private Tuple <object, GedParse> Make(GedRecord rec)
{
// 1. The first line in the rec should start with '0'
var head = rec.FirstLine();
gs.Split(head, ' ');
char lvl = gs.Level(head);
//int firstDex = LineUtil.FirstChar(head);
//if (head[firstDex] != '0')
if (lvl != '0')
{
var rec2 = new Unknown(rec, null, gs.Tag(head));
//rec2.Error = UnkRec.ErrorCode.InvLevel;
return(new Tuple <object, GedParse>(rec2, null));
//throw new Exception("record head not zero"); // TODO should this be an error record instead?
}
// 2. search for and find the tag
//LineUtil.LineData ld = new LineUtil.LineData(); // TODO static?
//LineUtil.LevelTagAndRemain(ld, head);
//gs.Split(head, ' ');
// 3. create a GedCommon derived class
var remain = new string(gs.Remain(head));
return(GedRecFactory(rec, gs.Ident(head), gs.Tag(head), remain));
//return GedRecFactory(rec, ld.Ident, ld.Tag, ld.Remain);
}
protected virtual void Dispose(bool disposing)
{
// TODO any explicit disposal required?
Parser = null;
Data = null;
Errors = null;
_currRec = null;
}
// Common method for unit testing (instream != null) or file reading (gedPath != null)
/// <summary>
/// Read GEDCOM data into memory.
/// </summary>
/// <param name="gedPath">The path to the file. If null, will attempt to read from instream instead.</param>
/// <param name="instream">An input stream to read from instead of a file path.</param>
public void ReadGed(string gedPath, StreamReader instream = null)
{
_emptyLineSeen = 0;
FilePath = gedPath;
Errors = new List <UnkRec>();
GedReader _reader = new GedReader();
_reader.BufferSize = _bufferSize;
_reader.ProcessALine = ProcessLine;
_reader.ErrorTracker = DoError;
// Processor context
Parser = new GedParser(FilePath);
Data = new List <GEDCommon>();
if (Errors == null)
{
Errors = new List <UnkRec>();
}
_currRec = new GedRecord();
#if SQLITE
var foo = SQLite.Instance;
#elif LITEDB
var foo = SharpGEDParser.Parser.LiteDB.Instance;
#elif NOTESTREAM
var foo = NoteStream.Instance;
#endif
#if XREFTRACK
var bar = XrefTrack.Instance;
#endif
try
{
if (gedPath == null)
{
_reader.ReadFile(instream);
}
else
{
_reader.ReadFile(gedPath);
}
EndOfFile();
}
catch (Exception)
{
UnkRec err = new UnkRec();
err.Error = UnkRec.ErrorCode.Exception;
err.Beg = _lineNum;
// TODO err.Error = string.Format("Exception: {0} line {1} | {2}", ex.Message, _lineNum, ex.StackTrace);
Errors.Add(err);
}
Parser.FinishUp();
GatherRecords();
GatherErrors();
_lineNum = _reader._lineNum;
_currRec = null;
}
private Tuple <object, GedParse> GedRecFactory(GedRecord rec, string ident, string tag, string remain)
{
if (string.IsNullOrWhiteSpace(tag))
{
var foo = new Unknown(rec, ident, "");
foo.Errors.Add(new UnkRec {
Error = UnkRec.ErrorCode.MissTag
});
return(new Tuple <object, GedParse>(foo, null));
}
// Parse 'top level' records. Parsing of some record types (e.g. NOTE, SOUR, etc) are likely to be in 'common' with sub-record parsing
// TODO Very much brute force. If/until this is found to be optimizable
switch (tag.ToUpper())
{
case "INDI":
return(new Tuple <object, GedParse>(new IndiRecord(rec, ident, remain), _IndiParseSingleton));
case "FAM":
return(new Tuple <object, GedParse>(new FamRecord(rec, ident, remain), _FamParseSingleton));
case "SOUR":
return(new Tuple <object, GedParse>(new SourceRecord(rec, ident, remain), _SourParseSingleton));
case "REPO":
return(new Tuple <object, GedParse>(new Repository(rec, ident, remain), _RepoParseSingleton));
case "NOTE":
{
string remainLS = gs.RemainLS(rec.FirstLine());
return(new Tuple <object, GedParse>(new NoteRecord(rec, ident, remainLS), _NoteParseSingleton));
}
case "OBJE":
return(new Tuple <object, GedParse>(new MediaRecord(rec, ident, remain), _MediaParseSingleton));
case "HEAD":
return(new Tuple <object, GedParse>(new HeadRecord(rec), _HeadParseSingleton));
case "TRLR":
return(null);
case "SUBM": // TODO temp ignore
case "SUBN": // TODO temp ignore
return(new Tuple <object, GedParse>(new DontCare(rec, tag), null));
default:
{
var foo = new Unknown(rec, ident, tag);
return(new Tuple <object, GedParse>(foo, null));
}
}
}
/// <summary>
/// Deal with a single line. It either starts with '0' and is to be a new record,
/// or is accumulated into a record.
