CharsetDetector · rstm-sf · Feb 25, 2023
diff --git a/src/CharsetDetector.cs b/src/CharsetDetector.cs
@@ -53,7 +53,7 @@ namespace UtfUnknown
     /// </summary>
     public class CharsetDetector
     {
-        internal InputState InputState;
+        private InputState _inputState;
 
         /// <summary>
         /// Start of the file
@@ -89,7 +89,7 @@ private IList<CharsetProber> CharsetProbers
         {
             get
             {
-                switch (InputState)
+                switch (_inputState)
                 {
                     case InputState.EscASCII:
                         return _escCharsetProber;
@@ -112,7 +112,7 @@ private IList<CharsetProber> CharsetProbers
         private CharsetDetector()
         {
             _start = true;
-            InputState = InputState.PureASCII;
+            _inputState = InputState.PureASCII;
             _lastChar = 0x00;
         }
 
@@ -350,9 +350,9 @@ private void FindInputState(byte[] buf, int offset, int len)
                 if ((buf[i] & 0x80) != 0 && buf[i] != 0xA0)
                 {
                     // we got a non-ascii byte (high-byte)
-                    if (InputState != InputState.Highbyte)
+                    if (_inputState != InputState.Highbyte)
                     {
-                        InputState = InputState.Highbyte;
+                        _inputState = InputState.Highbyte;
 
                         // kill EscCharsetProber if it is active
                         _escCharsetProber = null;
@@ -361,11 +361,11 @@ private void FindInputState(byte[] buf, int offset, int len)
                 }
                 else
                 {
-                    if (InputState == InputState.PureASCII &&
+                    if (_inputState == InputState.PureASCII &&
                         (buf[i] == 0x1B || (buf[i] == 0x7B && _lastChar == 0x7E)))
                     {
                         // found escape character or HZ "~{"
-                        InputState = InputState.EscASCII;
+                        _inputState = InputState.EscASCII;
                         _escCharsetProber = _escCharsetProber ?? GetNewProbers();
                     }
                     _lastChar = buf[i];
@@ -453,7 +453,7 @@ private DetectionResult DataEnd()
                 return new DetectionResult(_detectionDetail);
             }
 
-            if (InputState == InputState.Highbyte)
+            if (_inputState == InputState.Highbyte)
             {
                 var detectionResults = _charsetProbers
                     .Select(prober => new DetectionDetail(prober))
@@ -465,12 +465,12 @@ private DetectionResult DataEnd()
 
                 //TODO why done isn't true?
             }
-            else if (InputState == InputState.PureASCII)
+            else if (_inputState == InputState.PureASCII)
             {
                 //TODO why done isn't true?
                 return new DetectionResult(new DetectionDetail(CodepageName.ASCII, 1.0f));
             }
-            else if (InputState == InputState.EscASCII)
+            else if (_inputState == InputState.EscASCII)
             {
                 return new DetectionResult(new DetectionDetail(CodepageName.ASCII, 1.0f));
             }
@@ -480,7 +480,7 @@ private DetectionResult DataEnd()
 
         internal IList<CharsetProber> GetNewProbers()
         {
-            switch (InputState)
+            switch (_inputState)
             {
                 case InputState.EscASCII:
                     return new List<CharsetProber>() { new EscCharsetProber() };
@@ -499,4 +499,4 @@ internal IList<CharsetProber> GetNewProbers()
             }
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Chinese/BIG5DistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Chinese/BIG5DistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Chinese
 {
-    public class BIG5DistributionAnalyser : CharDistributionAnalyser
+    internal class BIG5DistributionAnalyser : CharDistributionAnalyser
     {
         // Big5 frequency table
         // by Taiwan's Mandarin Promotion Council
@@ -926,4 +926,4 @@ public override int GetOrder(byte[] buf, int offset)
             }
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Chinese/EUCTWDistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Chinese/EUCTWDistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Chinese
 {
-    public class EUCTWDistributionAnalyser : CharDistributionAnalyser
+    internal class EUCTWDistributionAnalyser : CharDistributionAnalyser
     {
         // EUCTW frequency table
         // Converted from big5 work
@@ -425,4 +425,4 @@ public override int GetOrder(byte[] buf, int offset)
                 return -1;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Chinese/GB18030DistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Chinese/GB18030DistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Chinese
 {
-    public class GB18030DistributionAnalyser : CharDistributionAnalyser
+    internal class GB18030DistributionAnalyser : CharDistributionAnalyser
     {
         // GB2312 most frequently used character table
         // Char to FreqOrder table, from hz6763
@@ -471,4 +471,4 @@ public override int GetOrder(byte[] buf, int offset)
                 return -1;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Japanese/EUCJPContextAnalyser.cs b/src/Core/Analyzers/MultiByte/Japanese/EUCJPContextAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Japanese
 {
-    public class EUCJPContextAnalyser : JapaneseContextAnalyser
+    internal class EUCJPContextAnalyser : JapaneseContextAnalyser
     {
         private const byte HIRAGANA_FIRST_BYTE = 0xA4;
 
