public void Can_collect_exact_phrase_joined_by_and()
{
var dir = Path.Combine(Setup.Dir, "Can_collect_exact_phrase_joined_by_and");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Dictionary <string, string> >
{
new Dictionary <string, string> {
{ "_id", "0" }, { "title", "rambo first blood" }
},
new Dictionary <string, string> {
{ "_id", "1" }, { "title", "rambo 2" }
},
new Dictionary <string, string> {
{ "_id", "2" }, { "title", "rocky 2" }
},
new Dictionary <string, string> {
{ "_id", "3" }, { "title", "the raiders of the lost ark" }
},
new Dictionary <string, string> {
{ "_id", "4" }, { "title", "the rain man" }
},
new Dictionary <string, string> {
{ "_id", "5" }, { "title", "the good, the bad and the ugly" }
}
};
string indexName;
using (var writer = new StreamWriteOperation(dir, new Analyzer(), docs.ToStream()))
{
indexName = writer.Execute();
}
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.That(scores.Count, Is.EqualTo(3));
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.That(scores.Count, Is.EqualTo(1));
Assert.IsTrue(scores.Any(d => d.DocumentId == 5));
}
}
public void Can_collect_near()
{
var dir = Path.Combine(Setup.Dir, "Can_collect_near");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Dictionary <string, string> >
{
new Dictionary <string, string> {
{ "_id", "0" }, { "title", "rambo" }
},
new Dictionary <string, string> {
{ "_id", "1" }, { "title", "rambo 2" }
},
new Dictionary <string, string> {
{ "_id", "2" }, { "title", "rocky 2" }
},
new Dictionary <string, string> {
{ "_id", "3" }, { "title", "raiders of the lost ark" }
},
new Dictionary <string, string> {
{ "_id", "4" }, { "title", "tomb raider" }
}
};
string indexName;
using (var writer = new StreamWriteOperation(dir, new Analyzer(), docs.ToStream()))
{
indexName = writer.Execute();
}
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.That(scores.Count, Is.EqualTo(1));
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.That(scores.Count, Is.EqualTo(2));
Assert.IsTrue(scores.Any(d => d.DocumentId == 3));
Assert.IsTrue(scores.Any(d => d.DocumentId == 4));
}
}
static void Write(string[] args)
{
var take = 1000;
if (Array.IndexOf(args, "--take") > 0)
{
take = int.Parse(args[Array.IndexOf(args, "--take") + 1]);
}
var fileName = args[Array.IndexOf(args, "--file") + 1];
string dir = null;
string indexName = null;
if (Array.IndexOf(args, "--dir") > 0)
{
dir = args[Array.IndexOf(args, "--dir") + 1];
}
if (Array.IndexOf(args, "--name") > 0)
{
indexName = args[Array.IndexOf(args, "--name") + 1];
}
var url = ConfigurationManager.AppSettings.Get("sir.endpoint");
var inproc = !string.IsNullOrWhiteSpace(dir);
Console.WriteLine("writing...");
var docs = new List <Dictionary <string, string> >();
var writeTimer = new Stopwatch();
writeTimer.Start();
if (inproc)
{
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
using (var writer = new StreamWriteOperation(dir, new Analyzer(), fileName, take))
{
writer.Execute();
}
}
else
{
Console.WriteLine("Executing HTTP POST");
using (var client = new WriterClient(indexName, url))
{
client.Write(docs);
}
}
Console.WriteLine("write operation took {0}", writeTimer.Elapsed);
}
public void Can_rank_near_term()
{
var dir = Path.Combine(Setup.Dir, "Can_rank_near_term");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Dictionary <string, string> >
{
new Dictionary <string, string> {
{ "_id", "0" }, { "title", "Gustav Horn, Count of Pori" }
},
new Dictionary <string, string> {
{ "_id", "1" }, { "title", "Port au Port Peninsula" }
},
new Dictionary <string, string> {
{ "_id", "2" }, { "title", "Pore" }
},
new Dictionary <string, string> {
{ "_id", "3" }, { "title", "Born 2.0" }
},
new Dictionary <string, string> {
{ "_id", "4" }, { "title", "P**n" }
}
};
string indexName;
using (var writer = new StreamWriteOperation(dir, new Analyzer(), docs.ToStream()))
{
indexName = writer.Execute();
}
var query = new QueryParser(new Analyzer()).Parse("+title:p**n~");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.That(scores.Count, Is.EqualTo(5));
Assert.IsTrue(scores.First().DocumentId.Equals(4));
Assert.IsTrue(scores[1].DocumentId.Equals(0));
Assert.IsTrue(scores[2].DocumentId.Equals(1));
Assert.IsTrue(scores[3].DocumentId.Equals(3));
Assert.IsTrue(scores[4].DocumentId.Equals(2));
}
}
public void Can_rank_near_phrase()
{
var dir = Path.Combine(Setup.Dir, "Can_rank_near_phrase");
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var docs = new List <Dictionary <string, string> >
{
new Dictionary <string, string> {
{ "_id", "0" }, { "title", "Tage Mage" }
},
new Dictionary <string, string> {
{ "_id", "1" }, { "title", "aye-aye" }
},
new Dictionary <string, string> {
{ "_id", "2" }, { "title", "Cage Rage Championships" }
},
new Dictionary <string, string> {
{ "_id", "3" }, { "title", "Page Up and Page Down keys" }
},
new Dictionary <string, string> {
{ "_id", "4" }, { "title", "Golden Age of P**n" }
}
};
string indexName;
using (var writer = new StreamWriteOperation(dir, new Analyzer(), docs.ToStream()))
{
indexName = writer.Execute();
}
var query = new QueryParser(new Analyzer()).Parse("+title:age of p**n~");
using (var collector = new Collector(dir, IxInfo.Load(Path.Combine(dir, indexName + ".ix")), new Tfidf()))
{
var scores = collector.Collect(query).ToList();
Assert.That(scores.Count, Is.EqualTo(5));
Assert.IsTrue(scores.First().DocumentId.Equals(4));
}
}
