public SearchEngineResult Parse(string html, Encoding encoding)
{
HTMLparser oP = HtmlParserFactory.GetInstance();
searchResult = new SearchEngineResult();
searchResult.SearchEngineType = SearchEngineType.Google;
item = new SearchEngineResult.ResultItem();
//item.Source = "Google";
oP.Init(encoding.GetBytes(html));
oP.SetEncoding(encoding);
HTMLchunk oChunk = null;
int state = 0;
bool bEncodingSet = false;
while ((oChunk = oP.ParseNext()) != null)
{
switch (oChunk.oType)
{
case HTMLchunkType.OpenTag:
HandleOpenTag(oChunk, ref state);
printParams:
if (oChunk.sTag == "meta")
{
HandleMetaEncoding(oP, oChunk, ref bEncodingSet);
};
HandleParam(oChunk, ref state);
break;
case HTMLchunkType.CloseTag:
HandleCloseTag(oChunk, ref state);
break;
case HTMLchunkType.Text:
HandleText(oChunk, ref state);
break;
default:
break;
}
}
return searchResult;
}
private static SearchEngineResult GetSearchResult()
{
SearchEngineResult result = new SearchEngineResult();
result.SearchEngineType = SearchEngineType.Other;
SearchEngineResult.ResultItem item1 = new SearchEngineResult.ResultItem();
item1.Title = "INFO";
item1.CacheUrl = "http://aaa.www.com/";
item1.Description = "LinTian" + " " + 1 + " Done.";
item1.SimilarUrl = "http://similar.www.com/";
item1.Url = "http://info.tsinghua.edu.cn";
result.Results.Add(item1);
SearchEngineResult.ResultItem item2 = new SearchEngineResult.ResultItem();
item2.Title = "INFO2";
item2.CacheUrl = "http://aaa.www.com22222/";
item2.Description = "LunaR" + " " + 2 + " Done2.";
item2.SimilarUrl = "http://similar.www.com2222/";
item2.Url = "http://info.tsinghua.edu.cn2";
result.Results.Add(item2);
return result;
}
private void HandleCloseTag(HTMLchunk oChunk, ref int state)
{
if (oChunk.sTag == "a" && state == 6)
{
state += 1;
}
else if (oChunk.sTag == "li" && state == 7)
{
state = 4;
if (item.Url != null && item.Url != "")
{
searchResult.Results.Add(item);
item = new SearchEngineResult.ResultItem();
}
}
else if (oChunk.sTag == "ul" && state == 4)
{
state = -1;
}
}
private void HandleParam(HTMLchunk oChunk, ref int state)
{
if (oChunk.iParams > 0)
{
for (int i = 0; i < oChunk.iParams; i++)
{
switch (oChunk.cParamChars[i])
{
default:
if (oChunk.sValues[i] == "g" && oChunk.sParams[i] == "class" && state == 2)
{
state = 3;
if (item.Url!=null && item.Url!="")
{
searchResult.Results.Add(item);
item = new SearchEngineResult.ResultItem();
//item.Source = "Google";
}
}else if(oChunk.sValues[i] == "r" && oChunk.sParams[i] == "class" && state == 3)
{
state = 4;
}else if(oChunk.sValues[i] == "s" && oChunk.sParams[i] == "class" && state == 6)
{
state = 7;
}
else if (oChunk.sValues[i] == "gl" && oChunk.sParams[i] == "class" && state == 7)
{
state = 8;
}
else if (oChunk.sParams[i] == "href")
{
if (state == 5)
{
item.Url = oChunk.sValues[i];
}
else if (state == 9 || state == 11)
{
if (oChunk.sValues[i].IndexOf("q=related")!=-1)
{
item.SimilarUrl = oChunk.sValues[i];
}
else if (oChunk.sValues[i].IndexOf("q=cache") != -1)
{
item.CacheUrl = oChunk.sValues[i];
}
}
}
break;
}
}
}
}
private void HandleParam(HTMLchunk oChunk, ref int state)
{
if (oChunk.iParams > 0)
{
for (int i = 0; i < oChunk.iParams; i++)
{
switch (oChunk.cParamChars[i])
{
default:
if (oChunk.sValues[i] == "f" && oChunk.sParams[i] == "class" && state == 2)
{
state = 3;
if (item.Url != null && item.Url != "")
{
searchResult.Results.Add(item);
item = new SearchEngineResult.ResultItem();
//item.Source = "Baidu";
}
}
else if (oChunk.sParams[i] == "href")
{
if (state == 4)
{
item.Url = oChunk.sValues[i];
}
else if (state == 6 || state == 8)
{
if (oChunk.sValues[i].IndexOf("cache") != -1)
{
item.CacheUrl = oChunk.sValues[i];
}
else if (oChunk.sValues[i].StartsWith("s?cl=2"))
{
item.SimilarUrl = oChunk.sValues[i];
}
}
}
break;
}
}
}
}
public static void Test()
{
QueryResultRecordManager manager = new QueryResultRecordManager("data", new TimeSpan(0, 1, 0));
Pipeline.Pipeline pipeline = new Pipeline.Pipeline();
pipeline.QueryResultSubscriberManager.AddSubscriber(manager);
