move vectorspace ns into search prj

kreeben · Apr 12, 2022 · 9fa07a2 · 9fa07a2
1 parent b86d2fe
commit 9fa07a2
Show file tree

Hide file tree

Showing 47 changed files with 29 additions and 63 deletions.
diff --git a/src/Sir.Cmd/BenchmarkCommand.cs b/src/Sir.Cmd/BenchmarkCommand.cs
@@ -17,13 +17,16 @@ public void Run(IDictionary<string, string> args, ILogger logger)
                 RunIndexBenchmark(args, logger);
         }
 
+        /// <summary>
+        /// E.g. benchmark --tokenize --file d:\enwiki-20211122-cirrussearch-content.json.gz --directory C:\projects\resin\src\Sir.HttpServer\AppData\database --skip 0 --take 10000
+        /// </summary>
         public void RunTokenizeBenchmark(IDictionary<string, string> args, ILogger logger)
         {
             const int numOfRuns = 10;
             int skip = int.Parse(args["skip"]);
             int take = int.Parse(args["take"]);
             var fileName = args["file"];
-            var model = new BagOfCharsModel();
+            var model = new NGramModel(new BagOfCharsModel());
             var documents = new List<Document>(WikipediaHelper.Read(fileName, skip, take, new HashSet<string> { "text" }));
             var timer = Stopwatch.StartNew();
 

diff --git a/src/Sir.Cmd/Program.cs b/src/Sir.Cmd/Program.cs
@@ -6,7 +6,6 @@
 using Newtonsoft.Json;
 using Sir.Documents;
 using Sir.Search;
-using Sir.VectorSpace;
 
 namespace Sir.Cmd
 {

diff --git a/src/Sir.Crawl/CrawlUserDirectoryCommand.cs b/src/Sir.Crawl/CrawlUserDirectoryCommand.cs
@@ -1,7 +1,6 @@
 using HtmlAgilityPack;
 using Microsoft.Extensions.Logging;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Collections.Generic;
 using System.Diagnostics;

diff --git a/src/Sir.HttpServer/HttpQueryParser.cs b/src/Sir.HttpServer/HttpQueryParser.cs
@@ -1,7 +1,6 @@
 using Microsoft.AspNetCore.Http;
 using Newtonsoft.Json;
 using Sir.Search;
-using Sir.VectorSpace;
 using System.Collections.Generic;
 using System.Dynamic;
 using System.IO;

diff --git a/src/Sir.HttpServer/HttpWriter.cs b/src/Sir.HttpServer/HttpWriter.cs
@@ -4,7 +4,6 @@
 using Microsoft.AspNetCore.Http;
 using Newtonsoft.Json;
 using Sir.Search;
-using Sir.VectorSpace;
 
 namespace Sir.HttpServer
 {

diff --git a/src/Sir.HttpServer/IHttpReader.cs b/src/Sir.HttpServer/IHttpReader.cs
@@ -1,6 +1,5 @@
 using Microsoft.AspNetCore.Http;
 using Sir.Search;
-using Sir.VectorSpace;
 using System.Threading.Tasks;
 
 namespace Sir.HttpServer

diff --git a/src/Sir.HttpServer/IHttpWriter.cs b/src/Sir.HttpServer/IHttpWriter.cs
@@ -1,5 +1,4 @@
 using Microsoft.AspNetCore.Http;
-using Sir.VectorSpace;
 
 namespace Sir.HttpServer
 {

diff --git a/src/Sir.HttpServer/ServiceConfiguration.cs b/src/Sir.HttpServer/ServiceConfiguration.cs
@@ -1,7 +1,6 @@
 using Microsoft.Extensions.DependencyInjection;
 using Microsoft.Extensions.Logging;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.IO;
 

diff --git a/src/Sir.Mnist/IndexMnistCommand.cs b/src/Sir.Mnist/IndexMnistCommand.cs
@@ -1,7 +1,6 @@
 using Microsoft.Extensions.Logging;
 using Sir.Documents;
 using Sir.Search;
-using Sir.VectorSpace;
 using System.Collections.Generic;
 using System.Diagnostics;
 using System.IO;

diff --git a/src/Sir.Search/Field.cs b/src/Sir.Search/Field.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System;
+using System;
 using System.Collections.Generic;
 using System.Diagnostics;
 

diff --git a/src/Sir.VectorSpace/Hit.cs → src/Sir.Search/Hit.cs b/src/Sir.VectorSpace/Hit.cs → src/Sir.Search/Hit.cs
@@ -1,6 +1,6 @@
 using System.Diagnostics;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     [DebuggerDisplay("{Score} {Node}")]
     public class Hit

diff --git a/src/Sir.VectorSpace/ColumnReader.cs → src/Sir.Search/Index/ColumnReader.cs b/src/Sir.VectorSpace/ColumnReader.cs → src/Sir.Search/Index/ColumnReader.cs
@@ -5,7 +5,7 @@
 using System.Diagnostics;
 using System.IO;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     /// <summary>
     /// Index bitmap reader. Each block is a <see cref="Sir.Search.VectorNode"/>.

