static Regex CreateHighlightingRegex( )
{
var pb = new PatternBuilder( );
pb.Add(@"\\.");
pb.AddGroup("left_par", @"(?'left_par'\()"); // '('
pb.AddGroup("right_par", @"(?'right_par'\))"); // ')'
return(pb.ToRegex( ));
}
Regex CreateColouringRegex(WebView2RegexOptions options)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\c[A-Za-z]"); // \cx control char
pb_escape.Add(@"\\x[0-9a-fA-F]{1,2}"); // hexadecimal char
pb_escape.Add(@"\\u[0-9a-fA-F]{1,4}"); // hexadecimal char
if (options.u)
{
// language=regex
pb_escape.Add(@"\\u\{[0-9a-fA-F]+(\}|$)"); // hexadecimal char
// language=regex
pb_escape.Add(@"\\(p|P)\{.*?(\}|$)"); // unicode property
}
pb_escape.Add(@"\\."); // \.
pb_escape.EndGroup( );
var pb = new PatternBuilder( );
pb.AddGroup(null, $@"\[\]?({pb_escape.ToPattern( )} |.)*?(\]|$)");
// language=regex
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))");
// language=regex
pb.Add(@"(?'name'\\k<.*?(>|$))");
pb.Add(pb_escape.ToPattern( ));
return(pb.ToRegex( ));
}
static Regex CreateColouringRegex(GrammarEnum grammar)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
if (grammar == GrammarEnum.ECMAScript)
{
pb_escape.Add(@"\\c[A-Za-z]");
}
if (grammar == GrammarEnum.ECMAScript)
{
pb_escape.Add(@"\\x[0-9A-Fa-f]{1,2}"); // (two digits required)
}
if (grammar == GrammarEnum.awk)
{
pb_escape.Add(@"\\[0-7]{1,3}"); // octal code
}
if (grammar == GrammarEnum.ECMAScript)
{
pb_escape.Add(@"\\u[0-9A-Fa-f]{1,4}"); // (four digits required)
}
if (grammar == GrammarEnum.basic ||
grammar == GrammarEnum.grep)
{
pb_escape.Add(@"(?!\\\( | \\\) | \\\{ | \\\})\\.");
}
else
{
pb_escape.Add(@"\\.");
}
pb_escape.EndGroup( );
//
var pb_class = new PatternBuilder( ).AddGroup("class", @"\[(?'c'[:=.]) .*? (\k<c>\] | $)");
//
var pb = new PatternBuilder( );
pb.Add(pb_escape.ToPattern( ));
pb.AddGroup(null, $@"( \[ ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$) )");
// (group names and comments are not supported by C++ Regex)
return(pb.ToRegex( ));
}
Regex CreateColouringRegex( )
{
bool is_literal = OptionsControl.IsOptionSelected("literal");
if (is_literal)
{
return(PatternBuilder.AlwaysFailsRegex);
}
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\[pP][A-Za-z]"); // Unicode character class (one-letter name)
pb_escape.Add(@"\\[pP]\{.*?(\}|$)"); // Unicode character class
pb_escape.Add(@"\\0[0-7]{1,2}"); // octal, two digits after 0
pb_escape.Add(@"\\[0-7]{1,3}"); // octal, three digits
pb_escape.Add(@"\\x[0-9a-fA-F]{1,2}"); // hexa, two digits
pb_escape.Add(@"\\x\{[0-9a-fA-F]*(\}|$)"); // hexa, error if empty
pb_escape.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
pb_escape.Add(@"\\.");
pb_escape.EndGroup( );
//
var pb_class = new PatternBuilder( ).AddGroup("class", @"\[(?'c'[:]) .*? (\k<c>\] | $)"); // only [: :], no [= =], no [. .]
