add: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Experimental/Regex/union.rb File: union.rb =================================================================== --- [no source file] +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Experimental/Regex/union.rb;regex3 @@ -1,0 +1,8 @@ +p Regexp.union(/a/mix, /b/i, /c/) +p Regexp.union(/a/mix, /b/i, "asdasd*") +p Regexp.union() +p Regexp.union(/a/mix) +p Regexp.union("a") +p Regexp.union("", /a/mix) +p Regexp.union(//, /a/mix) +p Regexp.union(nil) rescue p $! \ No newline at end of file =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/RubyTests.cs;C448505 File: RubyTests.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/RubyTests.cs;C448505 (server) 5/24/2008 10:11 PM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/RubyTests.cs;regex3 @@ -58,6 +58,7 @@ Regex2, RegexTransform1, RegexTransform2, + RegexEscape1, Scenario_RubyScopeParsing, Scenario_RubyScopes1, =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/ExceptionTests.cs;C448505 File: ExceptionTests.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/ExceptionTests.cs;C448505 (server) 5/25/2008 10:12 AM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/ExceptionTests.cs;regex3 @@ -579,7 +579,6 @@ }, @"42"); } - [Run] public void ExceptionArg2() { AssertOutput(delegate() { CompilerTest(@" =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/RegexTests.cs;C448505 File: RegexTests.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/RegexTests.cs;C448505 (server) 5/24/2008 10:12 PM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/IronRuby.Tests/Runtime/RegexTests.cs;regex3 @@ -114,6 +114,9 @@ @"\G", @"\G", + + @"[a\-z]", + @"[a\-z]", }; for (int i = 0; i < incorrectPatterns.Length; i += 2) { @@ -154,6 +157,28 @@ Assert(e == t); new Regex(t); } + + public void RegexEscape1() { + string[] patterns = new string[] { + @"", + @"", + + @"\", + @"\\", + + @"(*)", + @"\(\*\)", + + "$ ^ | [ ] ( ) \\ . # - { } * + ? \t \v \n \r \f \a \b", + @"\$ \^ \| \[ \] \( \) \\ \. \# \- \{ \} \* \+ \? \t \v \n \r \f " + "\a \b" + }; + + for (int i = 0; i < patterns.Length; i += 2) { + string expected = patterns[i + 1]; + string actual = StringRegex.Escape(patterns[i]); + Assert(actual == expected); + } + } } } =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Initializers.Generated.cs;C448505 File: Initializers.Generated.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Initializers.Generated.cs;C448505 (server) 5/25/2008 11:48 AM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Initializers.Generated.cs;regex3 @@ -3551,6 +3551,11 @@ new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.MatchIndex), }); + module.DefineLibraryMethod("==", 0x29, new System.Delegate[] { + new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.Equals), + new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.Equals), + }); + module.DefineLibraryMethod("===", 0x29, new System.Delegate[] { new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.CaseCompare), new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.CaseCompare), @@ -3560,6 +3565,15 @@ new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.IsCaseInsensitive), }); + module.DefineLibraryMethod("eql?", 0x29, new System.Delegate[] { + new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.Equals), + new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.Equals), + }); + + module.DefineLibraryMethod("hash", 0x29, new System.Delegate[] { + new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.GetHash), + }); + module.DefineLibraryMethod("inspect", 0x29, new System.Delegate[] { new Microsoft.Scripting.Utils.Function(Ruby.Builtins.RegexpOps.Inspect), }); =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Builtins/RubyRegexOps.cs;C448505 File: RubyRegexOps.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Builtins/RubyRegexOps.cs;C448505 (server) 5/25/2008 9:51 AM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Libraries.LCA_RESTRICTED/Builtins/RubyRegexOps.cs;regex3 @@ -28,10 +28,6 @@ //declared private instance methods: // initialize // initialize_copy - //declared public instance methods: - // == - // eql? - // hash //------------------------------------------------------------ Regexp::new // Regexp.new(string [, options [, lang]]) => regexp @@ -144,8 +140,13 @@ // Ruby: doesn't wrap if there is a single embedded expression that evaluates to non-nil: // puts(/#{nil}#{/a/}#{nil}/) // We don't do that. - - MutableString result = MutableString.CreateMutable(); + + return Append(self, MutableString.CreateMutable()); + } + + private static MutableString/*!