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( ));
}
Regex CreateHighlightingRegex(IcuRegexOptions options)
{
var pb = new PatternBuilder( );
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (options.UREGEX_COMMENTS)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.Add(@"\\Q.*?(\\E|$)"); // quoted part
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"\\[NpPx]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_brace'\{).*?((?'right_brace'\})|$)"); // '{...}'
string posix_bracket = @"(\[:.*?(:\]|$))"; // [:...:]
pb.Add($@"
(?'left_bracket'\[)
\]?
(?> {posix_bracket} | (?'left_bracket'\[)(?<c>) | (\\. | [^\[\]])+ | (?'right_bracket'\])(?<-c>))*
(?(c)(?!))
(?'right_bracket'\])?
|
(?'right_bracket'\])
" );
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
static Regex CreateHighlightingRegex(GrammarEnum grammar)
{
var pb = new PatternBuilder( );
if (grammar == GrammarEnum.extended ||
grammar == GrammarEnum.ECMAScript ||
grammar == GrammarEnum.egrep ||
grammar == GrammarEnum.awk)
{
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{).*?((?'right_brace'\})|$)"); // '{...}'
}
if (grammar == GrammarEnum.basic ||
grammar == GrammarEnum.grep)
{
pb.Add(@"(?'left_par'\\\()"); // '\)'
pb.Add(@"(?'right_par'\\\))"); // '\('
pb.Add(@"(?'left_brace'\\{).*?((?'right_brace'\\})|$)"); // '\{...\}'
}
pb.Add(@"((?'left_bracket'\[) ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
//public static implicit operator TokenFragment(PatternBuilder builder)
//{
// return new TokenPattern(builder.CreateArray());
//}
public static PatternBuilder operator +(PatternBuilder op1, PatternBuilder op2)
{
var result = new PatternBuilder();
for (var i = 0; i < op1._entryCount; i++)
{
result.Add(op1[i]);
}
for (var j = 0; j < op2._entryCount; j++)
{
result.Add(op2[j]);
}
return(result);
}
Regex CreateColouringRegex(bool isVerbose)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\x[0-9a-fA-F]{1,2}"); // hexa, two digits
pb_escape.Add(@"\\0[0-7]+"); // octal, after '0'
pb_escape.Add(@"\\[1-7][0-7]{2,}"); // octal, three digits
pb_escape.Add(@"\\N\{.+?(\} | $)"); // Unicode name, ex.: \N{DIGIT ONE}
pb_escape.Add(@"\\.");
pb_escape.EndGroup( );
//
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (isVerbose)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment*/
}
pb.EndGroup( );
//
pb.Add(@"\(\?P(?'name'<.*?(>|$))");
pb.Add(@"\(\?P=(?'name'.*?(\)|$))");
pb.Add(@"(?'name'\\[1-9][0-9]?(?![0-9]))");
//
pb.Add(pb_escape.ToPattern( ));
//
string char_group = @"( \[ \]? .*? (\]|$) )";
pb.Add(char_group);
//
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(DRegexOptions options)
{
var pb = new PatternBuilder( );
pb.Add(@"\(\?\#.*?(\)|$)"); // inline comment
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"\\[pP]\{.*?(\} | $)"); // property
pb.Add(@"(?'left_brace'\{) (\d+(,\d*)? | ,\d+) ((?'right_brace'\})|$)"); // '{...}'
string posix_bracket = ""; // Not supported: @"(\[:.*?(:\]|$))"; // [:...:]
pb.Add($@"
(?'left_bracket'\[)
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} (?'left_bracket'\[)(?<c>) | (\\. | [^\[\]])+ | (?'right_bracket'\])(?<-c>))*
(?(c)(?!))
