/// <summary>
/// register a synonym
/// </summary>
/// <param name="from">original word</param>
/// <param name="to">converted word</param>
public void Add(string from, string to)
{
var from2 = StrMatch.ReplaceMetaStr(from);
Regex reg;
if (from.EndsWith("#"))
{
from = from.Substring(0, from.Length - 1);
from2 = from2.Substring(0, from2.Length - 1);
reg = new Regex(string.Format("(^|[^A-Z0-9]){0}[0-9]+", from2), RegexOptions.IgnoreCase | RegexOptions.Compiled);
}
else
{
reg = new Regex(string.Format("(^|[^A-Z0-9]){0}($|[^A-Z0-9])", from2), RegexOptions.IgnoreCase | RegexOptions.Compiled);
}
_regs.Add(reg);
_froms.Add(from);
_tos.Add(to);
}
/// <summary>
/// make instance with specific match pattern
/// </summary>
/// <param name="pattern"></param>
public StrMatch(string pattern)
{
var reopt = RegexOptions.IgnoreCase | RegexOptions.Singleline;
var isRegularExpressioning = false;
var isDoubleQuatationing = false;
var isEscaping = false;
var expss = new List <List <string> >();
var str = new StringBuilder();
var exps = new List <string>();
expss.Add(exps);
pattern = pattern.Trim();
for (var ci = 0; ci < pattern.Length; ci++)
{
var c = pattern[ci];
if (isRegularExpressioning == false)
{
if (c == '\\' || isEscaping)
{
isEscaping = !isEscaping;
str.Append(c);
continue;
}
if (c == '{')
{
isRegularExpressioning = true;
}
if (c == '\"')
{
isDoubleQuatationing = !isDoubleQuatationing;
str.Append(c);
continue;
}
if (c == '|' && !isDoubleQuatationing)
{
var cc = str.ToString().Trim();
if (cc != null)
{
exps.Add(cc);
expss.Add(exps = new List <string>()); // prepare next OR operator
}
str = new StringBuilder();
continue;
}
if (char.IsWhiteSpace(c) && !isDoubleQuatationing)
{
var cc = str.ToString().Trim();
if (cc != "")
{
exps.Add(cc);
}
str = new StringBuilder();
}
else
{
str.Append(c);
}
}
else
{
str.Append(c);
if (c == '}')
{
isRegularExpressioning = false;
exps.Add(str.ToString());
str = new StringBuilder();
}
}
}
if (str.ToString().Trim() != "")
{
exps.Add(str.ToString());
}
foreach (var strs in expss)
{
var resRemove = new List <Regex>();
_resRemoves.Add(resRemove);
var resCollect = new List <Regex>();
_resCollects.Add(resCollect);
var nStartMark = 0; // number of ＾ character
var nEndMark = 0; // number of ＄ character
foreach (var com in strs)
{
var com2 = StrMatch.ConvertSimpleAsterisk(com);
if (com2.StartsWith("^"))
{
nStartMark++;
}
if (com2.EndsWith("$") && com2.StartsWith("-") == false)
{
nEndMark++;
}
}
if (nStartMark > 1)
{
Debug.WriteLine("Warning : Found ^ mark twice or more");
}
if (nEndMark > 1)
{
Debug.WriteLine("Warning : Found $ mark twice of more");
}
foreach (var com in strs)
{
if (com.Length < 1)
{
continue;
}
if ((com.StartsWith("{") || com.StartsWith("-{")) && com.EndsWith("}"))
{
var re = new Regex(com.Substring(com.IndexOf('{') + 1, com.Length - com.IndexOf('{') - 2), reopt);
if (com[0] == '-')
{
resRemove.Add(re);
NRemove++;
}
else
{
resCollect.Add(re);
}
}
else
{
var isRemove = false;
string restr;
if (com.StartsWith("-"))
{
restr = com.Substring(1, com.Length - 1);
if (restr == "")
{
restr = " "; // Restore space character that removed with Trim function
}
isRemove = true;
}
else
{
restr = com;
}
// Remove double quotations
var preC = '\0';
var restr2 = new StringBuilder();
for (var i = 0; i < restr.Length; i++)
{
var c = restr[i];
if (c == '\"' && preC != '\\')
{
preC = c;
continue;
}
restr2.Append(c);
preC = c;
}
restr = restr2.ToString();
// Support simple asterisk expression
restr = StrMatch.ConvertSimpleAsterisk(restr);
// Change * or ? character to regular expression
restr = restr.Replace(".", "\\.");
restr = restr.Replace("*", ".*");
restr = restr.Replace("?", ".?");
// Kaizen
restr = Regex.Replace(restr, "&", "\\s*&\\s*"); // Not care space character around & mark
// remove consecutive space
int ll;
do
{
ll = restr.Length;
restr = restr.Replace(" ", " ");
} while (ll != restr.Length);
restr = restr.Replace(" ", " +");
if (restr.Length > 0)
{
try
{
var re = new Regex(restr, reopt);
if (isRemove)
{
resRemove.Add(re);
NRemove++;
}
else
{
resCollect.Add(re);
}
}
catch (Exception)
{
}
}
}
}
}
if (NRemove == 0)
{
_isAllMatch = IsMatch("龗鱻麤");
}
else
{
_isAllMatch = false;
}
}
