[WIP] Prevent document reload with old content

src/Analysis/Ast/Impl/Documents/Definitions/IDocument.cs

@@ -63,9 +63,9 @@ public interface IDocument: IPythonModule, IDisposable {
         void Update(IEnumerable<DocumentChange> changes);
 
         /// <summary>
-        /// Resets document buffer to the provided content or tries to load it if content is null, then parses and analyzes document.
+        /// Forces parse and analysis of the document.
         /// </summary>
-        void Reset(string content);
+        void Invalidate();
 
         /// <summary>
         /// Provides collection of parsing errors, if any.

src/Analysis/Ast/Impl/Documents/DocumentBuffer.cs

@@ -16,13 +16,16 @@
 using System.Collections.Generic;
 using System.Linq;
 using System.Text;
+using Microsoft.Python.Core.Diagnostics;
 using Microsoft.Python.Parsing;
 
 namespace Microsoft.Python.Analysis.Documents {
     internal sealed class DocumentBuffer {
         private readonly object _lock = new object();
         private StringBuilder _sb = new StringBuilder();
         private string _content;
+        private bool _contentDropped;
+        private bool _populated;
 
         public int Version { get; private set; }
 
@@ -34,16 +37,48 @@ public string Text {
             }
         }
 
-        public void Reset(int version, string content) {
+        /// <summary>
+        /// Clear buffer content to save memory.
+        /// The buffer cannot be modified after this point.
+        /// </summary>
+        public void Clear() {
             lock (_lock) {
-                Version = version;
+                _content = string.Empty;
+                _sb = null;
+                _contentDropped = true;
+            }
+        }
+
+        /// <summary>
+        /// Advances content version of the buffer without changing the contents.
+        /// Typically used to invalidate analysis.
+        /// </summary>
+        public void MarkChanged() {
+            lock (_lock) {
+                Check.InvalidOperation(_populated, "Buffer is not populated.");
+                Check.InvalidOperation(!_contentDropped, "Buffer content was dropped and cannot be updated.");
+                Version++;
+            }
+        }
+
+        /// <summary>
+        /// Populates buffer with initial content. This can ony happen once.
+        /// </summary>
+        /// <param name="content"></param>
+        public void Populate(string content) {
+            lock (_lock) {
+                Check.InvalidOperation(!_populated, "Buffer is already populated.");
+                Check.InvalidOperation(!_contentDropped, "Buffer content was dropped and cannot be updated.");
+                Version = 0;
                 _content = content ?? string.Empty;
                 _sb = null;
+                _populated = true;
             }
         }
 
         public void Update(IEnumerable<DocumentChange> changes) {
             lock (_lock) {
+                Check.InvalidOperation(!_contentDropped, "Buffer content was dropped and cannot be updated.");
                 _sb = _sb ?? new StringBuilder(_content);
 
                 foreach (var change in changes) {
@@ -59,9 +94,11 @@ public void Update(IEnumerable<DocumentChange> changes) {
 
                     var start = NewLineLocation.LocationToIndex(lineLoc, change.ReplacedSpan.Start, _sb.Length);
                     var end = NewLineLocation.LocationToIndex(lineLoc, change.ReplacedSpan.End, _sb.Length);
+
                     if (end > start) {
                         _sb.Remove(start, end - start);
                     }
+
                     if (!string.IsNullOrEmpty(change.InsertedText)) {
                         _sb.Insert(start, change.InsertedText);
                     }
@@ -98,7 +135,6 @@ public IEnumerable<NewLineLocation> GetNewLineLocations() {
                             if (i == _sb.Length - 1) {
                                 yield return new NewLineLocation(i + 1, NewLineKind.None);
                             }
-
                             break;
                     }
                 }

src/Analysis/Ast/Impl/Documents/RunningDocumentTable.cs

@@ -106,7 +106,7 @@ public IDocument OpenDocument(Uri uri, string content, string filePath = null) {
                     };
                     entry = CreateDocument(mco);
                 }
-                justOpened = TryOpenDocument(entry, content);
+                justOpened = TryOpenDocument(entry);
                 document = entry.Document;
             }
 
