public static ParseResult <Image> Parse(GedcomLine first, ILineProvider lineProvider)
{
var image = new Image();
image.ID = ParserHelper.ParseID(first.GetFirstItem()).ToString();
var initialLevel = first.Level;
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
var splitLine = line.GetFirstItem();
if (ParserHelper.Equals(splitLine, "TITL"))
{
var title = line.GetLineContent(4);
image.Title = title.ToString();
}
else if (ParserHelper.Equals(splitLine, "FILE"))
{
var filePath = line.GetLineContent(4);
image.FilePath = filePath.ToString();
}
}
return(ParseResult.Create(image, line));
}
public static ParseResult <GedcomEvent> Parse(GedcomLine first, ILineProvider lineProvider)
{
GedcomEvent gedcomEvent = new GedcomEvent();
var initialLevel = first.Level;
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()).Length > 0)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level <= first.Level)
{
break;
}
ReadOnlySpan <char> tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "DATE"))
{
// If checks we're parsing actual date and not
// CREA or CHAN tags
// TODO: should actually put CREA and CHAN into different parser
if (line.Level == initialLevel + 1)
{
var dateContent = line.GetLineContent(4);
gedcomEvent.Date = dateContent.Length == 0 ? null : dateContent.ToString();
}
}
else if (ParserHelper.Equals(tag, "PLAC"))
{
// If checks we're parsing actual date and not
// CREA or CHAN tags
// TODO: should actually put CREA and CHAN into different parser
if (line.Level == initialLevel + 1)
{
var placContent = line.GetLineContent(4);
gedcomEvent.Location = placContent.Length == 0 ? null : placContent.ToString();
}
}
}
return(ParseResult.Create(gedcomEvent, line));
}
public static ParseResult <GedcomEvent> Parse(GedcomLine first, ILineProvider lineProvider)
{
GedcomEvent gedcomEvent = new GedcomEvent();
var initialLevel = first.Level;
GedcomLine line = default;
string currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level <= first.Level)
{
break;
}
switch (line.GetTagOrRef())
{
case "DATE":
// If checks we're parsing actual date and not
// CREA or CHAN tags
// TODO: should actually put CREA and CHAN into different parser
if (line.Level == initialLevel + 1)
{
gedcomEvent.Date = line.GetLineContent();
}
break;
case "PLAC":
// If checks we're parsing actual date and not
// CREA or CHAN tags
// TODO: should actually put CREA and CHAN into different parser
if (line.Level == initialLevel + 1)
{
gedcomEvent.Location = line.GetLineContent();
}
break;
}
}
return(ParseResult.Create(gedcomEvent, line));
}
public static ParseResult <Note> Parse(GedcomLine first, ILineProvider lineProvider)
{
string id = ParserHelper.ParseID(first.GetFirstItem()).ToString();
string text = null;
var initialLevel = first.Level;
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()).Length > 0)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
ReadOnlySpan <char> tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "CONT"))
{
string contText = line.GetLineContent(4).ToString();
text += Environment.NewLine + contText;
}
else if (ParserHelper.Equals(tag, "CONC"))
{
// TODO: is GenesReunited maintaining the trailing space?
// If so, is this correct?
string concText = line.GetLineContent(4).ToString();
text += concText;
}
}
return(ParseResult.Create(new Note(id, text), line));
}
public static ParseResult <Note> Parse(GedcomLine first, ILineProvider lineProvider)
{
string id = ParserHelper.ParseID(first.GetTagOrRef());
string text = null;
var initialLevel = first.Level;
GedcomLine line = default;
string currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
switch (line.GetTagOrRef())
{
case "CONT":
string contText = line.GetLineContent();
text += Environment.NewLine + contText;
break;
case "CONC":
// TODO: is GenesReunited maintaining the trailing space?
// If so, is this correct?
