static int ProcessTitle(IAdapter adapter, ColumnOrdering columnOrdering)
{
int row = 0;
string title = null;
string ministry = null;
int? year = null;
bool findTitle = false;
bool prevRowIsSection = false;
while (true)
{
var currRow = adapter.GetCells(row);
string section_text;
bool isSection = IAdapter.IsSectionRow(currRow, adapter.GetColsCount(), prevRowIsSection, out section_text);
if (isSection)
{
if (section_text.Length > 20)
{
if (GetValuesFromTitle(section_text, ref title, ref year, ref ministry))
{
findTitle = true;
}
}
else
{
columnOrdering.Section = section_text;
}
}
else if (WeakHeaderCheck(currRow))
{
break;
}
row += 1;
if (row >= adapter.GetRowsCount())
{
row = 0;
break;
throw new ColumnDetectorException(String.Format("Headers not found"));
}
prevRowIsSection = isSection;
}
if (!findTitle)
{
if (GetValuesFromTitle(adapter.GetTitleOutsideTheTable(), ref title, ref year, ref ministry))
{
findTitle = true;
}
}
if (findTitle)
{
columnOrdering.Title = title;
columnOrdering.Year = year;
columnOrdering.MinistryName = ministry;
}
return(row);
}
static List <Cell> FindSubcellsUnder(IAdapter adapter, Cell cell)
{
var subCells = new List <Cell>();
if (cell.Row + cell.MergedRowsCount >= adapter.GetRowsCount())
{
return(subCells);
}
if (cell.CellWidth == 0 && cell.GetText(true).Trim() == "")
{
return(subCells);
}
var undercCells = adapter.GetCells(cell.Row + cell.MergedRowsCount);
foreach (var underCell in undercCells)
{
if (underCell.Col < cell.Col)
{
continue;
}
if (underCell.Col >= cell.Col + cell.MergedColsCount)
{
break;
}
if (!underCell.IsEmpty)
{
subCells.Add(underCell);
}
}
return(subCells);
}
public static DeclarationField PredictEmptyColumnTitle(IAdapter adapter, Cell headerCell)
{
List <string> texts = new List <string>();
int rowIndex = headerCell.Row + headerCell.MergedRowsCount;
const int maxRowToCollect = 10;
for (int i = 0; i < maxRowToCollect; i++)
{
var cells = adapter.GetCells(rowIndex, IAdapter.MaxColumnsCount);
string dummy;
if (IAdapter.IsSectionRow(cells, adapter.GetColsCount(), false, out dummy))
{
rowIndex += 1;
}
else
{
var c = adapter.GetCell(rowIndex, headerCell.Col);
if (c != null)
{
texts.Add(c.GetText(true));
rowIndex += c.MergedRowsCount;
}
else
{
rowIndex += 1;
}
}
if (rowIndex >= adapter.GetRowsCount())
{
break;
}
}
var field = PredictByStrings(texts);
if (headerCell.TextAbove != null && ((field & DeclarationField.AllOwnTypes) > 0))
{
string h = headerCell.TextAbove;
// AllOwnTypes defined from
field &= ~DeclarationField.AllOwnTypes;
if (HeaderHelpers.IsMixedColumn(h))
{
field |= DeclarationField.Mixed;
}
else if (HeaderHelpers.IsStateColumn(h))
{
field |= DeclarationField.State;
}
else if (HeaderHelpers.IsOwnedColumn(h))
{
field |= DeclarationField.Owned;
}
}
Logger.Debug(string.Format("predict by {0} -> {1}",
String.Join("\\n", texts), field));
return(field);
}
static public List <Cell> GetColumnCells(IAdapter adapter, int headerStartRow, out int headerEndRow)
{
headerEndRow = headerStartRow + 1;
var firstRow = adapter.GetCells(headerStartRow);
List <Cell> columnCells = new List <Cell>();
bool headerCanHaveSecondLevel = true;
int maxMergedRows = 1;
var texts = new List <string>();
foreach (var cell in firstRow)
{
string text = cell.GetText(true);
if (adapter.GetRowsCount() == cell.MergedRowsCount)
{
continue;
}
if (cell.CellWidth == 0 && text.Trim() == "")
{
continue;
}
if (maxMergedRows < cell.MergedRowsCount)
{
maxMergedRows = cell.MergedRowsCount;
}
var underCells = FindSubcellsUnder(adapter, cell);
if (underCells.Count() <= 1 || !headerCanHaveSecondLevel)
{
headerEndRow = Math.Max(headerEndRow, cell.Row + cell.MergedRowsCount);
// иногда в двухярусном заголовке в верхней клетке пусто, а в нижней есть заголовок (TwoRowHeaderEmptyTopCellTest)
if (text.Trim() == "" && cell.MergedRowsCount < maxMergedRows && underCells.Count() == 1)
{
columnCells.Add(underCells.First());
}
else
{
columnCells.Add(cell);
}
texts.Add(cell.Text.NormSpaces());
// обработка ошибки документа DepEnergo2010
if (columnCells.Count == 1 && cell.MergedRowsCount == 1 && underCells.Count == 1)
{
string cellBelowName = underCells[0].GetText(true);
headerCanHaveSecondLevel = cellBelowName.Length < 5;
}
}
// current cell spans several columns, so the header probably occupies two rows instead of just one
// with the second row reserved for subheaders
else
{
foreach (var underCell in underCells)
{
underCell.TextAbove = cell.Text.NormSpaces();
columnCells.Add(underCell);
texts.Add(underCell.TextAbove + "^" + underCell.Text.NormSpaces());
}
headerEndRow = Math.Max(headerEndRow, underCells[0].Row + underCells[0].MergedRowsCount);
}
}
Logger.Debug("column titles: " + String.Join("|", texts));
return(columnCells);
}