public void ImportNewHumanApprovedByDefaultWordGloss()
{
var wsf = Cache.WritingSystemFactory;
const string xml = "<document><interlinear-text>" +
"<paragraphs><paragraph><phrases><phrase><words>" +
"<word>" +
"<item type='txt' lang='en'>supercalifragilisticexpialidocious</item>" +
"<item type='gls' lang='pt'>absurdo</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
FDO.IText importedText = null;
var options = CreateImportInterlinearOptions(xml);
li.ImportInterlinear(options, ref importedText);
using (var firstEntry = Cache.LanguageProject.Texts.GetEnumerator())
{
firstEntry.MoveNext();
var imported = firstEntry.Current;
Assert.IsNotNull(imported);
var para = imported.ContentsOA.ParagraphsOS[0] as IStTxtPara;
Assert.IsNotNull(para);
Assert.That(para.Analyses.Count(), Is.EqualTo(1));
var wfiWord = para.Analyses.First().Wordform;
int wsWordform = wsf.get_Engine("en").Handle;
Assert.That(wfiWord.Form.get_String(wsf.get_Engine("en").Handle).Text,
Is.EqualTo("supercalifragilisticexpialidocious"));
Assert.That(wfiWord.AnalysesOC.Count, Is.GreaterThan(0));
var wfiAnalysis = wfiWord.AnalysesOC.First();
// make sure we also created a morpheme form
AssertMorphemeFormMatchesWordform(wfiWord, wfiAnalysis, wsWordform);
// make sure we created a human approved opinion
AssertHumanApprovedOpinion(wfiWord, wfiAnalysis);
var at = new AnalysisTree(para.Analyses.First());
Assert.IsNotNull(at.Gloss, "IAnalysis should be WfiGloss");
Assert.That(at.Gloss.Form.get_String(wsf.get_Engine("pt").Handle).Text, Is.EqualTo("absurdo"));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiGlossRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiMorphBundleRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiAnalysisRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiWordformRepository>().Count, Is.EqualTo(1));
}
}
public void ImportWordsWithMultipleWss()
{
var wsEn = Cache.WritingSystemFactory.GetWsFromStr("en");
IWritingSystem wsWbl;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("wbl-Arab-AF", out wsWbl);
wsWbl.RightToLeftScript = true;
IWritingSystem wsWblIpa;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("wbl-Qaaa-AF-fonipa-x-Zipa", out wsWblIpa);
const string xml =
@"<document version='2'>
<interlinear-text guid='5eecc8be-f41b-4433-be94-8950a8ce75e5'>
<item type='title' lang='wbl-Arab-AF'>تست</item>
<item type='title' lang='en'>Test</item>
<item type='comment' lang='en'></item>
<paragraphs>
<paragraph guid='b21daced-5c85-4610-8023-8d7d4b3191f4'>
<phrases>
<phrase guid='0b0346e0-3bb8-40e7-a0a4-f7771d233e93'>
<item type='segnum' lang='en'>1</item>
<words>
<word guid='0b548dff-6a8e-4c21-a977-fcc4ddc268be'>
<item type='txt' lang='wbl-Arab-AF'>baseline</item>
<item type='txt' lang='wbl-Qaaa-AF-fonipa-x-Zipa'>beslain</item>
<item type='gls' lang='en'>gloss</item>
</word>
</words>
</phrase>
</phrases>
</paragraph>
</paragraphs>
<languages>
<language lang='wbl-Arab-AF' font='Times New Roman' vernacular='true' RightToLeft='true' />
<language lang='en' font='Times New Roman' />
<language lang='wbl-Qaaa-AF-fonipa-x-Zipa' font='Doulos SIL' vernacular='true' />
</languages>
</interlinear-text>
</document>";
using (var stream = new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())))
{
var options = new LinguaLinksImport.ImportInterlinearOptions
{
Progress = new DummyProgressDlg(),
AnalysesLevel = LinguaLinksImport.ImportAnalysesLevel.Wordform,
BirdData = stream,
AllottedProgress = 0,
CheckAndAddLanguages = DummyCheckAndAddLanguagesInternal
};
FDO.IText importedText = null;
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var result = li.ImportInterlinear(options, ref importedText);
Assert.True(result, "ImportInterlinear was not successful.");
Assert.That(importedText.ContentsOA.ParagraphsOS.Count, Is.EqualTo(1));
