public void ExtractTextByInChineseBracketsColonKeyWord(string filename)
{
var lines = new List <String>();
var sr = new StreamReader(filename);
while (!sr.EndOfStream)
{
var line = sr.ReadLine();
if (!String.IsNullOrEmpty(line))
{
lines.Add(line);
}
}
sr.Close();
for (int CurrentLineIdx = 0; CurrentLineIdx < lines.Count; CurrentLineIdx++)
{
var line = lines[CurrentLineIdx];
foreach (var word in LeadingColonKeyWordListInChineseBrackets)
{
var result = GetValueInChineseBracketsLeadingKeyWord(line, word);
foreach (var item in result)
{
CandidateWord.Add(new LocAndValue <string>()
{
Loc = CurrentLineIdx,
Value = item
});
}
}
}
}
void ExtractByStartEndStringFeature(string filename)
{
var lines = new List <String>();
var sr = new StreamReader(filename);
while (!sr.EndOfStream)
{
var line = sr.ReadLine();
if (!String.IsNullOrEmpty(line))
{
lines.Add(line);
}
}
sr.Close();
foreach (var word in StartEndFeature)
{
for (int CurrentLineIdx = 0; CurrentLineIdx < lines.Count; CurrentLineIdx++)
{
var line = lines[CurrentLineIdx];
var list = RegularTool.GetMultiValueBetweenString(line, word.StartWith, word.EndWith);
foreach (var item in list)
{
CandidateWord.Add(new LocAndValue <string>()
{
Value = item
});
}
}
;
}
}
/// <summary>
/// 使用开始结束关键字进行抽取
/// </summary>
/// <param name="keys"></param>
/// <param name="dplist"></param>
public void StartWithKey(List <DPKeyWord> keys, List <List <struWordDP> > dplist)
{
foreach (var key in keys)
{
foreach (var paragragh in dplist)
{
bool isStart = false;
string x = String.Empty;
foreach (var word in paragragh)
{
if (word.cont == "。" || word.cont == "：" || word.cont == "，" || word.cont == "；")
{
if (isStart)
{
x = String.Empty;
isStart = false;
continue;
}
}
if (isStart)
{
if (word.relate == LTPTrainingDP.右附加关系)
{
continue;
}
x += word.cont;
}
if (key.StartWord.Contains(word.cont))
{
if (key.StartDPValue.Length == 0 || (key.StartDPValue.Length != 0 && key.StartDPValue.Contains(word.relate)))
{
isStart = true;
}
}
if (key.EndWord.Contains(word.cont))
{
if (key.EndDPValue.Length == 0 || (key.EndDPValue.Length != 0 && key.EndDPValue.Contains(word.relate)))
{
if (isStart)
{
CandidateWord.Add(new LocAndValue <string>()
{
Value = x
});
}
isStart = false;
x = String.Empty;
}
}
}
}
}
}
/// <summary>
/// 检索流程方法
/// </summary>
/// <param name="root">HTML根</param>
/// <param name="ExtractMethod">特定检索方法(HTML内容，候补词列表)</param>
void SearchNormalContent(MyRootHtmlNode root, Func <String, List <String> > ExtractMethod)
{
foreach (var paragrah in root.Children)
{
//从各个段落的内容中取得：内容包含了内置列表，所以，这里不再重复
foreach (var contentNode in paragrah.Children)
{
if (contentNode.TableId == -1)
{
//非表格
var candidate = ExtractMethod(contentNode.Content);
foreach (var item in candidate)
{
CandidateWord.Add(new LocAndValue <String>()
{
Loc = contentNode.PositionId,
Value = item
});
}
}
}
}
}
public void ExtractTextByColonKeyWord(string filename)
{
var lines = new List <String>();
var sr = new StreamReader(filename);
while (!sr.EndOfStream)
{
var line = sr.ReadLine();
if (!String.IsNullOrEmpty(line))
{
lines.Add(line);
}
}
sr.Close();
for (int CurrentLineIdx = 0; CurrentLineIdx < lines.Count; CurrentLineIdx++)
{
var line = lines[CurrentLineIdx].Replace(" ", "");
foreach (var word in LeadingColonKeyWordList)
{
if (Utility.GetStringAfter(line, word) != String.Empty)
{
var result = Utility.GetStringAfter(line, word);
if (string.IsNullOrEmpty(result))
{
continue;
}
CandidateWord.Add(new LocAndValue <string>()
{
Loc = CurrentLineIdx,
Value = result
});
break;
}
}
}
}
