gc, penTextElem,
(int)(textElementPageX * scale),
(int)(textElementPageY * scale),
(int)(textElementWidth * scale),
(int)(textElementHeight * scale),
5);
}
using (Font font = new Font("Arial", (int)(textElementHeight * scale), GraphicsUnit.Pixel))
{
foreach (PdfCharElement c in textElement.Characters)
{
gc.DrawString(c.Char,
font,
brushText,
(int)((textElementPageX + c.Displacement) * scale),
(int)(textElementPageY * scale));
if (penCharElem != null)
{
DrawRoundedRectangle(gc, penCharElem,
(int)((textElementPageX + c.Displacement) * scale),
(int)(textElementPageY * scale),
(int)(c.Width * scale),
(int)(textElementHeight * scale),
5);
}
}
}
}
public PdfTextElementColumn GetColumn(string column, bool fuzzy = true)
{
PdfTextElement columnHead = FindElementByText(column, fuzzy);
if (columnHead == null)
{
return(PdfTextElementColumn.Empty);
}
double headY = columnHead.GetY() - columnHead.VisibleHeight;
double headX1 = columnHead.GetX();
double headX2 = headX1 + columnHead.VisibleWidth;
// Determine horizontal extent
double extentX1 = double.MinValue;
double extentX2 = double.MaxValue;
foreach (PdfTextElement elem in _textElements)
{
if (elem == columnHead)
{
continue;
}
if (TextElementHorizontalIntersection(columnHead, elem) == false)
{
continue;
}
double elemX1 = elem.GetX();
double elemX2 = elemX1 + elem.VisibleWidth;
if (elemX2 < headX1)
{
if (elemX2 > extentX1)
{
extentX1 = elemX2;
}
}
if (elemX1 > headX2)
{
if (elemX1 < extentX2)
{
extentX2 = elemX1;
}
}
}
PdfTextElementColumn columnData = GetColumn(columnHead, headY, headX1, headX2, extentX1, extentX2);
return(columnData);
}
private void FlushTextElement()
{
if (_sbText.Length == 0)
{
_textWidth = 0;
return;
}
PdfTextElement textElem = BuildTextElement();
AddTextElement(textElem);
_textMatrixCurrent.Matrix[0, 2] += _textWidth;
_sbText = new StringBuilder();
_listCharacters.Clear();
_textWidth = 0;
}
public PdfTextElement SubPart(int startIndex, int endIndex)
{
PdfTextElement blockElem = new PdfTextElement
{
Font = null,
FontSize = FontSize,
Matrix = Matrix.Copy(),
RawText = RawText.Substring(startIndex, endIndex - startIndex),
VisibleText = VisibleText.Substring(startIndex, endIndex - startIndex),
VisibleWidth = 0,
VisibleHeight = VisibleHeight,
Characters = new List <PdfCharElement>(),
Childs = new List <PdfTextElement>(),
};
double displacement = Characters[startIndex].Displacement;
blockElem.Matrix.Matrix[0, 2] += displacement;
for (int j = startIndex; j < endIndex; j++)
{
blockElem.Characters.Add(new PdfCharElement
{
Char = Characters[j].Char,
Displacement = Characters[j].Displacement - displacement,
Width = Characters[j].Width,
});
}
PdfCharElement lastChar = blockElem.Characters[blockElem.Characters.Count - 1];
blockElem.VisibleWidth = lastChar.Displacement + lastChar.Width;
foreach (PdfTextElement elem in Childs)
{
blockElem.Childs.Add(elem);
}
return(blockElem);
}
public PdfTextElementColumn GetColumn(PdfTextElement columnHead, double headY, double headX1, double headX2, double extentX1, double extentX2)
{
// Get all the elements that intersects vertically, are down and sort results
var columnDataRaw = new List <PdfTextElement>();
foreach (PdfTextElement elem in _textElements)
{
if (TextElementVerticalIntersection(elem, headX1, headX2) == false)
{
continue;
}
// Only intems down the column
double elemY = elem.GetY();
if (elemY >= headY)
{
continue;
}
columnDataRaw.Add(elem);
}
columnDataRaw = columnDataRaw.OrderByDescending(elem => elem.GetY()).ToList();
// Only items completelly inside extents, try spliting big elements and break on big elements that can't be splitted
var columnElements = new List <PdfTextElement>();
foreach (PdfTextElement elem in columnDataRaw)
{
double elemX1 = elem.GetX();
double elemX2 = elemX1 + elem.VisibleWidth;
// Add elements completely inside
if (elemX1 > extentX1 && elemX2 < extentX2)
{
columnElements.Add(elem);
continue;
}
// Try to split elements intersecting extents of the column
double maxSpacing = elem.Characters.Average(c => c.Width) / 10;
