/// <summary>
/// 找到字幕中的生词，先进行分词，然后取每个单词的原型，然后看每个单词是否认识，认识则跳过，不认识则注释。
/// </summary>
/// <param name="subtitles"></param>
/// <returns></returns>
private IDictionary <string, SubtitleWord> PickNewWords(IList <SubtitleLine> subtitles)
{
Dictionary <string, SubtitleWord> result = new Dictionary <string, SubtitleWord>();
var knownVocabulary = dbOperator.FindAll <UserVocabulary>(v => v.KnownStatus == KnownStatus.Known).Select(v => v.Word).ToList();
var texts = subtitles.Select(s => s.EnglishText).ToList();
foreach (var line in texts)
{
var array = line.Split(new char[] { ' ', ',', '.', '?', ':', '!' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string word in array)
{
if (IsEnglishName(word))
{
//英文名，忽略
continue;
}
var original = englishWordService.GetOriginalWord(word);
if (knownVocabulary.Contains(word) || knownVocabulary.Contains(original))
{
//认识的单词，忽略
continue;
}
if (result.ContainsKey(original))
{
//重复的单词
continue;
}
var mean = RemarkWord(original);
if (mean != null)
{
result.Add(original, new SubtitleWord()
{
Word = original, WordInSubitle = word, Means = mean.Means, SubtitleSentence = line
});
}
}
}
return(result);
}
public void TestVerbIng(string word, string result)
{
var o = service.GetOriginalWord(word);
Assert.AreEqual(o, result);
}