*/ Append(RubyRegex/*!*/ self, MutableString/*!*/ result) { + Assert.NotNull(self, result); + result.Append("(?"); if (AppendOptionString(result, self.Options, true, false) < 3) { result.Append('-'); @@ -248,6 +249,21 @@ return Match(context, self, Protocols.CastToString(context, str), 0); } + [RubyMethod("hash")] + public static int GetHash(RubyRegex/*!*/ self) { + return self.GetHashCode(); + } + + [RubyMethod("=="), RubyMethod("eql?")] + public static bool Equals(RubyRegex/*!*/ self, object other) { + return false; + } + + [RubyMethod("=="), RubyMethod("eql?")] + public static bool Equals(CodeContext/*!*/ context, RubyRegex/*!*/ self, [NotNull]RubyRegex/*!*/ other) { + return self.Equals(other); + } + //-------------------------------------------------------------- Regexp#=~ // rxp.match(str) => matchdata or nil //------------------------------------------------------------------------ @@ -376,17 +392,29 @@ [RubyMethod("union", RubyMethodAttributes.PublicSingleton)] public static RubyRegex/*!*/ Union(CodeContext/*!*/ context, object/*!*/ self, [NotNull]params object[]/*!*/ strings) { + if (strings.Length == 0) { + return new RubyRegex("(?!)", RubyRegexOptions.NONE); + } + MutableString result = MutableString.CreateMutable(); for (int i = 0; i < strings.Length; i++) { - MutableString str = Protocols.CastToString(context, strings[i]); if (i > 0) { result.Append('|'); } - result.Append(RubyRegex.Escape(str)); + + RubyRegex regex = strings[i] as RubyRegex; + if (regex != null) { + if (strings.Length == 1) { + return regex; + } + + Append(regex, result); + } else { + result.Append(RubyRegex.Escape(Protocols.CastToString(context, strings[i]))); + } } return new RubyRegex(result, RubyRegexOptions.NONE); } - } } =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.Content.cs;C448472 File: MutableString.Content.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.Content.cs;C448472 (server) 5/25/2008 12:46 PM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.Content.cs;regex3 @@ -61,6 +61,9 @@ public abstract byte[]/*!*/ ToByteArray(); public abstract GenericRegex/*!*/ ToRegularExpression(RubyRegexOptions options); + // returns self if there are no characters to be escaped: + public abstract Content/*!*/ EscapeRegularExpression(); + // read: public abstract bool IsBinary { get; } public abstract int Length { get; } @@ -173,6 +176,11 @@ return new StringRegex(_data, options); } + public override Content/*!*/ EscapeRegularExpression() { + StringBuilder sb = StringRegex.EscapeToStringBuilder(_data); + return (sb != null) ? new StringContent(_owner, sb.ToString()) : this; + } + private BinaryContent/*!*/ ToBinary() { return ToContent(_owner._encoding.GetBytes(_data)); } @@ -489,6 +497,11 @@ return new StringRegex(_data.ToString(), options); } + public override Content/*!*/ EscapeRegularExpression() { + StringBuilder sb = StringRegex.EscapeToStringBuilder(_data.ToString()); + return (sb != null) ? new StringBuilderContent(_owner, sb) : this; + } + private BinaryContent/*!*/ ToBinary() { return ToContent(_owner._encoding.GetBytes(_data.ToString())); } @@ -834,6 +847,10 @@ return new BinaryRegex(_data.ToArray(), options); } + public override Content/*!*/ EscapeRegularExpression() { + return new BinaryContent(_owner, new List(BinaryRegex.Escape(_data.ToArray()))); + } + private StringBuilderContent/*!*/ ToStringBuilder() { return ToStringBuilder(0); } =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.cs;C444052 File: MutableString.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.cs;C444052 (server) 5/25/2008 12:46 PM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/MutableString.cs;regex3 @@ -750,6 +750,10 @@ _content.GetDebugView(out value, out type); return type; } + + internal MutableString/*!*/ EscapeRegularExpression() { + return new MutableString(_content.EscapeRegularExpression(), _encoding); + } #endregion =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/RubyRegex.cs;C448505 File: RubyRegex.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/RubyRegex.cs;C448505 (server) 5/24/2008 10:07 PM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Builtins/RubyRegex.cs;regex3 @@ -25,7 +25,7 @@ using Ruby.Compiler; namespace Ruby.Builtins { - public abstract class GenericRegex { + public abstract class GenericRegex { private readonly RubyRegexOptions _options; public abstract bool IsEmpty { get; } @@ -36,6 +36,16 @@ public abstract MatchCollection/*!*/ Matches(MutableString/*!*/ input, int start); public abstract MutableString[]/*!