int indexStart = 0;
int indexEnd = elem.Characters.Count - 1;
bool indexStartValid = true;
bool indexEndValid = true;
if (elemX1 < extentX1)
{
// Search best start
int index = 0;
double characterPosition = elemX1 + elem.Characters[index].Displacement;
while (characterPosition < extentX1 && index < (elem.Characters.Count - 1))
{
index++;
characterPosition = elemX1 + elem.Characters[index].Displacement;
}
double spacing = elem.GetCharacterPreviousSpacing(index);
while (spacing < maxSpacing && index < (elem.Characters.Count - 1))
{
index++;
spacing = elem.GetCharacterPreviousSpacing(index);
}
if (spacing < maxSpacing)
{
indexStartValid = false;
}
indexStart = index;
}
if (elemX2 > extentX2)
{
// Search best end
int index = elem.Characters.Count - 1;
double characterPosition = elemX1 + elem.Characters[index].Displacement + elem.Characters[index].Width;
while (characterPosition > extentX2 && index > 0)
{
index--;
characterPosition = elemX1 + elem.Characters[index].Displacement + elem.Characters[index].Width;
}
double spacing = elem.GetCharacterPrecedingSpacing(index);
while (spacing < maxSpacing && index > 0)
{
index--;
spacing = elem.GetCharacterPrecedingSpacing(index);
}
if (spacing < maxSpacing)
{
indexEndValid = false;
}
indexEnd = index;
}
// Break when there is no good split, spaning all extent
if (indexStartValid == false && indexEndValid == false)
{
break;
}
// Continue when only one of the sides is invalid. (outside elements intersecting extents of the column)
if (indexStartValid == false || indexEndValid == false)
{
continue;
}
// Add splitted element
columnElements.Add(elem.SubPart(indexStart, indexEnd + 1));
}
var columnData = new PdfTextElementColumn(columnHead, columnElements, headY, extentX1, extentX2);
return(columnData);
}
private void JoinTextElements()
{
var textElementsCondensed = new List <PdfTextElement>();
while (_textElements.Count > 0)
{
PdfTextElement elem = _textElements[0];
_textElements.Remove(elem);
double blockY = elem.GetY();
double blockXMin = elem.GetX();
double blockXMax = blockXMin + elem.VisibleWidth;
// Prepare first neighbour
var textElementNeighbours = new List <PdfTextElement>();
textElementNeighbours.Add(elem);
// Search Neighbours
int i = 0;
while (i < _textElements.Count)
{
PdfTextElement neighbour = _textElements[i];
if (neighbour.Font != elem.Font || neighbour.FontSize != elem.FontSize)
{
i++;
continue;
}
double neighbourY = neighbour.GetY();
if (Math.Abs(neighbourY - blockY) > 0.001)
{
i++; continue;
}
double maxWidth = neighbour.MaxWidth();
double neighbourXMin = neighbour.GetX();
double neighbourXMax = neighbourXMin + neighbour.VisibleWidth;
double auxBlockXMin = blockXMin - maxWidth;
double auxBlockXMax = blockXMax + maxWidth;
if (auxBlockXMax >= neighbourXMin && neighbourXMax >= auxBlockXMin)
{
_textElements.Remove(neighbour);
textElementNeighbours.Add(neighbour);
if (blockXMax < neighbourXMax)
{
blockXMax = neighbourXMax;
}
if (blockXMin > neighbourXMin)
{
blockXMin = neighbourXMin;
}
i = 0;
continue;
}
i++;
}
if (textElementNeighbours.Count == 1)
{
textElementsCondensed.Add(elem);
continue;
}
// Join neighbours
var chars = new List <PdfCharElement>();
foreach (PdfTextElement neighbour in textElementNeighbours)
{
double neighbourXMin = neighbour.GetX();
foreach (PdfCharElement c in neighbour.Characters)
{
chars.Add(new PdfCharElement
{
Char = c.Char,
Displacement = (c.Displacement + neighbourXMin) - blockXMin,
Width = c.Width,
});
}
}
chars = chars.OrderBy(c => c.Displacement).ToList();
var sbText = new StringBuilder();
foreach (PdfCharElement c in chars)
{
sbText.Append(c.Char);
}
PdfTextElement blockElem = new PdfTextElement
{
Font = null,
FontSize = elem.FontSize,
Matrix = elem.Matrix.Copy(),
RawText = sbText.ToString(),
VisibleText = sbText.ToString(),
VisibleWidth = blockXMax - blockXMin,
VisibleHeight = elem.VisibleHeight,
Characters = chars,
Childs = textElementNeighbours,
};
blockElem.Matrix.Matrix[0, 2] = blockXMin;
textElementsCondensed.Add(blockElem);
}
_textElements = textElementsCondensed;
}