|
// 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.Collections.Generic;
using System.Collections.Immutable;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using Microsoft.CodeAnalysis.CodeCleanup.Providers;
using Microsoft.CodeAnalysis.Internal.Log;
using Microsoft.CodeAnalysis.Host;
using Microsoft.CodeAnalysis.PooledObjects;
using Microsoft.CodeAnalysis.Shared;
using Microsoft.CodeAnalysis.Shared.Collections;
using Microsoft.CodeAnalysis.Shared.Extensions;
using Microsoft.CodeAnalysis.Text;
using Roslyn.Utilities;
using Microsoft.CodeAnalysis.Formatting;
namespace Microsoft.CodeAnalysis.CodeCleanup;
internal abstract class AbstractCodeCleanerService : ICodeCleanerService
{
public abstract ImmutableArray<ICodeCleanupProvider> GetDefaultProviders();
protected abstract ImmutableArray<TextSpan> GetSpansToAvoid(SyntaxNode root);
public async Task<Document> CleanupAsync(Document document, ImmutableArray<TextSpan> spans, CodeCleanupOptions options, ImmutableArray<ICodeCleanupProvider> providers, CancellationToken cancellationToken)
{
using (Logger.LogBlock(FunctionId.CodeCleanup_CleanupAsync, cancellationToken))
{
// If there is no span to format...
if (!spans.Any())
{
// ... then return the Document unchanged
return document;
}
var codeCleaners = providers.IsDefault ? GetDefaultProviders() : providers;
var root = await document.GetRequiredSyntaxRootAsync(cancellationToken).ConfigureAwait(false);
var normalizedSpan = spans.ToNormalizedSpans();
if (CleanupWholeNode(root.FullSpan, normalizedSpan))
{
// We are cleaning up the whole document, so there is no need to do expansive span tracking between cleaners.
return await IterateAllCodeCleanupProvidersAsync(document, document, options, r => [r.FullSpan], codeCleaners, cancellationToken).ConfigureAwait(false);
}
// We need to track spans between cleaners. Annotate the tree with the provided spans.
var (newNode, annotations) = AnnotateNodeForTextSpans(root, normalizedSpan, cancellationToken);
// If it urns out we don't need to annotate anything since all spans are merged to one span that covers the whole node...
if (newNode == null)
{
// ... then we are cleaning up the whole document, so there is no need to do expansive span tracking between cleaners.
return await IterateAllCodeCleanupProvidersAsync(document, document, options, n => [n.FullSpan], codeCleaners, cancellationToken).ConfigureAwait(false);
}
// Replace the initial node and document with the annotated node.
var annotatedRoot = newNode;
var annotatedDocument = document.WithSyntaxRoot(annotatedRoot);
// Run the actual cleanup.
return await IterateAllCodeCleanupProvidersAsync(
document, annotatedDocument, options,
r => GetTextSpansFromAnnotation(r, annotations, cancellationToken),
codeCleaners, cancellationToken).ConfigureAwait(false);
}
}
public async Task<SyntaxNode> CleanupAsync(SyntaxNode root, ImmutableArray<TextSpan> spans, SyntaxFormattingOptions options, SolutionServices services, ImmutableArray<ICodeCleanupProvider> providers, CancellationToken cancellationToken)
{
using (Logger.LogBlock(FunctionId.CodeCleanup_Cleanup, cancellationToken))
{
// If there is no span to format...
if (!spans.Any())
{
// ... then return the Document unchanged
return root;
}
var codeCleaners = providers.IsDefault ? GetDefaultProviders() : providers;
var normalizedSpan = spans.ToNormalizedSpans();
if (CleanupWholeNode(root.FullSpan, normalizedSpan))
{
// We are cleaning up the whole document, so there is no need to do expansive span tracking between cleaners.
return await IterateAllCodeCleanupProvidersAsync(root, root, options, r => [r.FullSpan], services, codeCleaners, cancellationToken).ConfigureAwait(false);
}
// We need to track spans between cleaners. Annotate the tree with the provided spans.
var (newNode, annotations) = AnnotateNodeForTextSpans(root, normalizedSpan, cancellationToken);
// If it urns out we don't need to annotate anything since all spans are merged to one span that covers the whole node...
