public bool DoesWordCrossWord(CustomCell cell)
{
return((this.Top >= cell.Top && this.Top <= cell.Bottom) ||
(cell.Top >= this.Top && cell.Top <= this.Bottom) ||
(this.Bottom <= cell.Bottom && this.Bottom >= cell.Top) ||
(cell.Bottom <= this.Bottom && cell.Bottom >= this.Top));
}
private List <CustomRow> GetLinesFromWordsInRows(List <XmlElement> rows, int pageIndex)
{
List <CustomRow> lines = new List <CustomRow>();
foreach (XmlElement row in rows)
{
CustomRow line = new CustomRow()
{
IsTableRow = true,
Top = -1,
Bottom = -1
};
foreach (XmlElement xmlCell in row.GetElementsByTagName("cell"))
{
List <CustomCell> cells = xmlCell.GetElementsByTagName("line").Cast <XmlElement>().
Select(x => new CustomCell
{
Top = int.Parse(x.Attributes["t"].Value),
Bottom = int.Parse(x.Attributes["b"].Value),
Right = int.Parse(x.Attributes["r"].Value),
Left = int.Parse(x.Attributes["l"].Value),
BaseLine = int.Parse(x.Attributes["baseline"].Value),
Text = (x as XmlElement).InnerText,
}).ToList();
if (cells.Any())
{
cells = cells.OrderBy(x => x.Top).ToList();
int cellTop = cells.Min(x => x.Top);
int cellBottom = cells.Max(x => x.Bottom);
line.Top = (line.Top != -1) ? Math.Min(line.Top, cellTop) : cellTop;
line.Bottom = (line.Bottom != -1) ? Math.Max(line.Bottom, cellBottom) : cellBottom;
CustomCell newCellWord = new CustomCell()
{
Top = cellTop,
Bottom = cellBottom,
Left = cells.Min(x => x.Left),
Right = cells.Max(x => x.Right),
Text = string.Empty
};
foreach (CustomCell cell in cells)
{
newCellWord.Text += (cell != cells.Last()) ? $"{cell.Text} " : cell.Text;
}
line.Cells.Add(newCellWord);
}
}
if (line.Cells.Any())
{
lines.Add(line);
}
}
return(lines);
}
public bool AddWord(CustomCell cell)
{
if (Left == 0 && Right == 0)
{
Cells.Add(cell);
Left = cell.Left;
Right = cell.Right;
return(true);
}
else if ((cell.Left >= Left && cell.Left <= Right) ||
(cell.Right <= Right && cell.Right >= Left))
{
Cells.Add(cell);
Left = Math.Min(Left, cell.Left);
Right = Math.Max(Right, cell.Right);
return(true);
}
return(false);
}
private List <CustomRow> GetLinesFromWords(List <CustomCell> cellsNotInRows)
{
cellsNotInRows = cellsNotInRows.ToArray().ToList();
List <CustomRow> lines = new List <CustomRow>();
while (cellsNotInRows.Any())
{
CustomCell currentCell = cellsNotInRows.First();
System.Drawing.Rectangle currentLineBoundaries = System.Drawing.Rectangle.Empty;
// finding highest element in current line
foreach (CustomCell cell in cellsNotInRows)
{
System.Drawing.Rectangle boundaries = currentCell.GetUnionBoundaries(cell);
if (boundaries != System.Drawing.Rectangle.Empty &&
boundaries.Bottom - boundaries.Top > currentLineBoundaries.Bottom - currentLineBoundaries.Top)
{
currentLineBoundaries = boundaries;
}
}
// creating new line
CustomRow line = new CustomRow()
{
Top = currentLineBoundaries.Top,
Bottom = currentLineBoundaries.Bottom
};
lines.Add(line);
foreach (CustomCell cell in cellsNotInRows)
{
if (cell.IsInsideVerticalBoundaries(currentLineBoundaries))
{
line.Cells.Add(cell);
}
}
// removing words from words collection
foreach (CustomCell cell in line.Cells)
{
cellsNotInRows.Remove(cell);
}
}
// sorting words in line (by both x and y axis)
foreach (CustomRow line in lines)
{
List <CustomCell> cellsTmp = line.Cells.ToList();
while (cellsTmp.Any())
{
WordsGroup wordsGroup = new WordsGroup();
CustomCell cell = cellsTmp.First();
wordsGroup.AddWord(cell);
cellsTmp.Remove(cell);
bool addedWord = false;
while (addedWord)
{
foreach (CustomCell tmpCell in cellsTmp)
{
addedWord = wordsGroup.AddWord(tmpCell) ? true : addedWord;
cellsTmp.Remove(tmpCell);
if (addedWord)
{
break;
}
}
}
line.CellsGroups.Add(wordsGroup);
}
line.SortCells();
}
return(lines);
}
public System.Drawing.Rectangle GetUnionBoundaries(CustomCell cell)
{
return(DoesWordCrossWord(cell) ? System.Drawing.Rectangle.FromLTRB(Math.Min(cell.Left, Left), Math.Min(cell.Top, Top), Math.Max(cell.Right, Right), Math.Max(cell.Bottom, Bottom)) : System.Drawing.Rectangle.Empty);
}