public Language Read()
{
var sw = new Stopwatch();
sw.Start();
_orthography = ReadOrthography();
_trace.TraceInformation($"{_dirPath} orthography loading time is {sw.ElapsedMilliseconds} ms");
sw.Restart();
var morphotactics = ReadMorphotactics();
_trace.TraceInformation($"{_dirPath} morphotactics loading time is {sw.ElapsedMilliseconds} ms");
sw.Restart();
var roots = ReadRoots();
_trace.TraceInformation($"{_dirPath} roots loading time is {sw.ElapsedMilliseconds} ms");
sw.Restart();
var suffixes = ReadSuffixes();
_trace.TraceInformation($"{_dirPath} suffixes loading time is {sw.ElapsedMilliseconds} ms");
sw.Restart();
return(new Language(_languageType, _orthography, morphotactics, roots, suffixes));
}
public Language Read()
{
var sw = new Stopwatch();
sw.Start();
_orthography = ReadOrthography();
Debug.Print($"orthograpy: {sw.ElapsedMilliseconds} ms");
sw.Restart();
Morphotactics morphotactics = ReadMorphotactics();
Debug.Print($"morphotactics: {sw.ElapsedMilliseconds} ms");
sw.Restart();
MorphemeSurfaceDictionary <Root> roots = ReadRoots();
Debug.Print($"roots: {sw.ElapsedMilliseconds} ms");
sw.Restart();
Suffixes suffixes = ReadSuffixes();
Debug.Print($"suffixes: {sw.ElapsedMilliseconds} ms");
sw.Restart();
int index = _dirPath.LastIndexOf("\\", StringComparison.Ordinal);
string langCode = index > -1 ? _dirPath.Substring(index + 1) : _dirPath;
return(new Language(langCode, morphotactics, roots, suffixes));
}
private MutableLanguage(LanguageType type,
Orthography orthography,
Morphotactics morphotactics,
MorphemeContainer <Root> roots,
MorphemeContainer <Suffix> suffixes)
: base(type, orthography, morphotactics, roots, suffixes)
{
}
public Language Parse(LanguageData data)
{
_orthography = ParseOrthography(data.OrthographyXml);
var morphotactics = ParseMorphotactics(data.MorphotacticsXml);
var roots = ParseRoots(data.RootTxt);
var suffixes = ParseSuffixes(data.SuffixTxt);
return(new Language(data.Type, _orthography, morphotactics, roots, suffixes));
}
internal Language(LanguageType type,
Orthography orthography,
Morphotactics morphotactics,
MorphemeContainer <Root> roots,
MorphemeContainer <Suffix> suffixes)
{
Type = type;
Orthography = orthography;
Morphotactics = morphotactics;
Roots = roots;
Suffixes = suffixes;
Analyzer = new WordAnalyzer(this);
}
private void Add(RootEntry entry)
{
var rules = Orthography.GetRules(entry.Rules);
var root = new Root(entry.Pos, entry.Lex, entry.Surfaces, entry.Labels, rules);
Roots.ById.Add(entry.Id, root);
foreach (string surface in entry.Surfaces)
{
Roots.BySurface.Add(surface, root);
}
}
public SuffixLexiconReader(Orthography orthography)
{
_orthography = orthography;
}
public RootLexiconReader(Orthography orthography)
{
_orthography = orthography;
}
private void ParseText_DoWork(object sender, DoWorkEventArgs e)
{
Lexicon.Clear();
var expElementPattern = new Regex($"({PunctuationPattern.ToString()})|({WordPattern.ToString()})");
var whiteSpacePattern = new Regex(@"[\s\n\r]+", RegexOptions.Singleline | RegexOptions.Multiline);
var worker = sender as BackgroundWorker;
var text = e.Argument as string;
int progress = 0;
string state = string.Empty;
// TODO: handling of paragraph breaks and section headers, etc
foreach (Match p in ParagraphPattern.Matches(text))
{
string paragraphText = p.Value.Trim();
paragraphText = whiteSpacePattern.Replace(paragraphText, " ");
var paragraph = new Lx.Discourse();
Text.Discourse.AddLast(paragraph);
foreach (Match l in LinePattern.Matches(paragraphText))
{
//store line, section up into words and punctuation
string cleanedLine = l.Value.Trim();
//cleanedLine = whiteSpacePattern.Replace(cleanedLine, " ");
state = cleanedLine;
var expression = new Lx.Expression(cleanedLine);
paragraph.Expressions.AddLast(expression);
foreach (Match m in expElementPattern.Matches(expression.Graph))
{
if (m.Groups.Count > 0)
{
// string m => List<Glyphs>
var glyphs = Script.AddGlyphs(m.Value.ToCharArray());
// List<Glyph> => List<Grapheme>
// Pre-analysis, graphemes are 1:1 with glyphs
var graphemes = Orthography.AddGraphemes(glyphs);
// List<Grapheme> => Morpheme
if (string.IsNullOrEmpty(m.Groups[1].Value))
{
//var morph = Text.Lexicon.Add(m.Groups[2].Value);
//morph.GraphemeChain.Add(Lx.SegmentChain<Lx.Grapheme>.NewSegmentChain(graphemes));
Lx.Morpheme morph = Text.Lexicon.Add(graphemes);
expression.Sequence.AddLast(morph);
}
else
{
expression.Sequence.AddLast(Text.Paralexicon.Add(m.Groups[1].Value));
}
}
}
worker.ReportProgress(++progress, state);
}
}
UpdateLocalLexicon();
}