public void StoreDocument()
{
MockRepository mocks = new MockRepository();
IIndexConnector conn = mocks.StrictMock <IIndexConnector>();
ScrewTurn.Wiki.SearchEngine.IDocument doc = mocks.StrictMock <ScrewTurn.Wiki.SearchEngine.IDocument>();
string dummyState = "state";
string content = "This is some test content.";
string title = "My Document";
Expect.Call(doc.Title).Return(title).Repeat.AtLeastOnce();
Expect.Call(doc.Tokenize(content)).Return(ScrewTurn.Wiki.SearchEngine.Tools.Tokenize(content, ScrewTurn.Wiki.SearchEngine.WordLocation.Content));
Expect.Call(doc.Tokenize(title)).Return(ScrewTurn.Wiki.SearchEngine.Tools.Tokenize(title, ScrewTurn.Wiki.SearchEngine.WordLocation.Title));
Predicate <ScrewTurn.Wiki.SearchEngine.WordInfo[]> contentPredicate = (array) => {
return
(array.Length == 5 &&
array[0].Text == "this" &&
array[1].Text == "is" &&
array[2].Text == "some" &&
array[3].Text == "test" &&
array[4].Text == "content");
};
Predicate <ScrewTurn.Wiki.SearchEngine.WordInfo[]> titlePredicate = (array) => {
return
(array.Length == 2 &&
array[0].Text == "my" &&
array[1].Text == "document");
};
Predicate <ScrewTurn.Wiki.SearchEngine.WordInfo[]> keywordsPredicate = (array) => {
return
(array.Length == 1 &&
array[0].Text == "test");
};
conn.DeleteDataForDocument(doc, dummyState);
LastCall.On(conn);
Expect.Call(conn.SaveDataForDocument(null, null, null, null, null)).IgnoreArguments()
.Constraints(RC.Is.Same(doc), RC.Is.Matching(contentPredicate), RC.Is.Matching(titlePredicate), RC.Is.Matching(keywordsPredicate), RC.Is.Same(dummyState))
.Return(8);
mocks.ReplayAll();
SqlIndex index = new SqlIndex(conn);
Assert.AreEqual(8, index.StoreDocument(doc, new string[] { "test" }, content, dummyState), "Wrong occurrence count");
mocks.VerifyAll();
}
/// <summary>
/// Stores a document in the index.
/// </summary>
/// <param name="document">The document.</param>
/// <param name="keywords">The document keywords, if any, an empty array or <c>null</c> otherwise.</param>
/// <param name="content">The content of the document.</param>
/// <param name="state">A state object that is passed to the IndexStorer SaveDate/DeleteData function.</param>
/// <returns>The number of indexed words (including duplicates).</returns>
/// <remarks>
/// Indexing the content of the document is <b>O(n)</b>,
/// where <b>n</b> is the total number of words in the document.
/// </remarks>
public int StoreDocument(IDocument document, IList <string> keywords, string content, object state)
{
if (document == null)
{
throw new ArgumentNullException("document");
}
if (keywords == null)
{
keywords = new string[0];
}
if (content == null)
{
throw new ArgumentNullException("content");
}
RemoveDocument(document, state);
keywords = SearchEngine.Tools.CleanupKeywords(keywords);
// Prepare content words
var contentWords = document.Tokenize(content);
contentWords = SearchEngine.Tools.RemoveStopWords(contentWords, stopWords);
// Prepare title words
var titleWords = document.Tokenize(document.Title);
titleWords = SearchEngine.Tools.RemoveStopWords(titleWords, stopWords);
for (var i = 0; i < titleWords.Length; i++)
{
titleWords[i] = new WordInfo(titleWords[i].Text, titleWords[i].FirstCharIndex, titleWords[i].WordIndex,
WordLocation.Title);
}
// Prepare keywords
var words = new WordInfo[keywords.Count];
var count = 0;
for (var i = 0; i < words.Length; i++)
{
words[i] = new WordInfo(keywords[i], (ushort)count, (ushort)i, WordLocation.Keywords);
count += 1 + keywords[i].Length;
}
return(connector.SaveDataForDocument(document, contentWords, titleWords, words, state));
}