var paraImported = importedText.ContentsOA[0];
var testPara = paraImported.Contents;
Assert.That(testPara.Text, Is.EqualTo("baseline"));
Assert.That(TsStringUtils.GetWsAtOffset(testPara, 0), Is.EqualTo(wsWbl.Handle));
Assert.That(testPara.RunCount, Is.EqualTo(1));
// main writing system should be the first one in the import
var mainWs = importedText.ContentsOA.MainWritingSystem;
Assert.That(mainWs, Is.EqualTo(wsWbl.Handle));
// Next verify that the IPA content got added to the imported word form
var a0 = paraImported.SegmentsOS[0].AnalysesRS[0];
var wf0 = a0 as IWfiWordform;
Assert.That(wf0, Is.Not.Null);
Assert.That(wf0.Form.get_String(wsWbl.Handle).Text, Is.EqualTo("baseline"));
Assert.That(wf0.Form.get_String(wsWblIpa.Handle).Text, Is.EqualTo("beslain"));
}
}
public void ImportWordsWithMultipleWss()
{
var wsEn = Cache.WritingSystemFactory.GetWsFromStr("en");
IWritingSystem wsWbl;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("wbl-Arab-AF", out wsWbl);
wsWbl.RightToLeftScript = true;
IWritingSystem wsWblIpa;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("wbl-Qaaa-AF-fonipa-x-Zipa", out wsWblIpa);
const string xml =
@"<document version='2'>
<interlinear-text guid='5eecc8be-f41b-4433-be94-8950a8ce75e5'>
<item type='title' lang='wbl-Arab-AF'>تست</item>
<item type='title' lang='en'>Test</item>
<item type='comment' lang='en'></item>
<paragraphs>
<paragraph guid='b21daced-5c85-4610-8023-8d7d4b3191f4'>
<phrases>
<phrase guid='0b0346e0-3bb8-40e7-a0a4-f7771d233e93'>
<item type='segnum' lang='en'>1</item>
<words>
<word guid='0b548dff-6a8e-4c21-a977-fcc4ddc268be'>
<item type='txt' lang='wbl-Arab-AF'>baseline</item>
<item type='txt' lang='wbl-Qaaa-AF-fonipa-x-Zipa'>beslain</item>
<item type='gls' lang='en'>gloss</item>
</word>
</words>
</phrase>
</phrases>
</paragraph>
</paragraphs>
<languages>
<language lang='wbl-Arab-AF' font='Times New Roman' vernacular='true' RightToLeft='true' />
<language lang='en' font='Times New Roman' />
<language lang='wbl-Qaaa-AF-fonipa-x-Zipa' font='Doulos SIL' vernacular='true' />
</languages>
</interlinear-text>
</document>" ;
using (var stream = new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())))
{
var options = new LinguaLinksImport.ImportInterlinearOptions
{
Progress = new DummyProgressDlg(),
AnalysesLevel = LinguaLinksImport.ImportAnalysesLevel.Wordform,
BirdData = stream,
AllottedProgress = 0,
CheckAndAddLanguages = DummyCheckAndAddLanguagesInternal
};
FDO.IText importedText = null;
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var result = li.ImportInterlinear(options, ref importedText);
Assert.True(result, "ImportInterlinear was not successful.");
Assert.That(importedText.ContentsOA.ParagraphsOS.Count, Is.EqualTo(1));
var paraImported = importedText.ContentsOA[0];
var testPara = paraImported.Contents;
Assert.That(testPara.Text, Is.EqualTo("baseline"));
Assert.That(TsStringUtils.GetWsAtOffset(testPara, 0), Is.EqualTo(wsWbl.Handle));
Assert.That(testPara.RunCount, Is.EqualTo(1));
// main writing system should be the first one in the import
var mainWs = importedText.ContentsOA.MainWritingSystem;
Assert.That(mainWs, Is.EqualTo(wsWbl.Handle));
// Next verify that the IPA content got added to the imported word form
var a0 = paraImported.SegmentsOS[0].AnalysesRS[0];
var wf0 = a0 as IWfiWordform;
Assert.That(wf0, Is.Not.Null);
Assert.That(wf0.Form.get_String(wsWbl.Handle).Text, Is.EqualTo("baseline"));
Assert.That(wf0.Form.get_String(wsWblIpa.Handle).Text, Is.EqualTo("beslain"));
}
}
public void WordsFragDoc_OneWordPhraseAndOneGloss_AvoidDuplicates()
{
var wsEn = Cache.WritingSystemFactory.GetWsFromStr("en");
IWritingSystem wsKal;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("qaa-x-kal", out wsKal);
