public static SummarizedDocument Summarize(SummarizerArguments args)
{
if (args == null)
{
return(null);
}
Article article;
if (args.InputString.Length > 0 && args.InputFile.Length == 0)
{
article = ParseDocument(args.InputString, args);
}
else
{
article = ParseFile(args.InputFile, args);
}
Grader.Grade(article);
Highlighter.Highlight(article, args);
SummarizedDocument summarizedDocument = CreateSummarizedDocument(article);
return(summarizedDocument);
}
private static Article ParseDocument(string text, SummarizerArguments args)
{
Dictionary rules = Dictionary.LoadFromFile(args.DictionaryLanguage);
Article article = new Article(rules);
article.ParseText(text);
return(article);
}
private static SummarizedDocument CreateSummarizedDocument(Article article, SummarizerArguments args)
{
SummarizedDocument sumDoc = new SummarizedDocument();
sumDoc.Concepts = article.Concepts;
foreach (Sentence sentence in article.Sentences)
{
if (sentence.Selected)
{
sumDoc.Sentences.Add(sentence.OriginalSentence);
}
}
return(sumDoc);
}
public static String Summarize(String text)
{
var summaryArgs = new SummarizerArguments();
summaryArgs.DictionaryLanguage = "en";
//summaryArgs.DisplayLines = 3;
summaryArgs.DisplayPercent = 50;
summaryArgs.InputString = text;
SummarizedDocument doc = OpenTextSummarizer.Summarizer.Summarize(summaryArgs);
var output = doc.Sentences;
var summary = string.Concat(output.ToArray());
return summary;
}
internal static void Highlight(Article article, SummarizerArguments args)
{
if (args.DisplayPercent == 0 && args.DisplayLines == 0)
{
return;
}
if (args.DisplayPercent == 0)
{
//get the highest scored n lines, without reordering the list.
SelectNumberOfSentences(article, args.DisplayLines);
}
else
{
SelectSentencesByPercent(article, args.DisplayPercent);
}
}
/// <summary>
/// encapsulates the algorithm for summarizing the help files
/// </summary>
/// <param name="toSummarize">responses from data access tier that will be summarized</param>
public string[] summarize(string[] toSummarize)//List<QueryResponseList> toSummarize)
{
List <string> sumList = new List <string>();
/*foreach (string input in getInputStringList(toSummarize))
* {
* OpenTextSummarizer.SummarizerArguments args = new OpenTextSummarizer.SummarizerArguments();
* args.InputString = input;
* OpenTextSummarizer.SummarizedDocument sd = OpenTextSummarizer.Summarizer.Summarize(args);
* sumList.Add(sd.Sentences.Aggregate((x,y)=> { return x + y; }));
* }
* return sumList.ToArray();*/
OpenTextSummarizer.SummarizerArguments args = new OpenTextSummarizer.SummarizerArguments();
args.InputString = String.Join(" ", toSummarize);
OpenTextSummarizer.SummarizedDocument sd = OpenTextSummarizer.Summarizer.Summarize(args);
return(sd.Sentences.ToArray());
}
private DataFeed EnrichWithConcepts(DataFeed feed)
{
try
{
var args = new SummarizerArguments
{
DictionaryLanguage = "en",
DisplayLines = 1,
InputString = feed.Title
};
var summary = Summarizer.Summarize(args);
feed.Concepts = summary.Concepts.ToArray();
}
catch (Exception)
{
}
return feed;
}
internal static void Highlight(Article article, SummarizerArguments args)
{
switch (args.DisplayPercent)
{
case 0 when args.DisplayLines == 0:
return;
case 0:
//get the highest scored n lines, without reordering the list.
