Parse sourcetext endpoint (#16899)

* Update ParseFile in TransparentCompiler

* Add unit tests

* Decrease cache size

Author: nojaf

src/Compiler/Facilities/Hashing.fs

@@ -78,4 +78,11 @@ module internal Md5Hasher =
 
     let addDateTimes (dts: System.DateTime seq) (s: byte array) = s |> addSeq dts addDateTime
 
+    let addInt (i: int) (s: byte array) =
+        i |> BitConverter.GetBytes |> addBytes <| s
+
+    let addIntegers (items: int seq) (s: byte array) = addSeq items addInt s
+
+    let addBooleans (items: bool seq) (s: byte array) = addSeq items addBool s
+
     let toString (bytes: byte array) = bytes |> System.BitConverter.ToString

src/Compiler/Facilities/Hashing.fsi

@@ -43,4 +43,8 @@ module internal Md5Hasher =
 
     val addDateTimes: dts: System.DateTime seq -> s: byte array -> byte array
 
+    val addIntegers: items: int seq -> s: byte array -> byte array
+
+    val addBooleans: items: bool seq -> s: byte array -> byte array
+
     val toString: bytes: byte array -> string

src/Compiler/Service/TransparentCompiler.fs

@@ -267,6 +267,9 @@ type internal CompilerCaches(sizeFactor: int) =
 
     member val ParseFile = AsyncMemoize(keepStrongly = 50 * sf, keepWeakly = 20 * sf, name = "ParseFile")
 
+    member val ParseFileWithoutProject =
+        AsyncMemoize<string, string, FSharpParseFileResults>(keepStrongly = 5 * sf, keepWeakly = 2 * sf, name = "ParseFileWithoutProject")
+
     member val ParseAndCheckFileInProject = AsyncMemoize(sf, 2 * sf, name = "ParseAndCheckFileInProject")
 
     member val ParseAndCheckAllFilesInProject = AsyncMemoizeDisabled(sf, 2 * sf, name = "ParseAndCheckFullProject")
@@ -1999,6 +2002,70 @@ type internal TransparentCompiler
             return parseResult
         }
 
+    member _.ParseFileWithoutProject
+        (
+            fileName: string,
+            sourceText: ISourceText,
+            options: FSharpParsingOptions,
+            cache: bool,
+            flatErrors: bool,
+            userOpName: string
+        ) : Async<FSharpParseFileResults> =
+        let parseFileAsync =
+            async {
+                let! ct = Async.CancellationToken
+
+                let diagnostics, parsedInput, anyErrors =
+                    ParseAndCheckFile.parseFile (sourceText, fileName, options, userOpName, false, flatErrors, false, ct)
+
+                return FSharpParseFileResults(diagnostics, parsedInput, anyErrors, Array.empty)
+            }
+
+        if not cache then
+            parseFileAsync
+        else
+            let cacheKey =
+                let sourceText = SourceTextNew.ofISourceText sourceText
+
+                { new ICacheKey<_, _> with
+                    member _.GetLabel() = shortPath fileName
+
+                    member _.GetKey() = fileName
+
+                    member _.GetVersion() =
+                        Md5Hasher.empty
+                        |> Md5Hasher.addStrings
+                            [
+                                yield fileName
+                                yield! options.ConditionalDefines
+                                yield! options.SourceFiles
+                                yield options.LangVersionText
+                            ]
+                        |> Md5Hasher.addBytes (sourceText.GetChecksum().ToArray())
+                        |> Md5Hasher.addIntegers
+                            [
+                                yield options.DiagnosticOptions.WarnLevel
+                                yield! options.DiagnosticOptions.WarnOff
+                                yield! options.DiagnosticOptions.WarnOn
+                                yield! options.DiagnosticOptions.WarnAsError
+                                yield! options.DiagnosticOptions.WarnAsWarn
+                            ]
+                        |> Md5Hasher.addBooleans
+                            [
+                                yield options.ApplyLineDirectives
+                                yield options.DiagnosticOptions.GlobalWarnAsError
+                                yield options.IsInteractive
+                                yield! (Option.toList options.IndentationAwareSyntax)
+                                yield! (Option.toList options.StrictIndentation)
+                                yield options.CompilingFSharpCore
+                                yield options.IsExe
+                            ]
+                        |> Md5Hasher.toString
+                }
+
+            caches.ParseFileWithoutProject.Get(cacheKey, NodeCode.AwaitAsync parseFileAsync)
+            |> Async.AwaitNodeCode
+
     member _.ParseAndCheckFileInProject(fileName: string, projectSnapshot: ProjectSnapshot, userOpName: string) =
         ignore userOpName
         ComputeParseAndCheckFileInProject fileName projectSnapshot
@@ -2450,8 +2517,8 @@ type internal TransparentCompiler
                 cache: bool,
                 flatErrors: bool,
                 userOpName: string
-            ) =
-            backgroundCompiler.ParseFile(fileName, sourceText, options, cache, flatErrors, userOpName)
+            ) : Async<FSharpParseFileResults> =
+            this.ParseFileWithoutProject(fileName, sourceText, options, cache, flatErrors, userOpName)
 
