RegexClassifier.cs

// Licensed to the .NET Foundation under one or more agreements.
// The .NET Foundation licenses this file to you under the MIT license.
// See the LICENSE file in the project root for more information.

using System;
using System.Composition;
using Microsoft.CodeAnalysis.Classification;
using Microsoft.CodeAnalysis.EmbeddedLanguages;
using Microsoft.CodeAnalysis.EmbeddedLanguages.Common;
using Microsoft.CodeAnalysis.EmbeddedLanguages.RegularExpressions;
using Microsoft.CodeAnalysis.Host.Mef;
using Microsoft.CodeAnalysis.PooledObjects;
using Microsoft.CodeAnalysis.Shared.Extensions;

namespace Microsoft.CodeAnalysis.Features.EmbeddedLanguages.RegularExpressions.LanguageServices
{
    using static EmbeddedSyntaxHelpers;

    using RegexToken = EmbeddedSyntaxToken<RegexKind>;
    using RegexTrivia = EmbeddedSyntaxTrivia<RegexKind>;

    /// <summary>
    /// Classifier impl for embedded regex strings.
    /// </summary>
    [ExtensionOrder(Before = PredefinedEmbeddedLanguageNames.Json)]
    [ExportEmbeddedLanguageClassifier(
        PredefinedEmbeddedLanguageNames.Regex,
        new[] { LanguageNames.CSharp, LanguageNames.VisualBasic },
        supportsUnannotatedAPIs: true, "Regex", "Regexp"), Shared]
    internal sealed class RegexClassifier : IEmbeddedLanguageClassifier
    {
        private static readonly ObjectPool<Visitor> s_visitorPool = SharedPools.Default<Visitor>();

        [ImportingConstructor]
        [Obsolete(MefConstruction.ImportingConstructorMessage, error: true)]
        public RegexClassifier()
        {
        }

        public void RegisterClassifications(EmbeddedLanguageClassificationContext context)
        {
            var info = context.Project.GetRequiredLanguageService<IEmbeddedLanguagesProvider>().EmbeddedLanguageInfo;

            var token = context.SyntaxToken;
            if (!info.IsAnyStringLiteral(token.RawKind))
                return;

            if (!context.Options.ColorizeRegexPatterns)
                return;

            var semanticModel = context.SemanticModel;
            var cancellationToken = context.CancellationToken;

            var detector = RegexLanguageDetector.GetOrCreate(semanticModel.Compilation, info);
            var tree = detector.TryParseString(token, semanticModel, cancellationToken);
            if (tree == null)
                return;

            var visitor = s_visitorPool.Allocate();
            try
            {
                visitor.Context = context;
                AddClassifications(tree.Root, visitor, context);
            }
            finally
            {
                visitor.Context = default;
                s_visitorPool.Free(visitor);
            }
        }

        private static void AddClassifications(RegexNode node, Visitor visitor, EmbeddedLanguageClassificationContext context)
        {
            node.Accept(visitor);

            foreach (var child in node)
            {
                if (child.IsNode)
                {
                    AddClassifications(child.Node, visitor, context);
                }
                else
                {
                    AddTriviaClassifications(child.Token, context);
                }
            }
        }

        private static void AddTriviaClassifications(RegexToken token, EmbeddedLanguageClassificationContext context)
        {
            foreach (var trivia in token.LeadingTrivia)
                AddTriviaClassifications(trivia, context);
        }

        private static void AddTriviaClassifications(RegexTrivia trivia, EmbeddedLanguageClassificationContext context)
        {
            if (trivia.Kind == RegexKind.CommentTrivia &&
                trivia.VirtualChars.Length > 0)
            {
                context.AddClassification(ClassificationTypeNames.RegexComment, GetSpan(trivia.VirtualChars));
            }
        }

        private class Visitor : IRegexNodeVisitor
        {
            public EmbeddedLanguageClassificationContext Context;

            private void AddClassification(RegexToken token, string typeName)
            {
                if (!token.IsMissing)
                    Context.AddClassification(typeName, token.GetSpan());
            }

            private void ClassifyWholeNode(RegexNode node, string typeName)
            {
                foreach (var child in node)
                {
                    if (child.IsNode)
                    {
                        ClassifyWholeNode(child.Node, typeName);
                    }
                    else
                    {
                        AddClassification(child.Token, typeName);
                    }
                }
            }

            public void Visit(RegexCompilationUnit node)
            {
                // Nothing to highlight.
            }

            public void Visit(RegexSequenceNode node)
            {
                // Nothing to highlight.   
            }

            #region Character classes

            public void Visit(RegexWildcardNode node)
                => AddClassification(node.DotToken, ClassificationTypeNames.RegexCharacterClass);

            public void Visit(RegexCharacterClassNode node)
            {
                AddClassification(node.OpenBracketToken, ClassificationTypeNames.RegexCharacterClass);
                AddClassification(node.CloseBracketToken, ClassificationTypeNames.RegexCharacterClass);
            }

            public void Visit(RegexNegatedCharacterClassNode node)
            {
                AddClassification(node.OpenBracketToken, ClassificationTypeNames.RegexCharacterClass);
                AddClassification(node.CaretToken, ClassificationTypeNames.RegexCharacterClass);
                AddClassification(node.CloseBracketToken, ClassificationTypeNames.RegexCharacterClass);
            }

            public void Visit(RegexCharacterClassRangeNode node)
                => AddClassification(node.MinusToken, ClassificationTypeNames.RegexCharacterClass);

            public void Visit(RegexCharacterClassSubtractionNode node)
                => AddClassification(node.MinusToken, ClassificationTypeNames.RegexCharacterClass);

            public void Visit(RegexCharacterClassEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexCharacterClass);

            public void Visit(RegexCategoryEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexCharacterClass);

