public List <TablePos> GetOnlyTablePos(string tableName, LocateConfiguration locateConfiguration, bool isChineseTableName)
{
List <TablePos> tablePoses = new List <TablePos>();
LinePos linePos = null;
_locateConfiguration = locateConfiguration;
LinePos tableNamePos = GetSearchTextResult(tableName, tableName, ref linePos, true);
if (tableNamePos == null)
{
linePos = null;
tableNamePos = GetSearchTextResult(tableName.GetFirstLetter(isChineseTableName), tableName, ref linePos, true);
}
if (tableNamePos != null)
{
string tableEndRegex = locateConfiguration.TableEndRegex;
string tableEndFirstLetterRegex = locateConfiguration.TableEndFirstLetterRegex;
LinePos tableEndPos = GetSearchTextResult(tableEndRegex, tableEndRegex, ref linePos, false);
if (tableEndPos == null)
{
linePos = null;
tableEndPos = GetSearchTextResult(tableEndFirstLetterRegex, tableEndRegex, ref linePos, false);
}
if (tableEndPos == null)
{
tableEndPos = pdfTronHelper.GetBottomPosOfPage(tableNamePos.PageNum, false, -1, -1);
}
tablePoses = GetTablePoses(tableNamePos, tableEndPos);
}
return(tablePoses);
}
public int CompareTo(object obj)
{
LinePos linePos = obj as LinePos;
if (PageNum < linePos.PageNum)
{
return(-1);
}
if (PageNum == linePos.PageNum)
{
if (AxisValue == linePos.AxisValue)
{
return(0);
}
return(AxisValue < linePos.AxisValue ? 1 : -1);
}
if (PageNum > linePos.PageNum)
{
return(1);
}
return(0);
}
private LinePos GetSearchTextResult(string searchText, string verifyText, ref LinePos lowRange, bool isTableName)
{
try
{
bool isDone = false;
int startPage = pdfTronHelper.GetLowPage(lowRange);
while (!isDone)
{
List <LinePos> linePoses = pdfTronHelper.Search(searchText, lowRange, null, ref startPage);
if (linePoses.Count > 0)
{
foreach (LinePos linePos in linePoses)
{
if (!Regex.IsMatch(linePos.TrimText, verifyText))
{
continue;
}
if (isTableName)
{
string tableNameNearbyRegex = _locateConfiguration.TableNameNearbyRegex;
string tableNameNearbyFirstLetterRegex = _locateConfiguration.TableNameNearbyFirstLetterRegex;
LinePos linePosCopy = (LinePos)linePos.Clone();
string nearByRegexExpression = tableNameNearbyRegex;
LinePos tableNearbyPos = GetSearchTextResult(nearByRegexExpression, nearByRegexExpression, ref linePosCopy, false);
if (tableNearbyPos == null)
{
LinePos lowRangeCopy2 = (LinePos)linePos.Clone();
tableNearbyPos = GetSearchTextResult(tableNameNearbyFirstLetterRegex, nearByRegexExpression, ref lowRangeCopy2, false);
}
if (tableNearbyPos == null || !ValidateTableNameNearbyPos(linePos, tableNearbyPos))
{
continue;
}
}
lowRange = new LinePos()
{
PageNum = linePos.PageNum,
AxisValue = linePos.AxisValue + GetError(true, linePos.PageNum),
TrimText = linePos.TrimText
};
linePos.AxisValue = linePos.AxisValueWithLineHeight;
return(linePos);
}
}
isDone = startPage > pdfDoc.GetPageCount();
}
}
catch (Exception ex)
{
string err = ex.ToString();
throw;
}
return(null);
}
public List <TablePos> GetMultiTablePos(string tableName, LocateConfiguration locateConfiguration)
{
List <TablePos> tablePoses = new List <TablePos>();
_locateConfiguration = locateConfiguration;
LinePos linePos = null;
List <LinePos> tableNamePoses = new List <LinePos>();
List <LinePos> tableEndPoses = new List <LinePos>();
LinePos tableNamePos = null;
do
{
LinePos linePosCopy = linePos == null ? null : (LinePos)linePos.Clone();
tableNamePos = GetSearchTextResult(tableName, tableName, ref linePos, true);
if (tableNamePos == null)
{
linePos = linePosCopy;
tableNamePos = GetSearchTextResult(tableName.GetFirstLetter(true), tableName, ref linePos, true);
}
