static void test4()
{
string originString = (new StreamReader("03,073.txt")).ReadToEnd();
DateTime dt = DateTime.Now;
Reaper reaper = new Reaper(originString);
foreach (Reaper part in reaper.RemainBeforeFirst("<table width=\"100%\" summary=\"table used for formatting\"><tr><td>").ReapByProfix("<hr><big><b><i>"))
{
string partName = part.RemainBeforeFirst(":</i></b></big>").GetResult()[0];
foreach (Reaper table in part.ReapByProfix("<div align=\"center\"><table border=1 summary=\""))
{
string tableName = table.RemainBeforeFirst("\" width=\"95%\">").GetResult()[0];
foreach (Reaper line in table.ReapByProfix("<tr><td>").GiveUpContain("<td>Jan</td><td>Feb</td><td>Mar</td>"))
{
string lineName = line.RemainBeforeFirst("</td>").GetResult()[0];
/////////
foreach (Reaper data in line.ReapByProfix("<td align=\"center\" nowrap>").RemainBeforeFirst("</td>"))
{
/////////
}
/////////////
}
}
}
////////////
TimeSpan ts = DateTime.Now - dt;
Console.WriteLine(ts);
}
static void test3()
{
Reaper reaper = new Reaper((new StreamReader("03,073.html")).ReadToEnd());
StreamWriter outputFile = new StreamWriter("test3_output.txt");
foreach (Reaper part in reaper.RemainBeforeFirst("<table width=\"100%\" summary=\"table used for formatting\"><tr><td>").ReapByProfix("<hr><big><b><i>"))
{
string partName = part.RemainBeforeFirst(":</i></b></big>").GetResult()[0];
ShowStrings(part.RemainBeforeFirst(":</i></b></big>").GetResult());
foreach (Reaper table in part.ReapByProfix("<div align=\"center\"><table border=1 summary=\""))
{
string tableName = table.RemainBeforeFirst("\" width=\"95%\">").GetResult()[0];
int lineCount = 0;
foreach (Reaper line in table.ReapByProfix("<tr><td>").GiveUpContain("<td>Jan</td><td>Feb</td><td>Mar</td>"))
{
string lineName = line.RemainBeforeFirst("</td>").GetResult()[0];
outputFile.WriteLine("PART:" + partName);
outputFile.WriteLine("TABLE:" + tableName);
outputFile.WriteLine("LINE:" + lineName);
outputFile.WriteLine("NO.:" + ++lineCount);
foreach (Reaper data in line.ReapByProfix("<td align=\"center\" nowrap>").RemainBeforeFirst("</td>"))
{
outputFile.Write(data.GetResult()[0] + " ");
}
outputFile.WriteLine();
}
}
}
outputFile.Close();
}
static void test1()
{
Reaper reaper = new Reaper((new StreamReader("03,073.html")).ReadToEnd());
StreamWriter outputFile = new StreamWriter("test1_output.txt");
List <string> partResult =
reaper
.RemainBeforeFirst("<table width=\"100%\" summary=\"table used for formatting\"><tr><td>")
.ReapByProfix("<hr><big><b><i>")
.GetResult();
foreach (var part in partResult)
{
string partName = (new Reaper(part)).RemainBeforeFirst(":</i></b></big>").GetResult()[0];
ShowStrings((new Reaper(part)).RemainBeforeFirst(":</i></b></big>").GetResult());
outputFile.WriteLine("----------PART:" + partName);
List <string> tableResult =
(new Reaper(part))
.ReapByProfix("<div align=\"center\"><table border=1 summary=\"")
.GetResult();
foreach (var table in tableResult)
{
string tableName = (new Reaper(table)).RemainBeforeFirst("\" width=\"95%\">").GetResult()[0];
//ShowStrings((new Reaper(table)).RemainBeforeFirst("\" width=\"95%\">").GetResult());
outputFile.WriteLine("-----TABLE:" + tableName);
List <string> lineResult =
(new Reaper(table))
.ReapByProfix("<tr><td>")
.GiveUpContain("<td>Jan</td><td>Feb</td><td>Mar</td>")
.GetResult();
int lineCount = 0;
foreach (var line in lineResult)
{
string lineName = (new Reaper(line)).RemainBeforeFirst("</td>").GetResult()[0];
//ShowStrings((new Reaper(line)).RemainBeforeFirst("</td>").GetResult());
outputFile.WriteLine("LINE:" + lineName + " " + ++lineCount);
List <string> dataResult =
(new Reaper(line))
.ReapByProfix("<td align=\"center\" nowrap>")
.RemainBeforeFirst("</td>")
.GetResult();
foreach (var data in dataResult)
{
outputFile.Write(data + " ");
}
outputFile.WriteLine();
}
}
}
outputFile.Close();
}
public bool 提取()
{
StringWriter strWriter = new StringWriter();
//TimeSpan repearTime = new TimeSpan();
string originString = (new StreamReader(_inFileName)).ReadToEnd();
//处理重大漏洞:HTML代码中,部分行名是包含在</tr>\n<tr><td align=\"center\">中,而大多数是包含在</tr>\n<tr><td>中
originString = originString.Replace("</tr>\n<tr><td align=\"center\">", "</tr>\n<tr><td>");
//DateTime startTime = DateTime.Now;
Reaper reaper = new Reaper(originString);
string latStr = reaper.RemainAfterFirst("<br>Latitude <b>").RemainBeforeFirst("</b>").GetResult()[0];
string lonStr = reaper.RemainAfterFirst("<br>Longitude <b>").RemainBeforeFirst("</b>").GetResult()[0];
