private static void SetVernacularLanguagesByUsage(Interlineartext interlinText)
{
foreach (var para in interlinText.paragraphs)
{
foreach (var phrase in para.phrases)
{
foreach (var item in phrase.Items)
{
if (item.type == "txt")
{
EnsureVernacularLanguage(interlinText, item.lang);
}
}
if (phrase.WordsContent.Words != null)
{
foreach (var word in phrase.WordsContent.Words)
{
foreach (var item in word.Items)
{
if (item.type == "txt")
{
EnsureVernacularLanguage(interlinText, item.lang);
}
}
// We could dig into the morphemes, but any client generating morphemes probably
// does things right, and anyway we don't import that yet.
}
}
}
}
}
private static string GetInstructions(Interlineartext interlinText, String wsName, String instructions)
{
var strBldr = new StringBuilder(wsName);
strBldr.Append(instructions);
strBldr.Append(Environment.NewLine); strBldr.Append(Environment.NewLine);
strBldr.Append(GetPartOfPhrase(interlinText));
return(strBldr.ToString());
}
private static bool SomeLanguageSpecifiesVernacular(Interlineartext interlinText)
{
foreach (var lang in interlinText.languages.language)
{
if (lang.vernacularSpecified)
{
return(true);
}
}
return(false);
}
private static void EnsureVernacularLanguage(Interlineartext interlinText, string langName)
{
foreach (var lang in interlinText.languages.language)
{
if (lang.lang == langName)
{
lang.vernacularSpecified = true;
lang.vernacular = true;
return;
}
}
}
private static void SetTextMetaAndMedia(FdoCache cache, Interlineartext interlinText, ILgWritingSystemFactory wsFactory, FDO.IText newText)
{
if (interlinText.Items != null) // apparently it is null if there are no items.
{
foreach (var item in interlinText.Items)
{
switch (item.type)
{
case "title":
newText.Name.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "title-abbreviation":
newText.Abbreviation.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "source":
newText.Source.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "comment":
newText.Description.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
}
}
}
//handle media files section (ELAN initiated) needs to be processed before the paragraphs as segmenets could reference
//these parts.
if (interlinText.mediafiles != null)
{
var mediaFiles = newText.MediaFilesOA = cache.ServiceLocator.GetInstance <ICmMediaContainerFactory>().Create();
mediaFiles.OffsetType = interlinText.mediafiles.offsetType;
foreach (var mediaFile in interlinText.mediafiles.media)
{
var media = cache.ServiceLocator.GetInstance <ICmMediaURIFactory>().Create(cache, new Guid(mediaFile.guid));
mediaFiles.MediaURIsOC.Add(media);
media.MediaURI = mediaFile.location;
}
}
}
private static string GetPartOfPhrase(Interlineartext interlinText)
{
int i = 0;
var strBldr = new StringBuilder(ITextStrings.ksImportLangMissingTextStartsWith);
foreach (var paragraph in interlinText.paragraphs)
{
foreach (var phrase in paragraph.phrases)
{
foreach (var word in phrase.WordsContent.Words)
{
strBldr.Append(word.Items[0].Value);
strBldr.Append(" ");
i++;
if (i > 6)
{
strBldr.Append(" ...");
return(strBldr.ToString());
}
}
}
}
return(strBldr.ToString());
}
/// <summary>
/// This method will create a new Text document from the given BIRD format Interlineartext.
/// </summary>
/// <param name="newText">The text to populate, could be set to null.</param>
/// <param name="textParams"></param>
private static bool PopulateTextFromBIRDDoc(ref FDO.IText newText, TextCreationParams textParams)
{
s_importOptions = textParams.ImportOptions;
Interlineartext interlinText = textParams.InterlinText;
FdoCache cache = textParams.Cache;
IThreadedProgress progress = textParams.Progress;
if (s_importOptions.CheckAndAddLanguages == null)
{
s_importOptions.CheckAndAddLanguages = CheckAndAddLanguagesInternal;
}
ILgWritingSystemFactory wsFactory = cache.WritingSystemFactory;
char space = ' ';
//handle the languages(writing systems) section alerting the user if new writing systems are encountered
if (!s_importOptions.CheckAndAddLanguages(cache, interlinText, wsFactory, progress))
{
return(false);
}
//handle the header(info or meta) information
SetTextMetaAndMedia(cache, interlinText, wsFactory, newText);
//create all the paragraphs
foreach (var paragraph in interlinText.paragraphs)
{
if (newText.ContentsOA == null)
{
newText.ContentsOA = cache.ServiceLocator.GetInstance <IStTextFactory>().Create();
}
IStTxtPara newTextPara = newText.ContentsOA.AddNewTextPara("");
int offset = 0;
if (paragraph.phrases == null)
{
continue;
}
foreach (var phrase in paragraph.phrases)
{
ICmObject oldSegment = null;
//Try and locate a segment with this Guid.
