public void TestPack()
{
Assert.Equal(BitPackage.Pack4bits(0, 0, 0, 0, 0, 0, 0, 0), 0);
Assert.Equal(BitPackage.Pack4bits(1, 1, 1, 1, 1, 1, 1, 1), 286331153);
Assert.Equal(BitPackage.Pack4bits(2, 2, 2, 2, 2, 2, 2, 2), 572662306);
Assert.Equal(BitPackage.Pack4bits(15, 15, 15, 15, 15, 15, 15, 15), -1);
}
// 整型数
public static byte[] Bit6Compact(string text)
{
// 检查
CheckBit6(text);
BitPackage package = new BitPackage();
foreach (char ch in text)
{
#if DEBUG
int prefix = (ch >> 6) & 0x3;
Debug.Assert(prefix == 0x0 || prefix == 0x1);
#endif
int start = ch << 2;
for (int i = 0; i < 6; i++)
{
package.Add((start & 0x80) == 0x80 ? true : false);
start = start << 1;
}
}
// 尾部 bits 补齐 BIN 100000
if (package.Index != 0)
{
package.Add(true);
}
package.Flush();
return(package.Bytes);
}
// 整型数
public static byte[] Bit7Compact(string text)
{
// 检查
CheckBit7(text);
BitPackage package = new BitPackage();
foreach (char ch in text)
{
#if DEBUG
int prefix = (ch >> 7) & 0x1;
Debug.Assert(prefix == 0x0);
#endif
int start = ch << 1;
for (int i = 0; i < 7; i++)
{
package.Add((start & 0x80) == 0x80 ? true : false);
start = start << 1;
}
}
// 尾部 bits 补齐 BIN 1111111
if (package.Index != 0)
{
int count = 8 - package.Index;
for (int i = 0; i < count; i++)
{
package.Add(true);
}
}
package.Flush();
return(package.Bytes);
}
public void ShouldPackCorrectly()
{
Assert.Equal(BitPackage.Pack4bits(0, 0, 0, 0, 0, 0, 0, 0), 0);
Assert.Equal(BitPackage.Pack4bits(1, 1, 1, 1, 1, 1, 1, 1), 286331153);
Assert.Equal(BitPackage.Pack4bits(2, 2, 2, 2, 2, 2, 2, 2), 572662306);
Assert.Equal(BitPackage.Pack4bits(15, 15, 15, 15, 15, 15, 15, 15), -1);
}
protected StateMachineModel(BitPackage classTable, int classFactor, BitPackage stateTable, int[] charLenTable, string name)
{
this.ClassTable = classTable;
this.classFactor = classFactor;
this.StateTable = stateTable;
this.CharLenTable = charLenTable;
this.name = name;
}
protected StateMachineModel(BitPackage classTable, int classFactor, BitPackage stateTable, int[] charLenTable, string name)
{
this.ClassTable = classTable;
this.classFactor = classFactor;
this.StateTable = stateTable;
this.CharLenTable = charLenTable;
this.name = name;
}
public StateMachineModel(BitPackage classTable, int classFactor,
BitPackage stateTable, int[] charLenTable, String name)
{
this.classTable = classTable;
ClassFactor = classFactor;
this.stateTable = stateTable;
this.charLenTable = charLenTable;
Name = name;
}
public void TestUnpack()
{
int[] data = new int[] {
BitPackage.Pack4bits(0, 1, 2, 3, 4, 5, 6, 7),
BitPackage.Pack4bits(8, 9, 10, 11, 12, 13, 14, 15)
};
BitPackage pkg = new BitPackage(
BitPackage.INDEX_SHIFT_4BITS,
BitPackage.SHIFT_MASK_4BITS,
BitPackage.BIT_SHIFT_4BITS,
BitPackage.UNIT_MASK_4BITS,
data);
for (int i = 0; i < 16; i++)
{
int n = pkg.Unpack(i);
Assert.Equal(n, i);
}
}
public void ShouldUnpackCorrectly()
{
var data = new int[]
{
BitPackage.Pack4bits(0, 1, 2, 3, 4, 5, 6, 7),
BitPackage.Pack4bits(8, 9, 10, 11, 12, 13, 14, 15)
};
var package = new BitPackage(
BitPackage.IndexShift4Bits,
BitPackage.ShiftMask4Bits,
BitPackage.BitShift4Bits,
BitPackage.UnitMask4Bits,
data);
for (int i = 0; i < 16; i++)
{
int actual = package.Unpack(i);
Assert.Equal(actual, i);
}
}
static string AddToPackage(char ch, char charset, BitPackage package)
{
string bits = "";
if (charset == 'u')
{
bits = GetUpperSetBits(ch);
}
else if (charset == 'l')
{
bits = GetLowerSetBits(ch);
}
else if (charset == 'd')
{
bits = GetDigiSetBits(ch);
}
else
{
throw new ArgumentException($"无法识别的 charset '{charset}'");
}
package.AddRange(bits);
return(bits);
}
// 整型数
public static byte[] Bit5Compact(string text)
{
// 检查
CheckBit5(text);
BitPackage package = new BitPackage();
foreach (char ch in text)
{
#if DEBUG
int prefix = (ch >> 5) & 0x7;
