public RegularExpression(List<Expression>/*!*/ pattern, RubyRegexOptions options, bool isCondition, SourceSpan location)
: base(location) {
Assert.NotNull(pattern);
_pattern = pattern;
_options = options;
}
internal static string Transform(string/*!*/ rubyPattern, RubyRegexOptions options) {
if (rubyPattern == "\\Af(?=[[:xdigit:]]{2}+\\z)") {
// pp.rb uses this pattern. The real fix requires cracking the entire regexp and so is left for later
return "\\Af(?=(?:[[:xdigit:]]{2})+\\z)";
}
RegexpTransformer transformer = new RegexpTransformer(rubyPattern);
return transformer.Transform();
}
internal static string Transform(string/*!*/ rubyPattern, RubyRegexOptions options, out bool hasGAnchor) {
if (rubyPattern == "\\Af(?=[[:xdigit:]]{2}+\\z)") {
// pp.rb uses this pattern. The real fix requires cracking the entire regexp and so is left for later
hasGAnchor = false;
return "\\Af(?=(?:[[:xdigit:]]{2})+\\z)";
}
RegexpTransformer transformer = new RegexpTransformer(rubyPattern);
var result = transformer.Transform();
hasGAnchor = transformer._hasGAnchor;
return result;
}
internal static string Transform(string/*!*/ rubyPattern, RubyRegexOptions options, out bool hasGAnchor) {
// TODO: surrogates (REXML uses this pattern)
if (rubyPattern == "^[\t\n\r -\uD7FF\uE000-\uFFFD\uD800\uDC00-\uDBFF\uDFFF]*$") {
hasGAnchor = false;
return "^(?:[\t\n\r -\uD7FF\uE000-\uFFFD]|[\uD800-\uDBFF][\uDC00-\uDFFF])*$";
}
RegexpTransformer transformer = new RegexpTransformer(rubyPattern);
var result = transformer.Transform();
hasGAnchor = transformer._hasGAnchor;
return result;
}
public static RegexOptions ToClrOptions(RubyRegexOptions options)
{
RegexOptions result = RegexOptions.Multiline;
if ((options & RubyRegexOptions.IgnoreCase) != 0)
{
result |= RegexOptions.IgnoreCase;
}
if ((options & RubyRegexOptions.Extended) != 0)
{
result |= RegexOptions.IgnorePatternWhitespace;
}
if ((options & RubyRegexOptions.Multiline) != 0)
{
result |= RegexOptions.Singleline;
}
return(result);
}
public void Set(MutableString /*!*/ pattern, RubyRegexOptions options)
{
ContractUtils.RequiresNotNull(pattern, "pattern");
// RubyRegexOptions.Once is only used to determine how the Regexp object should be created and cached.
// It is not a property of the final object. /foo/ should compare equal with /foo/o.
_options = options & ~RubyRegexOptions.Once;
RubyEncoding encoding = RubyEncoding.GetRegexEncoding(options);
if (encoding != null)
{
_pattern = MutableString.CreateBinary(pattern.ToByteArray(), encoding ?? RubyEncoding.Binary).Freeze();
}
else
{
_pattern = pattern.PrepareForCharacterRead().Clone().Freeze();
}
TransformPattern(encoding, options & RubyRegexOptions.EncodingMask);
}
//[DebuggerHidden]
private void TestCorrectPatternTranslation(string /*!*/ pattern, RubyRegexOptions options, string /*!*/ expected, bool expectedGAnchor, IList <string> expectedWarnings)
{
bool hasGAnchor;
List <string> warnings;
string actual = RegexpTransformer.Transform(pattern, options, out hasGAnchor, out warnings);
AreEqual(expected, actual);
new Regex(expected);
Assert(hasGAnchor == expectedGAnchor);
if (warnings != null)
{
Assert(expectedWarnings != null);
Assert(expectedWarnings.Count == warnings.Count);
foreach (var warning in expectedWarnings)
{
Assert(warnings.IndexOf(warning) >= 0);
}
}
else
{
Assert(expectedWarnings == null || expectedWarnings.Count == 0);
}
}
private Regex /*!*/ TransformPattern(RubyEncoding encoding, RubyRegexOptions kc)
{
// We can reuse cached CLR regex if it was created for the same k-coding:
if (_cachedRegex != null && kc == _cachedKCode)
{
return(_cachedRegex);
}
string pattern;
if (kc != 0 || encoding == RubyEncoding.Binary)
{
pattern = _pattern.ToString(encoding.Encoding);
}
else
{
pattern = _pattern.ConvertToString();
}
List <string> warnings;
Regex result;
try {
result = new Regex(RegexpTransformer.Transform(pattern, _options, out _hasGAnchor, out warnings), ToClrOptions(_options));
} catch (Exception e) {
throw new RegexpError(e.Message);
}
// TODO: report
//if (warnings != null) {
// ...