(?'right_bracket'\])?
|
(?'right_bracket'\])
" );
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
Regex CreateColouringRegex(DRegexOptions options)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\c[A-Za-z]"); // Matches the control character corresponding to letter C
pb_escape.Add(@"\\x[0-9a-fA-F]{0,2}"); // Matches a character with hexadecimal value of XX.
pb_escape.Add(@"\\u[0-9a-fA-F]{0,4}"); // Matches a character with hexadecimal value of XXXX.
pb_escape.Add(@"\\U[0-9a-fA-F]{0,8}"); // Matches a character with hexadecimal value of YYYYYY.
pb_escape.Add(@"\\[pP]\{.*?(\}|$)");
pb_escape.Add(@"\\[pP].?");
pb_escape.Add(@"\\.");
pb_escape.EndGroup( );
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)"); // An inline comment that is ignored while matching.
pb.EndGroup( );
pb.Add(@"\(\?P(?'name'<.*?(>|$))"); // Matches named subexpression regex labeling it with name 'name'.
{
// (nested groups: https://stackoverflow.com/questions/546433/regular-expression-to-match-balanced-parentheses)
string posix_bracket = ""; // Not supported: @"(?'escape'\[:.*?(:\]|$))"; // [:...:], use escape colour
pb.Add($@"
\[
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} \[(?<c>) | ({pb_escape.ToPattern( )} | [^\[\]])+ | \](?<-c>))*
(?(c)(?!))
\]
" );
}
pb.Add(pb_escape.ToPattern( ));
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 CreateHighlightingRegex( )
{
var pb = new PatternBuilder( );
pb.Add(@"\\.");
pb.AddGroup("left_par", @"(?'left_par'\()"); // '('
pb.AddGroup("right_par", @"(?'right_par'\))"); // ')'
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(bool isXorXX)
{
var pb = new PatternBuilder( );
pb.Add(@"(\(\?\#.*?(\)|$))"); // comment
if (isXorXX)
{
pb.Add(@"(\#[^\n]*)"); // line comment
}
pb.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
pb.Add(@"\\[xNopPbBgk]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{) \s* \d+ \s* (,\s*\d*)? \s* ((?'right_brace'\})|$)"); // '{...}'
pb.Add(@"((?'left_bracket'\[) \]? ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
pb.Add(@"\\."); // '\...'
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 CreateHighlightingRegex( )
{
bool is_literal = OptionsControl.IsCompileOptionSelected("PCRE2_LITERAL");
if (is_literal)
{
return(PatternBuilder.AlwaysFailsRegex);
}
bool is_extended = OptionsControl.IsCompileOptionSelected("PCRE2_EXTENDED");
bool allow_empty_class = OptionsControl.IsCompileOptionSelected("PCRE2_ALLOW_EMPTY_CLASS");
var pb = new PatternBuilder( );
pb.Add(@"(\(\?\#.*?(\)|$))"); // comment
if (is_extended)
{
pb.Add(@"(\#[^\n]*)"); // line comment
}
pb.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
pb.Add(@"\\[oNxupP]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{) \d+ (,\d*)? ((?'right_brace'\})|$)"); // '{...}'
if (allow_empty_class)
{
pb.Add(@"((?'left_bracket'\[) ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
}
else
{
pb.Add(@"((?'left_bracket'\[) \]? ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
}
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(bool isVerbose)
{
var pb = new PatternBuilder( );
pb.Add(@"(\(\?\#.*?(\)|$))"); // comment
if (isVerbose)
{
pb.Add(@"(\#[^\n]*)"); // line comment
}
pb.Add(@"\\[N]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{) (\d+ | \d*,\d*) ((?'right_brace'\})|$)"); // '{...}'
pb.Add(@"((?'left_bracket'\[) ]? (\\. | .)*? ((?'right_bracket'\])|$) )"); // [...]