@@ -217,7 +217,7 @@ public void ReloadAll() {
             }
 
             foreach (var (_, entry) in opened) {
-                entry.Document.Reset(null);
+                entry.Document.Invalidate();
             }
         }
 
@@ -277,10 +277,9 @@ private bool TryAddModulePath(ModuleCreationOptions mco) {
             return true;
         }
 
-        private bool TryOpenDocument(DocumentEntry entry, string content) {
+        private bool TryOpenDocument(DocumentEntry entry) {
             if (!entry.Document.IsOpen) {
                 entry.Document.IsOpen = true;
-                entry.Document.Reset(content);
                 entry.LockCount++;
                 return true;
             }

src/Analysis/Ast/Impl/Modules/PythonModule.cs

@@ -119,7 +119,7 @@ internal PythonModule(ModuleCreationOptions creationOptions, IServiceContainer s
             }
 
             IsPersistent = creationOptions.IsPersistent;
-            InitializeContent(creationOptions.Content, 0);
+            InitializeContent(creationOptions.Content);
         }
 
         #region IPythonType
@@ -315,14 +315,12 @@ public void Update(IEnumerable<DocumentChange> changes) {
             Services.GetService<IPythonAnalyzer>().InvalidateAnalysis(this);
         }
 
-        public void Reset(string content) {
+        public void Invalidate() {
             lock (_syncObj) {
-                if (content != Content) {
-                    ContentState = State.None;
-                    InitializeContent(content, _buffer.Version + 1);
-                }
+                ContentState = State.None;
+                _buffer.MarkChanged();
+                Parse();
             }
-
             Services.GetService<IPythonAnalyzer>().InvalidateAnalysis(this);
         }
 
@@ -451,7 +449,7 @@ public void NotifyAnalysisComplete(IDocumentAnalysis analysis) {
                 ContentState = State.Analyzed;
 
                 if (ModuleType != ModuleType.User) {
-                    _buffer.Reset(_buffer.Version, string.Empty);
+                    _buffer.Clear();
                 }
             }
 
@@ -487,7 +485,7 @@ public void AddAstNode(object o, Node n) {
         public void ClearContent() {
             lock (_syncObj) {
                 if (ModuleType != ModuleType.User) {
-                    _buffer.Reset(_buffer.Version, string.Empty);
+                    _buffer.Clear();
                     _astMap.Clear();
                 }
             }
@@ -515,26 +513,29 @@ protected virtual string LoadContent() {
             return null; // Keep content as null so module can be loaded later.
         }
 
-        private void InitializeContent(string content, int version) {
+        /// <summary>
+        /// Populates buffer with content. Content can either be provided, such as when
+        /// user opens the document or generated such as when module is compiled and
+        /// scraper will generate the content in the overridden LoadContent().
+        /// </summary>
+        private void InitializeContent(string content) {
             lock (_syncObj) {
-                LoadContent(content, version);
-
-                var startParse = ContentState < State.Parsing && (_parsingTask == null || version > 0);
-                if (startParse) {
+                LoadContent(content);
+                if (ContentState < State.Parsing && _parsingTask == null) {
                     Parse();
                 }
             }
         }
 
-        private void LoadContent(string content, int version) {
+        private void LoadContent(string content) {
             if (ContentState < State.Loading) {
                 try {
                     if (IsPersistent) {
                         content = string.Empty;
                     } else {
                         content = content ?? LoadContent();
                     }
-                    _buffer.Reset(version, content);
+                    _buffer.Populate(content);
                     ContentState = State.Loaded;
                 } catch (IOException) { } catch (UnauthorizedAccessException) { }
             }

src/Analysis/Ast/Test/DocumentBufferTests.cs

@@ -29,7 +29,7 @@ public class DocumentBufferTests {
         [TestMethod, Priority(0)]
         public void BasicDocumentBuffer() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"def f(x):
+            doc.Populate(@"def f(x):
     return
 
 def g(y):
@@ -77,7 +77,7 @@ def g(y):
         public void ResetDocumentBuffer() {
             var doc = new DocumentBuffer();
 