if (!String.IsNullOrEmpty(phrase.guid))
{
if (cache.ServiceLocator.ObjectRepository.TryGetObject(new Guid(phrase.guid), out oldSegment))
{
//We aren't merging, but we have this guid in our system, ignore the file Guid
oldSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset);
}
else
{
//The segment is identified by a Guid, but apparently we don't have it in our current document, so make one with the guid
oldSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset, cache,
new Guid(phrase.guid));
}
}
//set newSegment to the old, or create a brand new one.
ISegment newSegment = oldSegment as ISegment ?? cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset);
var tsStrFactory = cache.ServiceLocator.GetInstance <ITsStrFactory>();
//Fill in the ELAN time information if it is present.
AddELANInfoToSegment(cache, phrase, newSegment);
ITsString phraseText = null;
bool textInFile = false;
//Add all of the data from <item> elements into the segment.
AddSegmentItemData(cache, wsFactory, phrase, newSegment, tsStrFactory, ref textInFile, ref phraseText);
bool lastWasWord = false;
if (phrase.WordsContent != null && phrase.WordsContent.Words != null)
{
if (textParams.Version == 0 && PhraseHasExactlyOneTxtItemNotAKnownWordform(newSegment.Cache, phrase))
{
// It might be a SayMore text that makes the whole segment a single txt item.
// We want to add the text anyway (unless a higher level did so), but we will skip making
// a wordform. Eventual parsing of the text will do so.
if (!textInFile)
{
UpdatePhraseTextForWordItems(wsFactory, tsStrFactory, ref phraseText, phrase.WordsContent.Words[0], ref lastWasWord, space);
}
}
else
{
foreach (var word in phrase.WordsContent.Words)
{
//If the text of the phrase was not given in the document build it from the words.
if (!textInFile)
{
UpdatePhraseTextForWordItems(wsFactory, tsStrFactory, ref phraseText, word, ref lastWasWord, space);
}
AddWordToSegment(newSegment, word, tsStrFactory);
}
}
}
UpdateParagraphTextForPhrase(newTextPara, ref offset, phraseText);
}
}
return(true);
}
private static bool CheckAndAddLanguagesInternal(FdoCache cache, Interlineartext interlinText, ILgWritingSystemFactory wsFactory, IThreadedProgress progress)
{
if (interlinText.languages != null)
{
if (!SomeLanguageSpecifiesVernacular(interlinText))
{
// Saymore file? something else that doesn't know to do this? We will confuse the user if we try to treat all as analysis.
SetVernacularLanguagesByUsage(interlinText);
}
foreach (var lang in interlinText.languages.language)
{
bool fIsVernacular;
var writingSystem = SafelyGetWritingSystem(cache, wsFactory, lang, out fIsVernacular);
if (fIsVernacular)
{
if (!cache.LanguageProject.CurrentVernacularWritingSystems.Contains(writingSystem.Handle))
{
//we need to invoke the dialog on the main thread so we can use the progress dialog as the parent.