if (newNode == null)
{
// ... then we are cleaning up the whole document, so there is no need to do expansive span tracking between cleaners.
return await IterateAllCodeCleanupProvidersAsync(root, root, options, n => [n.FullSpan], services, codeCleaners, cancellationToken).ConfigureAwait(false);
}
// Replace the initial node and document with the annotated node.
var annotatedRoot = newNode;
// Run the actual cleanup.
return await IterateAllCodeCleanupProvidersAsync(
root, annotatedRoot, options,
r => GetTextSpansFromAnnotation(r, annotations, cancellationToken),
services, codeCleaners, cancellationToken).ConfigureAwait(false);
}
}
private static ImmutableArray<TextSpan> GetTextSpansFromAnnotation(
SyntaxNode node,
List<(SyntaxAnnotation previousAnnotation, SyntaxAnnotation nextAnnotation)> annotations,
CancellationToken cancellationToken)
{
// Now try to retrieve the text span from the annotations injected into the node.
var builder = ArrayBuilder<TextSpan>.GetInstance();
foreach (var (previousAnnotation, nextAnnotation) in annotations)
{
cancellationToken.ThrowIfCancellationRequested();
var previousMarkerAnnotation = previousAnnotation;
var nextMarkerAnnotation = nextAnnotation;
var previousTokenMarker = SpanMarker.FromAnnotation(previousMarkerAnnotation);
var nextTokenMarker = SpanMarker.FromAnnotation(nextMarkerAnnotation);
var previousTokens = node.GetAnnotatedNodesAndTokens(previousMarkerAnnotation).Where(n => n.IsToken).Select(n => n.AsToken());
var nextTokens = node.GetAnnotatedNodesAndTokens(nextMarkerAnnotation).Where(n => n.IsToken).Select(n => n.AsToken());
// Check whether we can use the tokens we got back from the node.
if (TryGetTextSpanFromAnnotation(previousTokenMarker, nextTokenMarker, node, previousTokens, nextTokens,
out var span))
{
builder.Add(span);
}
}
return builder.ToImmutableAndFree();
}
private static bool TryGetTextSpanFromAnnotation(
SpanMarker previousTokenMarker,
SpanMarker nextTokenMarker,
SyntaxNode node,
IEnumerable<SyntaxToken> previousTokens,
IEnumerable<SyntaxToken> nextTokens,
out TextSpan span)
{
// Set initial value
span = default;
var previousToken = previousTokens.FirstOrDefault();
var nextToken = nextTokens.FirstOrDefault();
// Define all variables required to determine state
var hasNoPreviousToken = previousToken.RawKind == 0;
var hasNoNextToken = nextToken.RawKind == 0;
var hasMultiplePreviousToken = previousTokens.Skip(1).Any();
var hasMultipleNextToken = nextTokens.Skip(1).Any();
var hasOnePreviousToken = !hasNoPreviousToken && !hasMultiplePreviousToken;
var hasOneNextToken = !hasNoNextToken && !hasMultipleNextToken;
// Normal case
if (hasOnePreviousToken && hasOneNextToken)
{
return TryCreateTextSpan(GetPreviousTokenStartPosition(previousTokenMarker.Type, previousToken),
GetNextTokenEndPosition(nextTokenMarker.Type, nextToken),
out span);
}
// Quick error check * we don't have any token, ignore this one.
// This can't happen unless one of cleaners violated the contract of only changing things inside of the provided span.
if (hasNoPreviousToken && hasNoNextToken)
{
return false;
}
// We don't have the previous token, but we do have one next token.
// nextTokenMarker has hint to how to find the opposite marker.
// If we can't find the other side, then it means one of cleaners has violated the contract. Ignore span.
if (hasNoPreviousToken && hasOneNextToken)
{
if (nextTokenMarker.OppositeMarkerType == SpanMarkerType.BeginningOfFile)
{
// Okay, found right span
return TryCreateTextSpan(node.SpanStart, GetNextTokenEndPosition(nextTokenMarker.Type, nextToken), out span);
}
return false;
}
// One previous token but no next token with hint case
if (hasOnePreviousToken && hasNoNextToken)
{
if (previousTokenMarker.OppositeMarkerType == SpanMarkerType.EndOfFile)
{
// Okay, found right span
return TryCreateTextSpan(GetPreviousTokenStartPosition(previousTokenMarker.Type, previousToken), node.Span.End, out span);
}
return false;
}
// Now the simple cases are done. Now we need to deal with cases where annotations found more than one corresponding token.