if (tableNamePos != null)
{
string tableEndRegex = locateConfiguration.TableEndRegex;
string tableEndFirstLetterRegex = locateConfiguration.TableEndFirstLetterRegex;
linePosCopy = linePos == null ? null : (LinePos)linePos.Clone();
LinePos tableEndPos = GetSearchTextResult(tableEndRegex, tableEndRegex, ref linePos, false);
if (tableEndPos == null)
{
linePos = linePosCopy;
tableEndPos = GetSearchTextResult(tableEndFirstLetterRegex, tableEndRegex, ref linePos, false);
}
if (tableEndPos == null)
{
tableEndPos = pdfTronHelper.GetBottomPosOfPage(tableNamePos.PageNum, false, -1, -1);
}
tableNamePoses.Add(tableNamePos);
tableEndPoses.Add(tableEndPos);
}
else
{
break;
}
} while (tableNamePos != null);
for (int i = 0; i < tableNamePoses.Count; i++)
{
List <TablePos> tempPoses = GetTablePoses(tableNamePoses[i], tableEndPoses[i]);
tablePoses.AddRange(tempPoses);
}
return(tablePoses);
}
private List <TablePos> GetTablePoses(LinePos tableNamePos, LinePos tableEndPos)
{
List <TablePos> tablePoses = new List <TablePos>();
tableNamePos.AxisValue = tableNamePos.AxisValueWithLineHeight;
//There is only one page
if (tableEndPos.PageNum == tableNamePos.PageNum)
{
TablePos tablePos = GetTablePos(tableNamePos.PageNum, tableNamePos, tableEndPos, false, false, null);
if (tablePos != null)
{
tablePoses.Add(tablePos);
}
}
//The page amount is over one
else
{
//Get start page
//Mark
TablePos startTablePos = null;
LinePos endPos = pdfTronHelper.GetBottomPosOfPage(tableNamePos.PageNum, false, -1, -1);
startTablePos = GetTablePos(tableNamePos.PageNum, tableNamePos, endPos, false, false, null);
if (startTablePos != null)
{
tablePoses.Add(startTablePos);
}
bool isStartTablePosNull = startTablePos == null;
//Get medial page
for (int i = tableNamePos.PageNum + 1; i < tableEndPos.PageNum; i++)
{
TablePos intervalTablePos = GetTablePos(i, null, pdfTronHelper.GetBottomPosOfPage(i, false, -1, -1),
false,
!isStartTablePosNull, isStartTablePosNull ? null : startTablePos.VerticalLines);
if (intervalTablePos != null)
{
tablePoses.Add(intervalTablePos);
}
}
//Get end page
int endPageNum = tableEndPos.PageNum;
TablePos endTablePos = GetTablePos(endPageNum, null, tableEndPos, true,
!isStartTablePosNull, isStartTablePosNull ? null : startTablePos.VerticalLines);
if (endTablePos != null)
{
tablePoses.Add(endTablePos);
}
}
return(tablePoses);
}
private TablePos GetTablePos(int pageNum, LinePos startPos, LinePos endPos, bool isEndTablePos, bool isSubsequentPage, SortedDictionary <double, FormLineList> lastPageVerticalLines)
{
FormLineSearcher searcher = new FormLineSearcher(pdfDoc);
if (endPos != null)
{
endPos.AxisValue = endPos.AxisValueWithLineHeight;
}
SortedDictionary <double, FormLineList>[] formLines = searcher.GetFormLines(pageNum, startPos, endPos, isSubsequentPage, lastPageVerticalLines);
if (formLines == null)
{
return(null);
}
TablePos tablePos;
tablePos = new TablePos
{
PageNum = pageNum,
HorizontialLines = formLines[0],
VerticalLines = formLines[1]
};
//RevertAxisTransform(tablePos);
return(tablePos);
}
public bool IsBetween(LinePos lowRange, LinePos maxRange)
{
return((lowRange == null || this > lowRange) &&
(maxRange == null || this < maxRange));
}
private bool ValidateTableNameUnitInstance(LinePos tableNamePos, LinePos unitPos)
{
return(pdfTronHelper.ValidateInstance(tableNamePos, unitPos, maxInstance_Between_TableName_And_TabelNameNearbyInfo));
}
private bool ValidateTableNameNearbyPos(LinePos tableNamePos, LinePos nearbyPos)
{
return(nearbyPos != null && ValidateTableNameUnitInstance(tableNamePos, nearbyPos));
}