public void GetDistance_Different()
{
string query = "<query><RecordSource ComparisonType=\"one of\">ELibrary</RecordSource><Title ComparisonType=\"FreeTextSearchWithStemming\">Защита информации</Title></query>";
SearchRequest request = new SearchRequest(query, "s1");
ExternalCatalogBook book = new ExternalCatalogBook("1", "Методы верификации");
Assert.IsTrue(NGramAnalyzer.GetDistance(request, book) > 0.4);
}
public void GetDistance_Identical()
{
string query = "<query><RecordSource ComparisonType=\"one of\">ELibrary</RecordSource><Title ComparisonType=\"FreeTextSearchWithStemming\">Защита информации</Title></query>";
SearchRequest request = new SearchRequest(query, "s1");
ExternalCatalogBook book = new ExternalCatalogBook("1", "Защита информации");
Assert.AreEqual(NGramAnalyzer.GetDistance(request, book), 0);
}
public static List <ExternalCatalogBook> GetBooks(int reportValue,
ref BackgroundWorker worker, ref DoWorkEventArgs e)
{
index = new Dictionary <char, HashSet <ExternalCatalogBook> >();
List <ExternalCatalogBook> books = new List <ExternalCatalogBook>();
int count = 0;
try
{
using (var db = new ExternalCatalogModel())
{
var table = db.ExternalCatalog.AsNoTracking();
worker.ReportProgress(table.Count(), BackgroundWorkerState.ExternalCatalogInitialize);
foreach (var row in table)
{
if (worker.CancellationPending)
{
e.Cancel = true;
break;
}
if (!CatalogProxy.IsPresentInCatalog(row.Title))
{
ExternalCatalogBook book = new ExternalCatalogBook(row.Id.ToString(), row.Title, row.Author, row.Keywords);
books.Add(book);
char t = book.Title[0];
char a = book.Author[0];
if (index.ContainsKey(t))
{
index[t].Add(book);
}
else
{
index[t] = new HashSet <ExternalCatalogBook>();
index[t].Add(book);
}
if (index.ContainsKey(a))
{
index[a].Add(book);
}
else
{
index[a] = new HashSet <ExternalCatalogBook>();
index[a].Add(book);
}
}
count++;
if (count % reportValue == 0)
{
worker.ReportProgress(count, BackgroundWorkerState.ExternalCatalogReportProgress);
}
}
}
}
catch (Exception ignore)
{
}
return(books);
}
private void btnDelete_Click(object sender, EventArgs e)
{
if (gridResult.SelectedRows.Count > 0)
{
ExternalCatalogBook book = new ExternalCatalogBook(gridResult.SelectedRows[0].Cells[0].Value.ToString(),
gridResult.SelectedRows[0].Cells[1].Value.ToString(), gridResult.SelectedRows[0].Cells[2].Value.ToString(), null);
CustomKeyValuePair <ExternalCatalogBook, double> k = new CustomKeyValuePair <ExternalCatalogBook, double>(book,
double.Parse(gridResult.SelectedRows[0].Cells[3].Value.ToString()));
gridResult.Rows.RemoveAt(gridResult.SelectedRows[0].Index);
btnDelete.Enabled = gridResult.SelectedRows.Count > 0;
_generalSettings.Recommendations.Remove(k);
_generalSettings.DeletedRecommendations.Add(k);
}
}
public static List <ExternalCatalogBook> GetBooks(string xmlPath, int reportValue,
ref BackgroundWorker worker, ref DoWorkEventArgs e)
{
index = new Dictionary <char, HashSet <ExternalCatalogBook> >();
List <ExternalCatalogBook> books = new List <ExternalCatalogBook>();
XmlDocument xml = new XmlDocument();
xml.Load(xmlPath);
XmlNode root = xml.LastChild;
XmlNode shopNode = root.LastChild;
XmlNode offersNode = shopNode.LastChild;
int count = 0;
if (offersNode.HasChildNodes)
{
worker.ReportProgress(offersNode.ChildNodes.Count, BackgroundWorkerState.ExternalCatalogInitialize);
for (int i = 0; i < offersNode.ChildNodes.Count; i++)
{
if (worker.CancellationPending)
{
e.Cancel = true;
break;
}
XmlNode offerNode = offersNode.ChildNodes[i];
string id = offerNode.Attributes["id"].Value;
string title = null, author = null;
foreach (XmlNode node in offerNode.ChildNodes)
{
if (node.Name.Equals("author"))
{
author = node.InnerText;
}
if (node.Name.Equals("name"))
{
title = node.InnerText;
}
if (title != null && author != null)
{
break;
}
}
if (isDocumentValid(title, author) && !CatalogProxy.IsPresentInCatalog(title))
{
ExternalCatalogBook book = new ExternalCatalogBook(id, title, author, null);
books.Add(book);
foreach (var word in GetWords(book.Title))
{
char t = word[0];
if (index.ContainsKey(t))
{
index[t].Add(book);
}
else
{
index[t] = new HashSet <ExternalCatalogBook>();
index[t].Add(book);
}
}
foreach (var word in GetWords(book.Author))
{
char t = word[0];
if (index.ContainsKey(t))
{
index[t].Add(book);
}
else
{
index[t] = new HashSet <ExternalCatalogBook>();
index[t].Add(book);
}
}
}
if (count % reportValue == 0)
{
worker.ReportProgress(count, BackgroundWorkerState.ExternalCatalogReportProgress);
}
count++;
if (count == 400)
{
break;
}
}
}
return(books);
}
public static double GetDistance(SearchRequest request, ExternalCatalogBook book)
{
double distance = 0;
List <string> requestTitle = new List <string>();
List <string> requestAuthor = new List <string>();
List <string> requestSubject = new List <string>();
List <string> requestKeywords = new List <string>();
AddWordsIfPresent(request, SearchFields.Title, ref requestTitle);
AddWordsIfPresent(request, SearchFields.Responsible, ref requestAuthor);
AddWordsIfPresent(request, SearchFields.SubjectText, ref requestSubject);
AddWordsIfPresent(request, SearchFields.Keywords, ref requestKeywords);
if (requestTitle.Count > 0 && requestSubject.Count > 0)
{
requestKeywords.AddRange(requestSubject);
}
else if (requestSubject.Count > 0)
{
requestTitle.AddRange(requestSubject);
}
requestSubject.Clear();
if (requestTitle.Count == 0)
{
requestTitle.AddRange(requestKeywords);
requestKeywords.Clear();
}
string title = book.Title;
if (book.HasKeywords() && requestKeywords.Count > 0)
{
distance += Math.Pow(1.0 - GetSimilarityForFields(requestKeywords, book.Keywords), 2);
}
else if (requestKeywords.Count > 0)
{
requestTitle.AddRange(requestKeywords);
requestKeywords.Clear();
}
if (book.HasAuthor() && requestAuthor.Count > 0)
{
distance += Math.Pow(1.0 - GetSimilarityForFields(requestAuthor, book.Author), 2);
}
else if (requestAuthor.Count > 0)
{
requestTitle.AddRange(requestAuthor);
requestAuthor.Clear();
}
else if (book.HasAuthor())
{
title += " " + book.Author;
}
if (requestTitle.Count > 0)
{
distance += Math.Pow(1.0 - GetSimilarityForFields(requestTitle, title), 2);
}
if (distance == 0)
{
return(0);
}
return(Math.Sqrt(distance));
}
