public void Search( )
{
// Basic integration test: search algorithms are already extensively tested with InMemoryIndexBase
MockRepository mocks = new MockRepository( );
IIndexConnector conn = mocks.StrictMock <IIndexConnector>( );
IWordFetcher fetcher = mocks.StrictMock <IWordFetcher>( );
Word dummy;
Expect.Call(fetcher.TryGetWord("test", out dummy)).Return(false);
Expect.Call(fetcher.TryGetWord("query", out dummy)).Return(false);
fetcher.Dispose( );
LastCall.On(fetcher);
Expect.Call(conn.GetWordFetcher( )).Return(fetcher);
mocks.ReplayAll( );
SqlIndex index = new SqlIndex(conn);
Assert.AreEqual(0, index.Search(new SearchParameters("test query")).Count, "Wrong search result count");
mocks.VerifyAll( );
}
/// <summary>
/// Performs a search in the index.
/// </summary>
/// <param name="parameters">The search parameters.</param>
/// <returns>The results.</returns>
public SearchResultCollection Search(SearchParameters parameters)
{
if (parameters == null)
{
throw new ArgumentNullException("parameters");
}
using (IWordFetcher fetcher = connector.GetWordFetcher())
{
if (parameters.DocumentTypeTags == null)
{
return(ScrewTurn.Wiki.SearchEngine.Tools.SearchInternal(parameters.Query, null, false, parameters.Options, fetcher));
}
return(ScrewTurn.Wiki.SearchEngine.Tools.SearchInternal(parameters.Query, parameters.DocumentTypeTags, true, parameters.Options, fetcher));
}
}
/// <summary>
/// Performs a search in the index.
/// </summary>
/// <param name="query">The search query.</param>
/// <param name="documentTypeTags">The document type tags to include in the search.</param>
/// <param name="filterDocumentType"><c>true</c> to apply the filter on the document type.</param>
/// <param name="options">The search options.</param>
/// <param name="fetcher">An object that is able to fetch words.</param>
/// <returns>The results.</returns>
/// <exception cref="ArgumentNullException">If <paramref name="query"/> or <paramref name="fetcher"/> are <c>null</c>.</exception>
/// <exception cref="ArgumentException">If <paramref name="query"/> is empty.</exception>
/// <exception cref="ArgumentNullException">If <paramref name="filterDocumentType"/> is <c>true</c> and <paramref name="documentTypeTags"/> is <c>null</c>.</exception>
/// <exception cref="ArgumentException">If <paramref name="filterDocumentType"/> is <c>true</c> and <paramref name="documentTypeTags"/> is empty.</exception>
public static SearchResultCollection SearchInternal(string query, string[] documentTypeTags, bool filterDocumentType, SearchOptions options, IWordFetcher fetcher)
{
if (query == null)
{
throw new ArgumentNullException("query");
}
if (query.Length == 0)
{
throw new ArgumentException("Query cannot be empty", "query");
}
if (filterDocumentType && documentTypeTags == null)
{
throw new ArgumentNullException("documentTypeTags");
}
if (filterDocumentType && documentTypeTags.Length == 0)
{
throw new ArgumentException("documentTypeTags cannot be empty", "documentTypeTags");
}
if (fetcher == null)
{
throw new ArgumentNullException("fetcher");
}
SearchResultCollection results = new SearchResultCollection();
query = query.ToLowerInvariant();
string[] queryWords = query.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
float totalRelevance = 0;
Word word = null;
foreach (string q in queryWords)
{
if (fetcher.TryGetWord(q, out word))
{
foreach (IDocument doc in word.Occurrences.Keys)
{
// Skip documents with excluded tags
if (filterDocumentType &&
!IsDocumentTypeTagIncluded(doc.TypeTag, documentTypeTags))
{
continue;
}
foreach (BasicWordInfo info in word.Occurrences[doc])
{
// If a search result is already present, add a new match to it,
// otherwise create a new search result object
WordInfo mi = new WordInfo(q, info.FirstCharIndex, info.WordIndex, info.Location);
SearchResult res = results.GetSearchResult(doc);
if (res == null)
{
res = new SearchResult(doc);
res.Relevance.SetValue(info.Location.RelativeRelevance);
res.Matches.Add(mi);
results.Add(res);
}
else
{
// Avoid adding duplicate matches (happens when query contains the same word multiple times)
if (!res.Matches.ContainsOccurrence(mi.Text, mi.FirstCharIndex))
{
res.Matches.Add(mi);
}
res.Relevance.SetValue(res.Relevance.Value + info.Location.RelativeRelevance);
}
totalRelevance += info.Location.RelativeRelevance;
}
}
}
}
if (options == SearchOptions.AllWords)
{
totalRelevance -= PurgeResultsForAllWords(results, queryWords);
}
else if (options == SearchOptions.ExactPhrase)
{
totalRelevance -= PurgeResultsForExactPhrase(results, queryWords);
}
else if (options == SearchOptions.AtLeastOneWord)
{
// Nothing to do
}
else
{
throw new InvalidOperationException("Unsupported SearchOptions");
}
// Finalize relevance values
for (int i = 0; i < results.Count; i++)
{
results[i].Relevance.Finalize(totalRelevance);
}
return(results);
}
/// <summary>
/// Performs a search in the index.
