public void Read(string filename)
{
using (StreamReader reader = new StreamReader(filename))
{
string line = string.Empty;
ConvertPaper paper = null;
string field = string.Empty;
string prefix = string.Empty;
while ((line = reader.ReadLine()) != null)
{
if (line.Contains("题名："))
{
paper = new ConvertPaper();
paper.Title = line.Substring("题名：".Length).Trim();
}
if (line.Contains("作者："))
{
paper.Author = line.Substring("作者：".Length).Trim();
}
if (line.Contains("单位："))
{
paper.Institutes = line.Substring("单位：".Length).Trim().Split(';').ToList <string>();
}
if (line.Contains("关键词："))
{
paper.Keywords = line.Substring("关键词：".Length).Trim();
}
if (line.Contains("来源："))
{
paper.Source = line.Substring("来源：".Length).Trim();
string[] splits = paper.Source.Split(',');
paper.Year = splits[1].Trim();
Match match = regYear.Match(paper.Year);
if (match.Success)
{
paper.Year = match.Value;
}
}
if (line.Contains("被引频次："))
{
int cites = 0;
int.TryParse(line.Substring("被引频次：".Length).Trim(), out cites);
paper.TotalCites = cites;
cache.Add(paper);
paper = null;
}
}
reader.Close();
}
}
public void Read(string filename)
{
using (StreamReader reader = new StreamReader(filename))
{
string line = string.Empty;
ConvertPaper paper = null;
string field = string.Empty;
string prefix = string.Empty;
while ((line = reader.ReadLine()) != null)
{
if (line.Contains("【文件序号】"))
{
paper = new ConvertPaper();
paper.AccessionNumber = line.Substring(6).Trim();
}
if (line.Contains("【来源篇名】"))
{
paper.Title = line.Substring(6).Trim();
}
if (line.Contains("【来源作者】"))
{
paper.Author = line.Substring(6).Trim();
}
if (line.Contains("【关 键 词】"))
{
paper.Keywords = line.Substring("【关 键 词】".Length).Trim();
}
if (line.Contains("【机构名称】"))
{
paper.Institutes.Add(line.Substring(6).Trim());
}
if (line.Contains("【期 刊】"))
{
paper.Source = line.Substring("【期 刊】".Length).Trim();
}
if (line.Contains("【年代卷期】"))
{
paper.Volume = line.Substring("【年代卷期】".Length).Trim();
paper.Year = line.Substring("【年代卷期】".Length, 4).Trim();
}
if (line.Contains("----"))
{
cache.Add(paper);
paper = null;
}
}
reader.Close();
}
}
/// <summary>
/// 读取数据进入缓存
/// </summary>
/// <param name="filename"></param>
public void Read(string filename)
{
using (StreamReader reader = new StreamReader(filename))
{
string line = string.Empty;
ConvertPaper paper = null;
string field = string.Empty;
string prefix = string.Empty;
while ((line = reader.ReadLine()) != null)
{
if (line.Length == 0)
{
continue;
}
prefix = line.Substring(0, 2);
if (prefix == " ")
{
prefix = field;
}
switch (prefix)
{
case "PT":
paper = new ConvertPaper();
break;
case "TI":
field = "TI";
paper.Title += line.Substring(3) + " ";
break;
case "AU":
field = "AU";
paper.Author_ += "; " + line.Substring(3).Trim();
paper.Author_Count++;
break;
case "AF":
field = "AF";
paper.Author += "; " + line.Substring(3);
paper.AuthorCount++;
break;
case "DE":
field = "DE";
paper.Keywords += " " + line.Substring(3);
break;
case "ID":
field = "ID";
paper.Keywords_ += " " + line.Substring(3);
break;
case "SO":
field = "SO";
paper.Source = line.Substring(3).Trim();
break;
case "C1":
field = "C1";
paper.Institutes.Add(line.Substring(3));
break;
case "Z9":
field = "Z9";
int cites = 0;
int.TryParse(line.Substring(3).Trim(), out cites);
paper.Cites = cites;
break;
case "TC":
field = "TC";
int totalcites = 0;
int.TryParse(line.Substring(3).Trim(), out totalcites);
paper.TotalCites = totalcites;
paper.isCite = true;
break;
case "PY":
field = "PY";
paper.Year = line.Substring(3).Trim();
break;
case "VL":
field = "VL";
