public void ReferencesToDeletedWordformsShouldSwitch()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
var text = Cache.ServiceLocator.GetInstance <ITextFactory>().Create();
//Cache.LangProject.TextsOC.Add(text);
var stText = Cache.ServiceLocator.GetInstance <IStTextFactory>().Create();
text.ContentsOA = stText;
var para = Cache.ServiceLocator.GetInstance <IStTxtParaFactory>().Create();
stText.ParagraphsOS.Add(para);
var seg = Cache.ServiceLocator.GetInstance <ISegmentFactory>().Create();
para.SegmentsOS.Add(seg);
seg.AnalysesRS.Add(wf1);
seg.AnalysesRS.Add(wf2);
seg.AnalysesRS.Add(wf3);
seg.AnalysesRS.Add(wf2);
seg.AnalysesRS.Add(wf3);
var wordset = Cache.ServiceLocator.GetInstance <IWfiWordSetFactory>().Create();
Cache.LangProject.MorphologicalDataOA.TestSetsOC.Add(wordset);
wordset.CasesRC.Add(wf2);
ICmPossibilityList list = Cache.LangProject.KeyTermsList;
var term = Cache.ServiceLocator.GetInstance <IChkTermFactory>().Create();
list.PossibilitiesOS.Add(term);
var rendering = Cache.ServiceLocator.GetInstance <IChkRenderingFactory>().Create();
term.RenderingsOC.Add(rendering);
rendering.SurfaceFormRA = wf2;
var chkRef = Cache.ServiceLocator.GetInstance <IChkRefFactory>().Create();
term.OccurrencesOS.Add(chkRef);
chkRef.RenderingRA = wf2;
WfiWordformServices.FixDuplicates(Cache, m_progress);
// One but only one should survive.
var wf = Survivor(wf1, wf2, wf3);
Assert.That(seg.AnalysesRS.Count, Is.EqualTo(5));
Assert.That(seg.AnalysesRS[0], Is.EqualTo(wf));
Assert.That(seg.AnalysesRS[1], Is.EqualTo(wf));
Assert.That(seg.AnalysesRS[2], Is.EqualTo(wf));
Assert.That(seg.AnalysesRS[3], Is.EqualTo(wf));
Assert.That(seg.AnalysesRS[4], Is.EqualTo(wf));
// for several tests to prove anything, the wordform that survives must NOT be the
// one that was there already.
Assert.That(wf, Is.Not.EqualTo(wf2));
Assert.That(wordset.CasesRC, Has.Member(wf));
Assert.That(rendering.SurfaceFormRA, Is.EqualTo(wf));
Assert.That(chkRef.RenderingRA, Is.EqualTo(wf));
}
/// <summary>
/// Combine this instance (which must have a wordform) with the next (which must also have one, in the same WS).
/// Returns null if it can't be done.
/// </summary>
public AnalysisOccurrence MakePhraseWithNextWord()
{
if (!CanMakePhraseWithNextWord())
{
return(null);
}
int beginOffset = GetMyBeginOffsetInPara();
var oldWf1 = Analysis.Wordform;
var next = NextAnalysisOccurrence();
var oldWf2 = next.Analysis.Wordform;
int endOffset = next.GetMyEndOffsetInPara();
var newForm = Paragraph.Contents.Substring(beginOffset, endOffset - beginOffset);
var wordform = WfiWordformServices.FindOrCreateWordform(Segment.Cache, newForm);
var mainWs = newForm.get_WritingSystem(0);
var otherWss =
oldWf1.Form.AvailableWritingSystemIds.Union(oldWf2.Form.AvailableWritingSystemIds).Where(ws => ws != mainWs);
// If the wordforms we're merging have other writing system information, and the destination wordform doesn't
// already have values for those WSs, copy it over.
foreach (var ws in otherWss)
{
if (wordform.Form.get_String(ws).Length == 0)
{
var combined = oldWf1.Form.get_String(ws);
var other = oldWf2.Form.get_String(ws);
if (combined.Length == 0)
{
combined = other; // just copy oldWf2, nothing in oldWf1
}
else if (other.Length > 0)
{
// concatenate them, with a space between
var bldr = combined.GetBldr();
bldr.Replace(bldr.Length, bldr.Length, " ", null);
bldr.ReplaceTsString(bldr.Length, bldr.Length, other);
combined = bldr.GetString();
}
// else combined is already the whole value, from oldWf1
wordform.Form.set_String(ws, combined);
}
}
//We need to replace the next analysis and the current one, or possibly the next 2 if there is a
//zero width space.