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
static Regex CreateHighlightingRegex(RegexOptions options)
{
var pb = new PatternBuilder( );
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (options.HasFlag(RegexOptions.IgnorePatternWhitespace))
{
pb.Add(@"\#[^\n]*"); // line comment
}
pb.Add(@"\\[pP]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{) \d+(,(\d+)?)? ((?'right_brace'\})|$)"); // '{...}'
pb.Add(@"(?'left_bracket'\[) \]? (\\.|.)*? ((?'right_bracket'\])|$)"); // '[...]'
pb.Add(@"\\."); // (skip)
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(WebView2RegexOptions options)
{
var pb = new PatternBuilder( );
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"\\[pPu]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_brace'\{).*?((?'right_brace'\})|$)"); // '{...}'
pb.Add(@"(?'left_bracket'\[) \]? (\\.|.)*? ((?'right_bracket'\])|$)"); // '[...]'
pb.Add(@"\\."); // (skip)
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(RustRegexOptions options)
{
bool is_regex_builder = options.@struct == "RegexBuilder";
var pb = new PatternBuilder( );
if (is_regex_builder && options.ignore_whitespace)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"\\[xuU]\{.*?(\}|$)"); // \x{7HHHHHHH ...} etc.
if (is_regex_builder && options.unicode)
{
pb.Add(@"\\[pP]\{.*?(\} | $)"); // property
}
pb.Add(@"(?'left_brace'\{) (\d+(,\d*)? | ,\d+) ((?'right_brace'\})|$)"); // '{...}'
string posix_bracket = @"(\[:.*?(:\]|$))"; // [:...:]
pb.Add($@"
(?'left_bracket'\[)
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} (?'left_bracket'\[)(?<c>) | (\\. | [^\[\]])+ | (?'right_bracket'\])(?<-c>))*
(?(c)(?!))
(?'right_bracket'\])?
|
(?'right_bracket'\])
" );
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
Regex CreateColouringRegex(IcuRegexOptions options)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\c[A-Za-z]"); // \cx control char
pb_escape.Add(@"\\[NpP]\{.*?(\} | $)"); // named character, property
pb_escape.Add(@"\\[uUx][0-9a-fA-F]+"); // hexadecimal char
pb_escape.Add(@"\\x\{[0-9a-fA-F]+(\}|$)"); // hexadecimal char
pb_escape.Add(@"\\0[0-7]+"); // octal
pb_escape.Add(@"\\Q.*?(\\E|$)"); // quoted part
pb_escape.Add(@"\\."); // \.
pb_escape.EndGroup( );
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (options.UREGEX_COMMENTS)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.EndGroup( );
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))");
pb.Add(@"(?'name'\\k<.*?(>|$))");
string posix_bracket = @"(?'escape'\[:.*?(:\]|$))"; // [:...:], use escape colour
pb.Add($@"
\[
\]?
(?> {posix_bracket} | \[(?<c>) | ({pb_escape.ToPattern( )} | [^\[\]])+ | \](?<-c>))*
(?(c)(?!))
\]
" );
pb.Add(pb_escape.ToPattern( ));
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex( )
{
bool is_literal = OptionsControl.IsOptionSelected("literal");
if (is_literal)
{
return(PatternBuilder.AlwaysFailsRegex);
}
var pb = new PatternBuilder( );
pb.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
pb.Add(@"\\[pPx]\{.*?(\}|$)"); // (skip)
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
pb.Add(@"(?'left_brace'\{) \d+ (,\d*)* ((?'right_brace'\})|$)"); // '{...}'
pb.Add(@"((?'left_bracket'\[) \]? ((\[:.*? (:\]|$)) | \\. | .)*? ((?'right_bracket'\])|$) )");
pb.Add(@"\\."); // '\...'
