|
// 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.Runtime.CompilerServices;
using System.Threading;
using System.Threading.Tasks;
using Microsoft.CodeAnalysis.Serialization;
using Microsoft.CodeAnalysis.Shared.Extensions;
using Microsoft.CodeAnalysis.SpellCheck;
using Microsoft.CommonLanguageServerProtocol.Framework;
using Roslyn.LanguageServer.Protocol;
using Roslyn.Utilities;
namespace Microsoft.CodeAnalysis.LanguageServer.Handler.SpellCheck
{
/// <summary>
/// Root type for both document and workspace spell checking requests.
/// </summary>
internal abstract class AbstractSpellCheckHandler<TParams, TReport>
: ILspServiceRequestHandler<TParams, TReport[]?>, ITextDocumentIdentifierHandler<TParams, TextDocumentIdentifier?>
where TParams : IPartialResultParams<TReport[]>
where TReport : VSInternalSpellCheckableRangeReport
{
/// <summary>
/// Cache where we store the data produced by prior requests so that they can be returned if nothing of
/// significance changed. The version key is produced by combining the checksums for project options <see
/// cref="ProjectState.GetParseOptionsChecksum"/> and <see cref="DocumentStateChecksums.Text"/>
/// </summary>
private readonly VersionedPullCache<(Checksum parseOptionsChecksum, Checksum textChecksum)?> _versionedCache;
public bool MutatesSolutionState => false;
public bool RequiresLSPSolution => true;
protected AbstractSpellCheckHandler()
{
_versionedCache = new(this.GetType().Name);
}
public abstract TextDocumentIdentifier? GetTextDocumentIdentifier(TParams requestParams);
/// <summary>
/// Retrieve the previous results we reported. Used so we can avoid resending data for unchanged files. Also
/// used so we can report which documents were removed and can have all their spell checking results cleared.
/// </summary>
protected abstract ImmutableArray<PreviousPullResult>? GetPreviousResults(TParams requestParams);
/// <summary>
/// Returns all the documents that should be processed in the desired order to process them in.
/// </summary>
protected abstract ImmutableArray<Document> GetOrderedDocuments(RequestContext context, CancellationToken cancellationToken);
/// <summary>
/// Creates the <see cref="VSInternalSpellCheckableRangeReport"/> instance we'll report back to clients to let them know our
/// progress. Subclasses can fill in data specific to their needs as appropriate.
/// </summary>
protected abstract TReport CreateReport(TextDocumentIdentifier identifier, int[]? ranges, string? resultId);
public async Task<TReport[]?> HandleRequestAsync(
TParams requestParams, RequestContext context, CancellationToken cancellationToken)
{
context.TraceInformation($"{this.GetType()} started getting spell checking spans");
// The progress object we will stream reports to.
using var progress = BufferedProgress.Create(requestParams.PartialResultToken);
// Get the set of results the request said were previously reported. We can use this to determine both
// what to skip, and what files we have to tell the client have been removed.
var previousResults = GetPreviousResults(requestParams) ?? [];
context.TraceInformation($"previousResults.Length={previousResults.Length}");
// First, let the client know if any workspace documents have gone away. That way it can remove those for
// the user from squiggles or error-list.
HandleRemovedDocuments(context, previousResults, progress);
// Create a mapping from documents to the previous results the client says it has for them. That way as we
// process documents we know if we should tell the client it should stay the same, or we can tell it what
// the updated spans are.
var documentToPreviousParams = GetDocumentToPreviousParams(context, previousResults);
// Next process each file in priority order. Determine if spans are changed or unchanged since the
// last time we notified the client. Report back either to the client so they can update accordingly.
var orderedDocuments = GetOrderedDocuments(context, cancellationToken);
context.TraceInformation($"Processing {orderedDocuments.Length} documents");
foreach (var document in orderedDocuments)
{
context.TraceInformation($"Processing: {document.FilePath}");
var languageService = document.GetLanguageService<ISpellCheckSpanService>();
if (languageService == null)
{
context.TraceInformation($"Ignoring document '{document.FilePath}' because it does not support spell checking");
continue;
}
var newResultId = await _versionedCache.GetNewResultIdAsync(
documentToPreviousParams,
document,
computeVersionAsync: async () => await ComputeChecksumsAsync(document, cancellationToken).ConfigureAwait(false),
cancellationToken).ConfigureAwait(false);
if (newResultId != null)
{
context.TraceInformation($"Spans were changed for document: {document.FilePath}");
await foreach (var report in ComputeAndReportCurrentSpansAsync(
document, languageService, newResultId, cancellationToken).ConfigureAwait(false))
{
progress.Report(report);
}
}
else
{
context.TraceInformation($"Spans were unchanged for document: {document.FilePath}");
// Nothing changed between the last request and this one. Report a (null-spans, same-result-id)
// response to the client as that means they should just preserve the current spans they have for
// this file.
