private static void WriteLine(this LineTable lta, MWord.Range range, int fontsize = 12)
{
var oTable = range.Tables.Add(range.Bookmarks["\\endofdoc"].Range, lta.RowCount + 1, lta.ColCount + 1);
for (var r = 0; r < lta.RowCount + 1; r++)
{
for (var c = 0; c < lta.ColCount + 1; c++)
{
var cellRange = oTable.Cell(r + 1, c + 1).Range;
cellRange.Text = lta.Content[r, c];
cellRange.OMaths.Add(cellRange);
cellRange.OMaths[1].BuildUp();
cellRange.OMaths[1].Range.Font.Size = fontsize;
}
foreach (var merge in lta.CellRange2Merge)
{
var cell1 = oTable.Cell(merge[0, 0] + 1, merge[0, 1] + 1);
var cell2 = oTable.Cell(merge[1, 0] + 1, merge[1, 1] + 1);
cell1.Merge(cell2);
}
oTable.AutoFitBehavior(WdAutoFitBehavior.wdAutoFitContent);
oTable.Borders.OutsideLineStyle = WdLineStyle.wdLineStyleOutset;
oTable.Borders.InsideLineStyle = WdLineStyle.wdLineStyleSingle;
}
}
private static void CondenseTable(LineTable table)
{
var cMax = table.LineBoxes.Max(x => x.Col_Max);
var rMax = table.LineBoxes.Max(x => x.Row_Max);
//var keep_col_single = table.LineBoxes.Where(x => x.Column_Span == 1).Select(x => x.Column_Min);
//var keep_row_single = table.LineBoxes.Where(x => x.Row_Span == 1).Select(x => x.Row_Min);
//var keep_col_mult = table.LineBoxes.Where(x => x.Column_Span > 1).Where(x => !keep_col_single.Any(c => c >= x.Column_Min || c <= x.Column_Max)).Select(x => x.Column_Min);
//var keep_row_mult = table.LineBoxes.Where(x => x.Row_Span > 1).Where(x => !keep_row_single.Any(c => c >= x.Row_Min || c <= x.Row_Max)).Select(x => x.Row_Min);
//var keep_col = keep_col_mult.Concat(keep_col_single).Distinct().ToList();
//var keep_row = keep_row_mult.Concat(keep_row_single).Distinct().ToList();
var keep_col = table.LineBoxes.SelectMany(x => new[] { x.Col_Max }).Distinct();
var keep_row = table.LineBoxes.SelectMany(x => new[] { x.Row_Max }).Distinct();
var remove_col = Enumerable.Range(0, cMax + 1).Except(keep_col).ToList();
var remove_row = Enumerable.Range(0, rMax + 1).Except(keep_row).ToList();
table.LineBoxes.ForEach(x =>
{
x.Col_Min -= remove_col.Count(y => y < x.Col_Min);
x.Col_Max -= remove_col.Count(y => y < x.Col_Max);
x.Row_Min -= remove_row.Count(y => y < x.Row_Min);
x.Row_Max -= remove_row.Count(y => y < x.Row_Max);
});
}
public static TableInfo FindDataRowsAndColumns(LineTable table)
{
var rMax = table.LineBoxes.Max(x => x.Row_Max);
var cMax = table.LineBoxes.Max(x => x.Col_Max);
var cols = Enumerable.Range(0, cMax + 1).Select(c => new ColInfo {
Col = c, Cells = table.LineBoxes.Where(x => c >= x.Col_Min && c <= x.Col_Max).OrderBy(x => x.Row_Min).ToList()
}).ToList();
var rows = Enumerable.Range(0, rMax + 1).Select(r => new RowInfo {
Row = r, Cells = table.LineBoxes.Where(x => r >= x.Row_Min && r <= x.Row_Max).OrderBy(x => x.Col_Min).ToList()
}).ToList();
// var rowMatches = rows.Select((row, i) => new { row, matchingRows = rows.Skip(i).Where(r2 => IsSimilarRow(row, r2)).ToList() }).ToList();
// Add row values
foreach (var row in rows)
{
row.Values = row.Cells.Select(x => new ValueInfo
{
Cell = x,
Cols = cols.Where(y => y.Cells.Contains(x)).ToList()
}).ToList();
}
// Identify Data Rows
var rowClusters = rows.Select(x => new List <RowInfo>()
{
x
}).ToList();
for (var i = 0; i < rowClusters.Count; i++)
{
for (var j = i + 1; j < rowClusters.Count; j++)
{
if (rowClusters[i].Any(x => rowClusters[j].Any(y => IsSimilarRow(x, y))))
{
rowClusters[i].AddRange(rowClusters[j]);
rowClusters.RemoveAt(j);
j--;
}
}
}
var topCluster = rowClusters.OrderByDescending(x => x.Count).FirstOrDefault();
if (topCluster.Count > 1)
{
topCluster.ForEach(x => x.IsDataRow = true);