string concText = line.GetLineContent();
text += concText;
break;
}
}
return(ParseResult.Create(new Note(id, text), line));
}
public static ParseResult <Image> Parse(GedcomLine first, ILineProvider lineProvider)
{
var image = new Image();
image.ID = ParserHelper.ParseID(first.GetTagOrRef());
var initialLevel = first.Level;
GedcomLine line = default;
string currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
switch (line.GetTagOrRef())
{
case "TITL":
var title = line.GetLineContent();
image.Title = title;
break;
case "FILE":
var filePath = line.GetLineContent();
image.FilePath = filePath;
break;
}
}
return(ParseResult.Create(image, line));
}
public static ParseResult <GedcomHeader> Parse(GedcomLine first, ILineProvider lineProvider)
{
CurrentLevel currentLevel = CurrentLevel.None;
var header = new GedcomHeader();
GedcomLine line = default;
string currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == 0)
{
break;
}
if (line.Level == 1)
{
switch (line.GetTagOrRef())
{
case "SOUR":
currentLevel = CurrentLevel.Sour;
break;
case "GEDC":
currentLevel = CurrentLevel.Gedc;
break;
case "CHAR":
header.GedcomCharacterSet = line.GetLineContent();
break;
}
}
else if (line.Level == 2)
{
if (currentLevel == CurrentLevel.Sour)
{
switch (line.GetTagOrRef())
{
case "NAME":
header.SourceName = line.GetLineContent();
break;
case "VERS":
header.SourceVers = line.GetLineContent();
break;
case "CORP":
header.SourceCorp = line.GetLineContent();
break;
}
}
else if (currentLevel == CurrentLevel.Gedc)
{
if (line.GetTagOrRef() == "VERS")
{
header.GedcomVers = line.GetLineContent();
}
}
}
}
return(ParseResult.Create(header, line));
}
public static ParseResult <Family> Parse(GedcomLine first, ILineProvider lineProvider)
{
var family = new Family();
family.ID = ParserHelper.ParseID(first.GetFirstItem()).ToString();
bool inMarriage = false;
var initialLevel = first.Level;
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()).Length > 0)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
var tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "MARR"))
{
inMarriage = true;
}
else if (ParserHelper.Equals(tag, "DATE"))
{
if (inMarriage) // TODO: should have MARR parser
{
var date = line.GetLineContent(4);
if (family.Marriage == null)
{
family.Marriage = new GedcomEvent();
}
family.Marriage.Date = date.ToString();
}
}
else if (ParserHelper.Equals(tag, "PLAC"))
{
if (inMarriage) // Assume level + 1 is MARR
{
var place = line.GetLineContent(4);
if (family.Marriage == null)
{
family.Marriage = new GedcomEvent();
}
family.Marriage.Location = place.ToString();
}
}
else if (ParserHelper.Equals(tag, "HUSB"))
{
// Ignore any husband and wife information in the middle of a marriage tag.
// Present for torture test files - and info redundant?
// can have e.g. "2 HUSB", with no additional info
var husbContent = line.GetLineContent(4);
if (husbContent.Length > 0)
{
family.HusbandID = ParserHelper.ParseID(husbContent).ToString();
}
}
else if (ParserHelper.Equals(tag, "WIFE"))
{
var wifeContent = line.GetLineContent(4);
// Ignore any husband and wife information in the middle of a marriage tag.
// Present for torture test files - and info redundant?
// can have e.g. "2 HUSB", with no additional info
if (wifeContent.Length > 0)
{
family.WifeID = ParserHelper.ParseID(wifeContent).ToString();
}
}
else if (ParserHelper.Equals(tag, "CHIL"))
{
var childContent = line.GetLineContent(4);
family.ChildIDs.Add(ParserHelper.ParseID(childContent).ToString());
}
else
{
inMarriage = false;
}
}
return(ParseResult.Create(family, line));
}
public static ParseResult <Individual> Parse(GedcomLine first, ILineProvider lineProvider)
{
var individual = new Individual();
individual.ID = ParserHelper.ParseID(first.GetTagOrRef());
GedcomLine line = default;
string currentRawLine;
var newLine = false;
while (true)
{
// TODO: this loop is really messy, as some of the parsing
// works on lines as retrieved from the parser, where some
// other iterations call onto a sub-parser which return a final
// line. Should split this logic
if (!newLine)
{
currentRawLine = lineProvider.ReadLine();
if (currentRawLine == null)
{
break;
}
line = ParserHelper.ParseLine(currentRawLine);
}
newLine = false;
if (line.Level == 0)
{
break;
}
switch (line.GetTagOrRef())
{
case "NAME":
if (individual.LastName != null || individual.FirstNames != null)
{
break;
}
ParseNames(line, individual);
break;
case "SEX":
individual.Gender = (line.GetLineContent() == "M") ? Gender.Male : Gender.Female;
break;
case "FAMS":
individual.FamilyIDSpouse = ParserHelper.ParseID(line.GetLineContent());
break;
case "FAMC":
individual.FamilyIDChild = ParserHelper.ParseID(line.GetLineContent());
break;
case "OBJE":
var objeContent = line.GetLineContent();
if (!string.IsNullOrEmpty(objeContent)) // TODO: test for this - currently don't support Images directly in Individual
// should fix this
{
individual.ImageID = ParserHelper.ParseID(objeContent);
}
break;
case "NOTE":
string noteText = line.GetLineContent() ?? string.Empty;
// My be an empty NOTE tag, but will be concatenated onto
if (noteText.StartsWith("@"))
{
string noteId = ParserHelper.ParseID(noteText);
individual.NoteID = noteId;
}
else
{
individual.Note = noteText;
}
break;
case "CONT":
string contText = line.GetLineContent() ?? string.Empty;
individual.Note += Environment.NewLine + contText;
break;
case "CONC":
// TODO: is GenesReunited maintaining the trailing space?