//}
_cachedKCode = kc;
_cachedRegex = result;
return(result);
}
internal protected BinaryRegex(byte[]/*!*/ pattern, RubyRegexOptions options)
: base(options) {
Assert.NotNull(pattern);
_pattern = ArrayUtils.Copy(pattern);
}
internal protected StringRegex(string/*!*/ pattern, RubyRegexOptions options)
: base(options) {
Assert.NotNull(pattern);
_pattern = pattern;
string transformed = RegexpTransformer.Transform(pattern, options);
try {
_regex = new Regex(transformed, RubyRegex.ToClrOptions(options));
} catch (ArgumentException e) {
Utils.Log("-- original ---" + new String('-', 50), "REGEX_ERROR");
Utils.Log(pattern, "REGEX_ERROR");
Utils.Log("-- transformed " + new String('-', 50), "REGEX_ERROR");
Utils.Log(transformed, "REGEX_ERROR");
Utils.Log("---------------" + new String('-', 50), "REGEX_ERROR");
throw new RegexpError(e.Message, e);
}
}
internal void SetRegexOptions(RubyRegexOptions value) {
Integer1 = (int)value;
}
private RegexpTransformer(string/*!*/ rubyPattern, RubyRegexOptions options) {
_rubyPattern = rubyPattern;
_options = options;
}
private static RubyRegex/*!*/ CreateRegexWorker(
RubyRegexOptions options,
StrongBox<RubyRegex> regexpCache,
bool isLiteralWithoutSubstitutions,
Func<RubyRegex> createRegex) {
try {
bool once = ((options & RubyRegexOptions.Once) == RubyRegexOptions.Once) || isLiteralWithoutSubstitutions;
if (once) {
// Note that the user is responsible for thread synchronization
if (regexpCache.Value == null) {
regexpCache.Value = createRegex();
}
return regexpCache.Value;
} else {
// In the future, we can consider caching the last Regexp. For some regexp literals
// with substitution, the substition will be the same most of the time
return createRegex();
}
} catch (RegexpError e) {
if (isLiteralWithoutSubstitutions) {
// Ideally, this should be thrown during parsing of the source, even if the
// expression happens to be unreachable at runtime.
throw new SyntaxError(e.Message);
} else {
throw;
}
}
}
public override GenericRegex /*!*/ ToRegularExpression(RubyRegexOptions options)
{
return(new StringRegex(_data, options));
}
public RegularExpression(List<Expression>/*!*/ pattern, RubyRegexOptions options, SourceSpan location)
: this(pattern, options, false, location) {
}
public static RubyRegex/*!*/ CreateRegexE(string/*!*/ str1, int codepage, RubyRegexOptions options) {
return new RubyRegex(str1, options);
}
public static RubyRegex/*!*/ CreateRegexU(string/*!*/ str1, RubyRegexOptions options) {
return new RubyRegex(str1, options);
}
public static RubyRegex/*!*/ CreateRegexN(object[]/*!*/ strings, int codepage, RubyRegexOptions options) {
return new RubyRegex(ConcatenateStrings(strings), options);
}
public static RubyRegex/*!*/ CreateRegexME(MutableString str1, string/*!*/ str2, int codepage, RubyRegexOptions options) {
return new RubyRegex(MutableString.CreateInternal(str1).Append(str2), options);
}
public static RubyRegex/*!*/ CreateRegexUM(string/*!*/ str1, MutableString str2, RubyRegexOptions options) {
return new RubyRegex(MutableString.Create(str1).Append(str2), options);
}
public RubyRegex(MutableString /*!*/ pattern, RubyRegexOptions options)
{
ContractUtils.RequiresNotNull(pattern, "pattern");
_regex = pattern.ToRegularExpression(options);
}
public RubyRegex()
{
_pattern = MutableString.CreateEmpty();
_options = RubyRegexOptions.NONE;
}
internal void SetRegexOptions(RubyRegexOptions value)
{
Integer1 = (int)value;
}
public static RubyRegex/*!*/ CreateRegexM(MutableString str1, RubyRegexOptions options) {
return new RubyRegex(MutableString.CreateInternal(str1), options);
}
internal static string/*!*/ TransformPattern(string/*!*/ pattern, RubyRegexOptions options) {
return RegexpTransformer.Transform(pattern, options);
}
//[DebuggerHidden]
private void TestCorrectPatternTranslation(string/*!*/ pattern, RubyRegexOptions options, string/*!*/ expected, bool expectedGAnchor, IList<string> expectedWarnings)
