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($"无法识别的 charset '{charset}'");
}
package.AddRange(bits);
return(bits);
}
public static byte[] IsilCompact(string text,
StringBuilder debugInfo = null)
{
// 检查
CheckIsil(text);
BitPackage package = new BitPackage();
char prev_charset = 'u'; // u(大写) l(小写) d(数字)
for (int index = 0; index < text.Length; index++)
{
char ch = text[index];
if (debugInfo != null)
{
debugInfo.Append($"{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);
}
