public void Can_modify_doc_in_index_and_retrieve_by_secondary_key() {
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
Assert.AreEqual(doc.ToString(), d2.Get("doc"));
d2.RemoveField("doc");
d2.RemoveField("foo");
doc = new XDoc("doc").Elem("foo", "baz");
d2.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d2.Add(new Field("foo", "baz", Field.Store.NO, Field.Index.ANALYZED));
index.Update(new Term("id", "123"), d2);
hits = index.Search(index.Parse("foo:baz"));
Assert.AreEqual(1, hits.Count());
var d3 = hits.First().Document;
Assert.AreEqual("123", d3.Get("id"));
Assert.AreEqual(doc.ToString(), d3.Get("doc"));
}
public void Can_modify_doc_in_index_and_retrieve_by_secondary_key()
{
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
Assert.AreEqual(doc.ToString(), d2.Get("doc"));
d2.RemoveField("doc");
d2.RemoveField("foo");
doc = new XDoc("doc").Elem("foo", "baz");
d2.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d2.Add(new Field("foo", "baz", Field.Store.NO, Field.Index.ANALYZED));
index.Update(new Term("id", "123"), d2);
hits = index.Search(index.Parse("foo:baz"));
Assert.AreEqual(1, hits.Count());
var d3 = hits.First().Document;
Assert.AreEqual("123", d3.Get("id"));
Assert.AreEqual(doc.ToString(), d3.Get("doc"));
}
public void UntokenizedAnalyzer_field_treats_whitespace_as_part_of_token() {
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "foo bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("\"foo bar\"");
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
}
public void UntokenizedAnalyzer_field_treats_whitespace_as_part_of_token()
{
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "foo bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("\"foo bar\"");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}
public void UntokenizedAnalyzer_field_is_case_insensitive() {
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "FOO", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo");
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
hits = index.Search("FOO");
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
}
public void UntokenizedAnalyzer_field_accepts_whitespace_term_followed_by_wildcards()
{
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "FOO BAR", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo\\ ba*");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}
public void FilenameAnalyzer_field_treats_whitespace_underscore_and_dash_the_same()
{
var index = new TestIndex(new FilenameAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "foo bar-baz_boom", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo_bar\\ baz-boom");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}
public void Can_inverse_score_search() {
var index = new TestIndex();
var d = new Document();
d.Add(new Field(index.Default, "Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("Foo",new Sort(new SortField(SortField.FIELD_SCORE.GetField(),SortField.SCORE,true)));
Assert.AreEqual(3,hits.Length());
Assert.Less(hits.Score(0), hits.Score(1));
Assert.Less(hits.Score(1), hits.Score(2));
}
public void UntokenizedAnalyzer_field_is_case_insensitive()
{
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "FOO", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
hits = index.Search("FOO");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}
public void Can_use_update_for_new_document()
{
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Update(new Term("id", "123"), d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
}
public void Can_find_file_with_dash_in_it()
{
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "cube-teal", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
Console.WriteLine("query");
var query = index.Parse("CUBE-*");
Assert.AreEqual("content:cube-*", query.ToString());
var hits = index.Search(query);
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}
public void Can_inverse_score_search() {
var index = new TestIndex();
var d = new Document();
d.Add(new Field(index.Default, "Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("Foo", new Sort(new SortField(SortField.FIELD_SCORE.GetField(), SortField.SCORE, true)));
Assert.AreEqual(3, hits.Count());
var score0 = hits.ElementAt(0).Score;
var score1 = hits.ElementAt(1).Score;
var score2 = hits.ElementAt(2).Score;
Assert.Less(score0, score1);
Assert.Less(score1, score2);
}
public void Update_does_not_keep_unstored_fields()
{
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
index.Update(new Term("id", "123"), d2);
hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(0, hits.Count());
}
public void RemoveFields_removes_all_occurences()
{
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.YES, Field.Index.ANALYZED));
d.Add(new Field("foo", "baz", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
var fields = d2.GetFields("foo");
Assert.AreEqual(2, fields.Length);
d2.RemoveFields("foo");
fields = d2.GetFields("foo");
Assert.AreEqual(0, fields.Length);