/// </summary>
/// <param name="line"></param>
/// <param name="lineNum"></param>
private bool ProcessLine(char [] line, int lineNum)
{
int len = line.Length;
int dex = LineUtil.FirstChar(line, 0, len);
if (dex < 0)
{
if (_emptyLineSeen == 0)
{
DoError(UnkRec.ErrorCode.EmptyLine, lineNum);
}
_emptyLineSeen++;
return(true); // empty line
}
if (len > 255) // TODO anything special for UTF-16?
{
DoError(UnkRec.ErrorCode.LineTooLong, lineNum);
// proceed anyway
}
char level = line[dex];
if (level < '0' || level > '9')
{
DoError(UnkRec.ErrorCode.InvLevel, lineNum);
return(false); // cannot proceed
}
// NOTE: do NOT warn about leading spaces "GEDCOM readers should ignore it when it occurs".
if (level == '0' && _currRec.LineCount > 0)
{
// start of a new record. deal with the previous record first
// TODO records should go into a 'to parse' list and asynchronously turned into head/indi/fam/etc
var parsed = Parser.Parse(_currRec);
if (parsed == null)
{
return(false);
}
Data.Add(parsed);
_currRec = new GedRecord(lineNum, line);
}
else
{
_currRec.AddLine(line);
}
return(true);
}
public void Parse(GEDCommon rec, GedRecord Lines, GSFactory gsfact)
{
ParseContext2 ctx = new ParseContext2();
ctx.gs = gsfact.Alloc(); // new GEDSplitter(GedParser._masterTagCache);
ctx.tagCache = GedParser._masterTagCache;
ctx.Lines = Lines;
ctx.Parent = rec;
int max = Lines.Max;
for (int i = 1; i < max; i++)
{
var line = Lines.GetLine(i);
ctx.Begline = i;
ctx.Endline = i; // assume it is one line long, parser might change it
ctx.gs.LevelTagAndRemain(line, ctx);
TagProc2 tagProc;
if (ctx.Tag != null && _tagSet2.TryGetValue(ctx.Tag, out tagProc))
{
tagProc(ctx);
}
else
{
// Custom and invalid treated as 'unknowns': let the consumer figure it out
// TODO gedr5419_blood_type_events.ged has garbage characters in SOUR/ABBR tags: incorrect line terminator, blank lines etc.
LookAhead(ctx);
rec.Unknowns.Add(new UnkRec(ctx.Tag, Lines.Beg + ctx.Begline, Lines.Beg + ctx.Endline));
}
i = ctx.Endline;
}
// TODO post parse error checking on sub-structures
PostCheck(ctx.Parent); // post parse error checking
gsfact.Free(ctx.gs);
ctx.gs = null;
}
public GEDCommon Parse(GedRecord rec)
{
// Given a glop of lines which represent a 'record', parse it into GED data (INDI/FAM/NOTE/OBJE/REPO/SOUR/etc)
Tuple <object, GedParse> parseSet = Make(rec);
if (parseSet == null)
{
return(null); // EOF
}
if (parseSet.Item2 == null)
{
return(parseSet.Item1 as GEDCommon); // unknown or NYI record type
}
GEDCommon recC2 = parseSet.Item1 as GEDCommon;
#if PARALLEL
_allTasks.Add(Task.Run(() => parseSet.Item2.Parse(recC2, rec, _GedSplitFactory)));
#else
parseSet.Item2.Parse(recC2, rec, _GedSplitFactory);
#endif
return(parseSet.Item1 as GEDCommon);
}
