public ActionResult ProcessText([FromBody] string text)
{
// Arabizi to arabic script
// either direct call to perl script
var textConverter = new TextConverter();
// or via a php api
// var textConverter = new ApiTextConverter();
// SA & Entity
var textSentimentAnalyzer = new TextSentimentAnalyzer();
var textEntityExtraction = new TextEntityExtraction();
// Arabizi to arabic from perl script
var arabicText = textConverter.Convert(text);
// Sentiment analysis from watson https://gateway.watsonplatform.net/";
var sentiment = textSentimentAnalyzer.GetSentiment(arabicText);
// Entity extraction from rosette (https://api.rosette.com/rest/v1/)
var entities = textEntityExtraction.GetEntities(arabicText);
var result = new TextAnalyze
{
ArabicText = arabicText,
Entities = entities.ToList(),
Sentiment = sentiment
};
return(Json(result, JsonRequestBehavior.AllowGet));
}
void GetImageUrlsComplete(string sourceHtml, string baseUrl)
{
string regexString = RegexStringLib.GetFileUrlRegexString("jpg|jpeg|png|ico|bmp|gif");
List <string> imageUrls = TextAnalyze.GetUrlList(sourceHtml, regexString, baseUrl);
if (imageUrls.Count > 0)
{
this.downloadUrlList = new List <string>();
foreach (string url in imageUrls)
{
string fullUrl = Utils.CompleteRelativeUrl(baseUrl, url);
this.downloadUrlList.Add(fullUrl);
tbxMessage.Text += string.Format("{1}{0}", Environment.NewLine, fullUrl);
}
MessageBox.Show(string.Format("{0} images got!", imageUrls.Count));
}
else
{
tbxMessage.Text = "no regexed";
}
}
private void OnLoadTextCallback(string path, byte[] bytes)
{
if (!mLoadHandles.ContainsKey(path))
{
//出现严重错误
return;
}
string text = "";
LoadHandler handle = mLoadHandles[path] as LoadHandler;
if (handle.isXML)
{
text = System.Text.Encoding.UTF8.GetString(bytes);
}
else
{
text = System.Text.Encoding.Unicode.GetString(bytes);
}
if (string.IsNullOrEmpty(text))
{
return;
}
//Hashtable tb = new Hashtable(new CustomEqualityComparer());
DataTable tb = new DataTable();
if (!handle.isXML)
{
WDBData dbData = null;
try
{
dbData = TextAnalyze.Analyze(text);
}
catch (Exception analyzeException)
{
throw new Exception("加载" + path + "失败: " + analyzeException.Message);
}
for (int i = 0; i < dbData.GetRecordCount(); ++i)
{
WDBSheetLine line = dbData.GetDataByNumber(i);
if (line != null)
{
object item = System.Activator.CreateInstance(handle.type);
try
{
tb.Add(loadLine(line, item, handle.keyIdx), item);
}
catch (Exception exp)
{
throw new Exception("解析" + path + "失败: " + exp.Message);
}
}
}
}
else
{
XmlDocument xDoc = new XmlDocument();
xDoc.LoadXml(text);
XmlNode node = xDoc.FirstChild;
node = node.NextSibling;
XmlNodeList nodeList = node.ChildNodes;
for (int i = 0; i < nodeList.Count; ++i)
{
XmlNode childNode = nodeList[i];
if (!string.IsNullOrEmpty(childNode.Name) &&
childNode.LocalName == "#comment")
{
continue;
}
object item = System.Activator.CreateInstance(handle.type);
Type itemType = item.GetType();
System.Reflection.FieldInfo[] fields = itemType.GetFields();
foreach (System.Reflection.FieldInfo f in fields)
{
XmlNode nd = childNode.Attributes.GetNamedItem(f.Name);
if (nd != null)
{
if (f.FieldType.Name == "Int32")
{
f.SetValue(item, System.Convert.ToInt32(nd.Value));
}
if (f.FieldType.Name == "Single")
{
f.SetValue(item, System.Convert.ToSingle(nd.Value));
}
if (f.FieldType.Name == "String")
{
f.SetValue(item, nd.Value);
}
}
}
tb.Add(fields[0].GetValue(item), item);
}
}
mLoadHandles.Remove(path);
mTableList.Add((int)handle.dataType, tb);
if (mLoadHandles.Count <= 0)
{
OnAllLoad();
}
}