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'.*?('|$))");
pb.Add(escapes_pb.ToPattern( ));
var regex = pb.ToRegex( );
return(regex);
}
Regex CreateColouringRegex(RustRegexOptions options)
{
bool is_regex = options.@struct == "Regex";
bool is_regex_builder = options.@struct == "RegexBuilder";
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
if (is_regex || (is_regex_builder && options.unicode))
{
pb_escape.Add(@"\\[pP]\{.*?(\}|$)"); // Unicode character class (general category or script)
pb_escape.Add(@"\\[pP].?"); // One-letter name Unicode character class
}
if (is_regex_builder && options.octal)
{
pb_escape.Add(@"\\[0-7]{1,3}"); // octal character code (up to three digits) (when enabled)
}
pb_escape.Add(@"\\x\{[0-9a-fA-F]*(\}|$)?"); // any hex character code corresponding to a Unicode code point
pb_escape.Add(@"\\x[0-9a-fA-F]{0,2}"); // hex character code (exactly two digits)
// (only 2 digits if no 'options.unicode'
pb_escape.Add(@"\\u\{[0-9a-fA-F]*(\}|$)?"); // any hex character code corresponding to a Unicode code point
pb_escape.Add(@"\\u[0-9a-fA-F]{0,4}"); // hex character code (exactly four digits)
pb_escape.Add(@"\\U\{[0-9a-fA-F]*(\}|$)?"); // any hex character code corresponding to a Unicode code point
pb_escape.Add(@"\\U[0-9a-fA-F]{0,8}"); // hex character code (exactly eight digits)
string any_esc = "";
if (!(is_regex || (is_regex_builder && options.unicode)))
{
any_esc += @"(?!\\[pP])";
}
if (!(is_regex_builder && options.octal))
{
any_esc += @"(?!\\[0-7])";
}
any_esc += @"\\.";
pb_escape.Add(any_esc);
pb_escape.EndGroup( );
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
if (is_regex_builder && options.ignore_whitespace)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.EndGroup( );
pb.Add(@"\(\?P(?'name'<.*?(>|$))");
{
// (nested groups: https://stackoverflow.com/questions/546433/regular-expression-to-match-balanced-parentheses)
string posix_bracket = @"(?'escape'\[:.*?(:\]|$))"; // [:...:], use escape colour
pb.Add($@"
\[
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} \[(?<c>) | ({pb_escape.ToPattern( )} | [^\[\]])+ | \](?<-c>))*
(?(c)(?!))
\]
" );
}
pb.Add(pb_escape.ToPattern( ));
return(pb.ToRegex( ));
}
Regex CreateColouringRegex( )
{
bool is_literal = OptionsControl.IsCompileOptionSelected("PCRE2_LITERAL");
if (is_literal)
{
return(PatternBuilder.AlwaysFailsRegex);
}
bool is_extended = OptionsControl.IsCompileOptionSelected("PCRE2_EXTENDED");
bool allow_empty_class = OptionsControl.IsCompileOptionSelected("PCRE2_ALLOW_EMPTY_CLASS");
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\c[A-Za-z]"); // ASCII escape
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(@"\\o\{[0-9]+(\} | $)"); // octal; bad values give error
pb_escape.Add(@"\\N\{U\+[0-9a-fA-F]+(\} | $)"); // hexa, error if no 'PCRE2_UTF'
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(@"\\u[0-9a-fA-F]{1,4}"); // hexa, four digits, error if no 'PCRE2_ALT_BSUX', 'PCRE2_EXTRA_ALT_BSUX'