*/ Split(MutableString/*!*/ input, int count, int start); + public bool Equals(GenericRegex/*!*/ other) { + // TODO: + return Options == other.Options && GetPattern().Equals(other.GetPattern()); + } + + public override int GetHashCode() { + // TODO: + return (int)Options ^ GetPattern().GetHashCode(); + } + protected GenericRegex(RubyRegexOptions options) { _options = options; } @@ -69,6 +79,10 @@ public override MutableString[]/*!*/ Split(MutableString/*!*/ input, int count, int start) { throw new NotImplementedException(); } + + internal static byte[]/*!*/ Escape(byte[]/*!*/ pattern) { + throw new NotImplementedException(); + } } public class StringRegex : GenericRegex { @@ -122,6 +136,92 @@ return MutableString.MakeArray(_regex.Split(input.ConvertToString(), count, start)); } + private static int SkipNonSpecial(string/*!*/ pattern, int i, out char escaped) { + while (i < pattern.Length) { + char c = pattern[i]; + switch (c) { + case '$': + case '^': + case '|': + case '[': + case ']': + case '(': + case ')': + case '\\': + case '.': + case '#': + case '-': + + case '{': + case '}': + case '*': + case '+': + case '?': + escaped = c; + return i; + + case '\t': + escaped = 't'; + return i; + + case '\v': + escaped = 'v'; + return i; + + case '\n': + escaped = 'n'; + return i; + + case '\r': + escaped = 'r'; + return i; + + case '\f': + escaped = 'f'; + return i; + } + i++; + } + + escaped = '\0'; + return -1; + } + + internal static string/*!*/ Escape(string/*!*/ pattern) { + StringBuilder sb = EscapeToStringBuilder(pattern); + return (sb != null) ? sb.ToString() : pattern; + } + + internal static StringBuilder EscapeToStringBuilder(string/*!*/ pattern) { + int first = 0; + char escaped; + int i = SkipNonSpecial(pattern, 0, out escaped); + + if (i == -1) { + return null; + } + + StringBuilder result = new StringBuilder(pattern.Length + 1); + + do { + Debug.Assert(i < pattern.Length); + // pattern[i] needs escape + + result.Append(pattern, first, i - first); + result.Append('\\'); + result.Append(escaped); + i++; + + Debug.Assert(i <= pattern.Length); + + first = i; + i = SkipNonSpecial(pattern, i, out escaped); + } while (i >= 0); + + result.Append(pattern, first, pattern.Length - first); + return result; + } + private static int SkipWellEscaped(string/*!*/ pattern, int i) { while (i < pattern.Length - 1) { if (pattern[i] == '\\') { @@ -137,6 +237,7 @@ case '\\': case '.': case '#': + case '-': case '{': case '}': @@ -259,7 +360,7 @@ } } - public class RubyRegex : IDuplicable { + public class RubyRegex : IDuplicable, IEquatable { private GenericRegex/*!*/ _regex; #region Construction @@ -307,6 +408,18 @@ return _regex.GetPattern(); } + public bool Equals(RubyRegex other) { + return ReferenceEquals(this, other) || other != null && _regex.Equals(other._regex); + } + + public override bool Equals(object other) { + return _regex.Equals(other as RubyRegex); + } + + public override int GetHashCode() { + return _regex.GetHashCode(); + } + #region Match, ReverseMatch, Matches, Split public Match/*!*/ Match(MutableString/*!*/ input) { @@ -363,7 +476,7 @@ // TODO: public static MutableString/*!*/ Escape(MutableString/*!*/ str) { - return MutableString.Create(Regex.Escape(str.ConvertToString())); + return str.EscapeRegularExpression(); } #region IDuplicable Members =================================================================== edit: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Runtime/RubyUtils.cs;C447221 File: RubyUtils.cs =================================================================== --- $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Runtime/RubyUtils.cs;C447221 (server) 5/25/2008 11:57 AM +++ Shelved Change: $/Dev10/feature/vs_langs01/Merlin/Main/Languages/Ruby/Ruby/Runtime/RubyUtils.cs;regex3 @@ -527,7 +527,16 @@ options.LocalNames = targetScope.GetVisibleLocalNames(); SourceUnit source = ec.Context.CreateSnippet(code.ConvertToString()); - ScriptCode compiledCode = source.Compile(options, ec.RuntimeErrorSink); + ScriptCode compiledCode; + try { + compiledCode = source.Compile(options, ec.RuntimeErrorSink); + } catch (SyntaxError e) { + Debug.WriteLine(e.Message, "INSTANCE EVAL ERROR"); + Debug.WriteLine(new String('-', 50), "INSTANCE EVAL ERROR"); + Debug.WriteLine(source.GetCode(), "INSTANCE EVAL ERROR"); + Debug.WriteLine(new String('-', 50), "INSTANCE EVAL ERROR"); + throw; + } Debug.Assert(compiledCode != null); // TODO: this is hack - we need arbitrary signature of the Initialize method: ===================================================================