// Mostly it means one of cleaners violated the contract, so we can just ignore the span except in one cases where it involves the beginning and end of the tree.
Contract.ThrowIfFalse(hasMultiplePreviousToken || hasMultipleNextToken);
// Check whether it is one of special cases or not
if (hasMultiplePreviousToken && previousTokenMarker.Type == SpanMarkerType.BeginningOfFile)
{
// Okay, it is an edge case. Let's use the start of the node as the beginning of the span
span = TextSpan.FromBounds(node.SpanStart, GetNextTokenEndPosition(nextTokenMarker.Type, nextToken));
return true;
}
if (hasMultipleNextToken && nextTokenMarker.Type == SpanMarkerType.EndOfFile)
{
// Okay, it is an edge case. Let's use the end of the node as the end of the span
span = TextSpan.FromBounds(GetPreviousTokenStartPosition(previousTokenMarker.Type, previousToken), node.Span.End);
return true;
}
// All other cases are invalid cases where one of code cleaners messed things up by moving around things it shouldn't move.
return false;
}
/// <summary>
/// Get the proper start position based on the span marker type.
/// </summary>
private static int GetPreviousTokenStartPosition(SpanMarkerType spanMarkerType, SyntaxToken previousToken)
{
Contract.ThrowIfTrue(spanMarkerType == SpanMarkerType.EndOfFile);
Contract.ThrowIfTrue(previousToken.RawKind == 0);
if (spanMarkerType == SpanMarkerType.Normal)
{
return previousToken.GetNextToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true).SpanStart;
}
return previousToken.SpanStart;
}
/// <summary>
/// Get the proper end position based on the span marker type.
/// </summary>
private static int GetNextTokenEndPosition(SpanMarkerType spanMarkerType, SyntaxToken nextToken)
{
Contract.ThrowIfTrue(spanMarkerType == SpanMarkerType.BeginningOfFile);
Contract.ThrowIfTrue(nextToken.RawKind == 0);
if (spanMarkerType == SpanMarkerType.Normal)
{
return nextToken.GetPreviousToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true).Span.End;
}
return nextToken.Span.End;
}
/// <summary>
/// Inject annotations into the node so that it can re-calculate spans for each code cleaner after each tree transformation.
/// </summary>
private static (SyntaxNode newNode, List<(SyntaxAnnotation previous, SyntaxAnnotation next)> annotations) AnnotateNodeForTextSpans(
SyntaxNode root, ImmutableArray<TextSpan> spans, CancellationToken cancellationToken)
{
// Get spans where the tokens around the spans are not overlapping with the spans.
var nonOverlappingSpans = GetNonOverlappingSpans(root, spans, cancellationToken);
// Build token annotation map
var tokenAnnotationMap = new Dictionary<SyntaxToken, List<SyntaxAnnotation>>();
var annotations = new List<(SyntaxAnnotation previous, SyntaxAnnotation next)>();
foreach (var span in nonOverlappingSpans)
{
cancellationToken.ThrowIfCancellationRequested();
GetTokensAroundSpan(root, span,
out var previousToken, out var startToken, out var endToken, out var nextToken);
// Create marker to insert
var startMarker = new SpanMarker(type: (previousToken.RawKind == 0) ? SpanMarkerType.BeginningOfFile : SpanMarkerType.Normal,
oppositeMarkerType: (nextToken.RawKind == 0) ? SpanMarkerType.EndOfFile : SpanMarkerType.Normal);
var endMarker = new SpanMarker(type: (nextToken.RawKind == 0) ? SpanMarkerType.EndOfFile : SpanMarkerType.Normal,
oppositeMarkerType: (previousToken.RawKind == 0) ? SpanMarkerType.BeginningOfFile : SpanMarkerType.Normal);
// Set proper tokens
previousToken = (previousToken.RawKind == 0) ? root.GetFirstToken(includeZeroWidth: true) : previousToken;
nextToken = (nextToken.RawKind == 0) ? root.GetLastToken(includeZeroWidth: true) : nextToken;
// Build token to marker map
tokenAnnotationMap.GetOrAdd(previousToken, _ => []).Add(startMarker.Annotation);
tokenAnnotationMap.GetOrAdd(nextToken, _ => []).Add(endMarker.Annotation);
// Remember markers
annotations.Add((startMarker.Annotation, endMarker.Annotation));
}
// Do a quick check.