pb_escape.Add(@"\\u\{[0-9a-fA-F]*(\} | $)"); // hexa, error if empty or no 'PCRE2_ALT_BSUX', 'PCRE2_EXTRA_ALT_BSUX'
pb_escape.Add(@"\\[pP]\{.*?(\} | $)"); // property
pb_escape.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
// backreferences
pb_escape.Add(@"\\[0-9]+"); // unbiguous
// see also named groups
pb_escape.Add(@"\\.");
pb_escape.EndGroup( );
//
var pb_class = new PatternBuilder( ).AddGroup("class", @"\[(?'c'[:=.]) .*? (\k<c>\] | $)");
//
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (is_extended)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.EndGroup( );
//
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))");
pb.Add(@"\(\?(?'name''.*?('|$))");
pb.Add(@"\(\?P(?'name'<.*?(>|$))");
pb.Add(@"(?'name'\\g[+]?[0-9]+)");
pb.Add(@"(?'name'\\g\{[+]?[0-9]*(\} | $))");
pb.Add(@"(?'name'\\[gk]<.*?(>|$))");
pb.Add(@"(?'name'\\[gk]'.*?('|$))");
pb.Add(@"(?'name'\\[gk]\{.*?(\}|$))");
pb.Add(@"(?'name'\(\?P=.*?(\)|$))"); //
//
pb.Add(pb_escape.ToPattern( ));
//
string char_group;
if (allow_empty_class)
{
char_group = $@"\[ ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$)";
}
else
{
char_group = $@"\[ \]? ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$)";
}
pb.Add(char_group);
// TODO: add support for '(*...)' constructs
return(pb.ToRegex( ));
}
Regex CreateHighlightingRegex(OnigurumaRegexInterop.OnigurumaHelper helper)
{
if (helper.IsONIG_SYNTAX_ASIS)
{
return(PatternBuilder.AlwaysFailsRegex);
}
var pb = new PatternBuilder( );
if (helper.IsONIG_SYN_OP2_QMARK_GROUP_EFFECT)
{
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
}
if (helper.IsONIG_OPTION_EXTEND)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
if (helper.IsONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)
{
pb.Add(@"\\Q.*?(\\E|$)"); // quoted part
}
if (helper.IsONIG_SYN_OP_LPAREN_SUBEXP)
{
pb.Add(@"(?'left_par'\()"); // '('
pb.Add(@"(?'right_par'\))"); // ')'
}
if (helper.IsONIG_SYN_OP_ESC_LPAREN_SUBEXP)
{
pb.Add(@"(?'left_par'\\\()"); // '\('
pb.Add(@"(?'right_par'\\\))"); // '\)'
}
if (helper.IsONIG_SYN_OP_ESC_O_BRACE_OCTAL)
{
pb.Add(@"\\o\{.*?(\}|$)"); // \o{17777777777 ...} wide octal chars
}
if (helper.IsONIG_SYN_OP_ESC_X_BRACE_HEX8)
{
pb.Add(@"\\x\{.*?(\}|$)"); // \x{7HHHHHHH ...} wide hexadecimal chars
}
if (helper.IsONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY || helper.IsONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT)
{
pb.Add(@"\\[pP]\{.*?(\} | $)"); // property
}
if (helper.IsONIG_SYN_OP_BRACE_INTERVAL)
{
pb.Add(@"(?'left_brace'\{) (\d+(,\d*)? | ,\d+) ((?'right_brace'\})|$)"); // '{...}'
}
if (helper.IsONIG_SYN_OP_ESC_BRACE_INTERVAL)
{
pb.Add(@"(?'left_brace'\\{).*?((?'right_brace'\\})|$)"); // '\{...\}'
}
string posix_bracket = "";
if (helper.IsONIG_SYN_OP_POSIX_BRACKET)
{
posix_bracket = @"(\[:.*?(:\]|$))"; // [:...:]
}
if (helper.IsONIG_SYN_OP_BRACKET_CC)
{
pb.Add($@"
(?'left_bracket'\[)
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} (?'left_bracket'\[)(?<c>) | (\\. | [^\[\]])+ | (?'right_bracket'\])(?<-c>))*
(?(c)(?!))