//otherwise the message box can be displayed behind everything
IAsyncResult asyncResult = progress.ThreadHelper.BeginInvoke(
new ShowDialogAboveProgressbarDelegate(ShowDialogAboveProgressbar),
new object[]
{
progress,
writingSystem.LanguageName + ITextStrings.ksImportVernacLangMissing,
ITextStrings.ksImportVernacLangMissingTitle,
MessageBoxButtons.OKCancel
});
var result = (DialogResult)progress.ThreadHelper.EndInvoke(asyncResult);
if (result == DialogResult.OK)
{
cache.LanguageProject.AddToCurrentVernacularWritingSystems((IWritingSystem)writingSystem);
}
else
{
return(false);
}
}
}
else
{
if (!cache.LanguageProject.CurrentAnalysisWritingSystems.Contains(writingSystem.Handle))
{
IAsyncResult asyncResult = progress.ThreadHelper.BeginInvoke(
new ShowDialogAboveProgressbarDelegate(ShowDialogAboveProgressbar),
new object[]
{
progress,
writingSystem.LanguageName + ITextStrings.ksImportAnalysisLangMissing,
ITextStrings.ksImportAnalysisLangMissingTitle,
MessageBoxButtons.OKCancel
});
var result = (DialogResult)progress.ThreadHelper.EndInvoke(asyncResult);
//alert the user
if (result == DialogResult.OK)
{
//alert the user
cache.LanguageProject.AddToCurrentAnalysisWritingSystems((IWritingSystem)writingSystem);
// We already have progress indications up.
XmlTranslatedLists.ImportTranslatedListsForWs(writingSystem.Id, cache, null);
}
else
{
return(false);
}
}
}
}
}
return(true);
}
private static bool MergeTextWithBIRDDoc(ref FDO.IText newText, TextCreationParams textParams)
{
s_importOptions = textParams.ImportOptions;
Interlineartext interlinText = textParams.InterlinText;
FdoCache cache = textParams.Cache;
IThreadedProgress progress = textParams.Progress;
if (s_importOptions.CheckAndAddLanguages == null)
{
s_importOptions.CheckAndAddLanguages = CheckAndAddLanguagesInternal;
}
ILgWritingSystemFactory wsFactory = cache.WritingSystemFactory;
char space = ' ';
//handle the languages(writing systems) section alerting the user if new writing systems are encountered
if (!s_importOptions.CheckAndAddLanguages(cache, interlinText, wsFactory, progress))
{
return(false);
}
//handle the header(info or meta) information as well as any media-files sections
SetTextMetaAndMedia(cache, interlinText, wsFactory, newText);
IStText oldContents = newText.ContentsOA;
IStText newContents = null;
//create all the paragraphs NOTE: Currently the paragraph guids are being ignored, this might be wrong.
foreach (var paragraph in interlinText.paragraphs)
{
if (newContents == null)
{
newContents = cache.ServiceLocator.GetInstance <IStTextFactory>().Create();
newText.ContentsOA = newContents;
}
IStTxtPara newTextPara = newContents.AddNewTextPara("");
int offset = 0;
if (paragraph.phrases == null)
{
continue;
}
foreach (var phrase in paragraph.phrases)
{
ICmObject oldSegment = null;
//Try and locate a segment with this Guid.
if (!String.IsNullOrEmpty(phrase.guid))
{
if (cache.ServiceLocator.ObjectRepository.TryGetObject(new Guid(phrase.guid), out oldSegment))
{
if (oldSegment as ISegment != null) //The segment matches, add it into our paragraph.
{
newTextPara.SegmentsOS.Add(oldSegment as ISegment);
}
else if (oldSegment == null) //The segment is identified by a Guid, but apparently we don't have it in our current document, so make one
{
oldSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset, cache, new Guid(phrase.guid));
}
else //The Guid is in use, but not by a segment. This is bad.
{
return(false);
}
}
}
//set newSegment to the old, or create a brand new one.
ISegment newSegment = oldSegment as ISegment ?? cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset);
var tsStrFactory = cache.ServiceLocator.GetInstance <ITsStrFactory>();
//Fill in the ELAN time information if it is present.
AddELANInfoToSegment(cache, phrase, newSegment);
ITsString phraseText = null;
bool textInFile = false;
//Add all of the data from <item> elements into the segment.