diff --git a/src/Sir.VectorSpace/ColumnWriter.cs → src/Sir.Search/Index/ColumnWriter.cs b/src/Sir.VectorSpace/ColumnWriter.cs → src/Sir.Search/Index/ColumnWriter.cs
@@ -1,7 +1,7 @@
 using System;
 using System.IO;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public class ColumnWriter : IDisposable
     {

diff --git a/src/Sir.VectorSpace/GraphBuilder.cs → src/Sir.Search/Index/GraphBuilder.cs b/src/Sir.VectorSpace/GraphBuilder.cs → src/Sir.Search/Index/GraphBuilder.cs
@@ -2,7 +2,7 @@
 using System.Collections.Generic;
 using System.IO;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public static class GraphBuilder
     {

diff --git a/src/Sir.VectorSpace/IColumnReader.cs → src/Sir.Search/Index/IColumnReader.cs b/src/Sir.VectorSpace/IColumnReader.cs → src/Sir.Search/Index/IColumnReader.cs
@@ -1,6 +1,6 @@
 using System;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public interface IColumnReader : IDisposable
     {

diff --git a/src/Sir.VectorSpace/IReducer.cs → src/Sir.Search/Index/IReducer.cs b/src/Sir.VectorSpace/IReducer.cs → src/Sir.Search/Index/IReducer.cs
@@ -1,7 +1,7 @@
 using System;
 using System.Collections.Generic;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public interface IReducer
     {

diff --git a/src/Sir.VectorSpace/PageIndexReader.cs → src/Sir.Search/Index/PageIndexReader.cs b/src/Sir.VectorSpace/PageIndexReader.cs → src/Sir.Search/Index/PageIndexReader.cs
@@ -3,7 +3,7 @@
 using System.IO;
 using System.Runtime.InteropServices;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     /// <summary>
     /// Index segment address reader.

diff --git a/src/Sir.VectorSpace/PageIndexWriter.cs → src/Sir.Search/Index/PageIndexWriter.cs b/src/Sir.VectorSpace/PageIndexWriter.cs → src/Sir.Search/Index/PageIndexWriter.cs
@@ -1,7 +1,7 @@
 using System;
 using System.IO;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     /// <summary>
     /// Index segment address writer.

diff --git a/src/Sir.VectorSpace/PathFinder.cs → src/Sir.Search/Index/PathFinder.cs b/src/Sir.VectorSpace/PathFinder.cs → src/Sir.Search/Index/PathFinder.cs
@@ -3,7 +3,7 @@
 using System.IO;
 using System.Text;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public static class PathFinder
     {

diff --git a/src/Sir.VectorSpace/PostingsReader.cs → src/Sir.Search/Index/PostingsReader.cs b/src/Sir.VectorSpace/PostingsReader.cs → src/Sir.Search/Index/PostingsReader.cs
@@ -5,7 +5,7 @@
 using System.IO;
 using System.Runtime.InteropServices;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     /// <summary>
     /// Allocate postings in memory.

diff --git a/src/Sir.VectorSpace/PostingsResolver.cs → src/Sir.Search/Index/PostingsResolver.cs b/src/Sir.VectorSpace/PostingsResolver.cs → src/Sir.Search/Index/PostingsResolver.cs
@@ -1,4 +1,4 @@
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public class PostingsResolver
     {

diff --git a/src/Sir.VectorSpace/Scorer.cs → src/Sir.Search/Index/Scorer.cs b/src/Sir.VectorSpace/Scorer.cs → src/Sir.Search/Index/Scorer.cs
@@ -1,6 +1,6 @@
 using System.Collections.Generic;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     public class Scorer
     {

diff --git a/src/Sir.VectorSpace/SerializableVector.cs → src/Sir.Search/Index/SerializableVector.cs b/src/Sir.VectorSpace/SerializableVector.cs → src/Sir.Search/Index/SerializableVector.cs
@@ -6,9 +6,8 @@
 using System.IO.MemoryMappedFiles;
 using System.Runtime.InteropServices;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
-    [System.Diagnostics.DebuggerDisplay("{Label}")]
     public class SerializableVector : ISerializableVector
     {
         public object Label { get; private set; }
@@ -223,7 +222,7 @@ public static long Serialize(ISerializableVector vector, Stream vectorStream)
 
