private void ParseUniverPage(string pageRefer)
{
currentUniversity = new UniversityData();
string text;
HttpWebRequest req = (HttpWebRequest)HttpWebRequest.Create(pageRefer);
HttpWebResponse resp = (HttpWebResponse)req.GetResponse();
using (StreamReader stream = new StreamReader(resp.GetResponseStream(), Encoding.UTF8))
{
text = stream.ReadToEnd();
}
Regex regionRegex = new Regex(regionPattern);
Match regionMatch = regionRegex.Match(text);
string regionName = regionMatch.Groups[1].Value.ToString();
Regex reg = new Regex(bachelorPattern);
foreach (Match match in reg.Matches(text))
{
string tmp = match.Groups[0].Value.ToString();
Regex innerReg = new Regex(linkPattern);
Match found = innerReg.Match(tmp);
ParseSpecialityDetails(mainRefer + found.Groups[1].Value.ToString(), regionName);
}
}
public void StartParse()
{
if (FileExist())
{
ReadFromBinary();
}
for (var id = universities.Count == 0 ? startPageId : universities.Count + 1; id <= endPageId; id++)
{
try
{
ParseUniverPage(refer + id);
if (currentUniversity.UniversityName != null && IsUniversity)
{
universities.Add(currentUniversity);
currentUniversity = null;
}
if (id % 5 == 0)
{
Console.WriteLine("id%10==0");
SaveToBinary();
}
Console.WriteLine("Ссылка: " + refer + id);
if (!IsUniversity)
{
IsUniversity = true;
}
}
catch (Exception ex)
{
}
}
//SaveToXml();
//SaveToBinary();
}
public ParseManager()
{
IsUniversity = true;
universities = new List <UniversityData>();
currentUniversity = null;
}
public ParseManager()
{
IsUniversity = true;
universities = new List<UniversityData>();
currentUniversity = null;
}
private void ParseUniverPage(string pageRefer)
{
currentUniversity = new UniversityData();
string text;
HttpWebRequest req = (HttpWebRequest)HttpWebRequest.Create(pageRefer);
HttpWebResponse resp = (HttpWebResponse)req.GetResponse();
using (StreamReader stream = new StreamReader(resp.GetResponseStream(), Encoding.UTF8))
{
text = stream.ReadToEnd();
}
Regex regionRegex = new Regex(regionPattern);
Match regionMatch = regionRegex.Match(text);
string regionName = regionMatch.Groups[1].Value.ToString();
Regex reg = new Regex(bachelorPattern);
foreach (Match match in reg.Matches(text))
{
string tmp = match.Groups[0].Value.ToString();
Regex innerReg = new Regex(linkPattern);
Match found = innerReg.Match(tmp);
ParseSpecialityDetails(mainRefer + found.Groups[1].Value.ToString(), regionName);
}
}
public void StartParse()
{
if (FileExist())
{
ReadFromBinary();
}
for (var id = universities.Count == 0 ? startPageId : universities.Count + 1; id <= endPageId; id++)
{
try
{
ParseUniverPage(refer + id);
if (currentUniversity.UniversityName != null && IsUniversity)
{
universities.Add(currentUniversity);
currentUniversity = null;
}
if (id % 5 == 0)
{
Console.WriteLine("id%10==0");
SaveToBinary();
}
Console.WriteLine("Ссылка: " + refer + id);
if (!IsUniversity)
IsUniversity = true;
}
catch (Exception ex)
{
}
}
//SaveToXml();
//SaveToBinary();
}