/// <summary>
/// Reads the orthography conversion settings file and applies the specified mapping to a list of strings
/// </summary>
/// <param name="fragments">The texts (as an IEnumerable<string>) to apply the mapping to</param>
/// <param name="orthographyConversionFile">The filename containing the conversion settings. It should be tab-delimited with 2 columns. The 1st column is a sequence of 1 or more characters in the source language. The 2nd column contains a sequence of characteres to map to in the target language.</param>
/// <returns></returns>
public static IEnumerable <string> ApplyOrthographyConversion(IEnumerable <string> fragments, string orthographyConversionFile)
{
var converter = new OrthographyConverter(orthographyConversionFile);
foreach (var fragment in fragments)
{
string mappedFragment = converter.ApplyMappings(fragment);
yield return(mappedFragment);
}
}
/// <summary>
/// Reads the orthography conversion settings file and applies the specified mapping to a piece of text
/// </summary>
/// <param name="text">The text (as a scalar string) to apply the mapping to</param>
/// <param name="orthographyConversionFile">The filename containing the conversion settings. It should be tab-delimited with 2 columns. The 1st column is a sequence of 1 or more characters in the source language. The 2nd column contains a sequence of characteres to map to in the target language.</param>
/// <returns></returns>
public static string ApplyOrthographyConversion(string text, string orthographyConversionFile)
{
var converter = new OrthographyConverter(orthographyConversionFile);
return(converter.ApplyMappings(text));
}
// Determines which language code to use for eSpeak
private string GetBestSupportedLanguage(string requestedLangCode, out string stdOut, out string stdErr)
{
stdOut = "";
stdErr = "";
// Normally requestedLangCode should be under our control.
// But just do a quick and easy check to make sure it looks reasonable. (there are some highly contrived scenarios where a XSS injection would be possible with some social engineering.)
if (requestedLangCode.Contains('"') || requestedLangCode.Contains('\\'))
{
// This doesn't look like a lang code and has non-zero potential for injection. just return a default value instead
Debug.Assert(false);
return("eo");
}
// First try the requested langauge directly.
// (We need to test eSpeak directly instead of Aeneas because when using TTS overrides, there's no Aeneas error message that tells us if the language is unsupported.
// Therefore, we explicitly test if the language is supported by the dependency (eSpeak) before getting started.
if (!DoesCommandCauseError($"espeak -v {requestedLangCode} -q \"hello world\"", kWorkingDirectory, out stdOut, out stdErr))
{
return(requestedLangCode);
}
// Nope, looks like the requested language is not supported by the eSpeak installation.
// Let's check the fallback languages.
var potentialFallbackLangs = new List <string>();
// Check the orthography conversion files. If present, they specify the (first) fallback language to be used.
string collectionPath = _bookSelection.CurrentSelection.CollectionSettings.FolderPath;
var matchingFiles = Directory.EnumerateFiles(collectionPath, $"convert_{requestedLangCode}_to_*.txt");
foreach (var matchingFile in matchingFiles)
{
Tuple <string, string> sourceAndTargetTuple = OrthographyConverter.ParseSourceAndTargetFromFilename(matchingFile);
if (sourceAndTargetTuple != null)
{
string targetLang = sourceAndTargetTuple.Item2;
potentialFallbackLangs.Add(targetLang);
break;
}
}
// Add more default fallback languages to the end
potentialFallbackLangs.Add("eo"); // "eo" is Esperanto
potentialFallbackLangs.Add("en");
// Now go and try the fallback languages until we (possibly) find one that works
string langCode = null;
foreach (var langCodeToTry in potentialFallbackLangs)
{
if (!DoesCommandCauseError($"espeak -v {langCodeToTry} -q \"hello world\"", kWorkingDirectory, out stdOut, out stdErr))
{
langCode = langCodeToTry;
break;
}
}
return(langCode);
}