{
bool hasGAnchor;
List<string> warnings;
string actual = RegexpTransformer.Transform(pattern, options, out hasGAnchor, out warnings);
AreEqual(expected, actual);
new Regex(expected);
Assert(hasGAnchor == expectedGAnchor);
if (warnings != null) {
Assert(expectedWarnings != null);
Assert(expectedWarnings.Count == warnings.Count);
foreach (var warning in expectedWarnings) {
Assert(warnings.IndexOf(warning) >= 0);
}
} else {
Assert(expectedWarnings == null || expectedWarnings.Count == 0);
}
}
public static RubyRegex/*!*/ CreateRegexN(object[]/*!*/ strings, RubyEncoding/*!*/ encoding, RubyRegexOptions options, StrongBox<RubyRegex> regexpCache) {
Func<RubyRegex> createRegex = delegate { return new RubyRegex(CreateMutableStringN(strings, encoding), options); };
return CreateRegexWorker(options, regexpCache, false, createRegex);
}
public abstract GenericRegex/*!*/ ToRegularExpression(RubyRegexOptions options);
public RegularExpression(List <Expression> /*!*/ pattern, RubyRegexOptions options, SourceSpan location)
: this(pattern, options, false, location)
{
}
public static RubyEncoding GetRegexEncoding(RubyRegexOptions options) {
switch (options & RubyRegexOptions.EncodingMask) {
#if !SILVERLIGHT
case RubyRegexOptions.EUC: return RubyEncoding.EUC;
case RubyRegexOptions.SJIS: return RubyEncoding.SJIS;
#endif
case RubyRegexOptions.UTF8: return RubyEncoding.UTF8;
case RubyRegexOptions.FIXED: return RubyEncoding.Binary;
default: return null;
}
}
public static RubyEncoding GetKCoding(RubyRegexOptions options) {
#if SILVERLIGHT
return null;
#else
switch (options & RubyRegexOptions.EncodingMask) {
case RubyRegexOptions.EUC: return RubyEncoding.KCodeEUC;
case RubyRegexOptions.SJIS: return RubyEncoding.KCodeSJIS;
case RubyRegexOptions.UTF8: return RubyEncoding.KCodeUTF8;
default: return null;
}
#endif
}
internal void SetRegexOptions(RubyRegexOptions value)
{
Integer = (int)value;
_type = TokenValueType.RegexOptions;
}
public override GenericRegex/*!*/ ToRegularExpression(RubyRegexOptions options) {
return new StringRegex(_data, options);
}
public RubyRegex(byte[] /*!*/ pattern, RubyRegexOptions options)
{
ContractUtils.RequiresNotNull(pattern, "pattern");
// TODO: _regex = new BinaryRegex(pattern, options);
_regex = new StringRegex(BinaryEncoding.Obsolete.GetString(pattern, 0, pattern.Length), options);
}
//[DebuggerHidden]
private void TestCorrectPatternTranslation(string/*!*/ pattern, RubyRegexOptions options, string/*!*/ expected, bool expectedGAnchor) {
bool hasGAnchor;
string actual = RegexpTransformer.Transform(pattern, options, out hasGAnchor);
AreEqual(expected, actual);
new Regex(expected);
Assert(hasGAnchor == expectedGAnchor);
}
protected GenericRegex(RubyRegexOptions options) {
ContractUtils.Requires(RubyRegex.GetPersistedOptions(options) == options);
_options = options;
}
public override GenericRegex/*!*/ ToRegularExpression(RubyRegexOptions options) {
// TODO: Fix BinaryRegex and use instead
return new StringRegex(ToString(), options);
}
public override GenericRegex /*!*/ ToRegularExpression(RubyRegexOptions options)
{
// TODO: Fix BinaryRegex and use instead
return(new StringRegex(ToString(), options));
}
private static int AppendOptionString(MutableString/*!*/ result, RubyRegexOptions options, bool enabled, bool includeEncoding) {
int count = 0;
if (((options & RubyRegexOptions.Multiline) != 0) == enabled) {
result.Append('m');
count++;
}
if (((options & RubyRegexOptions.IgnoreCase) != 0) == enabled) {
result.Append('i');
count++;
}
if (((options & RubyRegexOptions.Extended) != 0) == enabled) {
result.Append('x');
count++;
}
if (includeEncoding) {
switch (options & RubyRegexOptions.EncodingMask) {
case RubyRegexOptions.NONE: break;
case RubyRegexOptions.EUC: result.Append('e'); break;
case RubyRegexOptions.FIXED: result.Append('n'); break;
