public static StringsTree ParseTreeStructure(string startTag, string endTag, string data)
{
if (startTag == null)
{
throw new ArgumentNullException("startTag");
}
if (endTag == null)
{
throw new ArgumentNullException("endTag");
}
if (data == null)
{
throw new ArgumentNullException("data");
}
int startTagLength = startTag.Length;
int endTagLength = endTag.Length;
List <TreeParseData> parseList = new List <TreeParseData>();
int idx = 0;
int ident = 0;
while (idx < data.Length)
{
int resultIdxStart = data.IndexOf(startTag, idx);
if (resultIdxStart < 0)
{
resultIdxStart = int.MaxValue;
}
int resultIdxEnd = data.IndexOf(endTag, idx);
if (resultIdxEnd < 0)
{
resultIdxEnd = int.MaxValue;
}
if (resultIdxStart < resultIdxEnd)
{
ident++;
parseList.Add(new TreeParseData(true, resultIdxStart));
idx = resultIdxStart + startTagLength;
}
else if (resultIdxStart > resultIdxEnd)
{
ident--;
if (ident < 0)
{
throw new ArgumentNullException("Wrong data structure", "data");
}
parseList.Add(new TreeParseData(false, resultIdxEnd));
idx = resultIdxEnd + endTagLength;
}
else
{ //equal, end found
break;
}
}
StringsTree current = new StringsTree(null, 0);
for (idx = 0; idx < parseList.Count; idx++)
{
if (parseList[idx].IsStartTag)
{
StringsTree child = new StringsTree(current, idx);
current.Childs.Add(child);
current = child;
}
else
{ //this is child
int childStart = parseList[current.Offset].Position;
int childEnd = parseList[idx].Position;
if (parseList[current.Offset + 1].IsStartTag)
{
childEnd = parseList[current.Offset + 1].Position;
}
current.Data = data.Substring(childStart + startTagLength, childEnd - childStart - startTagLength);
current = current.Parent;
}
}
return(current);
}
public StringsTree(StringsTree parent, int offset)
{
this.parent = parent;
this.offset = offset;
}
protected override void DoTranslate(string phrase, LanguagePair languagesPair, string subject, Result result, NetworkSetting networkSetting)
{
string query = "http://www.slovnenya.com/dictionary/{0}";
query = string.Format(query, HttpUtility.UrlEncode(phrase).Replace("+", "%20"));
WebRequestHelper helper =
new WebRequestHelper(result, new Uri(query),
networkSetting,
WebRequestContentType.UrlEncodedGet);
helper.AcceptLanguage = "en-us,en";
string responseFromServer = helper.GetResponse();
if (responseFromServer.Contains("did not return any results</div>") ||
responseFromServer.Contains("</span>` did not return any results"))
{
result.ResultNotFound = true;
throw new TranslationException("Nothing found");
}
else if (responseFromServer.IndexOf("Query contains extraneous symbol(s)<") >= 0)
{
throw new TranslationException("Query contains extraneous symbols");
}
else
{
result.ArticleUrl = query;
result.ArticleUrlCaption = phrase;
string translation = StringParser.Parse("<hr style=\"border:0;background-color:grey;height:1px;width:92%;text-align:center\" />", "<hr style=\"border:0;background-color:grey;height:1px;width:92%;text-align:center\" />", responseFromServer);
StringsTree tree = StringParser.ParseTreeStructure("<table", "</table>", translation);
if (tree.Childs.Count != 1)
{
throw new TranslationException("Wrong data structure");
}
tree = tree.Childs[0];
if (tree.Childs.Count != 1)
{
throw new TranslationException("Wrong data structure");
}
tree = tree.Childs[0];
Result wordres = result;
if (tree.Childs.Count == 0)
{
throw new TranslationException("Wrong data structure");
}
//get word
if (tree.Childs[0].Childs.Count != 1)
{
throw new TranslationException("Wrong data structure");
}
string word = StringParser.Parse("font-size:14pt\">", "<", tree.Childs[0].Childs[0].Data);
for (int i = 1; i < tree.Childs.Count; i++)
{
StringsTree abbr_tree = tree.Childs[i];
Result abbrres = null;
string abbr = StringParser.Parse("font-size:12pt\">", "<", abbr_tree.Data);
Result tmpRes = CreateNewResult(abbr, languagesPair, subject);
wordres.Childs.Add(tmpRes);
abbrres = tmpRes;
StringParser parser = new StringParser(abbr_tree.Childs[0].Data);
string[] translations = parser.ReadItemsList("font-size:12pt\">", "<");
foreach (string trans in translations)
{
abbrres.Translations.Add(trans);
}
}
}
}