         public override string ToString()
         {
-            return Value.ToString();
+            return Value==null?base.ToString() : Value.ToString();
         }
     }
 }
diff --git a/src/Sir.Search/Models/BagOfCharsModel.cs b/src/Sir.Search/Models/BagOfCharsModel.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System.Collections.Generic;
+using System.Collections.Generic;
 
 namespace Sir.Search
 {

diff --git a/src/Sir.Search/Models/LinearClassifierImageModel.cs b/src/Sir.Search/Models/LinearClassifierImageModel.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System;
+using System;
 using System.Collections.Generic;
 using System.Linq;
 

diff --git a/src/Sir.Search/Models/NGramModel.cs b/src/Sir.Search/Models/NGramModel.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System.Collections.Generic;
+using System.Collections.Generic;
 
 namespace Sir.Search
 {

diff --git a/src/Sir.Search/Parsers/IQueryFormatter.cs b/src/Sir.Search/Parsers/IQueryFormatter.cs
@@ -1,5 +1,4 @@
 using Microsoft.AspNetCore.Http;
-using Sir.VectorSpace;
 using System.Threading.Tasks;
 
 namespace Sir.Search

diff --git a/src/Sir.Search/Parsers/QueryParser.cs b/src/Sir.Search/Parsers/QueryParser.cs
@@ -1,6 +1,5 @@
 using Microsoft.Extensions.Logging;
 using Newtonsoft.Json;
-using Sir.VectorSpace;
 using System.Collections.Generic;
 
 namespace Sir.Search

diff --git a/src/Sir.VectorSpace/Query.cs → src/Sir.Search/Query.cs b/src/Sir.VectorSpace/Query.cs → src/Sir.Search/Query.cs
@@ -1,6 +1,7 @@
-using System.Collections.Generic;
+
+using System.Collections.Generic;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     /// <summary>
     /// A boolean query.

diff --git a/src/Sir.Search/Session/Debug/GraphInfo.cs b/src/Sir.Search/Session/Debug/GraphInfo.cs
@@ -1,6 +1,4 @@
-using Sir.VectorSpace;
-
-namespace Sir.Search
+namespace Sir.Search
 {
     public class GraphInfo
     {

diff --git a/src/Sir.Search/Session/DocumentStreamSession.cs b/src/Sir.Search/Session/DocumentStreamSession.cs
@@ -1,5 +1,4 @@
 using Sir.Documents;
-using Sir.VectorSpace;
 using System;
 using System.Collections.Generic;
 using System.IO;

diff --git a/src/Sir.Search/Session/ISearchSession.cs b/src/Sir.Search/Session/ISearchSession.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System;
+using System;
 
 namespace Sir.Search
 {

diff --git a/src/Sir.Search/Session/IndexSession.cs b/src/Sir.Search/Session/IndexSession.cs
@@ -1,5 +1,4 @@
-using Sir.VectorSpace;
-using System;
+using System;
 using System.Collections.Generic;
 
 namespace Sir.Search

diff --git a/src/Sir.Search/Session/IndexWriter.cs b/src/Sir.Search/Session/IndexWriter.cs
@@ -1,5 +1,4 @@
 using Microsoft.Extensions.Logging;
-using Sir.VectorSpace;
 using System;
 using System.Collections.Generic;
 using System.Diagnostics;

diff --git a/src/Sir.Search/Session/SearchSession.cs b/src/Sir.Search/Session/SearchSession.cs
@@ -1,5 +1,4 @@
 using Microsoft.Extensions.Logging;
-using Sir.VectorSpace;
 using System;
 using System.Collections.Generic;
 using System.Diagnostics;

diff --git a/src/Sir.Search/Sir.Search.csproj b/src/Sir.Search/Sir.Search.csproj
@@ -25,7 +25,6 @@
   <ItemGroup>
     <ProjectReference Include="..\Sir.Core\Sir.Core.csproj" />
     <ProjectReference Include="..\Sir.Document\Sir.Documents.csproj" />
-    <ProjectReference Include="..\Sir.VectorSpace\Sir.VectorSpace.csproj" />
   </ItemGroup>
 
   <ItemGroup>

diff --git a/src/Sir.VectorSpace/Term.cs → src/Sir.Search/Term.cs b/src/Sir.VectorSpace/Term.cs → src/Sir.Search/Term.cs
@@ -1,6 +1,6 @@
 using System.Collections.Generic;
 