@@ -36,4 +36,4 @@ protected override int GetOrder(byte[] buf, int offset)
             return -1;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Japanese/EUCJPDistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Japanese/EUCJPDistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Japanese
 {
-    public class EUCJPDistributionAnalyser : SJISDistributionAnalyser
+    internal class EUCJPDistributionAnalyser : SJISDistributionAnalyser
     {
         /// <summary>
         /// first  byte range: 0xa0 -- 0xfe
@@ -15,4 +15,4 @@ public override int GetOrder(byte[] buf, int offset)
                 return -1;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Japanese/JapaneseContextAnalyser.cs b/src/Core/Analyzers/MultiByte/Japanese/JapaneseContextAnalyser.cs
@@ -38,7 +38,7 @@
 
 namespace UtfUnknown.Core.Analyzers.Japanese
 {
-    public abstract class JapaneseContextAnalyser
+    internal abstract class JapaneseContextAnalyser
     {
         protected const int CATEGORIES_NUM = 6;
         protected const int ENOUGH_REL_THRESHOLD = 100;

diff --git a/src/Core/Analyzers/MultiByte/Japanese/SJISContextAnalyser.cs b/src/Core/Analyzers/MultiByte/Japanese/SJISContextAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Japanese
 {
-    public class SJISContextAnalyser : JapaneseContextAnalyser
+    internal class SJISContextAnalyser : JapaneseContextAnalyser
     {
         private const byte HIRAGANA_FIRST_BYTE = 0x82;
 
@@ -33,4 +33,4 @@ protected override int GetOrder(byte[] buf, int offset)
             return -1;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Japanese/SJISDistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Japanese/SJISDistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Japanese
 {
-    public class SJISDistributionAnalyser : CharDistributionAnalyser
+    internal class SJISDistributionAnalyser : CharDistributionAnalyser
     {
         // Sampling from about 20M text materials include literature and computer technology
         // Japanese frequency table, applied to both S-JIS and EUC-JP
@@ -575,4 +575,4 @@ public override int GetOrder(byte[] buf, int offset)
             return order;
         }
     }
-}
+}
diff --git a/src/Core/Analyzers/MultiByte/Korean/EUCKRDistributionAnalyser.cs b/src/Core/Analyzers/MultiByte/Korean/EUCKRDistributionAnalyser.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Analyzers.Korean
 {
-    public class EUCKRDistributionAnalyser : CharDistributionAnalyser
+    internal class EUCKRDistributionAnalyser : CharDistributionAnalyser
     {
         // Sampling from about 20M text materials include literature and computer technology
 

diff --git a/src/Core/BitPackage.cs b/src/Core/BitPackage.cs
@@ -38,23 +38,23 @@
 
 namespace UtfUnknown.Core
 {
-    public class BitPackage
+    internal class BitPackage
     {
-        public static int INDEX_SHIFT_4BITS  = 3;
-        public static int INDEX_SHIFT_8BITS  = 2;
-        public static int INDEX_SHIFT_16BITS = 1;
-        public static int SHIFT_MASK_4BITS  = 7;
-        public static int SHIFT_MASK_8BITS  = 3;
+        public static int INDEX_SHIFT_4BITS => 3;
+        public static int INDEX_SHIFT_8BITS => 2;
+        public static int INDEX_SHIFT_16BITS => 1;
+        public static int SHIFT_MASK_4BITS => 7;
+        public static int SHIFT_MASK_8BITS => 3;
 
-        public static int SHIFT_MASK_16BITS = 1;
-        public static int BIT_SHIFT_4BITS  = 2;
-        public static int BIT_SHIFT_8BITS  = 3;
+        public static int SHIFT_MASK_16BITS => 1;
+        public static int BIT_SHIFT_4BITS  => 2;
+        public static int BIT_SHIFT_8BITS  => 3;
 
-        public static int BIT_SHIFT_16BITS = 4;
-        public static int UNIT_MASK_4BITS  = 0x0000000F;
-        public static int UNIT_MASK_8BITS  = 0x000000FF;
+        public static int BIT_SHIFT_16BITS => 4;
+        public static int UNIT_MASK_4BITS  => 0x0000000F;
+        public static int UNIT_MASK_8BITS  => 0x000000FF;
 