var previousParams = documentToPreviousParams[document];
progress.Report(CreateReport(previousParams.TextDocument, ranges: null, previousParams.PreviousResultId));
}
}
// If we had a progress object, then we will have been reporting to that. Otherwise, take what we've been
// collecting and return that.
context.TraceInformation($"{this.GetType()} finished getting spans");
return progress.GetFlattenedValues();
}
private static Dictionary<Document, PreviousPullResult> GetDocumentToPreviousParams(
RequestContext context, ImmutableArray<PreviousPullResult> previousResults)
{
Contract.ThrowIfNull(context.Solution);
var result = new Dictionary<Document, PreviousPullResult>();
foreach (var requestParams in previousResults)
{
if (requestParams.TextDocument != null)
{
var document = context.Solution.GetDocument(requestParams.TextDocument);
if (document != null)
result[document] = requestParams;
}
}
return result;
}
private async IAsyncEnumerable<TReport> ComputeAndReportCurrentSpansAsync(
Document document,
ISpellCheckSpanService service,
string resultId,
[EnumeratorCancellation] CancellationToken cancellationToken)
{
var textDocumentIdentifier = ProtocolConversions.DocumentToTextDocumentIdentifier(document);
var spans = await service.GetSpansAsync(document, cancellationToken).ConfigureAwait(false);
// protocol requires the results be in sorted order
spans = spans.Sort(static (s1, s2) => s1.TextSpan.CompareTo(s2.TextSpan));
if (spans.Length == 0)
{
yield return CreateReport(textDocumentIdentifier, [], resultId);
yield break;
}
// break things up into batches of 1000 items. That way we can send smaller messages to the client instead
// of one enormous one.
const int chunkSize = 1000;
var lastSpanEnd = 0;
for (var batchStart = 0; batchStart < spans.Length; batchStart += chunkSize)
{
var batchEnd = Math.Min(batchStart + chunkSize, spans.Length);
var batchSize = batchEnd - batchStart;
// Each span is encoded as a triple of ints. The 'kind', the 'relative start', and the 'length'.
// 'relative start' is the absolute-start for the first span, and then the offset from the end of the
// last span for all others.
var triples = new int[batchSize * 3];
var triplesIndex = 0;
for (var i = batchStart; i < batchEnd; i++)
{
var span = spans[i];
var kind = ProtocolConversions.SpellCheckSpanKindToSpellCheckableRangeKind(span.Kind);
triples[triplesIndex++] = (int)kind;
triples[triplesIndex++] = span.TextSpan.Start - lastSpanEnd;
triples[triplesIndex++] = span.TextSpan.Length;
lastSpanEnd = span.TextSpan.End;
}
Contract.ThrowIfTrue(triplesIndex != triples.Length);
yield return CreateReport(textDocumentIdentifier, triples, resultId);
}
}
private void HandleRemovedDocuments(
RequestContext context, ImmutableArray<PreviousPullResult> previousResults, BufferedProgress<TReport[]> progress)
{
Contract.ThrowIfNull(context.Solution);
foreach (var previousResult in previousResults)
{
var textDocument = previousResult.TextDocument;
if (textDocument != null)
{
var document = context.Solution.GetDocument(textDocument);
if (document == null)
{
context.TraceInformation($"Clearing spans for removed document: {textDocument.Uri}");
// Client is asking server about a document that no longer exists (i.e. was removed/deleted from
// the workspace). Report a (null-spans, null-result-id) response to the client as that means
// they should just consider the file deleted and should remove all spans information they've
// cached for it.
progress.Report(CreateReport(textDocument, ranges: null, resultId: null));
}
}
}
}
private static async Task<(Checksum parseOptionsChecksum, Checksum textChecksum)> ComputeChecksumsAsync(Document document, CancellationToken cancellationToken)
{
var project = document.Project;
var parseOptionsChecksum = project.State.GetParseOptionsChecksum();
var documentChecksumState = await document.State.GetStateChecksumsAsync(cancellationToken).ConfigureAwait(false);
var textChecksum = documentChecksumState.Text;
return (parseOptionsChecksum, textChecksum);
}
}
}
|