/// </summary>
/// <param name="query">The search query.</param>
/// <param name="documentTypeTags">The document type tags to include in the search.</param>
/// <param name="filterDocumentType"><c>true</c> to apply the filter on the document type.</param>
/// <param name="options">The search options.</param>
/// <param name="fetcher">An object that is able to fetch words.</param>
/// <returns>The results.</returns>
/// <exception cref="ArgumentNullException">If <paramref name="query"/> or <paramref name="fetcher"/> are <c>null</c>.</exception>
/// <exception cref="ArgumentException">If <paramref name="query"/> is empty.</exception>
/// <exception cref="ArgumentNullException">If <paramref name="filterDocumentType"/> is <c>true</c> and <paramref name="documentTypeTags"/> is <c>null</c>.</exception>
/// <exception cref="ArgumentException">If <paramref name="filterDocumentType"/> is <c>true</c> and <paramref name="documentTypeTags"/> is empty.</exception>
public static SearchResultCollection SearchInternal(string query, string[] documentTypeTags, bool filterDocumentType, SearchOptions options, IWordFetcher fetcher)
{
if(query == null) throw new ArgumentNullException("query");
if(query.Length == 0) throw new ArgumentException("Query cannot be empty", "query");
if(filterDocumentType && documentTypeTags == null) throw new ArgumentNullException("documentTypeTags");
if(filterDocumentType && documentTypeTags.Length == 0) throw new ArgumentException("documentTypeTags cannot be empty", "documentTypeTags");
if(fetcher == null) throw new ArgumentNullException("fetcher");
SearchResultCollection results = new SearchResultCollection();
query = query.ToLowerInvariant();
string[] queryWords = query.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
float totalRelevance = 0;
Word word = null;
foreach(string q in queryWords) {
if(fetcher.TryGetWord(q, out word)) {
foreach(IDocument doc in word.Occurrences.Keys) {
// Skip documents with excluded tags
if(filterDocumentType &&
!IsDocumentTypeTagIncluded(doc.TypeTag, documentTypeTags)) continue;
foreach(BasicWordInfo info in word.Occurrences[doc]) {
// If a search result is already present, add a new match to it,
// otherwise create a new search result object
WordInfo mi = new WordInfo(q, info.FirstCharIndex, info.WordIndex, info.Location);
SearchResult res = results.GetSearchResult(doc);
if(res == null) {
res = new SearchResult(doc);
res.Relevance.SetValue(info.Location.RelativeRelevance);
res.Matches.Add(mi);
results.Add(res);
}
else {
// Avoid adding duplicate matches (happens when query contains the same word multiple times)
if(!res.Matches.ContainsOccurrence(mi.Text, mi.FirstCharIndex)) {
res.Matches.Add(mi);
}
res.Relevance.SetValue(res.Relevance.Value + info.Location.RelativeRelevance);
}
totalRelevance += info.Location.RelativeRelevance;
}
}
}
}
if(options == SearchOptions.AllWords) {
totalRelevance -= PurgeResultsForAllWords(results, queryWords);
}
else if(options == SearchOptions.ExactPhrase) {
totalRelevance -= PurgeResultsForExactPhrase(results, queryWords);
}
else if(options == SearchOptions.AtLeastOneWord) {
// Nothing to do
}
else throw new InvalidOperationException("Unsupported SearchOptions");
// Finalize relevance values
for(int i = 0; i < results.Count; i++) {
results[i].Relevance.Finalize(totalRelevance);
}
return results;
}