-namespace Sir.VectorSpace
+namespace Sir.Search
 {
     [System.Diagnostics.DebuggerDisplay("{Key}:{Label}")]
     public class Term : BooleanStatement, ITerm

diff --git a/src/Sir.Store.Tests/BagOfCharsModelTests.cs b/src/Sir.Store.Tests/BagOfCharsModelTests.cs
@@ -1,7 +1,6 @@
 using Microsoft.Extensions.Logging;
 using NUnit.Framework;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Diagnostics;
 using System.IO;

diff --git a/src/Sir.Store.Tests/ImageModelTests.cs b/src/Sir.Store.Tests/ImageModelTests.cs
@@ -1,7 +1,6 @@
 using NUnit.Framework;
 using Sir.Mnist;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Diagnostics;
 using System.IO;

diff --git a/src/Sir.Store.Tests/IndexSessionTests.cs b/src/Sir.Store.Tests/IndexSessionTests.cs
@@ -2,7 +2,6 @@
 using NUnit.Framework;
 using Sir.Documents;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Diagnostics;
 using System.Linq;

diff --git a/src/Sir.Store.Tests/NGramModelTests.cs b/src/Sir.Store.Tests/NGramModelTests.cs
@@ -1,7 +1,6 @@
 using Microsoft.Extensions.Logging;
 using NUnit.Framework;
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Diagnostics;
 using System.IO;

diff --git a/src/Sir.StringCompare/Program.cs b/src/Sir.StringCompare/Program.cs
@@ -1,5 +1,4 @@
 using Sir.Search;
-using Sir.VectorSpace;
 using System;
 using System.Linq;
 

diff --git a/src/Sir.WebFront/Features/CrawlJob.cs b/src/Sir.WebFront/Features/CrawlJob.cs
@@ -9,7 +9,6 @@
 using System.Threading;
 using Microsoft.Extensions.Logging;
 using Sir.Search;
-using Sir.VectorSpace;
 
 namespace Sir.HttpServer.Features
 {

diff --git a/src/Sir.WebFront/Features/SaveAsJob.cs b/src/Sir.WebFront/Features/SaveAsJob.cs
@@ -3,7 +3,6 @@
 using Microsoft.Extensions.Logging;
 using Sir.Documents;
 using Sir.Search;
-using Sir.VectorSpace;
 
 namespace Sir.HttpServer.Features
 {

diff --git a/src/Sir.WebFront/HttpQueryParser.cs b/src/Sir.WebFront/HttpQueryParser.cs
@@ -1,7 +1,6 @@
 using Microsoft.AspNetCore.Http;
 using Newtonsoft.Json;
 using Sir.Search;
-using Sir.VectorSpace;
 using System.Collections.Generic;
 using System.Dynamic;
 using System.IO;

diff --git a/src/Sir.Wikipedia/IndexWikipediaCommand.cs b/src/Sir.Wikipedia/IndexWikipediaCommand.cs
@@ -1,7 +1,6 @@
 using Microsoft.Extensions.Logging;
 using Sir.Documents;
 using Sir.Search;
-using Sir.VectorSpace;
 using System.Collections.Generic;
 using System.IO;
 

diff --git a/src/Sir.sln b/src/Sir.sln
@@ -41,8 +41,6 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Sir.WebFront", "Sir.WebFron
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Sir.Search", "Sir.Search\Sir.Search.csproj", "{AB275A5B-E72E-475A-8E1A-FFA7A5F9C932}"
 EndProject
-Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Sir.VectorSpace", "Sir.VectorSpace\Sir.VectorSpace.csproj", "{9791D56B-F5A5-448D-A615-612585CA6C37}"
-EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Sir.StringCompare", "Sir.StringCompare\Sir.StringCompare.csproj", "{C6050E65-9411-41E3-A6EE-0A45E6FFB4FC}"
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Sir.Cmd", "Sir.Cmd\Sir.Cmd.csproj", "{CEDD3CA9-D38D-43BF-9013-212AE6332CE0}"
@@ -97,10 +95,6 @@ Global
 		{AB275A5B-E72E-475A-8E1A-FFA7A5F9C932}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{AB275A5B-E72E-475A-8E1A-FFA7A5F9C932}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{AB275A5B-E72E-475A-8E1A-FFA7A5F9C932}.Release|Any CPU.Build.0 = Release|Any CPU
-		{9791D56B-F5A5-448D-A615-612585CA6C37}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
-		{9791D56B-F5A5-448D-A615-612585CA6C37}.Debug|Any CPU.Build.0 = Debug|Any CPU
-		{9791D56B-F5A5-448D-A615-612585CA6C37}.Release|Any CPU.ActiveCfg = Release|Any CPU
-		{9791D56B-F5A5-448D-A615-612585CA6C37}.Release|Any CPU.Build.0 = Release|Any CPU
 		{C6050E65-9411-41E3-A6EE-0A45E6FFB4FC}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
 		{C6050E65-9411-41E3-A6EE-0A45E6FFB4FC}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{C6050E65-9411-41E3-A6EE-0A45E6FFB4FC}.Release|Any CPU.ActiveCfg = Release|Any CPU