paper.Volume = line.Substring(3).Trim();
break;
case "IS":
field = "IS";
paper.Issue = line.Substring(3).Trim();
break;
case "AR":
field = "AR";
paper.ArticleNumber = line.Substring(3).Trim();
break;
case "DI":
field = "DI";
paper.Doi = line.Substring(3).Trim();
break;
case "BP":
field = "BP";
paper.BeginPage = line.Substring(3).Trim();
break;
case "EP":
field = "EP";
paper.EndPage = line.Substring(3).Trim();
break;
case "CT":
field = "CT";
paper.ConferenceTitle += line.Substring(3).Trim() + " ";
break;
case "CL":
field = "CL";
paper.ConferenceAddress += line.Substring(3).Trim() + " ";
break;
case "CY":
field = "CY";
paper.ConferenceDate += line.Substring(3).Trim() + " ";
break;
case "UT":
field = "UT";
paper.AccessionNumber += line.Substring(3).Trim() + " ";
break;
case "ER":
if (!string.IsNullOrEmpty(paper.Author))
{
paper.Author = paper.Author.Substring(1);
}
else if (!string.IsNullOrEmpty(paper.Author_))
{
paper.Author = paper.Author_.Substring(1);
paper.AuthorCount = paper.Author_Count;
}
if (!string.IsNullOrEmpty(paper.Keywords))
{
paper.Keywords = paper.Keywords.Substring(1);
}
else if (!string.IsNullOrEmpty(paper.Keywords_))
{
paper.Keywords = paper.Keywords_.Substring(1);
}
cache.Add(paper);
break;
default:
field = "";
break;
}
}
reader.Close();
}
}
public void Read(string filename)
{
using (StreamReader reader = new StreamReader(filename))
{
string line = string.Empty;
ConvertPaper paper = null;
string field = string.Empty;
string prefix = string.Empty;
while ((line = reader.ReadLine()) != null)
{
if (line.Contains("Accession number:"))
{
paper = new ConvertPaper();
paper.AccessionNumber = line.Substring("Accession number:".Length).Trim();
}
if (line.Contains("Title:"))
{
paper.Title = line.Substring("Title:".Length).Trim();
}
if (line.Contains("Authors:"))
{
paper.Author = line.Substring("Authors:".Length).Trim();
if (!string.IsNullOrEmpty(paper.Author))
{
string[] authors = paper.Author.Split(';');
if (authors != null)
{
paper.AuthorCount = authors.Length;
}
}
}
if (line.Contains("Author affiliation:"))
{
string institutes = line.Substring("Author affiliation:".Length).Trim();
//机构需要根据 (1)这种标记匹配
MatchCollection matches = regSplitor.Matches(institutes);
List <string> splitors = new List <string>();
foreach (Match match in matches)
{
splitors.Add(match.Value);
}
//string[] _institutes = institutes.Split(';');
string[] _institutes = institutes.Split(splitors.ToArray(), StringSplitOptions.RemoveEmptyEntries);
for (int index = 0; index < _institutes.Length; index++)
{
_institutes[index] = "(" + (index + 1) + ")" + _institutes[index];
}
List <string> instituteslist = new List <string>();
if (_institutes != null && _institutes.Length > 0)
{
for (int index = 0; index < _institutes.Length; index++)
{
if (_institutes[index].Trim().Length > 0)
{
string temp = _institutes[index].Trim();
paper.Institutes.Add(temp.Trim());
}
}
}
}
if (line.Contains("Source title:"))
{
paper.Source = line.Substring("Source title:".Length).Trim();
}
if (line.Contains("Volume:"))
{
paper.Volume = line.Substring("Volume:".Length).Trim();
}
if (line.Contains("Issue:"))
{
paper.Issue = line.Substring("Issue:".Length).Trim();
}
if (line.Contains("Pages:"))
{
paper.BeginPage = line.Substring("Pages:".Length).Trim();
}
if (line.Contains("Publication year:"))
{
paper.Year = line.Substring("Publication year:".Length).Trim();
}
if (line.Contains("DOI:"))
{
paper.Doi = line.Substring("DOI:".Length).Trim();
}
if (line.Contains("Compilation and indexing terms"))
{
cache.Add(paper);
}
}
reader.Close();
}
}