const string xml =
@"<document>
<word>
<item type='txt' lang='qaa-x-kal'>support a phrase</item>
<item type='gls' lang='en'>phrase gloss</item>
</word>
</document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
// First Import
Assert.DoesNotThrow(() => li.ImportWordsFrag(
() => new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())),
LinguaLinksImport.ImportAnalysesLevel.WordGloss));
// Second Import
Assert.DoesNotThrow(() => li.ImportWordsFrag(
() => new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())),
LinguaLinksImport.ImportAnalysesLevel.WordGloss));
var wordsRepo = Cache.ServiceLocator.GetInstance<IWfiWordformRepository>();
var wff1 = wordsRepo.GetMatchingWordform(wsKal.Handle, "support a phrase");
Assert.That(wff1, Is.Not.Null);
Assert.That(wff1.AnalysesOC, Has.Count.EqualTo(1));
Assert.That(wff1.AnalysesOC.ElementAt(0).MeaningsOC, Has.Count.EqualTo(1));
Assert.That(wff1.AnalysesOC.ElementAt(0).MeaningsOC.ElementAt(0).Form.get_String(wsEn).Text, Is.EqualTo("phrase gloss"));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiGlossRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiAnalysisRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiWordformRepository>().Count, Is.EqualTo(1));
}
public void WordsFragDoc_OneWordAndMultiGloss()
{
var wsEn = Cache.WritingSystemFactory.GetWsFromStr("en");
IWritingSystem wsKal;
Cache.ServiceLocator.WritingSystemManager.GetOrSet("qaa-x-kal", out wsKal);
const string xml =
@"<document>
<word>
<item type='txt' lang='qaa-x-kal'>glossedtwice</item>
<item type='gls' lang='en'>firstgloss</item>
<item type='gls' lang='en'>secondgloss</item>
</word>
</document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
Assert.DoesNotThrow(() => li.ImportWordsFrag(
() => new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())),
LinguaLinksImport.ImportAnalysesLevel.WordGloss));
var wordsRepo = Cache.ServiceLocator.GetInstance<IWfiWordformRepository>();
var wff1 = wordsRepo.GetMatchingWordform(wsKal.Handle, "glossedtwice");
Assert.That(wff1, Is.Not.Null);
Assert.That(wff1.AnalysesOC, Has.Count.EqualTo(2), "multiple word glosses (without specifying morphology) should create separate WfiAnalyses with separate glosses");
Assert.That(wff1.AnalysesOC.ElementAt(0).MeaningsOC, Has.Count.EqualTo(1));
Assert.That(wff1.AnalysesOC.ElementAt(0).MeaningsOC.ElementAt(0).Form.get_String(wsEn).Text, Is.EqualTo("firstgloss"));
Assert.That(wff1.AnalysesOC.ElementAt(1).MeaningsOC, Has.Count.EqualTo(1));
Assert.That(wff1.AnalysesOC.ElementAt(1).MeaningsOC.ElementAt(0).Form.get_String(wsEn).Text, Is.EqualTo("secondgloss"));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiGlossRepository>().Count, Is.EqualTo(2));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiAnalysisRepository>().Count, Is.EqualTo(2));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiWordformRepository>().Count, Is.EqualTo(1));
}
public void DeserializeWordsFragDocument()
{
const string xml =
@"<document>
<word>
<item type='txt' lang='qaa-x-kal'>glossedonce</item>
<item type='gls' lang='en'>onlygloss</item>
</word>
<word>
<item type='txt' lang='qaa-x-kal'>glossedtwice</item>
<item type='gls' lang='en'>firstgloss</item>
<item type='gls' lang='en'>secondgloss</item>
</word>
<word>
<item type='txt' lang='qaa-x-kal'>support a phrase</item>
<item type='gls' lang='en'>phrase gloss</item>
</word>
</document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
Assert.DoesNotThrow(() => li.ImportWordsFrag(
() => new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())),
LinguaLinksImport.ImportAnalysesLevel.WordGloss));
}
public void ImportUnknownPhraseWholeSegmentVersion_MakesPhrase()
{
// import an analysis with word gloss
string xml = "<document version=\"2\"><interlinear-text guid='AAAAAAAA-AAAA-AAAA-AAAA-AAAAAAAAAAAA'>" +