-        public static int UNIT_MASK_16BITS = 0x0000FFFF;
+        public static int UNIT_MASK_16BITS => 0x0000FFFF;
 
         private int indexShift;
         private int shiftMask;
@@ -95,4 +95,4 @@ public int Unpack(int i)
                     ((i & shiftMask) << bitShift)) & unitMask;
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Chinese/BIG5SMModel.cs b/src/Core/Models/MultiByte/Chinese/BIG5SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Chinese
 {
-    public class BIG5SMModel : StateMachineModel
+    internal class BIG5SMModel : StateMachineModel
     {
         private readonly static int[] BIG5_cls = {
             BitPackage.Pack4bits(1,1,1,1,1,1,1,1),  // 00 - 07
@@ -59,4 +59,4 @@ public BIG5SMModel() : base(
         {
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Chinese/EUCTWSMModel.cs b/src/Core/Models/MultiByte/Chinese/EUCTWSMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Chinese
 {
-    public class EUCTWSMModel : StateMachineModel
+    internal class EUCTWSMModel : StateMachineModel
     {
         private readonly static int[] EUCTW_cls = {
             BitPackage.Pack4bits(2,2,2,2,2,2,2,2),  // 00 - 07
@@ -62,4 +62,4 @@ public EUCTWSMModel() : base(
         {
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Chinese/GB18030_SMModel.cs b/src/Core/Models/MultiByte/Chinese/GB18030_SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Chinese
 {
-    public class GB18030_SMModel : StateMachineModel
+    internal class GB18030_SMModel : StateMachineModel
     {
         private readonly static int[] GB18030_cls = {
             BitPackage.Pack4bits(1,1,1,1,1,1,1,1),  // 00 - 07
@@ -67,4 +67,4 @@ public GB18030_SMModel() : base(
         {
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Chinese/HZ_GB_2312_SMModel.cs b/src/Core/Models/MultiByte/Chinese/HZ_GB_2312_SMModel.cs
@@ -39,7 +39,7 @@
 //Escaped charsets state machines
 namespace UtfUnknown.Core.Models.MultiByte.Chinese
 {
-    public class HZ_GB_2312_SMModel : StateMachineModel
+    internal class HZ_GB_2312_SMModel : StateMachineModel
     {
         private readonly static int[] HZ_cls = {
             BitPackage.Pack4bits(1,0,0,0,0,0,0,0),  // 00 - 07
@@ -101,4 +101,4 @@ public HZ_GB_2312_SMModel() : base(
         {
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Chinese/Iso_2022_CN_SMModel.cs b/src/Core/Models/MultiByte/Chinese/Iso_2022_CN_SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Chinese
 {
-    public class Iso_2022_CN_SMModel : StateMachineModel
+    internal class Iso_2022_CN_SMModel : StateMachineModel
     {
         private readonly static int[] ISO2022CN_cls = {
             BitPackage.Pack4bits(2,0,0,0,0,0,0,0), // 00 - 07
@@ -64,4 +64,4 @@ public Iso_2022_CN_SMModel() : base(
         {
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Japanese/EUCJPSMModel.cs b/src/Core/Models/MultiByte/Japanese/EUCJPSMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Japanese
 {
-    public class EUCJPSMModel : StateMachineModel
+    internal class EUCJPSMModel : StateMachineModel
     {
         private readonly static int[] EUCJP_cls = {
             //BitPacket.Pack4bits(5,4,4,4,4,4,4,4),  // 00 - 07
@@ -63,4 +63,4 @@ public EUCJPSMModel() : base(
 
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Japanese/Iso_2022_JP_SMModel.cs b/src/Core/Models/MultiByte/Japanese/Iso_2022_JP_SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Japanese
 {
-    public class Iso_2022_JP_SMModel : StateMachineModel
+    internal class Iso_2022_JP_SMModel : StateMachineModel
     {
         private readonly static int[] ISO2022JP_cls = {
             BitPackage.Pack4bits(2,0,0,0,0,0,0,0), // 00 - 07
@@ -67,4 +67,4 @@ public Iso_2022_JP_SMModel() : base(
         }
 
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Japanese/SJIS_SMModel.cs b/src/Core/Models/MultiByte/Japanese/SJIS_SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Japanese
 {
-    public class SJIS_SMModel : StateMachineModel
+    internal class SJIS_SMModel : StateMachineModel
     {
         private readonly static int[] SJIS_cls = {
             //BitPacket.Pack4bits(0,1,1,1,1,1,1,1),  // 00 - 07
@@ -63,4 +63,4 @@ public SJIS_SMModel() : base(
 
         }
     }
-}
+}
diff --git a/src/Core/Models/MultiByte/Korean/CP949SMModel.cs b/src/Core/Models/MultiByte/Korean/CP949SMModel.cs
@@ -1,6 +1,6 @@
 namespace UtfUnknown.Core.Models.MultiByte.Korean
 {
-    public class CP949SMModel : StateMachineModel
+    internal class CP949SMModel : StateMachineModel
     {
         /*
         * 0: Unused