// If so, is this correct?
var concText = line.GetLineContent();
individual.Note += concText;
break;
case "BIRT":
var birthParseResult = EventParser.Parse(line, lineProvider);
if (individual.Birth == null)
{
individual.Birth = birthParseResult.Result;
}
line = birthParseResult.Line;
newLine = true;
break;
case "DEAT":
var deathParseResult = EventParser.Parse(line, lineProvider);
if (individual.Death == null)
{
individual.Death = deathParseResult.Result;
}
line = deathParseResult.Line;
newLine = true;
break;
case "BAPM":
var baptismParseResult = EventParser.Parse(line, lineProvider);
if (individual.Baptism == null)
{
individual.Baptism = baptismParseResult.Result;
}
line = baptismParseResult.Line;
newLine = true;
break;
case "RESI":
var residenceParseResult = EventParser.Parse(line, lineProvider);
individual.Residences.Add(residenceParseResult.Result);
line = residenceParseResult.Line;
newLine = true;
break;
case "CENS":
var censusParseResult = EventParser.Parse(line, lineProvider);
individual.Census.Add(censusParseResult.Result);
line = censusParseResult.Line;
newLine = true;
break;
case "MARR":
var marriageParseResult = EventParser.Parse(line, lineProvider);
if (marriageParseResult.Result != null)
{
individual.Marriages.Add(marriageParseResult.Result);
}
line = marriageParseResult.Line;
newLine = true;
// TODO: cope with multiple marriage records
// It seems that ancestry only outputs multiple marriage records in an individual
// if there's conflicting information, not for multiple marriages
// Multiple marriages for an individual are put into the FAMs tag
break;
//case "ENGA":
// // TODO:
// // individual.Engagement = EventParser.Parse(parserLinesEnga);
// break;
}
if (line.Level == 0)
{
break;
}
}
if (individual.LastName == null)
{
individual.LastName = string.Empty;
}
if (individual.FirstName == null)
{
individual.FirstName = string.Empty;
}
if (individual.FirstNames == null)
{
individual.FirstNames = string.Empty;
}
return(ParseResult.Create(individual, line));
}
public static ParseResult <Family> Parse(GedcomLine first, ILineProvider lineProvider)
{
var family = new Family();
family.ID = ParserHelper.ParseID(first.GetTagOrRef());
bool inMarriage = false;
var initialLevel = first.Level;
GedcomLine line = default;
string currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()) != null)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == first.Level)
{
break;
}
switch (line.GetTagOrRef())
{
case "MARR":
{
inMarriage = true;
break;
}
case "DATE":
if (inMarriage) // TODO: should have MARR parser
{
var date = line.GetLineContent();
if (family.Marriage == null)
{
family.Marriage = new GedcomEvent();
}
family.Marriage.Date = date;
}
break;
case "PLAC":
if (inMarriage) // Assume level + 1 is MARR
{
var place = line.GetLineContent();
if (family.Marriage == null)
{
family.Marriage = new GedcomEvent();
}
family.Marriage.Location = place;
}
break;
case "HUSB":
// Ignore any husband and wife information in the middle of a marriage tag.
// Present for torture test files - and info redundant?
// can have e.g. "2 HUSB", with no additional info
var contentHusb = line.GetLineContent();
if (!string.IsNullOrEmpty(contentHusb))
{
family.HusbandID = ParserHelper.ParseID(contentHusb);
}
break;
case "WIFE":
// Ignore any husband and wife information in the middle of a marriage tag.
// Present for torture test files - and info redundant?
