/// <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(ICollection <SubtitleLine> subtitles)
{
Dictionary <string, SubtitleWord> result = new Dictionary <string, SubtitleWord>();
var texts = subtitles.Select(s => s.EnglishText).ToList();
foreach (var line in texts)
{
var lineResult = sentenceParse.Pickup(line);
foreach (KeyValuePair <string, string> keyValuePair in lineResult)
{
if (result.ContainsKey(keyValuePair.Key))
{
continue;
}
string original = keyValuePair.Key;
string word = keyValuePair.Value;
var mean = sentenceParse.RemarkWord(line, word, original);
if (mean != null)
{
var wd = new SubtitleWord()
{
Word = mean.Word,
WordInSubitle = word,
Means = mean.Means,
SubtitleSentence = line,
SelectMean = mean.DefaultMean == null ? mean.Means[0].ToString() : mean.DefaultMean.ToString()
};
result.Add(original, wd);
}
}
}
return(result);
}
/// <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(ICollection <SubtitleLine> subtitles)
{
Dictionary <string, SubtitleWord> result = new Dictionary <string, SubtitleWord>();
var unknownWords = DbOperator.Instance.GetAllUserUnKnownVocabulary().ToDictionary(s => s.Word, s => s.IsStar);
var texts = subtitles.Select(s => s.EnglishText).ToList();
foreach (var line in texts)
{
var lineResult = sentenceParse.Pickup(line);
foreach (KeyValuePair <string, string> keyValuePair in lineResult)
{
string original = keyValuePair.Key;
//if (knownWords.Contains(original)) continue;
string word = keyValuePair.Value;
//if(knownWords.Contains(word)) continue;
if (result.ContainsKey(original))
{
result[original].ShowCount++;
if (!result[original].WordInSubtitle.Contains(word))
{
result[original].WordInSubtitle.Add(word);
}
continue;
}
var mean = sentenceParse.RemarkWord(line, word, original);
if (mean != null)
{
var wd = new SubtitleWord()
{
Word = mean.Word,
ShowCount = 1,
WordInSubtitle = new List <string>()
{
word
},
Means = mean.Means,
SubtitleSentence = line,
SelectMean = mean.DefaultMean == null ? mean.Means[0].ToString() : mean.DefaultMean.ToString()
};
if (unknownWords.ContainsKey(mean.Word))
{
wd.IsStar = unknownWords[mean.Word];
}
result.Add(original, wd);
}
}
}
return(result);
}
/// <summary>
/// 将一个句子中的生词进行注释
/// </summary>
/// <param name="line"></param>
/// <param name="words"></param>
/// <returns></returns>
private string ReplaceSubtitleLineByVocabulary(string line, IDictionary <string, SubtitleWord> words)
{
StringBuilder sb = new StringBuilder();
foreach (string s in SentenceParse.SplitSentence(line))
{
if (s.Length == 1)
{
sb.Append(s);
}
else
{
var word = s.ToLower();
string mean = "";
SubtitleWord wordWithMean = null;
if (words.ContainsKey(s))//这个词需要注释
{
wordWithMean = words[s];
}
else if (words.ContainsKey(word))
{
wordWithMean = words[word];
}
mean = wordWithMean?.SelectMean;
if (!String.IsNullOrEmpty(mean))
{
if (Global.ShortMean)
{
var meanarray = mean.Split(new char[] { ';', ',' }, StringSplitOptions.RemoveEmptyEntries);
mean = meanarray[0];
mean = mean.Substring(mean.IndexOf(' ') + 1);
}
var formatted = string.Format("{0}({1})", s, mean.Trim());
if (wordWithMean.IsStar)
{
//标星的单词，需要突出显示
formatted = String.Format("<font color='{0}'>{1}</font>", ColorTranslator.ToHtml(meanColor), formatted);
}
sb.Append(formatted);
}
else
{
sb.Append(s);
}
}
}
return(sb.ToString());
//var array = SentenceParse.SplitSentence2Words(line);
//foreach (string oword in array)
//{
// var word = oword.ToLower();
// //var w = englishWordService.GetOriginalWord(word);
// //var mean = words.ContainsKey(w) ? words[w].SelectMean : "";
// //if (!string.IsNullOrEmpty(mean))
// //{
// // if(shortMean)
// // {
// // var meanarray = mean.Split(new char[]{';',','},StringSplitOptions.RemoveEmptyEntries);
// // mean = meanarray[0];
// // }
// // line = ReplaceNewWord(line, word, mean);
// //}
//}
//return line;
}
/// <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(ICollection<SubtitleLine> subtitles)
{
Dictionary<string, SubtitleWord> result = new Dictionary<string, SubtitleWord>();
var unknownWords = DbOperator.Instance.GetAllUserUnKnownVocabulary().ToDictionary(s=>s.Word,s=>s.IsStar);
var texts = subtitles.Select(s => s.EnglishText).ToList();
foreach (var line in texts)
{
var lineResult = sentenceParse.Pickup(line);
foreach (KeyValuePair<string, string> keyValuePair in lineResult)
{
string original = keyValuePair.Key;
//if (knownWords.Contains(original)) continue;
string word = keyValuePair.Value;
//if(knownWords.Contains(word)) continue;
if (result.ContainsKey(original))
{
result[original].ShowCount++;
if (!result[original].WordInSubtitle.Contains(word))
{
result[original].WordInSubtitle.Add(word);
}
continue;
}
var mean = sentenceParse.RemarkWord(line, word, original);
if (mean != null)
{
var wd = new SubtitleWord()
{
Word = mean.Word,
ShowCount = 1,
WordInSubtitle = new List<string>() {word},
Means = mean.Means,
SubtitleSentence = line,
SelectMean = mean.DefaultMean == null ? mean.Means[0].ToString() : mean.DefaultMean.ToString()
};
if (unknownWords.ContainsKey(mean.Word))
{
wd.IsStar = unknownWords[mean.Word];
}
result.Add(original, wd);
}
}
}
return result;
}