//
var pb = new PatternBuilder( );
pb.Add(pb_escape.ToPattern( ));
//
pb.AddGroup(null, $@"\[ \]? ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$)"); // TODO: check 'escape' part
//
pb.Add(@"\(\?P(?'name'<.*?>)");
return(pb.ToRegex( ));
}
static Regex CreateCachedColouringRegex(RegexOptions options)
{
// (some patterns includes incomplete constructs)
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)");
if (options.HasFlag(RegexOptions.IgnorePatternWhitespace))
{
pb.Add(@"\#[^\n]*");
}
pb.EndGroup( );
var escapes_pb = new PatternBuilder( );
escapes_pb.BeginGroup("escape");
escapes_pb.Add(@"\\[0-7]{2,3}");
escapes_pb.Add(@"\\x[0-9A-Fa-f]{1,2}");
escapes_pb.Add(@"\\c[A-Za-z]");
escapes_pb.Add(@"\\u[0-9A-Fa-f]{1,4}");
escapes_pb.Add(@"\\(p|P)\{.*?(\}|$)");
escapes_pb.Add(@"\\k<([A-Za-z]+>)?");
escapes_pb.Add(@"\\.");
escapes_pb.EndGroup( );
pb.AddGroup(null, $@"\[\]?({escapes_pb.ToPattern( )} |.)*?(\]|$)");
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))"); // (balancing groups covered too)
pb.Add(@"\(\?(?'name''.*?('|$))");
pb.Add(@"(?'name'\\k<.*?(>|$))");
pb.Add(@"(?'name'\\k'.*?('|$))");
pb.Add(escapes_pb.ToPattern( ));
var regex = pb.ToRegex( );
return(regex);
}
static Regex CreateHighlightingRegex(GrammarEnum grammar, bool modX)
{
bool is_perl =
grammar == GrammarEnum.perl ||
grammar == GrammarEnum.ECMAScript ||
grammar == GrammarEnum.normal ||
grammar == GrammarEnum.JavaScript ||
grammar == GrammarEnum.JScript;
bool is_POSIX_extended =
grammar == GrammarEnum.extended ||
grammar == GrammarEnum.egrep ||
grammar == GrammarEnum.awk;
bool is_POSIX_basic =
grammar == GrammarEnum.basic ||
grammar == GrammarEnum.sed ||
grammar == GrammarEnum.grep ||
grammar == GrammarEnum.emacs;
bool is_emacs =
grammar == GrammarEnum.emacs;
var pb = new PatternBuilder( );
if (is_perl)
{
pb.Add(@"(\(\?\#.*?(\)|$))"); // comment
}
if (is_perl && modX)
{
pb.Add(@"(\#[^\n]*)"); // line comment
}
if (is_perl || is_POSIX_extended)
{
pb.Add(@"\\Q.*?(\\E|$)"); // skip \Q...\E
pb.Add(@"\\[xNpPgk]\{.*?(\}|$)"); // (skip)
}
if (is_perl || is_POSIX_extended)
{
pb.AddGroup("left_par", @"\("); // '('
pb.AddGroup("right_par", @"\)"); // ')'
pb.Add(@"(?'left_brace'\{) \s* \d+ \s* (, \s* \d*)? \s* ((?'right_brace'\})|$)"); // '{...}' (spaces are allowed)
}
if (is_POSIX_basic)
{
pb.AddGroup("left_par", @"\\\("); // '\('
pb.AddGroup("right_par", @"\\\)"); // '\)'
pb.Add(@"(?'left_brace'\\{).*?((?'right_brace'\\})|$)"); // '\{...\}'
}
if (is_perl || is_POSIX_extended || is_POSIX_basic)
{
pb.Add(@"((?'left_bracket'\[) \]? ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
pb.Add(@"\\."); // '\...'
}
return(pb.ToRegex( ));
}
static Regex CreateColouringRegex(GrammarEnum grammar, bool modX)
{
bool is_perl =
grammar == GrammarEnum.perl ||
grammar == GrammarEnum.ECMAScript ||
grammar == GrammarEnum.normal ||
grammar == GrammarEnum.JavaScript ||
grammar == GrammarEnum.JScript;
bool is_POSIX_extended =
grammar == GrammarEnum.extended ||
grammar == GrammarEnum.egrep ||
grammar == GrammarEnum.awk;
bool is_POSIX_basic =
grammar == GrammarEnum.basic ||
grammar == GrammarEnum.sed ||
grammar == GrammarEnum.grep ||
grammar == GrammarEnum.emacs;
bool is_emacs =
grammar == GrammarEnum.emacs;
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
if (is_perl || is_POSIX_extended || is_POSIX_basic)
{
pb_escape.Add(@"\\[1-9]"); // back reference
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\c[A-Za-z]"); // ASCII escape
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\x[0-9A-Fa-f]{1,2}"); // hex, two digits
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\x\{[0-9A-Fa-f]+(\}|$)"); // hex, four digits
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\0[0-7]{1,3}"); // octal, three digits
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\N\{.*?(\}|$)"); // symbolic name
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\[pP]\{.*?(\}|$)"); // property
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\[pP]."); // property, short name
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\Q.*?(\\E|$)");
}
; // quoted sequence
if (is_emacs)
{
pb_escape.Add(@"\\[sS]."); // syntax group
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\."); // various
}
if (is_POSIX_basic)
{
pb_escape.Add(@"(?!\\\( | \\\) | \\\{ | \\\})\\."); // various
}
pb_escape.EndGroup( );
var pb_class = new PatternBuilder( );
pb_class.BeginGroup("class");
if (is_perl || is_POSIX_extended || is_POSIX_basic)
{
pb_class.Add(@"\[(?'c'[:=.]) .*? (\k<c>\] | $)");
}
pb_class.EndGroup( );
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
if (is_perl)
{
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
}
if (is_perl && modX)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment*/
}
pb.EndGroup( );
if (is_perl)
{
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$)) | \(\?(?'name''.*?('|$))");
}
if (is_perl)
{
pb.Add(@"(?'name'\\g-?[1-9]) | (?'name'\\g\{.*?(\}|$))"); // back reference
}
if (is_perl)
{
pb.Add(@"(?'name'\\[gk]<.*?(>|$)) | (?'name'\\[gk]'.*?('|$))"); // back reference
}
if (is_perl || is_POSIX_extended || is_POSIX_basic)
{
pb.AddGroup(null, $@"\[ \]? ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$)");
}
pb.Add(pb_escape.ToPattern( ));
return(pb.ToRegex( ));
}