"<paragraphs><paragraph><phrases><phrase><words>" +
"<word>" +
"<item type='txt' lang='en'>this is not a phrase</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var options = CreateImportInterlinearOptions(xml);
FDO.IText importedText = null;
li.ImportInterlinear(options, ref importedText);
var stText = importedText.ContentsOA;
var para = (IStTxtPara)stText.ParagraphsOS[0];
var seg = para.SegmentsOS[0];
Assert.That(seg.AnalysesRS.Count, Is.EqualTo(1));
}
public void ImportKnownPhraseWholeSegmentNoVersion_MakesPhrase()
{
// import an analysis with word gloss
string xml = "<document><interlinear-text guid='AAAAAAAA-AAAA-AAAA-AAAA-AAAAAAAAAAAA'>" +
"<paragraphs><paragraph><phrases><phrase><words>" +
"<word>" +
"<item type='txt' lang='en'>this is a phrase</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
UndoableUnitOfWorkHelper.Do("undo", "redo", m_actionHandler,
() =>
{
var wf = Cache.ServiceLocator.GetInstance<IWfiWordformFactory>().Create();
int wsEn = Cache.WritingSystemFactory.GetWsFromStr("en");
wf.Form.set_String(wsEn, Cache.TsStrFactory.MakeString("this is a phrase", wsEn));
});
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var options = CreateImportInterlinearOptions(xml);
FDO.IText importedText = null;
li.ImportInterlinear(options, ref importedText);
var stText = importedText.ContentsOA;
var para = (IStTxtPara)stText.ParagraphsOS[0];
var seg = para.SegmentsOS[0];
Assert.That(seg.AnalysesRS.Count, Is.EqualTo(1));
}
public void ImportUnknownPhraseWholeSegmentNoVersion_MakesSeparateWords()
{
// import an analysis with word gloss
string xml = "<document><interlinear-text guid='AAAAAAAA-AAAA-AAAA-AAAA-AAAAAAAAAAAA'>" +
"<paragraphs><paragraph><phrases><phrase><words>" +
"<word>" +
"<item type='txt' lang='en'>this is not a phrase</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var options = CreateImportInterlinearOptions(xml);
FDO.IText importedText = null;
li.ImportInterlinear(options, ref importedText);
var stText = importedText.ContentsOA;
var para = (IStTxtPara)stText.ParagraphsOS[0];
var seg = para.SegmentsOS[0];
Assert.That(para.Contents.Text, Is.EqualTo("this is not a phrase"));
// It's acceptable either that it hasn't been parsed at all (and will be when we look at it) and so
// has no analyses, or that it's been parsed into five words. The other likely outcome is one phrase,
// which is not acceptable for parsing Saymore output (LT-12621).
Assert.That(seg.AnalysesRS.Count, Is.EqualTo(5).Or.EqualTo(0));
}
public void ImportNewUserConfirmedWordGlossSeparatedFromToExistingWfiAnalysis()
{
// build pre-existing data
var sl = Cache.ServiceLocator;
var wsf = Cache.WritingSystemFactory;
FDO.IText text;
IWfiWordform word = null;
ITsString paraContents = null;
var segGuid = new Guid();
NonUndoableUnitOfWorkHelper.Do(Cache.ActionHandlerAccessor, () =>
{
text = sl.GetInstance<ITextFactory>().Create(Cache, new Guid("AAAAAAAA-AAAA-AAAA-AAAA-AAAAAAAAAAAA"));
//Cache.LangProject.TextsOC.Add(text);
var sttext = sl.GetInstance<IStTextFactory>().Create();
text.ContentsOA = sttext;
IStTxtPara para = sl.GetInstance<IStTxtParaFactory>().Create();
sttext.ParagraphsOS.Add(para);
para.Contents = Cache.TsStrFactory.MakeString("supercalifragilisticexpialidocious", wsf.get_Engine("en").Handle);
paraContents = para.Contents;
ISegment segment = sl.GetInstance<ISegmentFactory>().Create();
para.SegmentsOS.Add(segment);
ITsString wform = TsStringUtils.MakeTss("supercalifragilisticexpialidocious",
wsf.get_Engine("en").Handle);
segGuid = segment.Guid;
word = sl.GetInstance<IWfiWordformFactory>().Create(wform);
var newWfiAnalysis = sl.GetInstance<IWfiAnalysisFactory>().Create();
word.AnalysesOC.Add(newWfiAnalysis);
segment.AnalysesRS.Add(word);