            #endregion

            #region Quantifiers

            public void Visit(RegexZeroOrMoreQuantifierNode node)
                => AddClassification(node.AsteriskToken, ClassificationTypeNames.RegexQuantifier);

            public void Visit(RegexOneOrMoreQuantifierNode node)
                => AddClassification(node.PlusToken, ClassificationTypeNames.RegexQuantifier);

            public void Visit(RegexZeroOrOneQuantifierNode node)
                => AddClassification(node.QuestionToken, ClassificationTypeNames.RegexQuantifier);

            public void Visit(RegexLazyQuantifierNode node)
                => AddClassification(node.QuestionToken, ClassificationTypeNames.RegexQuantifier);

            public void Visit(RegexExactNumericQuantifierNode node)
            {
                AddClassification(node.OpenBraceToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.FirstNumberToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.CloseBraceToken, ClassificationTypeNames.RegexQuantifier);
            }

            public void Visit(RegexOpenNumericRangeQuantifierNode node)
            {
                AddClassification(node.OpenBraceToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.FirstNumberToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.CommaToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.CloseBraceToken, ClassificationTypeNames.RegexQuantifier);
            }

            public void Visit(RegexClosedNumericRangeQuantifierNode node)
            {
                AddClassification(node.OpenBraceToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.FirstNumberToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.CommaToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.SecondNumberToken, ClassificationTypeNames.RegexQuantifier);
                AddClassification(node.CloseBraceToken, ClassificationTypeNames.RegexQuantifier);
            }

            #endregion

            #region Groupings

            public void Visit(RegexSimpleGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexSimpleOptionsGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexNestedOptionsGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexNonCapturingGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexPositiveLookaheadGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexNegativeLookaheadGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexPositiveLookbehindGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexNegativeLookbehindGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexAtomicGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexCaptureGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexBalancingGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexConditionalCaptureGroupingNode node)
                => ClassifyGrouping(node);

            public void Visit(RegexConditionalExpressionGroupingNode node)
                => ClassifyGrouping(node);

            // Captures and backreferences refer to groups.  So we classify them the same way as groups.
            public void Visit(RegexCaptureEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexGrouping);

            public void Visit(RegexKCaptureEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexGrouping);

            public void Visit(RegexBackreferenceEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexGrouping);

            private void ClassifyGrouping(RegexGroupingNode node)
            {
                foreach (var child in node)
                {
                    if (!child.IsNode)
                    {
                        AddClassification(child.Token, ClassificationTypeNames.RegexGrouping);
                    }
                }
            }

            #endregion

            #region Other Escapes

            public void Visit(RegexControlEscapeNode node)
                => ClassifyOtherEscape(node);

            public void Visit(RegexHexEscapeNode node)
                => ClassifyOtherEscape(node);

            public void Visit(RegexUnicodeEscapeNode node)
                => ClassifyOtherEscape(node);

            public void Visit(RegexOctalEscapeNode node)
                => ClassifyOtherEscape(node);

            public void ClassifyOtherEscape(RegexNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexOtherEscape);

            #endregion 

            #region Anchors

            public void Visit(RegexAnchorNode node)
                => AddClassification(node.AnchorToken, ClassificationTypeNames.RegexAnchor);

            public void Visit(RegexAnchorEscapeNode node)
                => ClassifyWholeNode(node, ClassificationTypeNames.RegexAnchor);

            #endregion

            public void Visit(RegexTextNode node)
                => AddClassification(node.TextToken, ClassificationTypeNames.RegexText);

            public void Visit(RegexPosixPropertyNode node)
            {
                // The .NET parser just interprets the [ of the node, and skips the rest. So
                // classify the end part as a comment.
                Context.AddClassification(ClassificationTypeNames.RegexText, node.TextToken.VirtualChars[0].Span);
                Context.AddClassification(
                    ClassificationTypeNames.RegexComment,
                    GetSpan(node.TextToken.VirtualChars[1], node.TextToken.VirtualChars.Last()));
            }

            public void Visit(RegexAlternationNode node)
            {
                for (var i = 1; i < node.SequenceList.NodesAndTokens.Length; i += 2)
                    AddClassification(node.SequenceList.NodesAndTokens[i].Token, ClassificationTypeNames.RegexAlternation);
            }

            public void Visit(RegexSimpleEscapeNode node)
                => ClassifyWholeNode(node, node.IsSelfEscape()
                    ? ClassificationTypeNames.RegexSelfEscapedCharacter
                    : ClassificationTypeNames.RegexOtherEscape);
        }
    }
}