Segment.AnalysesRS.Replace(Index, 1 + next.Index - Index, new ICmObject[] { wordform });
// We may be able to make guesses of this phrase elsewhere.
// Review JohnT: this is needed to make a 6.0 test pass, but it's a bit strange: in principle,
// this new phrase could be discovered and guessed anywhere, not just in this paragraph.
Paragraph.ParseIsCurrent = false;
// One of the old wordforms may now be able to be deleted (especially if it was a shorter phrase).
DeleteWordformIfPossible(oldWf1);
if (oldWf1 != oldWf2) // otherwise it may already have been deleted
{
DeleteWordformIfPossible(oldWf2);
}
return(this); // still has the correct segment and index, no reason to make a new one.
}
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);
}
public void WordformsWithNothingSpecialMerged()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
WfiWordformServices.FixDuplicates(Cache, m_progress);
// One but only one should survive.
Survivor(wf1, wf2, wf3);
}
public void CopesWithNullVernacular()
{
var wf1 = MakeWordform("");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
WfiWordformServices.FixDuplicates(Cache, m_progress);
Survivor(wf2, wf3);
Assert.That(wf1.IsValidObject, Is.True);
}
public void IdenticalTrivialAnalyses_AreMerged()
{
var wf = MakeWordform("bank");
var wa1 = MakeAnalysis(wf);
var wa2 = MakeAnalysis(wf);
WfiWordformServices.MergeDuplicateAnalyses(Cache, m_progress);
Assert.That(!wa1.IsValidObject || !wa2.IsValidObject); // one should be deleted
Assert.That(wa1.IsValidObject || wa2.IsValidObject); // and one should survive!
}
public void ExtraWritingSystemsAreCopied()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
int wsSpn = Cache.WritingSystemFactory.GetWsFromStr("es");
wf2.Form.set_String(wsSpn, TsStringUtils.MakeString("Spanish", wsSpn));
WfiWordformServices.FixDuplicates(Cache, m_progress);
var wf = Survivor(wf1, wf2, wf3);
Assert.That(wf, Is.Not.EqualTo(wf2)); // otherwise it doesn't prove much
Assert.That(wf.Form.get_String(wsSpn).Text, Is.EqualTo("Spanish"));
}
public void SpellingInCorrectBeatsUnknown()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
wf2.SpellingStatus = (int)SpellingStatusStates.incorrect;
WfiWordformServices.FixDuplicates(Cache, m_progress);
// One but only one should survive.
var wf = Survivor(wf1, wf2, wf3);
Assert.That(wf, Is.Not.EqualTo(wf2)); // otherwise it doesn't prove much
Assert.That(wf.SpellingStatus, Is.EqualTo((int)SpellingStatusStates.incorrect));
}
public void ChecksumIsReset()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordXX");
wf1.Checksum = 5;
wf2.Checksum = 6;
wf3.Checksum = 7;
WfiWordformServices.FixDuplicates(Cache, m_progress);
var wf = Survivor(wf1, wf2, wf3);
Assert.That(wf, Is.Not.EqualTo(wf2)); // otherwise it doesn't prove much
Assert.That(wf.Checksum, Is.EqualTo(0));
}
public void AnalysesPreserved()
{
var wf1 = MakeWordform("wordXX");
var wa1 = MakeAnalysis(wf1);
var wf2 = MakeWordform("wordXX");
var wa2 = MakeAnalysis(wf2);
var wf3 = MakeWordform("wordXX");
var wa3 = MakeAnalysis(wf3);
WfiWordformServices.FixDuplicates(Cache, m_progress);
var wf = Survivor(wf1, wf2, wf3);
Assert.That(wa1.Owner, Is.EqualTo(wf));
Assert.That(wa2.Owner, Is.EqualTo(wf));
Assert.That(wa3.Owner, Is.EqualTo(wf));
}
public void InconsistentAlternativesPreventMergeAndAreReported()
{
var wf1 = MakeWordform("wordXX");
var wf2 = MakeWordform("wordXX");
var wf3 = MakeWordform("wordZZ");
var wf4 = MakeWordform("wordZZ");
var wf5 = MakeWordform("wordZZ");
int wsSpn = Cache.WritingSystemFactory.GetWsFromStr("es");
wf1.Form.set_String(wsSpn, TsStringUtils.MakeString("Spanish", wsSpn));
wf2.Form.set_String(wsSpn, TsStringUtils.MakeString("SpanishOther", wsSpn));
wf3.Form.set_String(wsSpn, TsStringUtils.MakeString("Spanish", wsSpn));
wf4.Form.set_String(wsSpn, TsStringUtils.MakeString("SpanishOther", wsSpn));
wf5.Form.set_String(wsSpn, TsStringUtils.MakeString("SpanishYetAnother", wsSpn));
var failureList = WfiWordformServices.FixDuplicates(Cache, m_progress);
// None should be deleted.