// can have e.g. "2 HUSB", with no additional info
var contentWife = line.GetLineContent();
if (!string.IsNullOrEmpty(contentWife))
{
family.WifeID = ParserHelper.ParseID(contentWife);
}
break;
case "CHIL":
family.ChildIDs.Add(ParserHelper.ParseID(line.GetLineContent()));
break;
default:
inMarriage = false;
break;
}
}
return(ParseResult.Create(family, line));
}
public static ParseResult <Individual> Parse(GedcomLine first, ILineProvider lineProvider)
{
var individual = new Individual();
individual.ID = ParserHelper.ParseID(first.GetFirstItem()).ToString();
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
var newLine = false;
while (true)
{
// TODO: this loop is really messy, as some of the parsing
// works on lines as retrieved from the parser, where some
// other iterations call onto a sub-parser which return a final
// line. Should split this logic
if (!newLine)
{
currentRawLine = lineProvider.ReadLine();
if (currentRawLine == null)
{
break;
}
line = ParserHelper.ParseLine(currentRawLine);
}
newLine = false;
if (line.Level == 0)
{
break;
}
var tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "NAME") &&
individual.LastName == null && individual.FirstNames == null)
{
ParseNames(line, individual);
}
else if (ParserHelper.Equals(tag, "SEX"))
{
individual.Gender = ParserHelper.Equals(line.GetLineContent(3), "M") ? Gender.Male : Gender.Female;
}
else if (ParserHelper.Equals(tag, "FAMS"))
{
individual.FamilyIDSpouse = ParserHelper.ParseID(line.GetLineContent(4)).ToString();
}
else if (ParserHelper.Equals(tag, "FAMC"))
{
individual.FamilyIDChild = ParserHelper.ParseID(line.GetLineContent(4)).ToString();
}
else if (ParserHelper.Equals(tag, "OBJE"))
{
var content = line.GetLineContent(4);
if (content.Length > 0)
{
individual.ImageID = ParserHelper.ParseID(content).ToString();
}
}
else if (ParserHelper.Equals(tag, "NOTE"))
{
var noteContent = line.GetLineContent(4);
if (noteContent.Length > 0)
{
if (noteContent[0] == '@')
{
individual.NoteID = ParserHelper.ParseID(noteContent).ToString();
}
else
{
individual.Note = noteContent.ToString();
}
}
}
else if (ParserHelper.Equals(tag, "CONT"))
{
var contContent = line.GetLineContent(4);
individual.Note += Environment.NewLine + contContent.ToString();
}
else if (ParserHelper.Equals(tag, "CONC"))
{
// TODO: is GenesReunited maintaining the trailing space?
// If so, is this correct?
var concContent = line.GetLineContent(4);
if (concContent.Length > 0)
{
individual.Note += concContent.ToString();
}
}
else if (ParserHelper.Equals(tag, "BIRT"))
{
var birthParseResult = EventParser.Parse(line, lineProvider);
if (individual.Birth == null)
{
individual.Birth = birthParseResult.Result;
}
line = birthParseResult.Line;
newLine = true;
}
else if (ParserHelper.Equals(tag, "DEAT"))
{
var deathParseResult = EventParser.Parse(line, lineProvider);
if (individual.Death == null)
{
individual.Death = deathParseResult.Result;
}
line = deathParseResult.Line;
newLine = true;
}
else if (ParserHelper.Equals(tag, "BAPM"))
{
var baptismParseResult = EventParser.Parse(line, lineProvider);
if (individual.Baptism == null)
{
individual.Baptism = baptismParseResult.Result;
}
line = baptismParseResult.Line;
newLine = true;
}
else if (ParserHelper.Equals(tag, "RESI"))
{
var residenceParseResult = EventParser.Parse(line, lineProvider);
individual.Residences.Add(residenceParseResult.Result);
line = residenceParseResult.Line;
newLine = true;
}
else if (ParserHelper.Equals(tag, "CENS"))
{
var censusParseResult = EventParser.Parse(line, lineProvider);
individual.Census.Add(censusParseResult.Result);
line = censusParseResult.Line;
newLine = true;
}
else if (ParserHelper.Equals(tag, "MARR"))
{
var marriageParseResult = EventParser.Parse(line, lineProvider);
if (marriageParseResult.Result != null)
{
individual.Marriages.Add(marriageParseResult.Result);
}
line = marriageParseResult.Line;
newLine = true;
// TODO: cope with multiple marriage records
// It seems that ancestry only outputs multiple marriage records in an individual
// if there's conflicting information, not for multiple marriages