(?'right_bracket'\])?
|
(?'right_bracket'\])
" );
}
pb.Add(@"\\."); // '\...'
return(pb.ToRegex( ));
}
Regex CreateColouringRegex(OnigurumaRegexInterop.OnigurumaHelper helper)
{
if (helper.IsONIG_SYNTAX_ASIS)
{
return(PatternBuilder.AlwaysFailsRegex);
}
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\0[0-7]{1,2}"); // octal, two digits after 0
pb_escape.Add(@"\\[0-7]{1,3}"); // octal, three digits
if (helper.IsONIG_SYN_OP_ESC_O_BRACE_OCTAL)
{
pb_escape.Add(@"\\o\{[0-7]+ (\s+ [0-7]+)* (\}|$)"); // \o{17777777777 ...} wide octal chars
}
pb_escape.Add(@"\\u[0-9a-fA-F]+"); // \uHHHH wide hexadecimal char
if (helper.IsONIG_SYN_OP_ESC_X_HEX2)
{
pb_escape.Add(@"\\x[0-9a-fA-F]+"); // \xHH hexadecimal char
}
if (helper.IsONIG_SYN_OP_ESC_X_BRACE_HEX8)
{
pb_escape.Add(@"\\x\{[0-9a-fA-F]+ (\s+ [0-9a-fA-F]+)* (\}|$)"); // \x{7HHHHHHH ...} wide hexadecimal chars
}
if (helper.IsONIG_SYN_OP_ESC_C_CONTROL)
{
pb_escape.Add(@"\\c[A-Za-z]"); // \cx control char
pb_escape.Add(@"\\C-([A-Za-z])?"); // \C-x control char
}
pb_escape.Add(@"\\M-([A-Za-z])?"); // \M-x meta (x|0x80)
pb_escape.Add(@"\\M-(\\C-([A-Za-z])?)?"); // \M-x meta control char
pb_escape.Add(@"\\[pP]\{.*?(\} | $)"); // property
/*
* Probably not useful
*
* if( helper.IsONIG_SYN_OP_ESC_ASTERISK_ZERO_INF )
* {
* pb_escape.Add( @"(?!\\\*)");
* }
*
* if( helper.IsONIG_SYN_OP_ESC_PLUS_ONE_INF )
* {
* pb_escape.Add( @"(?!\\\+)");
* }
*
* if( helper.IsONIG_SYN_OP_ESC_QMARK_ZERO_ONE )
* {
* pb_escape.Add( @"(?!\\\?)");
* }
*
* if( helper.IsONIG_SYN_OP_ESC_BRACE_INTERVAL )
* {
* pb_escape.Add( @"(?!\\[{}])");
* }
*/
pb_escape.Add(@"\\.");
if (helper.IsONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE)
{
pb_escape.Add(@"\\Q.*?(\\E|$)"); // quoted part; use 'escape' name to take its colour
}
pb_escape.EndGroup( );
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
if (helper.IsONIG_SYN_OP2_QMARK_GROUP_EFFECT)
{
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
}
if (helper.IsONIG_OPTION_EXTEND)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment
}
pb.EndGroup( );
if (helper.IsONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)
{
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))");
pb.Add(@"\(\?(?'name''.*?('|$))");
}
if (helper.IsONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY)
{
pb.Add(@"\(\?@(?'name'<.*?(>|$))");
pb.Add(@"\(\?@(?'name''.*?('|$))");
}
if (helper.IsONIG_SYN_OP2_ESC_K_NAMED_BACKREF)
{
pb.Add(@"(?'name'\\k<.*?(>|$))");
pb.Add(@"(?'name'\\k'.*?('|$))");;
}
if (helper.IsONIG_SYN_OP2_ESC_G_SUBEXP_CALL)
{
pb.Add(@"(?'name'\\g<.*?(>|$))");
pb.Add(@"(?'name'\\g'.*?('|$))");
}
// (nested groups: https://stackoverflow.com/questions/546433/regular-expression-to-match-balanced-parentheses)
string posix_bracket = "";
if (helper.IsONIG_SYN_OP_POSIX_BRACKET)
{
posix_bracket = @"(?'escape'\[:.*?(:\]|$))"; // [:...:], use escape colour
}
pb.Add($@"
\[
\]?
(?> {posix_bracket}{( posix_bracket.Length == 0 ? "" : " |" )} \[(?<c>) | ({pb_escape.ToPattern( )} | [^\[\]])+ | \](?<-c>))*
(?(c)(?!))