AddSegmentItemData(cache, wsFactory, phrase, newSegment, tsStrFactory, ref textInFile, ref phraseText);
bool lastWasWord = false;
if (phrase.WordsContent != null && phrase.WordsContent.Words != null)
{
foreach (var word in phrase.WordsContent.Words)
{
//If the text of the phrase was not found in a "txt" item for this segment then build it from the words.
if (!textInFile)
{
UpdatePhraseTextForWordItems(wsFactory, tsStrFactory, ref phraseText, word, ref lastWasWord, space);
}
MergeWordToSegment(newSegment, word, tsStrFactory);
}
}
UpdateParagraphTextForPhrase(newTextPara, ref offset, phraseText);
}
}
return(true);
}
bool DummyCheckAndAddLanguagesInternal(FdoCache cache, Interlineartext interlinText, ILgWritingSystemFactory wsFactory, IThreadedProgress progress)
{
return(true);
}
/// <summary>
/// Set text metadata, create or merge media file URI's.
/// <note>media files (ELAN initiated) need to be processed before the paragraphs, as segments could reference these parts.</note>
/// </summary>
/// <param name="cache"></param>
/// <param name="interlinText">The source text</param>
/// <param name="wsFactory"></param>
/// <param name="newText">The target text</param>
/// <param name="merging">True if we are merging into an existing text; False if we are creating everything new</param>
private static void SetTextMetaAndMergeMedia(FdoCache cache, Interlineartext interlinText, ILgWritingSystemFactory wsFactory,
FDO.IText newText, bool merging)
{
if (interlinText.Items != null) // apparently it is null if there are no items.
{
foreach (var item in interlinText.Items)
{
switch (item.type)
{
case "title":
newText.Name.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "title-abbreviation":
newText.Abbreviation.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "source":
newText.Source.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
case "comment":
newText.Description.set_String(GetWsEngine(wsFactory, item.lang).Handle, item.Value);
break;
}
}
}
if (interlinText.mediafiles != null)
{
if (newText.MediaFilesOA == null)
{
newText.MediaFilesOA = cache.ServiceLocator.GetInstance <ICmMediaContainerFactory>().Create();
}
newText.MediaFilesOA.OffsetType = interlinText.mediafiles.offsetType;
foreach (var mediaFile in interlinText.mediafiles.media)
{
ICmObject extantObject;
cache.ServiceLocator.ObjectRepository.TryGetObject(new Guid(mediaFile.guid), out extantObject);
var media = extantObject as ICmMediaURI;
if (media == null)
{
media = cache.ServiceLocator.GetInstance <ICmMediaURIFactory>().Create(cache, new Guid(mediaFile.guid));
newText.MediaFilesOA.MediaURIsOC.Add(media);
}
else if (!merging)
{
// If a media URI with the same GUID exists, and we are not merging, create a new media URI with a new GUID
media = cache.ServiceLocator.GetInstance <ICmMediaURIFactory>().Create();
newText.MediaFilesOA.MediaURIsOC.Add(media);
// Update references to this Media URI
foreach (var phrase in interlinText.paragraphs.SelectMany(para => para.phrases))
{
if (mediaFile.guid.Equals(phrase.mediaFile))
{
phrase.mediaFile = media.Guid.ToString();
}
}
}
// else, the media URI already exists and we are merging; simply update the location
media.MediaURI = mediaFile.location;
}
}
}
private static bool SomeLanguageSpecifiesVernacular(Interlineartext interlinText)
{
// return true if any language in the languages section is vernacular
return(interlinText.languages.language.Any(lang => lang.vernacularSpecified));
}
/// <summary>
/// Merge the contents of the given Text into the exising one. If this fails
/// for some reason then return false to tell the calling method to abort the import.
/// </summary>
/// <param name="newText"></param>
/// <param name="textParams"></param>
/// <returns>The imported text may be in a writing system that is not part of this project. Return false if the user
/// rejects the text which tells the caller of this method to abort the import.</returns>
private static bool MergeTextWithBIRDDoc(ref LCModel.IText newText, TextCreationParams textParams)
{
s_importOptions = textParams.ImportOptions;
Interlineartext interlinText = textParams.InterlinText;
LcmCache cache = textParams.Cache;
IThreadedProgress progress = textParams.Progress;
if (s_importOptions.CheckAndAddLanguages == null)
{
s_importOptions.CheckAndAddLanguages = CheckAndAddLanguagesInternal;
}
ILgWritingSystemFactory wsFactory = cache.WritingSystemFactory;
char space = ' ';
//handle the languages(writing systems) section alerting the user if new writing systems are encountered
if (!s_importOptions.CheckAndAddLanguages(cache, interlinText, wsFactory, progress))
{
return(false);
}
//handle the header(info or meta) information as well as any media-files sections
SetTextMetaAndMergeMedia(cache, interlinText, wsFactory, newText, true);
IStText newContents = newText.ContentsOA;
//create or reuse the paragraphs available. NOTE: Currently the paragraph guids are being ignored, this might be wrong.