-            doc.Reset(0, string.Empty);
+            doc.Populate(string.Empty);
             Assert.AreEqual(string.Empty, doc.Text);
 
             doc.Update(new[] {
@@ -86,18 +86,13 @@ public void ResetDocumentBuffer() {
 
             Assert.AreEqual("text", doc.Text);
             Assert.AreEqual(1, doc.Version);
-
-            doc.Reset(0, @"abcdef");
-
-            Assert.AreEqual(@"abcdef", doc.Text);
-            Assert.AreEqual(0, doc.Version);
         }
 
         [TestMethod, Priority(0)]
         public void ReplaceAllDocumentBuffer() {
             var doc = new DocumentBuffer();
 
-            doc.Reset(0, string.Empty);
+            doc.Populate(string.Empty);
             Assert.AreEqual(string.Empty, doc.Text);
 
             doc.Update(new[] {
@@ -134,7 +129,7 @@ public void ReplaceAllDocumentBuffer() {
         [TestMethod, Priority(0)]
         public void DeleteMultipleDisjoint() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"
+            doc.Populate(@"
 line1
 line2
 line3
@@ -157,7 +152,7 @@ public void DeleteMultipleDisjoint() {
         [TestMethod, Priority(0)]
         public void InsertMultipleDisjoint() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"
+            doc.Populate(@"
 line
 line
 line
@@ -180,7 +175,7 @@ public void InsertMultipleDisjoint() {
         [TestMethod, Priority(0)]
         public void DeleteAcrossLines() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"
+            doc.Populate(@"
 line1
 line2
 line3
@@ -199,7 +194,7 @@ public void DeleteAcrossLines() {
         [TestMethod, Priority(0)]
         public void SequentialChanges() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"
+            doc.Populate(@"
 line1
 line2
 line3
@@ -218,7 +213,7 @@ public void SequentialChanges() {
         [TestMethod, Priority(0)]
         public void InsertTopToBottom() {
             var doc = new DocumentBuffer();
-            doc.Reset(0, @"linelinelineline");
+            doc.Populate(@"linelinelineline");
             doc.Update(new[] {
                 DocumentChange.Insert("\n", new SourceLocation(1, 1)),
                 DocumentChange.Insert("1\n", new SourceLocation(2, 5)),
@@ -233,7 +228,7 @@ public void InsertTopToBottom() {
         [DataTestMethod, Priority(0)]
         public void NewLines(string s, NewLineLocation[] expected) {
             var doc = new DocumentBuffer();
-            doc.Reset(0, s);
+            doc.Populate(s);
             var nls = doc.GetNewLineLocations().ToArray();
             for (var i = 0; i < nls.Length; i++) {
                 Assert.AreEqual(nls[i].Kind, expected[i].Kind);
@@ -281,7 +276,18 @@ public IEnumerable<object[]> GetData(MethodInfo methodInfo) =>
                             new NewLineLocation(12, NewLineKind.LineFeed),
                             new NewLineLocation(13, NewLineKind.None)
                         }
+                    },
+                    new object[] {
+                    "\r\na\r\n\r\n\r\n \r\n",
+                    new NewLineLocation[] {
+                        new NewLineLocation(2, NewLineKind.CarriageReturnLineFeed),
+                        new NewLineLocation(5, NewLineKind.CarriageReturnLineFeed),
+                        new NewLineLocation(7, NewLineKind.CarriageReturnLineFeed),
+                        new NewLineLocation(9, NewLineKind.CarriageReturnLineFeed),
+                        new NewLineLocation(12, NewLineKind.CarriageReturnLineFeed),
+                        new NewLineLocation(13, NewLineKind.None)
                     }
+                }
                 };
             public string GetDisplayName(MethodInfo methodInfo, object[] data)
                 => data != null ? $"{methodInfo.Name} ({FormatName((string)data[0])})" : null;