         member this.TryGetRecentCheckResultsForFile
             (

src/Compiler/Service/TransparentCompiler.fsi

@@ -125,6 +125,8 @@ type internal CompilerCaches =
     member ParseFile:
         AsyncMemoize<((string * string) * string), (string * string * bool), ProjectSnapshot.FSharpParsedFile>
 
+    member ParseFileWithoutProject: AsyncMemoize<string, string, FSharpParseFileResults>
+
     member ProjectExtras: AsyncMemoizeDisabled<obj, obj, obj>
 
     member SemanticClassification: AsyncMemoize<(string * (string * string)), string, SemanticClassificationView option>

src/Compiler/Service/service.fsi

@@ -97,7 +97,7 @@ type public FSharpChecker =
     /// <param name="fileName">The path for the file. The file name is used as a module name for implicit top level modules (e.g. in scripts).</param>
     /// <param name="sourceText">The source to be parsed.</param>
     /// <param name="options">Parsing options for the project or script.</param>
-    /// <param name="cache">Store the parse in a size-limited cache assocaited with the FSharpChecker. Default: true</param>
+    /// <param name="cache">Store the parse in a size-limited cache associated with the FSharpChecker. Default: true</param>
     /// <param name="userOpName">An optional string used for tracing compiler operations associated with this request.</param>
     member ParseFile:
         fileName: string * sourceText: ISourceText * options: FSharpParsingOptions * ?cache: bool * ?userOpName: string ->

tests/FSharp.Compiler.ComponentTests/FSharpChecker/TransparentCompiler.fs

@@ -1125,4 +1125,45 @@ let ``The script load closure should always be evaluated`` useTransparentCompile
             | FSharpCheckFileAnswer.Succeeded checkFileResults -> Assert.Equal(0, checkFileResults.Diagnostics.Length)
         finally
             FileSystemAutoOpens.FileSystem <- currentFileSystem
-    }
+    }
+
+[<Fact>]
+let ``Parsing without cache and without project snapshot`` () =
+    async {
+        let checker = FSharpChecker.Create(useTransparentCompiler = true)
+        let fileName = "Temp.fs"
+        let sourceText = "let a = 0" |> SourceText.ofString
+        let parsingOptions = { FSharpParsingOptions.Default with SourceFiles = [| fileName |]; IsExe = true }
+        let! parseResult = checker.ParseFile(fileName, sourceText, parsingOptions, cache = false)
+        Assert.False(parseResult.ParseHadErrors)
+        Assert.True(Array.isEmpty parseResult.Diagnostics)
+        Assert.Equal(0, checker.Caches.ParseFile.Count)
+        Assert.Equal(0, checker.Caches.ParseFileWithoutProject.Count)
+    }
+
+// In this scenario, the user is typing something in file B.fs.
+// The idea is that the IDE will introduce an additional (fake) identifier in order to have a potential complete syntax tree.
+// The user never wrote this code so we need to ensure nothing is added to checker.Caches.ParseFile
+[<Fact>]
+let ``Parsing with cache and without project snapshot`` () =
+    async {
+        let checker = FSharpChecker.Create(useTransparentCompiler = true)
+        let fileName = "B.fs"
+        let parsingOptions = { FSharpParsingOptions.Default with SourceFiles = [| "A.fs"; fileName; "C.fs" |] }        
+        let sourceText =
+            SourceText.ofString """
+module B
+
+let b : int = ExtraIdentUserNeverWroteRulezzz
+"""
+        let! parseResult = checker.ParseFile(fileName, sourceText, parsingOptions, cache = true)
+        Assert.False(parseResult.ParseHadErrors)
+        Assert.True(Array.isEmpty parseResult.Diagnostics)
+        
+        let! parseAgainResult = checker.ParseFile(fileName, sourceText, parsingOptions, cache = true)
+        Assert.False(parseAgainResult.ParseHadErrors)
+        Assert.True(Array.isEmpty parseAgainResult.Diagnostics)
+
+        Assert.Equal(0, checker.Caches.ParseFile.Count)
+        Assert.Equal(1, checker.Caches.ParseFileWithoutProject.Count)
+    }