public void Approve_TwoSegmentsUnknownWord_LearnsUnknownWord()
{
using (var smtModel = new ThotSmtModel(TestHelpers.ToyCorpusConfigFileName))
using (IInteractiveSmtEngine engine = smtModel.CreateInteractiveEngine())
{
using (IInteractiveTranslationSession session = engine.TranslateInteractively(1,
"hablé con recepción .".Split()))
{
TranslationResult result = session.CurrentResults[0];
Assert.That(result.TargetSegment, Is.EqualTo("hablé with reception .".Split()));
result = session.SetPrefix("i talked".Split(), true)[0];
Assert.That(result.TargetSegment, Is.EqualTo("i talked with reception .".Split()));
session.SetPrefix("i talked with reception .".Split(), true);
session.Approve();
}
using (IInteractiveTranslationSession session = engine.TranslateInteractively(1,
"hablé hasta cinco en punto .".Split()))
{
TranslationResult result = session.CurrentResults[0];
Assert.That(result.TargetSegment, Is.EqualTo("talked until five o ' clock .".Split()));
}
}
}
private void TestSegment(IInteractiveSmtEngine engine, ITranslationSuggester suggester, int n,
ParallelTextSegment segment, StreamWriter traceWriter)
{
traceWriter?.WriteLine($"Segment: {segment.SegmentRef}");
IReadOnlyList <string> sourceSegment = segment.SourceSegment.Preprocess(Preprocessors.Lowercase);
traceWriter?.WriteLine($"Source: {string.Join(" ", sourceSegment)}");
IReadOnlyList <string> targetSegment = segment.TargetSegment.Preprocess(Preprocessors.Lowercase);
traceWriter?.WriteLine($"Target: {string.Join(" ", targetSegment)}");
traceWriter?.WriteLine(new string('=', 120));
string[][] prevSuggestionWords = null;
bool isLastWordSuggestion = false;
string suggestionResult = null;
using (IInteractiveTranslationSession session = engine.TranslateInteractively(n, sourceSegment))
{
while (session.Prefix.Count < targetSegment.Count || !session.IsLastWordComplete)
{
int targetIndex = session.Prefix.Count;
if (!session.IsLastWordComplete)
{
targetIndex--;
}
bool match = false;
TranslationSuggestion[] suggestions = suggester.GetSuggestions(session).ToArray();
string[][] suggestionWords = suggestions.Select((s, k) =>
s.TargetWordIndices.Select(j =>
session.CurrentResults[k].TargetSegment[j]).ToArray()).ToArray();
if (prevSuggestionWords == null || !SuggestionsAreEqual(prevSuggestionWords, suggestionWords))
{
WritePrefix(traceWriter, suggestionResult, session.Prefix);
WriteSuggestions(traceWriter, session, suggestions);
suggestionResult = null;
if (suggestions.Any(s => s.TargetWordIndices.Count > 0))
{
_totalSuggestionCount++;
}
}
for (int k = 0; k < suggestions.Length; k++)
{
TranslationSuggestion suggestion = suggestions[k];
var accepted = new List <int>();
for (int i = 0, j = targetIndex; i < suggestionWords[k].Length && j < targetSegment.Count; i++)
{
if (suggestionWords[k][i] == targetSegment[j])
{
accepted.Add(suggestion.TargetWordIndices[i]);
j++;
}
else if (accepted.Count == 0)
{
j = targetIndex;
}
else
{
break;
}
}
if (accepted.Count > 0)
{
session.AppendSuggestionToPrefix(k, accepted);
isLastWordSuggestion = true;
_actionCount++;
_totalAcceptedSuggestionCount++;
if (accepted.Count == suggestion.TargetWordIndices.Count)
{
suggestionResult = "ACCEPT_FULL";
_fullSuggestionCount++;
}
else if (accepted[0] == suggestion.TargetWordIndices[0])
{
suggestionResult = "ACCEPT_INIT";
_initSuggestionCount++;
}
else if (accepted[accepted.Count - 1]
== suggestion.TargetWordIndices[suggestion.TargetWordIndices.Count - 1])
{
suggestionResult = "ACCEPT_FIN";
_finalSuggestionCount++;
}
else
{
suggestionResult = "ACCEPT_MID";
_middleSuggestionCount++;
}
_acceptedSuggestionCounts[k]++;
match = true;
break;
}
}
if (!match)
{
if (isLastWordSuggestion)
{
_actionCount++;
isLastWordSuggestion = false;
WritePrefix(traceWriter, suggestionResult, session.Prefix);
suggestionResult = null;
}
int len = session.IsLastWordComplete ? 0 : session.Prefix[session.Prefix.Count - 1].Length;
string targetWord = targetSegment[targetIndex];
if (len == targetWord.Length)
{
session.AppendToPrefix("", true);
}
else
{
string c = targetWord.Substring(len, 1);
session.AppendToPrefix(c, false);
}
suggestionResult = suggestions.Any(s => s.TargetWordIndices.Count > 0) ? "REJECT" : "NONE";
_actionCount++;
}
prevSuggestionWords = suggestionWords;
}
WritePrefix(traceWriter, suggestionResult, session.Prefix);
session.Approve(_approveAlignedOption.HasValue());
}
_charCount += targetSegment.Sum(w => w.Length + 1);
traceWriter?.WriteLine();
}