public void DifferentAnalyses_AreNotMerged()
{
var wf = MakeWordform("bank");
var wa1 = MakeAnalysis(wf);
var entryBank = MakeEntry("bank", "side of river");
MakeBundle(wa1, entryBank.SensesOS[0]);
var wa2 = MakeAnalysis(wf);
var entryKick = MakeEntry("bank", "place for money");
MakeBundle(wa2, entryKick.SensesOS[0]);
var wa3 = MakeAnalysis(wf);
MakeBundle(wa3, entryBank.SensesOS[0]);
var verb = MakePartOfSpeech("verb");
wa3.CategoryRA = verb;
var wa4 = MakeAnalysis(wf);
MakeBundle(wa4, "bank"); // analysis only to morpheme level
var wa5 = MakeAnalysis(wf);
MakeBundle(wa5, "bank0"); // analysis only to morpheme level
var wa6 = MakeAnalysis(wf);
var bundle6 = MakeBundle(wa6, "bank");
bundle6.MsaRA = MakeMsa(entryBank);
var wa7 = MakeAnalysis(wf);
var bundle7 = MakeBundle(wa7, "bank0");
bundle7.MorphRA = MorphServices.MakeMorph(entryBank, TsStringUtils.MakeString("bank", Cache.DefaultVernWs));
var wa8 = MakeAnalysis(wf);
MakeBundle(wa8, entryBank.SensesOS[0]);
MakeBundle(wa8, entryBank.SensesOS[0]); // makes sure it handles different number of bundles
WfiWordformServices.MergeDuplicateAnalyses(Cache, m_progress);
// We could try for something stronger; the basic idea is that it should change nothing.
// But the kind of changes it makes is deleting objects, so this should cover it.
Assert.That(wa1.IsValidObject);
Assert.That(wa2.IsValidObject);
Assert.That(wa3.IsValidObject);
Assert.That(wa4.IsValidObject);
Assert.That(wa5.IsValidObject);
Assert.That(wa6.IsValidObject);
Assert.That(wa7.IsValidObject);
Assert.That(wa8.IsValidObject);
}
/// <summary>
/// This overload finds guesses for wordforms specified in a dictionary that maps a wordform to the
/// string that we want to match (might be a different case form).
/// </summary>
/// <param name="wordsToMatch"></param>
/// <returns></returns>
private Dictionary <IWfiWordform, EmptyWwfInfo> MapWordsForComputerGuessesToBestMatchingEntry(Dictionary <IWfiWordform, ITsString> wordsToMatch)
{
var matchingMorphs = new Dictionary <ITsString, IMoStemAllomorph>(new TsStringEquator());
foreach (var tssWord in wordsToMatch.Values)
{
matchingMorphs[tssWord] = null;
}
MorphServices.GetMatchingMonomorphemicMorphs(Cache, matchingMorphs);
var mapEmptyWfInfo = new Dictionary <IWfiWordform, EmptyWwfInfo>();
foreach (var kvp in wordsToMatch)
{
var word = kvp.Key;
var tssWord = kvp.Value;
var bestMatchingMorph = matchingMorphs[tssWord];
if (bestMatchingMorph != null)
{
var entryOrVariant = bestMatchingMorph.OwnerOfClass <ILexEntry>();
ILexEntry mainEntry;
ILexSense sense;
GetMainEntryAndSense(entryOrVariant, out mainEntry, out sense);
if (sense == null && mainEntry.SensesOS.Count > 0)
{
sense = mainEntry.SensesOS.Where(s => s.MorphoSyntaxAnalysisRA is IMoStemMsa)
.FirstOrDefault();
}
IMoStemMsa msa = null;
IPartOfSpeech pos = null;
if (sense != null)
{
msa = (IMoStemMsa)sense.MorphoSyntaxAnalysisRA;
pos = msa.PartOfSpeechRA;
}
// map the word to its best entry.
var entryInfo = new EmptyWwfInfo(bestMatchingMorph, msa, pos, sense);
mapEmptyWfInfo.Add(word, entryInfo);
}
}
return(mapEmptyWfInfo);
}
/// <summary>
/// Confirm that we can break the specified string into the two parts needed for a circumfix.
/// Return true (and the two parts, not stripped of affix markers) if successful.
/// </summary>
public static bool GetCircumfixLeftAndRightParts(LcmCache cache, ITsString tssLexemeForm,
out string sLeftMember, out string sRightMember)
{
// split citation form into left and right parts
sLeftMember = null;
sRightMember = null;
var aSpacePeriod = new[] { ' ', '.' };
var lexemeForm = tssLexemeForm.Text;
var iLeftEnd = lexemeForm.IndexOfAny(aSpacePeriod);
if (iLeftEnd < 0)
{
return(false);
}
sLeftMember = lexemeForm.Substring(0, iLeftEnd);
int iRightBegin = lexemeForm.LastIndexOfAny(aSpacePeriod);
if (iRightBegin < 0)
{
return(false);
}
sRightMember = lexemeForm.Substring(iRightBegin + 1);
int clsidForm;
try
{
var temp = sLeftMember;
MorphServices.FindMorphType(cache, ref temp, out clsidForm);
temp = sRightMember;
MorphServices.FindMorphType(cache, ref temp, out clsidForm);
}
catch (Exception)
{
return(false);
}
return(true);
}
