From a410043fd4d681b52d7beb7b0e88a7c2b5964278 Mon Sep 17 00:00:00 2001
From: guigang xie <xie.guigang@gcmodeller.org>
Date: Thu, 21 Sep 2023 15:51:10 +0800
Subject: [PATCH] make improvements of the resource hash algorithm

---
 src/LINQ/RQL/RQL.vbproj  |  1 +
 src/LINQ/RQL/Resource.vb | 57 ++++++++++++++++++++++++++++++++++++++--
 2 files changed, 56 insertions(+), 2 deletions(-)

diff --git a/src/LINQ/RQL/RQL.vbproj b/src/LINQ/RQL/RQL.vbproj
index 59331d6..8baa142 100644
--- a/src/LINQ/RQL/RQL.vbproj
+++ b/src/LINQ/RQL/RQL.vbproj
@@ -4,6 +4,7 @@
     <RootNamespace>RQL</RootNamespace>
     <TargetFrameworks>net6.0-windows;net6.0;net48</TargetFrameworks>
     <Platforms>AnyCPU;x64</Platforms>
+    <Configurations>Debug;Release;Rsharp_app_release;mzkit_win32;gcmodeller_desktop;Docker</Configurations>
     <AutoGenerateBindingRedirects>True</AutoGenerateBindingRedirects>
     <UseApplicationFramework>True</UseApplicationFramework>
     <GeneratePackageOnBuild>True</GeneratePackageOnBuild>
diff --git a/src/LINQ/RQL/Resource.vb b/src/LINQ/RQL/Resource.vb
index 93b3c14..ea6e6f9 100644
--- a/src/LINQ/RQL/Resource.vb
+++ b/src/LINQ/RQL/Resource.vb
@@ -1,8 +1,11 @@
 Imports System.IO
+Imports System.Runtime.CompilerServices
 Imports System.Text
 Imports Microsoft.VisualBasic.ComponentModel.Ranges
 Imports Microsoft.VisualBasic.Data.GraphTheory
+Imports Microsoft.VisualBasic.Data.Repository
 Imports Microsoft.VisualBasic.DataStorage.HDSPack.FileSystem
+Imports Microsoft.VisualBasic.SecurityString
 
 ''' <summary>
 ''' Resource query language
@@ -22,25 +25,60 @@ Public Class Resource : Implements IDisposable
         index = parser.Read
     End Sub
 
+    <MethodImpl(MethodImplOptions.AggressiveInlining)>
     Public Function Add(key As String, str As String)
         Return Add(key, Encoding.UTF8.GetBytes(str))
     End Function
 
+    <MethodImpl(MethodImplOptions.AggressiveInlining)>
     Public Function ReadString(map As String) As String
+        Return Encoding.UTF8.GetString(ReadBuffer(map))
+    End Function
+
+    Public Function ReadBuffer(map As String) As Byte()
         Dim path As String = URL(map)
         Dim file As Stream = buf.OpenFile(path, FileMode.Open, FileAccess.Read)
         Dim bytes As Byte() = New Byte(file.Length - 1) {}
         Call file.Read(bytes, Scan0, bytes.Length)
-        Return Encoding.UTF8.GetString(bytes)
+        Return bytes
     End Function
 
+    <MethodImpl(MethodImplOptions.AggressiveInlining)>
     Private Shared Function URL(map As String) As String
         Return $"/pool/{map.Substring(4, 2)}/{map.Substring(16, 6)}/{map}"
     End Function
 
+    Public Shared Function GetHashKey(data As Byte()) As String
+        Static md5 As New Md5HashProvider
+
+        ' all null/empty data point to ZERO location
+        If data.IsNullOrEmpty Then
+            Return New String("0"c, 32)
+        End If
+
+        ' combine two hash algorithm for avoid the hash confliction
+        Dim key1 As String = md5.GetMd5Hash(data.ToArray)
+        Dim firstByte = data(0).ToString
+        Dim lastByte = data(data.Length - 1).ToString
+        Dim middleByte = data((data.Length - 1) / 2).ToString
+        Dim fnv = FNV1a.GetHashCode({firstByte, lastByte, middleByte}).ToHexString
+        Dim hashcode As String = md5.GetMd5Hash(key1 & fnv)
+
+        Return hashcode
+    End Function
+
+    ''' <summary>
+    ''' 
+    ''' </summary>
+    ''' <param name="key">The query key, could be any text</param>
+    ''' <param name="data">the data for store in the database and associated with
+    ''' given query text data <paramref name="key"/>, the unique reference key of
+    ''' this resource data is generated via a specific hash algorithm based on 
+    ''' this data payload.</param>
+    ''' <returns></returns>
     Public Function Add(key As String, data As Byte()) As Boolean
         Dim tokens As String() = Strings.LCase(key).Split
-        Dim map As String = key.MD5
+        Dim map As String = GetHashKey(data)
         Dim path As String = URL(map)
 
         For Each si As String In tokens
@@ -63,6 +101,21 @@ Public Class Resource : Implements IDisposable
         Return True
     End Function
 
+    ''' <summary>
+    ''' Query resources matches
+    ''' </summary>
+    ''' <param name="query">any query term</param>
+    ''' <returns>
+    ''' A collection of the query result key with score value,
+    ''' the numeric tag in this collection is the query matches
+    ''' score and the key string value could be used for read
+    ''' resource data via the <see cref="ReadBuffer(String)"/> 
+    ''' function.
+    ''' </returns>
+    ''' <remarks>
+    ''' the result data of the query result has already been re-order
+    ''' via the matches score desc
+    ''' </remarks>
     Public Function [Get](query As String) As IEnumerable(Of NumericTagged(Of String))
         Dim tokens As String() = Strings.LCase(query).Split
         Dim maps As New Dictionary(Of String, Double)