Assert.That(wf1.IsValidObject && wf2.IsValidObject && wf3.IsValidObject && wf4.IsValidObject && wf5.IsValidObject);
Assert.That(failureList, Is.EqualTo("wordXX wordZZ"));
}
public void ComplexMerge_SucceedsAndFixesRefs()
{
var wf = MakeWordform("bank");
var wa1 = MakeAnalysis(wf);
var entryBankRiver = MakeEntry("bank", "side of river");
MakeBundle(wa1, entryBankRiver.SensesOS[0]);
var wa2 = MakeAnalysis(wf);
var entryBankMoney = MakeEntry("bank", "place for money");
MakeBundle(wa2, entryBankMoney.SensesOS[0]);
var wadup1 = MakeAnalysis(wf);
MakeBundle(wadup1, entryBankMoney.SensesOS[0]);
var wadup2 = MakeAnalysis(wf);
MakeBundle(wadup2, entryBankMoney.SensesOS[0]);
var wgMoney = MakeGloss(wa2, "money");
var wgPlaceMoney = MakeGloss(wadup1, "place for money");
var wgdup1 = MakeGloss(wa2, "money");
var wgdup2 = MakeGloss(wadup1, "money");
var wgdup3 = MakeGloss(wadup2, "money");
var text = MakeText("");
var seg = MakeSegment(text.ContentsOA, "bank bank bank bank bank bank bank bank");
seg.AnalysesRS.Add(wa1);
seg.AnalysesRS.Add(wa2);
seg.AnalysesRS.Add(wadup1);
seg.AnalysesRS.Add(wadup2);
seg.AnalysesRS.Add(wgMoney);
seg.AnalysesRS.Add(wgPlaceMoney);
seg.AnalysesRS.Add(wgdup1);
seg.AnalysesRS.Add(wgdup2);
seg.AnalysesRS.Add(wgdup3);
WfiWordformServices.MergeDuplicateAnalyses(Cache, m_progress);
// Theoretically, wa3 or wa4 could legitimately be the survivor. But in fact it is now wa1.
var survivor = wa2;
Assert.That(survivor.IsValidObject);
Assert.That(wa2.IsValidObject);
Assert.That(wadup1.IsValidObject, Is.False);
Assert.That(wadup2.IsValidObject, Is.False);
var survivingGloss = wgMoney; // Enhance: any other duplicate surviving is OK.
Assert.That(survivingGloss.IsValidObject);
Assert.That(wgPlaceMoney.IsValidObject);
Assert.That(wgPlaceMoney.Owner, Is.EqualTo(survivor));
Assert.That(wgdup1.IsValidObject, Is.False);
Assert.That(wgdup2.IsValidObject, Is.False);
Assert.That(wgdup3.IsValidObject, Is.False);
Assert.That(seg.AnalysesRS[0], Is.EqualTo(wa1));
Assert.That(seg.AnalysesRS[1], Is.EqualTo(survivor));
Assert.That(seg.AnalysesRS[2], Is.EqualTo(survivor));
Assert.That(seg.AnalysesRS[3], Is.EqualTo(survivor));
Assert.That(seg.AnalysesRS[4], Is.EqualTo(survivingGloss));
Assert.That(seg.AnalysesRS[5], Is.EqualTo(wgPlaceMoney));
Assert.That(seg.AnalysesRS[6], Is.EqualTo(survivingGloss));
Assert.That(seg.AnalysesRS[7], Is.EqualTo(survivingGloss));
Assert.That(seg.AnalysesRS[8], Is.EqualTo(survivingGloss));
}