}
public void Can_add_same_field_multiple_times()
{
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
d.Add(new Field("foo", "baz", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
hits = index.Search(index.Parse("foo:baz"));
Assert.AreEqual(1, hits.Count());
var d3 = hits.First().Document;
Assert.AreEqual("123", d3.Get("id"));
}
public void Can_inverse_score_search()
{
var index = new TestIndex();
var d = new Document();
d.Add(new Field(index.Default, "Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo Foo", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("Foo", new Sort(new SortField(SortField.FIELD_SCORE.GetField(), SortField.SCORE, true)));
Assert.AreEqual(3, hits.Count());
var score0 = hits.ElementAt(0).Score;
var score1 = hits.ElementAt(1).Score;
var score2 = hits.ElementAt(2).Score;
Assert.Less(score0, score1);
Assert.Less(score1, score2);
}
public void Can_query_for_all_documents()
{
var index = new TestIndex();
var d = new Document();
d.Add(new Field(index.Default, "Foo", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "bar", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(new MatchAllDocsQuery());
Assert.AreEqual(3, hits.Count());
var matches = new List <string>();
foreach (var result in hits)
{
matches.Add(result.Document.Get(index.Default));
}
Assert.AreEqual(new[] { "bar", "Foo", "Foo Foo" }, matches.OrderBy(x => x).ToArray());
}
public void Can_use_update_for_new_document() {
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Update(new Term("id", "123"), d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
}
public void RemoveFields_removes_all_occurences() {
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.YES, Field.Index.ANALYZED));
d.Add(new Field("foo", "baz", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
var fields = d2.GetFields("foo");
Assert.AreEqual(2, fields.Length);
d2.RemoveFields("foo");
fields = d2.GetFields("foo");
Assert.AreEqual(0, fields.Length);
}
public void FilenameAnalyzer_field_treats_whitespace_underscore_and_dash_the_same() {
var index = new TestIndex(new FilenameAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "foo bar-baz_boom", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo_bar\\ baz-boom");
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
}
public void Can_find_file_with_dash_in_it() {
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "cube-teal", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
Console.WriteLine("query");
var query = index.Parse("CUBE-*");
Assert.AreEqual("content:cube-*",query.ToString());
var hits = index.Search(query);
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
}
public void UntokenizedAnalyzer_field_accepts_whitespace_term_followed_by_wildcards() {
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "FOO BAR", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo\\ ba*");
Assert.AreEqual(1, hits.Length());
Assert.AreEqual("a", hits.Doc(0).Get("id"));
}
public void Can_add_same_field_multiple_times() {
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
d.Add(new Field("foo", "baz", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
Assert.AreEqual("123", d2.Get("id"));
hits = index.Search(index.Parse("foo:baz"));
Assert.AreEqual(1, hits.Count());
var d3 = hits.First().Document;
Assert.AreEqual("123", d3.Get("id"));
}
public void Can_query_for_all_documents() {
var index = new TestIndex();
var d = new Document();
d.Add(new Field(index.Default, "Foo", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "Foo Foo", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
d = new Document();
d.Add(new Field(index.Default, "bar", Field.Store.YES, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(new MatchAllDocsQuery());
Assert.AreEqual(3, hits.Count());
var matches = new List<string>();
foreach(var result in hits) {
matches.Add(result.Document.Get(index.Default));
}
Assert.AreEqual(new[] { "bar", "Foo", "Foo Foo" }, matches.OrderBy(x => x).ToArray());
}
public void Update_does_not_keep_unstored_fields() {
var index = new TestIndex();
var doc = new XDoc("doc").Elem("foo", "bar");
var d = new Document();
d.Add(new Field("id", "123", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("doc", doc.ToString(), Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field("foo", "bar", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(1, hits.Count());
var d2 = hits.First().Document;
index.Update(new Term("id", "123"), d2);
hits = index.Search(index.Parse("foo:bar"));
Assert.AreEqual(0, hits.Count());
}
public void UntokenizedAnalyzer_field_accepts_wildcards() {
var index = new TestIndex(new UntokenizedAnalyzer());
var d = new Document();
d.Add(new Field("id", "a", Field.Store.YES, Field.Index.UN_TOKENIZED));
d.Add(new Field(index.Default, "FOO BAR", Field.Store.NO, Field.Index.ANALYZED));
index.Add(d);
var hits = index.Search("foo*");
Assert.AreEqual(1, hits.Count());
Assert.AreEqual("a", hits.First().Document.Get("id"));
}