\]
" );
if (helper.IsONIG_SYN_OP_ESC_LPAREN_SUBEXP)
{
pb.Add(@"\\\( | \\\)"); // (skip)
}
if (helper.IsONIG_SYN_OP_ESC_BRACE_INTERVAL)
{
pb.Add(@"\\\{ | \\\}"); // (skip)
}
pb.Add(pb_escape.ToPattern( ));
return(pb.ToRegex( ));
}
Regex CreateColouringRegex(bool isXorXx)
{
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
pb_escape.Add(@"\\c[A-Za-z]"); // control char
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(@"\\N\{.*?(\} | $)"); // Unicode name or hexa
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(@"\\o\{[0-9]+(\} | $)"); // octal
pb_escape.Add(@"\\[pP]([a-zA-Z] | $)"); // property
pb_escape.Add(@"\\[pP]\{.*?(\} | $)"); // property
pb_escape.Add(@"\\Q.*?(\\E|$)"); // quoted sequence, \Q...\E
pb_escape.Add(@"\\[bB]\{.*?(\} | $)"); // Unicode boundary
pb_escape.Add(@"\\.");
pb_escape.EndGroup( );
//
var pb_class = new PatternBuilder( ).AddGroup("class", @"\[(?'c'[:]) .*? (\k<c>\] | $)"); // [: ... :]
//
var pb = new PatternBuilder( );
pb.BeginGroup("comment");
pb.Add(@"\(\?\#.*?(\)|$)"); // comment
if (isXorXx)
{
pb.Add(@"\#.*?(\n|$)"); // line-comment*/
}
pb.EndGroup( );
//
pb.Add(@"\(\?(?'name'<(?![=!]).*?(>|$))");
pb.Add(@"\(\?(?'name''.*?('|$))");
pb.Add(@"\(\?P(?'name'<.*?(>|$))");
pb.Add(@"\(\?P(?'name'[=>].*?(\)|$))");
pb.Add(@"(?'name'\\g[0-9]+)");
pb.Add(@"(?'name'\\[gk]\{.*?(\}|$))");
pb.Add(@"(?'name'\\[gk]<.*?(>|$))");
pb.Add(@"(?'name'\\k'.*?('|$))");
//
pb.Add(pb_escape.ToPattern( ));
//
string char_group = $@"( \[ \]? ({pb_class.ToPattern( )} | {pb_escape.ToPattern( )} | . )*? (\]|$) )";
pb.Add(char_group);
//
return(pb.ToRegex( ));
}
static Regex CreateColouringRegex(GrammarEnum grammar, bool modX)
{
bool is_perl =
grammar == GrammarEnum.perl ||
grammar == GrammarEnum.ECMAScript ||
grammar == GrammarEnum.normal ||
grammar == GrammarEnum.JavaScript ||
grammar == GrammarEnum.JScript;
bool is_POSIX_extended =
grammar == GrammarEnum.extended ||
grammar == GrammarEnum.egrep ||
grammar == GrammarEnum.awk;
bool is_POSIX_basic =
grammar == GrammarEnum.basic ||
grammar == GrammarEnum.sed ||
grammar == GrammarEnum.grep ||
grammar == GrammarEnum.emacs;
bool is_emacs =
grammar == GrammarEnum.emacs;
var pb_escape = new PatternBuilder( );
pb_escape.BeginGroup("escape");
if (is_perl || is_POSIX_extended || is_POSIX_basic)
{
pb_escape.Add(@"\\[1-9]"); // back reference
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\c[A-Za-z]"); // ASCII escape
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\x[0-9A-Fa-f]{1,2}"); // hex, two digits
}
if (is_perl || is_POSIX_extended)
{
pb_escape.Add(@"\\x\{[0-9A-Fa-f]+(\}|$)"); // 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( ));
}
internal void AddSpecial(string input, string internal_)
{
pattern.Add(input, internal_);
}
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( ));
}
