public uint ReadLine(ref GedcomParserProgress progress)
{
ResetLine();
bool endOfLine = false;
uint noOfChars = 0;
int lfCnt = 0;
bool lineEndStarted = false;
while ((progress.position < progress.size) && !endOfLine)
{
char ch;
ch = (char)progress.data[progress.position++];
if (lineEndStarted || parser.IsNewLine(ch))
{
if (lineFeedString[lfCnt++] != ch)
{
DebugStringAdd("Line:" + lineNo + " Inconsistent line feeds!");
if ((progress.data.Length > progress.position) && (parser.IsNewLine((char)progress.data[progress.position + 1])))
{
progress.position++;
}
endOfLine = true;
}
if (lfCnt == lineFeedString.Length)
{
endOfLine = true;
}
lineEndStarted = true;
}
else
{
AddChar(ch);
noOfChars++;
}
}
//trace.TraceInformation("ReadLine end");
return(noOfChars);
}
private void Parse(ref FileImportResult importResult)
{
bool bomFound = false;
GedcomLineObject rootLineObject = new GedcomLineObject(0);
GedcomParserProgress progress = new GedcomParserProgress(fileBuffer.GetBuffer(), fileBuffer.GetSize());
double lastPrintPercent = 0.0;
double printPercent;
GedcomLineObject currentLineObject = rootLineObject;
GedcomLineData prevLineData = null;
int lineDiff = 0;
uint lineLength;
//string importResult = "";
bomFound = CheckBomMark(ref progress, ref importResult);
GedcomTreeDecoderClass treeDecoder = new GedcomTreeDecoderClass(ref familyTree, ref importResult);
treeDecoder.SetCharacterSet(characterSet);
while (!progress.IsEndOfFile())
{
GedcomLineData lineData;
line.SetDebugString(importResult);
lineLength = line.ReadLine(ref progress);
lineData = line.DecodeLine();
importResult = line.GetDebugString();
parsedLines++;
if ((lineData != null) && lineData.valid)
{
if (lineData.level == (currentLineObject.GetLevel() + 1))
{
GedcomLineObject subLineObject = new GedcomLineObject(lineData.level);
subLineObject.parent = currentLineObject;
if (prevLineData != null)
{
prevLineData.child = subLineObject;
}
currentLineObject = subLineObject;
}
/*else if (currentLineObject.GetLevel() == lineData.level)
* {
* }*/
else if (lineData.level < currentLineObject.GetLevel())
{
bool decodeDone = false;
printPercent = 100.0 * (double)progress.position / (double)progress.size;
if ((printPercent - lastPrintPercent) > 0.10)
{
if (backgroundWorker != null)
{
backgroundWorker.ReportProgress((int)printPercent, "Working...");
}
lastPrintPercent = printPercent;
if (trace.Switch.Level.HasFlag(SourceLevels.Information))
{
trace.TraceInformation("Decode position 1 " + progress.position + " (" + progress.size + ") " + DateTime.Now.ToString());
trace.TraceInformation("Lines " + parsedLines + " (" + treeDecoder.GetDecodedLines() + ") " + printPercent.ToString("F") + "%");
trace.TraceData(TraceEventType.Verbose, 0, familyTree.GetShortTreeInfo());
//familyTree.PrintShort();
}
}
do
{
if (currentLineObject.parent != null)
{
currentLineObject = currentLineObject.parent;
if ((currentLineObject.GetLevel() == 0) && (parsedLines > 0))
{
treeDecoder.DecodeObject(currentLineObject);
currentLineObject.Clear();
if (treeDecoder.GetCharacterSet() != characterSet)
{
if (bomFound)
{
treeDecoder.DebugStringAdd("Warning! BOM and character set in Gedcom part mismatches! " + treeDecoder.GetCharacterSet() + "," + characterSet);
}
SetCharacterSet(treeDecoder.GetCharacterSet());
}
}
}
else
{
if ((parsedLines - 1) - treeDecoder.GetDecodedLines() - 1 > lineDiff)
{
treeDecoder.DebugStringAdd("Decode position " + progress.position + " (" + progress.size + ")");