});
// import an analysis with word gloss
string xml = "<document><interlinear-text guid='AAAAAAAA-AAAA-AAAA-AAAA-AAAAAAAAAAAA'>" +
"<paragraphs><paragraph><phrases><phrase guid='" + segGuid + "'><words>" +
"<word>" +
"<item type='txt' lang='en'>supercalifragilisticexpialidocious</item>" +
"<item type='gls' lang='pt'>absurdo</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
var options = CreateImportInterlinearOptions(xml);
FDO.IText importedText = null;
li.ImportInterlinear(options, ref importedText);
using (var firstEntry = Cache.LanguageProject.Texts.GetEnumerator())
{
firstEntry.MoveNext();
var imported = firstEntry.Current;
Assert.IsNotNull(imported);
var importedPara = imported.ContentsOA.ParagraphsOS[0] as IStTxtPara;
Assert.IsNotNull(importedPara);
// assert that new Analysis was created
Assert.That(Cache.ServiceLocator.GetInstance<IWfiAnalysisRepository>().Count, Is.EqualTo(2));
// make sure imported word gloss is correct
Assert.That(importedPara.SegmentsOS[0].AnalysesRS.Count, Is.EqualTo(1));
var importedAnalysis = importedPara.SegmentsOS[0].AnalysesRS[0];
var skippedWord = importedAnalysis.Wordform;
var at = new AnalysisTree(importedAnalysis);
Assert.IsNotNull(at.Gloss, "IAnalysis should be WfiGloss");
var newGloss = at.Gloss;
Assert.That(newGloss.Form.get_String(wsf.get_Engine("pt").Handle).Text, Is.EqualTo("absurdo"));
Assert.That(skippedWord.Guid, Is.EqualTo(word.Guid));
// make sure nothing else has changed:
Assert.That(Cache.LanguageProject.Texts.Count, Is.EqualTo(1));
Assert.That(imported.ContentsOA.ParagraphsOS.Count, Is.EqualTo(1));
Assert.AreEqual(paraContents.Text, importedPara.Contents.Text, "Imported Para contents differ from original");
Assert.IsTrue(paraContents.Equals(importedPara.Contents), "Ws mismatch between imported and original paragraph");
Assert.That(skippedWord.Form.get_String(wsf.get_Engine("en").Handle).Text,
Is.EqualTo("supercalifragilisticexpialidocious"));
Assert.That(skippedWord.Guid, Is.EqualTo(word.Guid));
// make sure nothing else changed
Assert.That(Cache.ServiceLocator.GetInstance<IWfiGlossRepository>().Count, Is.EqualTo(1));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiWordformRepository>().Count, Is.EqualTo(1));
}
}
public void SkipNewGuessedWordGloss()
{
var wsf = Cache.WritingSystemFactory;
const string xml = "<document><interlinear-text>" +
"<paragraphs><paragraph><phrases><phrase><words>" +
"<word>" +
"<item type='txt' lang='en'>supercalifragilisticexpialidocious</item>" +
"<item type='gls' lang='pt' analysisStatus='guessByHumanApproved'>absurdo</item>" +
"</word>" +
"</words></phrase></phrases></paragraph></paragraphs></interlinear-text></document>";
var li = new BIRDFormatImportTests.LLIMergeExtension(Cache, null, null);
FDO.IText importedText = null;
var options = CreateImportInterlinearOptions(xml);
li.ImportInterlinear(options, ref importedText);
using (var firstEntry = Cache.LanguageProject.Texts.GetEnumerator())
{
firstEntry.MoveNext();
var imported = firstEntry.Current;
Assert.IsNotNull(imported);
var para = imported.ContentsOA.ParagraphsOS[0] as IStTxtPara;
Assert.IsNotNull(para);
Assert.That(para.Analyses.Count(), Is.EqualTo(1));
Assert.That(para.Analyses.First().Wordform.Form.get_String(wsf.get_Engine("en").Handle).Text,
Is.EqualTo("supercalifragilisticexpialidocious"));
var at = new AnalysisTree(para.Analyses.First());
Assert.IsNull(at.Gloss, "Analysis should not be WfiGloss");
// assert that nothing else was created
Assert.That(Cache.ServiceLocator.GetInstance<IWfiGlossRepository>().Count, Is.EqualTo(0));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiAnalysisRepository>().Count, Is.EqualTo(0));
Assert.That(Cache.ServiceLocator.GetInstance<IWfiWordformRepository>().Count, Is.EqualTo(1));
}
}