SelectNumberOfSentences(article, args.DisplayLines);
break;
default:
SelectSentencesByPercent(article, args.DisplayPercent);
break;
}
}
private static Article ParseFile(string fileName, SummarizerArguments args)
{
string text = LoadFile(fileName);
return(ParseDocument(text, args));
}
static void summarizeData(string query)
{
outputLabelText(" SUMMARIZING:", query);
string finalSummary = "";
string outputFileSummaryFinal = System.IO.Path.GetFullPath(query.Replace(":", "-")) + "\\index.html";
string outputFileSummaryRaw = System.IO.Path.GetFullPath(query.Replace(":", "-")) + "\\__summaryRaw.txt";
System.IO.StreamReader file = new System.IO.StreamReader(outputFileSummaryRaw);
string row = "";
//string summary = "";
while ((row = file.ReadLine()) != null)
{
if (row != cr.ToString() || row != cr.ToString() + lf.ToString()) {
//summary += row + System.Environment.NewLine;
string[] split = row.Split(delimiter.ToCharArray());
string g = split[0];
string url = split[1];
string title = split[2];
string data = split[3];
var uri = new Uri(url);
var host = uri.Host;
SummarizerArguments sumargs = new SummarizerArguments();
sumargs.DictionaryLanguage = "en";
sumargs.DisplayLines = 10;
sumargs.DisplayPercent = 0;
sumargs.InputString = data;
SummarizedDocument doc = Summarizer.Summarize(sumargs);
var concepts = doc.Concepts.Select(p => p + ", ");
finalSummary += "<h1>" + title + "<span class='datasource'>versions: <a href='pages\\" + g + ".html' target='_blank'>local</a> | <a href='" + url + "' target='_blank'>web</a></span></h1>" + System.Environment.NewLine;
finalSummary += "<blockquote>";
finalSummary += "<p><strong>Site:</strong> " + host + "</p>";
var conceptSumary = doc.Concepts
.Where(p => p.ToString() != "" && p.Length >= 3)
.Select(p => "<li>'" + p + "'</li>");
finalSummary += "<h2>Concepts</h2><ul>" + string.Concat(conceptSumary.ToArray()) + "</ul>";
finalSummary += System.Environment.NewLine + System.Environment.NewLine;
var sentenceSummary = doc.Sentences
.Select(p => "<li>" + p.Trim() + "</li>");
string oneMore = string.Concat(sentenceSummary.ToArray());
foreach (string c in doc.Concepts) {
if (c.Length > 3) {
oneMore = oneMore.Replace(c, "<span class='hlPink'>" + c + "</span>");
}
}
foreach (string c in query.Split(' '))
{
if ((c.Length > 3) && (!c.StartsWith("-")))
{
oneMore = oneMore.Replace(c, "<span class='hlYellow'>" + c + "</span>");
}
}
finalSummary += "<h2>Summary</h2><ul>" + oneMore + "</ul>";
finalSummary += "</blockquote>";
finalSummary += System.Environment.NewLine + System.Environment.NewLine + System.Environment.NewLine + "<hr>";
}
}
#region
//SummarizerArguments sumargs = new SummarizerArguments();
//sumargs.DictionaryLanguage = "en";
//sumargs.DisplayLines = 10;
//sumargs.DisplayPercent = 0;
//sumargs.InputString = summary;
//SummarizedDocument doc = Summarizer.Summarize(sumargs);
//var concepts = doc.Concepts.Select(p => p + ", ");
//finalSummary += "Concepts - [" + string.Concat(concepts.ToArray()) + "]";
//finalSummary += System.Environment.NewLine + System.Environment.NewLine;
//var sentenceSummary = doc.Sentences.Select(p => p + System.Environment.NewLine + System.Environment.NewLine);
//finalSummary += "Sentence Summary - " + System.Environment.NewLine + string.Concat(sentenceSummary.ToArray()) + "]";
#endregion
string outputTemplate = htmlTemplate;
outputTemplate = outputTemplate
.Replace("{{query}}", query)
.Replace("{{timeStamp}}", DateTime.Now.ToString("dd MMM yyyy HH:mm"))
.Replace("{{output}}", finalSummary);
//string finalSummary = string.Join("\r\n\r\n", doc.Sentences.ToArray());
// do some stuff with summary. It is your result.
System.IO.File.WriteAllText(outputFileSummaryFinal, outputTemplate);
Process.Start(outputFileSummaryFinal);
}