treeDecoder.DebugStringAdd("Lines " + parsedLines + " (" + treeDecoder.GetDecodedLines() + "," + lineDiff + ")");
treeDecoder.DebugStringAdd("New undecoded lines: " + (parsedLines - treeDecoder.GetDecodedLines() - lineDiff) + "!");
lineDiff = parsedLines - treeDecoder.GetDecodedLines();
}
if (trace.Switch.Level.HasFlag(SourceLevels.Information))
{
trace.TraceInformation("Decode position 2 " + progress.position + " (" + progress.size + ")");
trace.TraceInformation("Lines " + parsedLines + " (" + treeDecoder.GetDecodedLines() + ")");
trace.TraceData(TraceEventType.Information, 0, familyTree.GetShortTreeInfo());
//familyTree.PrintShort();
}
currentLineObject.gedcomLines.Clear();
decodeDone = true;
}
} while ((currentLineObject.GetLevel() > lineData.level) && !decodeDone);
}
prevLineData = lineData;
currentLineObject.gedcomLines.Add(lineData);
}
else
{
if (lineData != null)
{
treeDecoder.DebugStringAdd("Line:" + lineData.lineNo + " Error bad gedcom line [" + lineData + "]");
}
else
{
treeDecoder.DebugStringAdd("Line: " + parsedLines + ": Error bad gedcom line:no data found ");
}
}
if (progress.IsEndOfFile())
{
treeDecoder.DecodeObject(currentLineObject);
if (treeDecoder.GetCharacterSet() != characterSet)
{
if (bomFound)
{
treeDecoder.DebugStringAdd("Warning! BOM and character set in Gedcom part mismatches! " + treeDecoder.GetCharacterSet() + "," + characterSet);
}
SetCharacterSet(treeDecoder.GetCharacterSet());
}
treeDecoder.DebugStringAdd("Line:" + lineData.lineNo + " end of file " + parsedLines);
if (lineData.level != 0)
{
treeDecoder.DebugStringAdd("Line:" + lineData.lineNo + "Error: The Gedcom file did not end correctly! Was it inomplete? = " + currentLineObject.gedcomLines.Count);
}
if (!treeDecoder.DecodingCompleted())
{
treeDecoder.DebugStringAdd("Line:" + lineData.lineNo + "Error: The Gedcom file did not end correctly! No trailer detected = " + currentLineObject.gedcomLines.Count);
}
treeDecoder.ShowUnknownTags();
}
}
backgroundWorker = null;
treeDecoder.DebugStringAdd("Gedcom file parsing finished at " + currentLineObject.gedcomLines.Count);
importResult = treeDecoder.GetImportResult();
}
private bool CheckBomMark(ref GedcomParserProgress progress, ref FileImportResult importResult)
{
Byte[] fileDataBuffer = fileBuffer.GetBuffer();
Byte[] UTF_8_BOM = { 0xEF, 0xBB, 0xBF };
Byte[] UTF_16BE_BOM = { 0xFE, 0xFF };
Byte[] UTF_16LE_BOM = { 0xFF, 0xFE };
bool match;
match = true;
for (int i = 0; match && (i < UTF_8_BOM.Length); i++)
{
if (UTF_8_BOM[i] != fileDataBuffer[i])
{
match = false;
}
}
if (match)
{
importResult.AddString("BOM says UTF-8!");
//trace.TraceInformation();
SetCharacterSet(GedcomFileCharacterSet.Utf8);
progress.position = UTF_8_BOM.Length;
return(true);
}
match = true;
for (int i = 0; match && (i < UTF_16BE_BOM.Length); i++)
{
if (UTF_16BE_BOM[i] != fileDataBuffer[i])
{
match = false;
}
}
if (match)
{
importResult.AddString("BOM says UTF-8!");
//trace.TraceInformation("BOM says UTF-16-BE!");
SetCharacterSet(GedcomFileCharacterSet.Utf16BE);
progress.position = UTF_16BE_BOM.Length;
return(true);
}
match = true;
for (int i = 0; match && (i < UTF_16LE_BOM.Length); i++)
{
if (UTF_16LE_BOM[i] != fileDataBuffer[i])
{
match = false;
}
}
if (match)
{
importResult.AddString("BOM says UTF-16-LE!");
//trace.TraceInformation("BOM says UTF-16-LE!");
SetCharacterSet(GedcomFileCharacterSet.Utf16LE);
progress.position = UTF_16LE_BOM.Length;
return(true);
}
importResult.AddString("BOM not found!");
return(false);
}