public void Can_search_exact()
{
var dir = Path.Combine(Dir, "SearcherTests.Can_search_exact");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <List <Field> >
{
new List <Field> {
new Field("_id", "0"), new Field("title", "Rambo First Blood")
},
new List <Field> {
new Field("_id", "1"), new Field("title", "rambo 2")
},
new List <Field> {
new Field("_id", "2"), new Field("title", "rocky 2")
},
new List <Field> {
new Field("_id", "3"), new Field("title", "the raiders of the lost ark")
},
new List <Field> {
new Field("_id", "4"), new Field("title", "the rain man")
},
new List <Field> {
new Field("_id", "5"), new Field("title", "the good, the bad and the ugly")
}
};
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.NoCompression, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
using (var searcher = new Searcher(dir))
{
var result = searcher.Search("title:rambo");
Assert.AreEqual(2, result.Total);
Assert.AreEqual(2, result.Docs.Count);
Assert.IsTrue(result.Docs.Any(d => d.Document.Id == 0));
Assert.IsTrue(result.Docs.Any(d => d.Document.Id == 1));
Assert.AreEqual(
"Rambo First Blood",
result.Docs.First(d => d.Document.Id == 0).Document.Fields.First(f => f.Key == "title").Value);
}
using (var searcher = new Searcher(dir))
{
var result = searcher.Search("title:the");
Assert.AreEqual(3, result.Total);
Assert.AreEqual(3, result.Docs.Count);
Assert.IsTrue(result.Docs.Any(d => d.Document.Id == 3));
Assert.IsTrue(result.Docs.Any(d => d.Document.Id == 4));
Assert.IsTrue(result.Docs.Any(d => d.Document.Id == 5));
}
}
public void Can_collect_prefixed()
{
var dir = Path.Combine(Dir, "Can_collect_prefixed");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Field>
{
new Field(0, "_id", "0"), new Field(0, "title", "rambo"),
new Field(1, "_id", "1"), new Field(1, "title", "rambo 2"),
new Field(2, "_id", "2"), new Field(2, "title", "rocky 2"),
new Field(3, "_id", "3"), new Field(3, "title", "raiders of the lost ark"),
new Field(4, "_id", "4"), new Field(4, "title", "rain man")
}.GroupBy(f => f.DocumentId).Select(g => new Document(g.Key, g.ToList()));
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.Lz, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(new QueryContext("title", "ra")
{
Prefix = true
}).ToList();
Assert.AreEqual(4, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 0));
Assert.IsTrue(scores.Any(d => d.DocumentId == 1));
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
}
public void Can_collect_exact_phrase_joined_by_not()
{
var dir = Path.Combine(Dir, "Can_collect_exact_phrase_joined_by_not");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <List <Field> >
{
new List <Field> {
new Field("_id", "0"), new Field("title", "rambo first blood")
},
new List <Field> {
new Field("_id", "1"), new Field("title", "rambo 2")
},
new List <Field> {
new Field("_id", "2"), new Field("title", "rocky 2")
},
new List <Field> {
new Field("_id", "3"), new Field("title", "raiders of the lost ark")
},
new List <Field> {
new Field("_id", "4"), new Field("title", "the rain man")
},
new List <Field> {
new Field("_id", "5"), new Field("title", "the good, the bad and the ugly")
}
};
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.QuickLz, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
var query = new QueryParser(new Analyzer()).Parse("+title:the");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.AreEqual(3, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
Assert.IsTrue(scores.Any(d => d.DocumentId == 5));
}
query = new QueryParser(new Analyzer()).Parse("+title:the -title:ugly");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.AreEqual(2, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
}
public void Can_collect_near()
{
var dir = Path.Combine(Dir, "Can_collect_near");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <List <Field> >
{
new List <Field> {
new Field("_id", "0"), new Field("title", "rambo")
},
new List <Field> {
new Field("_id", "1"), new Field("title", "rambo 2")
},
new List <Field> {
new Field("_id", "2"), new Field("title", "rocky 2")
},
new List <Field> {
new Field("_id", "3"), new Field("title", "raiders of the lost ark")
},
new List <Field> {
new Field("_id", "4"), new Field("title", "tomb raider")
}
};
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.QuickLz, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(new QueryContext("title", "raider")
{
Fuzzy = false, Edits = 1
}).ToList();
Assert.AreEqual(1, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(new QueryContext("title", "raider")
{
Fuzzy = true, Edits = 1
}).ToList();
Assert.AreEqual(2, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
}
public void Can_collect_near_phrase()
{
var dir = Path.Combine(Dir, "Can_collect_near_phrase_joined_by_and");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Field>
{
new Field(0, "_id", "0"), new Field(0, "title", "rambo first blood"),
new Field(1, "_id", "1"), new Field(1, "title", "rambo 2"),
new Field(2, "_id", "2"), new Field(2, "title", "rocky 2"),
new Field(3, "_id", "3"), new Field(3, "title", "the raid"),
new Field(4, "_id", "4"), new Field(4, "title", "the rain man"),
new Field(5, "_id", "5"), new Field(5, "title", "the good, the bad and the ugly")
}.GroupBy(f => f.DocumentId).Select(g => new Document(g.Key, g.ToList()));
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.Lz, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
var query = new QueryParser(new Analyzer()).Parse("+title:rain man");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.AreEqual(1, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
query = new QueryParser(new Analyzer(), 0.75f).Parse("+title:rain man~");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.AreEqual(1, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
}
public void Can_collect_by_id()
{
var dir = Path.Combine(Dir, "Can_collect_by_id");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Field>
{
new Field(0, "_id", "abc0123"), new Field(0, "title", "rambo first blood"),
new Field(1, "_id", "1"), new Field(1, "title", "rambo 2"),
new Field(2, "_id", "2"), new Field(2, "title", "rocky 2"),
new Field(3, "_id", "3"), new Field(3, "title", "the raiders of the lost ark"),
new Field(4, "_id", "four"), new Field(4, "title", "the rain man"),
new Field(5, "_id", "5five"), new Field(5, "title", "the good, the bad and the ugly")
}.GroupBy(f => f.DocumentId).Select(g => new Document(g.Key, g.ToList())).OrderBy(d => d.Id);
var writer = new DocumentUpsertOperation(dir, new Analyzer(), compression: Compression.Lz, primaryKey: "_id", documents: docs);
long indexName = writer.Commit();
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(new QueryContext("_id", "3")).ToList();
Assert.AreEqual(1, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
}
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(new QueryContext("_id", "5five")).ToList();
Assert.AreEqual(1, scores.Count);
Assert.IsTrue(scores.Any(d => d.DocumentId == 5));
}
}