topCluster.ForEach(x => x.Cells.ForEach(y => y.IsDataCell = true));
}
return(new TableInfo()
{
Cols = cols, Rows = rows
});
}
public static string ToClipboard_HtmlTableFormat(this LineTable table)
{
var sb = new StringBuilder();
sb.Append("<table>");
var cMax = table.LineBoxes.Max(x => x.Col_Max);
var rMax = table.LineBoxes.Max(x => x.Row_Max);
// Reverse Rows
for (var r = rMax; r >= 0; r--)
{
sb.Append("<tr>");
for (var c = 0; c <= cMax; c++)
{
var v = table.LineBoxes.Where(x => x.Col_Min <= c && x.Col_Max >= c && x.Row_Min <= r && x.Row_Max >= r).OrderBy(x => x.Col_Min).ThenBy(x => x.Row_Min).FirstOrDefault();
// Skip if in span
if (v?.Col_Min != c || v?.Row_Max != r)
{
continue;
}
var span = $"{(v?.Row_Span > 1 ? $"rowspan={v.Row_Span}" : "")} {(v?.Col_Span > 1 ? $"colspan={v.Col_Span}" : "")}";
// Force excel text format
sb.Append($"<td {span} style='mso-number-format:\"\\@\"'>");
sb.Append(v?.CellText ?? "");
sb.Append("</td>");
}
sb.Append("</tr>");
}
sb.Append("</table>");
var html = sb.ToString();
return(html);
}
public ITable <T> Table <T>() where T : ITableRow
{
ITable <T> table = null;
IConnector connector = new SqlServerConnector(ConnectionString);
if (typeof(T) == typeof(Station))
{
table = new StationTable(connector) as ITable <T>;
}
else if (typeof(T) == typeof(Line))
{
table = new LineTable(connector) as ITable <T>;
}
else if (typeof(T) == typeof(StationLine))
{
table = new StationLineTable(connector) as ITable <T>;
}
else if (typeof(T) == typeof(MetroTransfer))
{
table = new MetroTransferTable(connector) as ITable <T>;
}
return(table);
}
public static TableData ExtractTable(LineTable table, TablePattern pattern)
{
var tableInfo = TableDataRowFinder.FindDataRowsAndColumns(table);
var columnPatterns = AssignColumns(pattern, tableInfo);
if (columnPatterns.Any(x => x.Key.IsRequired && x.Value == null))
{
return(null);
}
var columns = columnPatterns.Select(x => new TableDataColumn()
{
Name = x.Key.Name,
SourceBounds = x.Value?.Cells.Where(c => c.Col_Span == 1).Select(c => c.Box.Bounds).Where(b => b != new Bounds()).UnionBounds() ?? new Bounds(),
}).ToList();
var rows = tableInfo.Rows.Select(row =>
{
var rowOut = new TableDataRow()
{
Values = columnPatterns.Select(col =>
{
var v = row.Values.Where(x => x.Cols.Contains(col.Value)).FirstOrDefault();
if (v == null)
{
return(null);
}
// Ensure value matches pattern
if (!col.Key.ValueMatchRegex.IsMatch(v.Cell.CellText))
{
return(null);
}
var column = columns.Where(c => c.Name == col.Key.Name).FirstOrDefault();
return(new TableDataValue()
{
Column = column,
Value = v.Cell.CellText,
MergeId = v.Cell.Col_Span > 1 || v.Cell.Row_Span > 1 ? "" + v.Cell.CellId : null as string,
SourceBounds = v.Cell.Box.Bounds,
FontHeight = v.Cell.Box.Texts.Average(x => x.FontHeight),
});
}).Where(x => x != null).ToList()
};
if (!columnPatterns.Keys.Where(x => x.IsRequired).All(x => rowOut.Values.Any(y => y.Column.Name == x.Name)))
{
return(null);
}
return(rowOut);
}).Where(x => x != null).ToList();
// columns.ForEach(col => col.SourceBounds = rows.SelectMany(x => x.Values).Where(v => v.Column == col).Select(x => x.SourceBounds).UnionBounds());
// columns.ForEach(col => col.SourceHeaderText = rows.SelectMany(x => x.Values).Where(v => v.Column == col).Select(x => x.SourceBounds).UnionBounds());
return(new TableData()
{
SourceBounds = table.Bounds,
// SourceBounds = rows.SelectMany(x => x.Values.Select(v => v.SourceBounds)).UnionBounds(),
SourceBounds_Cropped = rows.SelectMany(x => x.Values.Select(v => v.SourceBounds)).UnionBounds(),
TableName = pattern.Name,
Rows = rows,
Columns = columns,
});
}