foreach (var paragraph in interlinText.paragraphs)
{
IStTxtPara newTextPara = null;
if (newContents == null)
{
newContents = cache.ServiceLocator.GetInstance <IStTextFactory>().Create();
newText.ContentsOA = newContents;
newTextPara = newContents.AddNewTextPara("");
}
int offset = 0;
if (paragraph.phrases == null)
{
continue;
}
foreach (var phrase in paragraph.phrases)
{
ICmObject oldSegment = null;
//Try and locate a segment with this Guid. Assign newTextPara to the paragraph we're working on if we haven't already
if (!String.IsNullOrEmpty(phrase.guid))
{
if (cache.ServiceLocator.ObjectRepository.TryGetObject(new Guid(phrase.guid), out oldSegment))
{
if (oldSegment as ISegment != null) //The segment matches, add it into our paragraph.
{
IStTxtPara segmentOwner = newContents.ParagraphsOS.FirstOrDefault(para =>
para.Guid.Equals((oldSegment as ISegment).Owner.Guid)) as IStTxtPara;
if (segmentOwner != null && newTextPara == null) //We found the StTxtPara that correspond to this paragraph
{
newTextPara = segmentOwner;
}
}
else if (oldSegment == null) //The segment is identified by a Guid, but apparently we don't have it in our current document, so make one
{
if (newTextPara == null)
{
newTextPara = newContents.AddNewTextPara("");
}
oldSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset, cache, new Guid(phrase.guid));
}
else //The Guid is in use, but not by a segment. This is bad.
{
return(false);
}
}
}
//If newTextPara is null, try to use a paragraph that a sibling phrase belongs to.
//Note: newTextPara is only assigned once, and won't be reassigned until we iterate through a new paragraph.
if (newTextPara == null)
{
var phraseGuids = paragraph.phrases.Select(p => p.guid);
foreach (IStTxtPara para in newContents.ParagraphsOS)
{
if (para.SegmentsOS.Any(seg => phraseGuids.Contains(seg.Guid.ToString())))
{
newTextPara = para;
break;
}
}
}
//Can't find any paragraph for our phrase, create a brand new paragraph
if (newTextPara == null)
{
newTextPara = newContents.AddNewTextPara("");
}
//set newSegment to the old, or create a brand new one.
ISegment newSegment = oldSegment as ISegment;
if (newSegment == null)
{
if (!string.IsNullOrEmpty(phrase.guid))
{
//The segment is identified by a Guid, but apparently we don't have it in our current document, so make one with the guid
newSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset, cache, new Guid(phrase.guid));
}
else
{
newSegment = cache.ServiceLocator.GetInstance <ISegmentFactory>().Create(newTextPara, offset);
}
}
//Fill in the ELAN time information if it is present.
AddELANInfoToSegment(cache, phrase, newSegment);
ITsString phraseText = null;
bool textInFile = false;
//Add all of the data from <item> elements into the segment.
AddSegmentItemData(cache, wsFactory, phrase, newSegment, ref textInFile, ref phraseText);
bool lastWasWord = false;
if (phrase.WordsContent != null && phrase.WordsContent.Words != null)
{
//Rewrite our analyses
newSegment.AnalysesRS.Clear();
foreach (var word in phrase.WordsContent.Words)
{
//If the text of the phrase was not found in a "txt" item for this segment then build it from the words.
if (!textInFile)
{
UpdatePhraseTextForWordItems(wsFactory, ref phraseText, word, ref lastWasWord, space);
}
MergeWordToSegment(newSegment, word);
}
}
UpdateParagraphTextForPhrase(newTextPara, ref offset, phraseText);
}
}
return(true);
}