QueryResult result = new QueryResult(new InputQuery("Bill Gates"));
SearchEngineResult resultItem = new SearchEngineResult();
resultItem.SearchEngineType = SearchEngineType.Google;
resultItem.SearchUrl = "http://www.google.com/query.jsp";
SearchEngineResult.ResultItem item = new SearchEngineResult.ResultItem();
item.Title = "ddd";
item.Url = "http://www.gfw.com/";
item.SimilarUrl = "http://www.g.com/ddd";
item.CacheUrl = "http://www.g.com/cache";
item.Description = "Who cares?";
resultItem.Results.Add(item);
SuggestionResult resultItem1 = new SuggestionResult();
resultItem1.SuggestionType = SuggestionType.Google;
resultItem1.SearchUrl = "json";
SuggestionResult.ResultItem item1 = new SuggestionResult.ResultItem();
item1.Index = "1";
item1.Number = "2";
item1.Suggestion = "haha";
resultItem1.Results.Add( item1 );
result.SearchEngineResultItems.Add(resultItem);
result.SuggestionResultItems.Add(resultItem1);
pipeline.OnQueryResultReady(result);
Thread.Sleep(1000);
List<QueryResult> resultList = manager.GetResultList(DateTime.Today, DateTime.Today);
foreach (QueryResult queryResult in resultList)
{
Console.WriteLine(queryResult);
}
Console.WriteLine("----------------------------------------------");
manager.RemoveAllResultList();
resultList = manager.GetResultList(DateTime.Today, DateTime.Today);
foreach (QueryResult queryResult in resultList)
{
Console.WriteLine(queryResult);
}
Console.WriteLine("----------------------------------------------");
result = new QueryResult(new InputQuery("Bill Gates"));
resultItem = new SearchEngineResult();
resultItem.SearchEngineType = SearchEngineType.Google;
resultItem.SearchUrl = "http://www.google.com/query.jsp";
item = new SearchEngineResult.ResultItem();
item.Title = "ddd";
item.Url = "http://www.gfw.com/";
item.SimilarUrl = "http://www.g.com/ddd";
item.CacheUrl = "http://www.g.com/cache";
item.Description = "Who cares?";
resultItem.Results.Add(item);
result.Items.Add(resultItem);
pipeline.OnQueryResultReady(result);
Thread.Sleep(1000);
manager.RemoveResultListFromDate(DateTime.Today);
resultList = manager.GetResultList(DateTime.Today - new TimeSpan(1, 0, 0, 0), DateTime.Today);
foreach (QueryResult queryResult in resultList)
{
Console.WriteLine(queryResult);
}
Console.WriteLine("----------------------------------------------");
manager.RemoveResultListFromDate(DateTime.Today + new TimeSpan(1, 0, 0, 0));
resultList = manager.GetResultList(DateTime.Today - new TimeSpan(1, 0, 0, 0), DateTime.Today);
foreach (QueryResult queryResult in resultList)
{
Console.WriteLine(queryResult);
}
Console.WriteLine("----------------------------------------------");
}
public IQueryResultItem Search( InputQuery query )
{
SearchEngineResult result = new SearchEngineResult();
result.SearchEngineType = SearchEngineType.Other;
SearchEngineResult.ResultItem item1 = new SearchEngineResult.ResultItem();
item1.Title = "INFO";
item1.CacheUrl = "http://aaa.www.com/";
item1.Description = query + " " + _count + " Done.";
item1.SimilarUrl = "http://similar.www.com/";
item1.Url = "http://info.tsinghua.edu.cn";
result.Results.Add(item1);
_count++;
Thread.Sleep(_waitTime);
return result;
}
private void HandleParam(HTMLchunk oChunk, ref int state)
{
if (oChunk.iParams > 0)
{
for (int i = 0; i < oChunk.iParams; i++)
{
switch (oChunk.cParamChars[i])
{
default:
if (oChunk.sValues[i] == "f" && oChunk.sParams[i] == "class" && state == 2)
{
state = 3;
if (item.Url != null && item.Url != "")
{
searchResult.Results.Add(item);
item = new SearchEngineResult.ResultItem();
//item.Source = "Sogou";
}
}
else if (oChunk.sParams[i] == "href")
{
if (state == 4)
{
item.Url = oChunk.sValues[i];
}
else if (state == 7 )
{
item.CacheUrl = oChunk.sValues[i];
}else if(state==10)
{
item.SimilarUrl = oChunk.sValues[i];
}
}
else if (oChunk.sParams[i] == "id" && (state == 6 || state == 9))
{
if (oChunk.sValues[i].StartsWith("sogou_snapshot"))
{
state = 7;
}
else if (oChunk.sValues[i].StartsWith("sogou_sis"))
{
state = 10;
}
}
break;
}
}
}
}