// If, after all merges, the spans are merged into one span that covers the whole tree, return right away.
if (CleanupWholeNode(annotations))
{
// This will indicate that no annotation is needed.
return default;
}
// Inject annotations
var newNode = InjectAnnotations(root, tokenAnnotationMap);
return (newNode, annotations);
}
/// <summary>
/// Make sure annotations are positioned outside of any spans. If not, merge two adjacent spans to one.
/// </summary>
private static ImmutableArray<TextSpan> GetNonOverlappingSpans(
SyntaxNode root, ImmutableArray<TextSpan> spans, CancellationToken cancellationToken)
{
// Create interval tree for spans
var intervalTree = SimpleMutableIntervalTree.Create(new TextSpanIntervalIntrospector(), spans);
// Find tokens that are outside of spans
var tokenSpans = new List<TextSpan>();
foreach (var span in spans)
{
cancellationToken.ThrowIfCancellationRequested();
_ = GetSpanAlignedToTokens(root, span, out var startToken, out var endToken);
var previousToken = startToken.GetPreviousToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
var nextToken = endToken.GetNextToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
// Make sure the previous and next tokens we found do not overlap with any existing spans. If they do, merge two spans.
previousToken = (previousToken.RawKind == 0) ? root.GetFirstToken(includeZeroWidth: true) : previousToken;
var start = intervalTree.HasIntervalThatOverlapsWith(previousToken.SpanStart, previousToken.Span.Length)
? previousToken.SpanStart
: startToken.SpanStart;
nextToken = (nextToken.RawKind == 0) ? root.GetLastToken(includeZeroWidth: true) : nextToken;
var end = intervalTree.HasIntervalThatOverlapsWith(nextToken.SpanStart, nextToken.Span.Length)
? nextToken.Span.End
: endToken.Span.End;
tokenSpans.Add(TextSpan.FromBounds(start, end));
}
return tokenSpans.ToNormalizedSpans().ToImmutableArray();
}
/// <summary>
/// Retrieves four tokens around span like below.
///
/// [previousToken][startToken][SPAN][endToken][nextToken]
/// </summary>
private static void GetTokensAroundSpan(
SyntaxNode root, TextSpan span,
out SyntaxToken previousToken,
out SyntaxToken startToken,
out SyntaxToken endToken,
out SyntaxToken nextToken)
{
// Get tokens at the edges of the span
startToken = root.FindToken(span.Start, findInsideTrivia: true);
endToken = root.FindTokenFromEnd(span.End, findInsideTrivia: true);
// There must be tokens at each edge
Contract.ThrowIfTrue(startToken.RawKind == 0 || endToken.RawKind == 0);
// Get the previous and next tokens around span
previousToken = startToken.GetPreviousToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
nextToken = endToken.GetNextToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
}
/// <summary>
/// Adjust provided span to align to either token's start position or end position.
