/// <summary>
/// Joins the horizontal and vertical lines.
/// </summary>
/// <param name="allLines">All the lines.</param>
/// <returns>The orizontal and the vertical lines (eventually joined)</returns>
private List <Line> JoinLines(List <Line> allLines)
{
JoinedVerticalLines = JoinVerticalLines(allLines);
JoinedHorizontalLines = JoinHorizontalLines(allLines);
return(JoinedHorizontalLines.Union(JoinedVerticalLines).ToList());
}
/// <summary>
/// Determines the table structures.
/// </summary>
public void DetermineTableStructures()
{
JoinedLines = JoinLines(AllLines);
// Find table borders
foreach (Line horizontalLine in JoinedHorizontalLines.OrderBy(_ => _.StartPoint.Y))
{
// We consider that this line is a top line of a table if
// 1. There is not a table with this line inside
// 2. There is a vertical line starting from this line
if (Tables.Any(_ => _.Contains(horizontalLine.StartPoint.Y)))
{
continue;
}
Line?tableLine = JoinedVerticalLines
.Where(_ => _.StartPoint == horizontalLine.StartPoint || _.StartPoint == horizontalLine.EndPoint)
.OrderByDescending(_ => _.EndPoint.Y - _.StartPoint.Y)
.Cast <Line?>()
.FirstOrDefault();
if (tableLine == null)
{
continue;
}
Table tableStructure = new Table()
{
TopLeftPoint = horizontalLine.StartPoint,
BottomRightPoint = new Point(horizontalLine.EndPoint.X, tableLine.Value.EndPoint.Y)
};
Tables.Add(tableStructure);
}
// Add the first row and the first column to all tables
foreach (Table tableStructure in Tables)
{
tableStructure.Rows.Add(new Row()
{
BeginY = tableStructure.TopLeftPoint.Y
});
tableStructure.Columns.Add(new Column()
{
BeginX = tableStructure.TopLeftPoint.X
});
}
// Find rows
foreach (Line horizontalLine in JoinedHorizontalLines.OrderBy(_ => _.StartPoint.Y))
{
var tableStructure = Tables.FirstOrDefault(_ => _.Contains(horizontalLine));
// No table contains this line
if (tableStructure == null)
{
continue;
}
// Check if the row already belongs to the table
if (tableStructure.Rows.Any(_ => Math.Abs(_.BeginY - horizontalLine.StartPoint.Y) < ContentExtractor.Tolerance))
{
continue;
}
// Check if the row is the bottom edge of the table
if (tableStructure.BottomRightPoint.Y - horizontalLine.StartPoint.Y < ContentExtractor.Tolerance)
{
continue;
}
tableStructure.Rows.Add(new Row()
{
BeginY = horizontalLine.StartPoint.Y
});
}
// Find columns
foreach (Line verticalLine in JoinedVerticalLines.OrderBy(_ => _.StartPoint.X))
{
var tableStructure = Tables.FirstOrDefault(_ => _.Contains(verticalLine));
// No table contains this line
if (tableStructure == null)
{
continue;
}
// The row already belongs to the table
if (tableStructure.Columns.Any(_ => Math.Abs(_.BeginX - verticalLine.StartPoint.X) < ContentExtractor.Tolerance))
{
continue;
}
// Check if the row is the bottom edge of the table
if (tableStructure.BottomRightPoint.X - verticalLine.StartPoint.X < ContentExtractor.Tolerance)
{
continue;
}
tableStructure.Columns.Add(new Column()
{
BeginX = verticalLine.StartPoint.X
});
}
// Fix EndX and EndY and indexes
foreach (Table tableStructure in Tables)
{
// Fix EndYs
for (int i = 0; i < tableStructure.Rows.Count - 1; i++)
{
tableStructure.Rows[i].EndY = tableStructure.Rows[i + 1].BeginY - ContentExtractor.Tolerance * 0.1f;
}
tableStructure.Rows[tableStructure.Rows.Count - 1].EndY = tableStructure.BottomRightPoint.Y;
// Fix EndXs
for (int i = 0; i < tableStructure.Columns.Count - 1; i++)
{
tableStructure.Columns[i].EndX = tableStructure.Columns[i + 1].BeginX - ContentExtractor.Tolerance * 0.1f;
}
tableStructure.Columns[tableStructure.Columns.Count - 1].EndX = tableStructure.BottomRightPoint.X;
int index;
index = 0;
foreach (var column in tableStructure.Columns.OrderBy(_ => _.BeginX))
{
column.Index = index;
index++;
}
index = 0;
foreach (var row in tableStructure.Rows.OrderByDescending(_ => _.BeginY))
{
row.Index = index;
index++;
}
tableStructure.CreateContent();
}
}