case RubyRegexOptions.UTF8: result.Append('u'); break;
case RubyRegexOptions.SJIS: result.Append('s'); break;
default: throw Assert.Unreachable;
}
}
return count;
}
public RubyRegex(string /*!*/ pattern, RubyRegexOptions options)
{
ContractUtils.RequiresNotNull(pattern, "pattern");
_regex = new StringRegex(pattern, options);
}
internal void SetRegexOptions(RubyRegexOptions value) {
Integer = (int)value;
_type = TokenValueType.RegexOptions;
}
public AssertTokenizer/*!*/ Read(RubyRegexOptions expected) {
Next();
_tests.Assert(_actualToken == Tokens.RegexpEnd);
_tests.Assert(expected == _actualValue.RegExOptions);
return this;
}
private static RubyRegex ConstructRubyRegexp(RubyConstructor/*!*/ ctor, Node node) {
ScalarNode scalar = node as ScalarNode;
if (node == null) {
throw RubyExceptions.CreateTypeError("Can only create regex from scalar node");
}
Match match = _regexPattern.Match(scalar.Value);
if (!match.Success) {
throw new ConstructorException("Invalid Regular expression: \"" + scalar.Value + "\"");
}
RubyRegexOptions options = new RubyRegexOptions();
foreach (char c in match.Groups["opts"].Value) {
switch (c) {
case 'i': options |= RubyRegexOptions.IgnoreCase; break;
case 'x': options |= RubyRegexOptions.Extended; break;
case 'm': options |= RubyRegexOptions.Multiline; break;
case 'o': break;
case 'n': options |= RubyRegexOptions.FIXED; break;
case 'e': options |= RubyRegexOptions.EUC; break;
case 's': options |= RubyRegexOptions.SJIS; break;
case 'u': options |= RubyRegexOptions.UTF8; break;
default:
throw new ConstructorException("Unknown regular expression option: '" + c + "'");
}
}
// TODO: encoding (ignore kcode on 1.9, string enc?):
return new RubyRegex(MutableString.CreateMutable(match.Groups["expr"].Value, RubyEncoding.UTF8), options);
}
// fixes escapes
// - unescapes non-special characters
// - fixes \xF -> \x0F
internal static string /*!*/ TransformPattern(string /*!*/ pattern, RubyRegexOptions options)
{
int first = 0;
int i = SkipWellEscaped(pattern, 0);
// trailing backslash is an error in both MRI, .NET
if (i == -1)
{
return(pattern);
}
StringBuilder result = new StringBuilder(pattern.Length);
do
{
Debug.Assert(i + 1 < pattern.Length);
Debug.Assert(pattern[i] == '\\');
result.Append(pattern, first, i - first);
i++;
char c = pattern[i++];
switch (c)
{
case 'x':
result.Append('\\');
result.Append('x');
// error:
if (i == pattern.Length)
{
break;
}
// fix single digit:
c = pattern[i++];
if (i == pattern.Length || !Tokenizer.IsHexadecimalDigit(pattern[i]))
{
result.Append('0');
}
result.Append(c);
break;
case 'h': // Oniguruma only: [0-9A-Fa-f]
case 'H': // Oniguruma only: [^0-9A-Fa-f]
case 'g': // Oniguruma only
case 'k': // Oniguruma, .NET: named backreference, MRI not supported
// remove backslash
default:
if (Tokenizer.IsDecimalDigit(c))
{
// TODO:
// \([1-9][0-9]*) where there is no group of such number (replace by an empty string)
result.Append('\\');
}
// .NET throws invalid escape exception, remove backslash:
result.Append(c);
break;
}
Debug.Assert(i <= pattern.Length);
first = i;
i = SkipWellEscaped(pattern, i);
} while (i >= 0);
result.Append(pattern, first, pattern.Length - first);
return(result.ToString());
}
public RubyRegex(MutableString /*!*/ pattern, RubyRegexOptions options)
{
Set(pattern, options);
}
protected GenericRegex(RubyRegexOptions options)
{
_options = options;
}
public abstract GenericRegex /*!*/ ToRegularExpression(RubyRegexOptions options);
internal protected BinaryRegex(byte[] /*!*/ pattern, RubyRegexOptions options)
: base(options)
{
Assert.NotNull(pattern);
_pattern = ArrayUtils.Copy(pattern);
}
public GenericRegex /*!*/ ToRegularExpression(RubyRegexOptions options)
{
return(_content.ToRegularExpression(options));
}
public void Set(MutableString /*!*/ pattern, RubyRegexOptions options)
{
_regex = pattern.ToRegularExpression(options);
}