/// </summary>
private static TextSpan GetSpanAlignedToTokens(
SyntaxNode root, TextSpan span,
out SyntaxToken startToken, out SyntaxToken endToken)
{
startToken = FindTokenOnLeftOfPosition(root, span.Start);
endToken = FindTokenOnRightOfPosition(root, span.End);
// Found two different tokens
if (startToken.Span.End <= endToken.SpanStart)
{
return TextSpan.FromBounds(startToken.SpanStart, endToken.Span.End);
}
else
{
// Found one token inside of the span
endToken = startToken;
return startToken.Span;
}
}
/// <summary>
/// Find closest token (including one in structured trivia) right of given position
/// </summary>
private static SyntaxToken FindTokenOnRightOfPosition(SyntaxNode root, int position)
{
var token = root.FindTokenOnRightOfPosition(position, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
if (token.RawKind == 0)
{
return root.GetLastToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
}
return token;
}
/// <summary>
/// Find closest token (including one in structured trivia) left of given position
/// </summary>
private static SyntaxToken FindTokenOnLeftOfPosition(SyntaxNode root, int position)
{
// find token on left
var token = root.FindTokenOnLeftOfPosition(position, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
if (token.RawKind == 0)
{
// if there is no token on left, return the first token
return root.GetFirstToken(includeZeroWidth: true, includeSkipped: true, includeDirectives: true, includeDocumentationComments: true);
}
return token;
}
private static bool CleanupWholeNode(List<(SyntaxAnnotation previous, SyntaxAnnotation next)> annotations)
{
if (annotations.Count != 1)
{
return false;
}
var startMarker = SpanMarker.FromAnnotation(annotations[0].previous);
var endMarker = SpanMarker.FromAnnotation(annotations[0].next);
return startMarker.Type == SpanMarkerType.BeginningOfFile && endMarker.Type == SpanMarkerType.EndOfFile;
}
private static bool CleanupWholeNode(TextSpan nodeSpan, ImmutableArray<TextSpan> spans)
{
if (spans.Length > 1)
{
return false;
}
return spans[0].Contains(nodeSpan);
}
private async Task<Document> IterateAllCodeCleanupProvidersAsync(
Document originalDocument,
Document annotatedDocument,
CodeCleanupOptions options,
Func<SyntaxNode, ImmutableArray<TextSpan>> spanGetter,
ImmutableArray<ICodeCleanupProvider> codeCleaners,
CancellationToken cancellationToken)
{
using (Logger.LogBlock(FunctionId.CodeCleanup_IterateAllCodeCleanupProviders, cancellationToken))
{
var currentDocument = annotatedDocument;
Document? previousDocument = null;
var spans = ImmutableArray<TextSpan>.Empty;
#if DEBUG
var originalDocHasErrors = await annotatedDocument.HasAnyErrorsAsync(cancellationToken).ConfigureAwait(false);
#endif
var current = 0;
var count = codeCleaners.Length;
foreach (var codeCleaner in codeCleaners)
{
cancellationToken.ThrowIfCancellationRequested();
current++;
if (previousDocument != currentDocument)
{
// Document was changed by the previous code cleaner, compute new spans.
var root = await currentDocument.GetRequiredSyntaxRootAsync(cancellationToken).ConfigureAwait(false);
previousDocument = currentDocument;
spans = GetSpans(root, spanGetter);
}
// If we are at the end and there were no changes to the document, use the original document for the cleanup.
if (current == count && currentDocument == annotatedDocument)
{
currentDocument = originalDocument;
}
using (Logger.LogBlock(FunctionId.CodeCleanup_IterateOneCodeCleanup, GetCodeCleanerTypeName, codeCleaner, cancellationToken))
{
currentDocument = await codeCleaner.CleanupAsync(currentDocument, spans, options, cancellationToken).ConfigureAwait(false);
}
#if DEBUG
if (!originalDocHasErrors && currentDocument != annotatedDocument)
{
await currentDocument.VerifyNoErrorsAsync("Pretty-listing introduced errors in error-free code", cancellationToken).ConfigureAwait(false);
}
#endif
}
// If none of the cleanup operations changed the document, we should return the original document
// rather than the one that has our annotations.
if (currentDocument != annotatedDocument)
{
return currentDocument;
}
else
{
return originalDocument;
}
}
}
private ImmutableArray<TextSpan> GetSpans(
SyntaxNode root, Func<SyntaxNode, ImmutableArray<TextSpan>> spanGetter)
{
// Get all the spans we've been requested to clean up.
var requestedSpans = new NormalizedTextSpanCollection(spanGetter(root));
// See if there are any spans we should not touch.
var spansToAvoid = new NormalizedTextSpanCollection(GetSpansToAvoid(root));
// Remove the spans we should not touch from the requested spans and return that final set.