Debug.Assert(prefix == 2);
#endif
int start = ch << 3;
for (int i = 0; i < 5; i++)
{
package.Add((start & 0x80) == 0x80 ? true : false);
start = start << 1;
}
}
package.Flush();
return(package.Bytes);
}
public static byte[] IsilCompact(string text,
StringBuilder debugInfo = null)
{
// 检查
CheckIsil(text);
BitPackage package = new BitPackage();
char prev_charset = 'u'; // u(大写) l(小写) d(数字)
for (int index = 0; index < text.Length; index++)
{
char ch = text[index];
if (debugInfo != null)
{
debugInfo.Append($"{ch}");
}
char next_ch = (char)0;
if (index + 1 <= text.Length - 1)
{
next_ch = text[index + 1];
}
// 获得当前字符可能处于的字符集
string current_charsets = GetCharsets(ch);
string action = "";
string targets = "";
// 如果当前字符的字符集和 prev_charset 不同
// 就必须 switch 或者 shift
if (current_charsets.IndexOf(prev_charset) == -1)
{
// 1) 如果 next_ch 在 prev_charset 以内，则用 shift
if (next_ch == 0 ||
GetCharsets(next_ch).IndexOf(prev_charset) != -1)
{
action = "shift";
targets = current_charsets[0].ToString();
}
else
{
action = "switch";
targets = GetCharsets(ch);
}
if (debugInfo != null)
{
debugInfo.Append($" {action}:{prev_charset}-{targets}");
}
string output = "";
if (action == "switch")
{
output = GetSwitchBits(prev_charset, targets[0]);
package.AddRange(output);
prev_charset = targets[0];
output += "," + AddToPackage(ch, prev_charset, package);
}
else
{
Debug.Assert(action == "shift");
output = GetShiftBits(prev_charset, targets[0]);
package.AddRange(output);
output += "," + AddToPackage(ch, targets[0], package);
}
if (debugInfo != null)
{
debugInfo.Append($" output:{output}\r\n");
}
}
else
{
// 直接输出字符
string output = AddToPackage(ch, prev_charset, package);
if (debugInfo != null)
{
debugInfo.Append($" output:{output}\r\n");
}
}
}
// 尾部 bits 补齐 BIN 1111111
if (package.Index != 0)
{
int count = 8 - package.Index;
for (int i = 0; i < count; i++)
{
package.Add(true);
}
}
package.Flush();
return(package.Bytes);
}
public static byte[] IsilCompress(string text)
{
// 检查
CheckIsil(text);
BitPackage package = new BitPackage();
char current_charset = 'u'; // u(大写) l(小写) d(数字)
for (int index = 0; index < text.Length; index++)
{
char ch = text[index];
if (ch == '-')
{
if (current_charset == 'u' || current_charset == 'l')
{
package.AddRange("00000");
}
else
{
Debug.Assert(current_charset == 'd');
package.AddRange("1010");
}
continue;
}
if (ch == ':')
{
if (current_charset == 'u')
{
package.AddRange("11011");
}
else if (current_charset == 'd')
{
package.AddRange("1011");
}
else
{
Debug.Assert(current_charset == 'l');
// TODO: 切换到 upper 或者 digit。最好预测一下，下一个字符，以便切换最优化
package.AddRange(GetSwitchBits(current_charset, 'u'));
current_charset = 'u';
package.AddRange("11011");
}
continue;
}
if (ch == '/')
{
if (current_charset == 'l')
{
package.AddRange("11011");
}
else
{
Debug.Assert(current_charset == 'd' || current_charset == 'u');
// 切换到 lower
package.AddRange(GetSwitchBits(current_charset, 'l'));
current_charset = 'l';
package.AddRange("11011");
}
continue;
}
if (ch >= 'A' && ch <= 'Z')
{
if (current_charset != 'u')
{
package.AddRange(GetSwitchBits(current_charset, 'u'));
current_charset = 'u';
}
AddToPackage(ch, current_charset, package);
continue;
}
if (ch >= 'a' && ch <= 'z')
{
if (current_charset != 'l')
{
package.AddRange(GetSwitchBits(current_charset, 'l'));
current_charset = 'l';
}
AddToPackage(ch, current_charset, package);
continue;
}
if (ch >= '0' && ch <= '9')
{
if (current_charset != 'd')
{
package.AddRange(GetSwitchBits(current_charset, 'd'));
current_charset = 'd';
}
AddToPackage(ch, current_charset, package);
continue;
}
throw new Exception($"出现了超出范围的字符 '{ch}'");
}
// 尾部 bits 补齐 BIN 1111111
if (package.Index != 0)
{
int count = 8 - package.Index;
for (int i = 0; i < count; i++)
{
package.Add(true);
}
}
package.Flush();
return(package.Bytes);
}
