/// <summary>
/// Gather and sort list of DeGeWo base forms that are omitted from OT-derived word list.
/// </summary>
private void gatherOmitted()
{
foreach (var x in baseFormToClass)
{
if (wordToWWC.ContainsKey(x.Key))
{
continue;
}
WordWithClass wwc = new WordWithClass
{
Word = x.Key,
FreqClass = x.Value,
OrigRank = baseFormToRank[x.Key]
};
corpusOmitted.Add(wwc);
}
corpusOmitted.Sort((x, y) => x.OrigRank.CompareTo(y.OrigRank));
}
/// <summary>
/// Records OT word with the provided class+rank, or updates earlier class+rank if this one's lower.
/// </summary>
private void addUpdateWordWithClass(string word, int freqClass, int origRank)
{
// Not recorded before: add.
if (!wordToWWC.ContainsKey(word))
{
WordWithClass wwc = new WordWithClass
{
Word = word,
FreqClass = freqClass,
OrigRank = origRank
};
wordToWWC[word] = wwc;
wordsWithClass.Add(wwc);
}
// Seen before: update if new class is lower.
else if (freqClass < wordToWWC[word].FreqClass)
{
wordToWWC[word].FreqClass = freqClass;
wordToWWC[word].OrigRank = origRank;
}
}