var result = NormalizedTextSpanCollection.Difference(requestedSpans, spansToAvoid);
return [.. result];
}
private async Task<SyntaxNode> IterateAllCodeCleanupProvidersAsync(
SyntaxNode originalRoot,
SyntaxNode annotatedRoot,
SyntaxFormattingOptions options,
Func<SyntaxNode, ImmutableArray<TextSpan>> spanGetter,
SolutionServices services,
ImmutableArray<ICodeCleanupProvider> codeCleaners,
CancellationToken cancellationToken)
{
using (Logger.LogBlock(FunctionId.CodeCleanup_IterateAllCodeCleanupProviders, cancellationToken))
{
var currentRoot = annotatedRoot;
SyntaxNode? previousRoot = null;
var spans = ImmutableArray<TextSpan>.Empty;
var current = 0;
var count = codeCleaners.Length;
foreach (var codeCleaner in codeCleaners)
{
cancellationToken.ThrowIfCancellationRequested();
current++;
if (previousRoot != currentRoot)
{
// The root was changed by the previous code cleaner, compute new spans.
previousRoot = currentRoot;
spans = GetSpans(currentRoot, spanGetter);
}
// If we are at the end and there were no changes to the document, use the original document for the cleanup.
if (current == count && currentRoot == annotatedRoot)
{
currentRoot = originalRoot;
}
using (Logger.LogBlock(FunctionId.CodeCleanup_IterateOneCodeCleanup, GetCodeCleanerTypeName, codeCleaner, cancellationToken))
{
currentRoot = await codeCleaner.CleanupAsync(currentRoot, spans, options, services, cancellationToken).ConfigureAwait(false);
}
}
// If none of the cleanup operations changed the root, we should return the original root
// rather than the one that has our annotations.
if (currentRoot != annotatedRoot)
{
return currentRoot;
}
else
{
return originalRoot;
}
}
}
private string GetCodeCleanerTypeName(ICodeCleanupProvider codeCleaner)
=> codeCleaner.ToString() ?? "";
private static SyntaxNode InjectAnnotations(SyntaxNode node, Dictionary<SyntaxToken, List<SyntaxAnnotation>> map)
{
var tokenMap = map.ToDictionary(p => p.Key, p => p.Value);
return node.ReplaceTokens(tokenMap.Keys, (o, n) => o.WithAdditionalAnnotations([.. tokenMap[o]]));
}
private static bool TryCreateTextSpan(int start, int end, out TextSpan span)
{
span = default;
if (start < 0 || end < start)
{
return false;
}
span = TextSpan.FromBounds(start, end);
return true;
}
/// <summary>
/// Enum that indicates type of span marker
/// </summary>
private enum SpanMarkerType
{
/// <summary>
/// Normal case
/// </summary>
Normal,
/// <summary>
/// Span starts at the beginning of the tree
/// </summary>
BeginningOfFile,
/// <summary>
/// Span ends at the end of the tree
/// </summary>
EndOfFile
}
/// <summary>
/// Internal annotation type to mark span location in the tree.
/// </summary>
private class SpanMarker
{
/// <summary>
/// Indicates the current marker type
/// </summary>
public SpanMarkerType Type { get; }
/// <summary>
/// Indicates how to find the other side of the span marker if it is missing
/// </summary>
public SpanMarkerType OppositeMarkerType { get; }
public SyntaxAnnotation Annotation { get; }
public const string AnnotationId = "SpanMarker";
private SpanMarker(SpanMarkerType type, SpanMarkerType oppositeMarkerType, SyntaxAnnotation annotation)
{
this.Type = type;
this.OppositeMarkerType = oppositeMarkerType;
this.Annotation = annotation;
}
public SpanMarker(SpanMarkerType type = SpanMarkerType.Normal, SpanMarkerType oppositeMarkerType = SpanMarkerType.Normal)
: this(type, oppositeMarkerType, new SyntaxAnnotation(AnnotationId, string.Format("{0} {1}", type, oppositeMarkerType)))
{
}
private static readonly char[] s_separators = [' '];
public static SpanMarker FromAnnotation(SyntaxAnnotation annotation)
{
Contract.ThrowIfNull(annotation.Data);
var types = annotation.Data.Split(s_separators).Select(s => (SpanMarkerType)Enum.Parse(typeof(SpanMarkerType), s)).ToArray();
return new SpanMarker(types[0], types[1], annotation);
}
}
}
|