//repearTime += DateTime.Now - startTime;
strWriter.Write("LATLON:");
strWriter.WriteLine(latStr + " " + lonStr);
//startTime = DateTime.Now;
foreach (Reaper part in reaper.RemainBeforeFirst("<table width=\"100%\" summary=\"table used for formatting\"><tr><td>").ReapByProfix("<hr><big><b><i>"))
{
string partName = part.RemainBeforeFirst(":</i></b></big>").GetResult()[0];
//ShowStrings(part.RemainBeforeFirst(":</i></b></big>").GetResult());
foreach (Reaper table in part.ReapByProfix("<div align=\"center\"><table border=1 summary=\""))
{
string tableName = table.RemainBeforeFirst("\" width=\"95%\">").GetResult()[0];
int lineCount = 0;
foreach (Reaper line in table.ReapByProfix("<tr><td>").GiveUpContain("<td>Jan</td><td>Feb</td><td>Mar</td>"))
{
string lineName = line.RemainBeforeFirst("</td>").GetResult()[0];
//repearTime += DateTime.Now - startTime;
strWriter.WriteLine("PART:" + partName);
strWriter.WriteLine("TABLE:" + tableName);
strWriter.WriteLine("LINE:" + lineName);
strWriter.WriteLine("NUM:" + ++lineCount);
strWriter.Write("DATA:");
//startTime = DateTime.Now;
foreach (Reaper data in line.ReapByProfix("<td align=\"center\" nowrap>").RemainBeforeFirst("</td>"))
{
//repearTime += DateTime.Now - startTime;
strWriter.Write(data.GetResult()[0] + " ");
//startTime = DateTime.Now;
}
strWriter.WriteLine();
}
}
}
//处理HTML的转义符
string result = strWriter.ToString();
result = result.Replace("<", "<");
result = result.Replace(">", ">");
result = result.Replace("°", "°");
//处理Average Daily Temperature Range 多出的*
result = result.Replace("* ", "");
try
{
StreamWriter _outFile = new StreamWriter(_outFileName);
_outFile.Write(result);
_outFile.Close();
}
catch (Exception ex)
{
Program.errLog.WriteLine(ex.Message);
Program.errLog.Flush();
}
//Console.WriteLine("完成:" + _inFileName + " -> " + _outFileName);
//Console.WriteLine("Reaper\n耗时" + repearTime);
return(true);
}
public bool 提取()
{
StringWriter strWriter = new StringWriter();
//TimeSpan repearTime = new TimeSpan();
string originString = (new StreamReader(_inFileName)).ReadToEnd();
//处理重大漏洞:HTML代码中,部分行名是包含在</tr>\n<tr><td align=\"center\">中,而大多数是包含在</tr>\n<tr><td>中
originString = originString.Replace("</tr>\n<tr><td align=\"center\">", "</tr>\n<tr><td>");
//DateTime startTime = DateTime.Now;
Reaper reaper = new Reaper(originString);
string latStr = reaper.RemainAfterFirst("<br>Latitude <b>").RemainBeforeFirst("</b>").GetResult()[0];
string lonStr = reaper.RemainAfterFirst("<br>Longitude <b>").RemainBeforeFirst("</b>").GetResult()[0];
//repearTime += DateTime.Now - startTime;
strWriter.Write("LATLON:");
strWriter.WriteLine(latStr + " " + lonStr);
//startTime = DateTime.Now;
foreach (Reaper part in reaper.RemainBeforeFirst("<table width=\"100%\" summary=\"table used for formatting\"><tr><td>").ReapByProfix("<hr><big><b><i>"))
{
string partName = part.RemainBeforeFirst(":</i></b></big>").GetResult()[0];
//ShowStrings(part.RemainBeforeFirst(":</i></b></big>").GetResult());
foreach (Reaper table in part.ReapByProfix("<div align=\"center\"><table border=1 summary=\""))
{
string tableName = table.RemainBeforeFirst("\" width=\"95%\">").GetResult()[0];
int lineCount = 0;
foreach (Reaper line in table.ReapByProfix("<tr><td>").GiveUpContain("<td>Jan</td><td>Feb</td><td>Mar</td>"))
{
string lineName = line.RemainBeforeFirst("</td>").GetResult()[0];
//repearTime += DateTime.Now - startTime;
strWriter.WriteLine("PART:" + partName);
strWriter.WriteLine("TABLE:" + tableName);
strWriter.WriteLine("LINE:" + lineName);
strWriter.WriteLine("NUM:" + ++lineCount);
strWriter.Write("DATA:");
//startTime = DateTime.Now;
foreach (Reaper data in line.ReapByProfix("<td align=\"center\" nowrap>").RemainBeforeFirst("</td>"))
{
//repearTime += DateTime.Now - startTime;
strWriter.Write(data.GetResult()[0] + " ");
//startTime = DateTime.Now;
}
strWriter.WriteLine();
}
}
}
//处理HTML的转义符
string result = strWriter.ToString();
result = result.Replace("<", "<");
result = result.Replace(">", ">");
result = result.Replace("°", "°");
//处理Average Daily Temperature Range 多出的*
result = result.Replace("* ", "");
try
{
StreamWriter _outFile = new StreamWriter(_outFileName);
_outFile.Write(result);
_outFile.Close();
}
catch(Exception ex)
{
Program.errLog.WriteLine(ex.Message);
Program.errLog.Flush();
}
//Console.WriteLine("完成:" + _inFileName + " -> " + _outFileName);
//Console.WriteLine("Reaper\n耗时" + repearTime);
return true;
}