// Multiple marriages for an individual are put into the FAMs tag
}
//case "ENGA":
// // TODO:
// // individual.Engagement = EventParser.Parse(parserLinesEnga);
// break;
if (line.Level == 0)
{
break;
}
}
if (individual.LastName == null)
{
individual.LastName = string.Empty;
}
if (individual.FirstName == null)
{
individual.FirstName = string.Empty;
}
if (individual.FirstNames == null)
{
individual.FirstNames = string.Empty;
}
return(ParseResult.Create(individual, line));
}
public static ParseResult <GedcomHeader> Parse(GedcomLine first, ILineProvider lineProvider)
{
CurrentLevel currentLevel = CurrentLevel.None;
var header = new GedcomHeader();
GedcomLine line = default;
ReadOnlySpan <char> currentRawLine;
while ((currentRawLine = lineProvider.ReadLine()).Length > 0)
{
line = ParserHelper.ParseLine(currentRawLine);
if (line.Level == 0)
{
break;
}
if (line.Level == 1)
{
var tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "SOUR"))
{
currentLevel = CurrentLevel.Sour;
}
else if (ParserHelper.Equals(tag, "GEDC"))
{
currentLevel = CurrentLevel.Gedc;
}
else if (ParserHelper.Equals(tag, "CHAR"))
{
header.GedcomCharacterSet = line.GetLineContent(4).ToString();
}
}
else if (line.Level == 2)
{
if (currentLevel == CurrentLevel.Sour)
{
var tag = line.GetFirstItem();
if (ParserHelper.Equals(tag, "NAME"))
{
header.SourceName = line.GetLineContent(4).ToString();
}
else if (ParserHelper.Equals(tag, "VERS"))
{
header.SourceVers = line.GetLineContent(4).ToString();
}
else if (ParserHelper.Equals(tag, "CORP"))
{
header.SourceCorp = line.GetLineContent(4).ToString();
}
}
else if (currentLevel == CurrentLevel.Gedc)
{
if (ParserHelper.Equals(line.GetFirstItem(), "VERS"))
{
header.GedcomVers = line.GetLineContent(4).ToString();
}
}
}
}
return(ParseResult.Create(header, line));
}
public void Parse(ILineProvider lineProvider)
{
GedcomLine lastLine = default;
var firstRawLine = lineProvider.ReadLine();
if (firstRawLine == null)
{
throw new InvalidOperationException("File empty");
}
var firstLine = ParserHelper.ParseLine(firstRawLine);
if (firstLine.Level != 0 && !ParserHelper.Equals(firstLine.GetFirstItem(), "HEAD"))
{
throw new InvalidOperationException("GEDCOM Header Not Found");
}
var gedcomHeaderParse = HeaderParser.Parse(firstLine, lineProvider);
_headerCallback?.Invoke(gedcomHeaderParse.Result);
var newLine = gedcomHeaderParse.Line;
lastLine = newLine;
while (newLine.LineContent.Length > 0)
{
var content = newLine.GetLineContent();
if (content.Length == 0)
{
var unrecognisedRawLine = lineProvider.ReadLine();
if (unrecognisedRawLine != null)
{
newLine = ParserHelper.ParseLine(unrecognisedRawLine);
lastLine = newLine;
}
else
{
newLine = default;
}
continue;
}
var unknown = false;
if (ParserHelper.Equals(content, "INDI"))
{
var individualParseResult = IndividualParser.Parse(newLine, lineProvider);
_individualCallback?.Invoke(individualParseResult.Result);
newLine = individualParseResult.Line;
lastLine = newLine;
}
else if (ParserHelper.Equals(content, "FAM"))
{
var familyParseResult = FamilyParser.Parse(newLine, lineProvider);
_familyCallback?.Invoke(familyParseResult.Result);
newLine = familyParseResult.Line;
lastLine = newLine;
}
else if (ParserHelper.Equals(content, "NOTE"))
{
var noteParseResult = NoteParser.Parse(newLine, lineProvider);
_noteCallback?.Invoke(noteParseResult.Result);
newLine = noteParseResult.Line;
lastLine = newLine;
}
else if (ParserHelper.Equals(content, "OBJE"))
{
var objParserResult = ObjectParser.Parse(newLine, lineProvider);
_imageCallback?.Invoke(objParserResult.Result);
newLine = objParserResult.Line;
lastLine = newLine;
}
else
{
var unrecognisedRawLine = lineProvider.ReadLine();
if (unrecognisedRawLine != null)
{
newLine = ParserHelper.ParseLine(unrecognisedRawLine);
lastLine = newLine;
}
else
{
newLine = default;
}
unknown = true;
}
if (unknown)
{
continue;
}
}
if (lastLine.LineContent.Length == 0)
{
throw new InvalidOperationException("file contains no content");
}
if (lastLine.Level != 0 || !ParserHelper.Equals(lastLine.GetFirstItem(), "TRLR"))
{
throw new InvalidOperationException("GEDCOM TRLR not found");
}
}
