public override object TransformData(IFreeDocument datas)
{
var text = datas[Column];
if (text == null)
return null;
var pattern = "all";
var format = "plain";
switch (ResultType)
{
case ContentType.Json:
format = "json";
break;
case ContentType.Text:
format = "plain";
break;
case ContentType.XML:
format = "xml";
break;
case ContentType.Byte:
format = "conll";
break;
}
switch (Pattern)
{
case Pattern.分词:
pattern = "ws";
break;
case Pattern.词性标注:
pattern = "pos";
break;
case Pattern.依存句法分析:
pattern = "dp";
break;
case Pattern.语义依存分析:
pattern = "sdp";
break;
case Pattern.语义角色标注:
pattern = "srl";
break;
case Pattern.命名实体识别:
pattern = "ner";
break;
}
var param = ("api_key=" + apiKey +
"&pattern=" + pattern +
"&format=" + format+
"&text=" + HttpUtility.UrlEncode(text.ToString())
);
var docs = buffHelper.Get(param);
if (docs == null)
{
var item = new HttpItem();
item.URL = uriBase+'?'+ param;
item.Method = MethodType.GET;
item.Encoding = EncodingType.UTF8;
var helper = new HttpHelper();
HttpStatusCode code;
var result = helper.GetHtml(item,out code);
if(code==HttpStatusCode.OK)
buffHelper.Set(param, result);
return result;
}
return docs;
}
public static HtmlDocument GetHtmlDocument(string url)
{
var httpitem = new HttpItem {URL = url};
var helper = new HttpHelper();
HttpStatusCode statusCode;
var doc2 = helper.GetHtml(httpitem, out statusCode);
if (statusCode != HttpStatusCode.OK)
return null;
var htmldoc = new HtmlDocument();
htmldoc.LoadHtml(doc2);
return htmldoc;
}
public static IEnumerable<List<FreeDocument>> GetMultiDataFromURL(string url)
{
var httpitem = new HttpItem {URL = url};
var helper = new HttpHelper();
HttpStatusCode statusCode;
var doc2 = helper.GetHtml(httpitem, out statusCode);
if (statusCode != HttpStatusCode.OK)
yield break;
if (doc2 == null)
yield return new List<FreeDocument>();
var htmldoc = new HtmlDocument();
htmldoc.LoadHtml(doc2);
foreach (var item in htmldoc.GetDataFromHtml())
{
yield return item;
}
}
public static HtmlDocument GetDocumentFromURL(string url, EncodingType encoding = EncodingType.Unknown)
{
var httpitem = new HttpItem();
httpitem.URL = url;
httpitem.Encoding = encoding;
var helper = new HttpHelper();
HttpStatusCode code;
var doc = new HtmlDocument();
var result = helper.GetHtml(httpitem, out code);
if (!HttpHelper.IsSuccess(code))
return doc;
doc.LoadHtml(result);
return doc;
}