[incrParse] Add validation of incremental parsing

If enabled using the environment variable
SOURCEKIT_INCREMENTAL_PARSE_VALIDATION, the incrementally parsed syntax
tree will be compared to the from-scratch parsing syntax tree. If they
differ a warning is emitted and log files showing the difference written
to a temporary directory.

include/swift/Syntax/Serialization/SyntaxSerialization.h

@@ -27,6 +27,10 @@
 namespace swift {
 namespace json {
 
+/// The associated value will be interpreted as \c bool. If \c true the node IDs
+/// will not be included in the serialized JSON.
+static void *DontSerializeNodeIdsUserInfoKey = &DontSerializeNodeIdsUserInfoKey;
+
 /// Serialization traits for SourcePresence.
 template <>
 struct ScalarEnumerationTraits<syntax::SourcePresence> {
@@ -141,8 +145,12 @@ struct ObjectTraits<syntax::RawSyntax> {
     }
     auto presence = value.getPresence();
     out.mapRequired("presence", presence);
-    auto nodeId = value.getId();
-    out.mapRequired("id", nodeId);
+
+    bool omitNodeId = (bool)out.getUserInfo()[DontSerializeNodeIdsUserInfoKey];
+    if (!omitNodeId) {
+      auto nodeId = value.getId();
+      out.mapRequired("id", nodeId);
+    }
   }
 };
 

tools/SourceKit/lib/SwiftLang/SwiftEditor.cpp

@@ -41,6 +41,7 @@
 #include "swift/Syntax/SyntaxNodes.h"
 
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/FileSystem.h"
 #include "llvm/Support/MemoryBuffer.h"
 #include "llvm/Support/Mutex.h"
 
@@ -2007,10 +2008,16 @@ const SourceManager &SwiftEditorDocument::getSourceManager() const {
   return Impl.SyntaxInfo->getSourceManager();
 }
 
+SourceManager &SwiftEditorDocument::getSourceManager() {
+  return Impl.SyntaxInfo->getSourceManager();
+}
+
 unsigned SwiftEditorDocument::getBufferID() const {
   return Impl.SyntaxInfo->getBufferID();
 }
 
+std::string SwiftEditorDocument::getFilePath() const { return Impl.FilePath; }
+
 void SwiftEditorDocument::formatText(unsigned Line, unsigned Length,
                                      EditorConsumer &Consumer) {
   auto SyntaxInfo = Impl.getSyntaxInfo();
@@ -2234,11 +2241,97 @@ void SwiftLangSupport::editorClose(StringRef Name, bool RemoveCache) {
 // EditorReplaceText
 //===----------------------------------------------------------------------===//
 
+void verifyIncrementalParse(SwiftEditorDocumentRef EditorDoc,
+                            unsigned EditOffset, unsigned EditLength,
+                            StringRef PreEditText, StringRef ReplaceText) {
+  swift::json::Output::UserInfoMap JsonUserInfo;
+  JsonUserInfo[swift::json::DontSerializeNodeIdsUserInfoKey] =
+      reinterpret_cast<void *>(true);
+
+  // Dump the incremental syntax tree
+  std::string IncrTreeString;
+  llvm::raw_string_ostream IncrTreeStream(IncrTreeString);
+  swift::json::Output IncrTreeOutput(IncrTreeStream, JsonUserInfo);
+  IncrTreeOutput << *EditorDoc->getSyntaxTree()->getRaw();
+
+  // Reparse the file from scratch
+  CompilerInvocation Invocation;
+  Invocation.getLangOptions().BuildSyntaxTree = true;
+  std::vector<std::string> Args;
+  SwiftDocumentSyntaxInfo ScratchSyntaxInfo(Invocation,
+                                            EditorDoc->getLatestSnapshot(),
+                                            Args, EditorDoc->getFilePath());
+  ScratchSyntaxInfo.parse();
+
+  // Dump the from-scratch syntax tree
+  std::string FromScratchTreeString;
+  llvm::raw_string_ostream ScratchTreeStream(FromScratchTreeString);
+  swift::json::Output ScratchTreeOutput(ScratchTreeStream, JsonUserInfo);
+  auto SyntaxRoot = ScratchSyntaxInfo.getSourceFile().getSyntaxRoot();
+  ScratchTreeOutput << *SyntaxRoot.getRaw();
+
+  // If the serialized format of the two trees doesn't match incremental parsing
+  // we have found an error.
+  if (IncrTreeStream.str().compare(ScratchTreeStream.str())) {
+    LOG_SECTION("Incremental Parsing", Warning) {
+      Log->getOS() << "Incremental parsing different to from scratch parsing\n";
+      Log->getOS() << "Edit was " << EditOffset << "-"
+                   << (EditOffset + EditLength) << "='" << ReplaceText << "'"
+                   << " pre-edit-text: '" << PreEditText << "'\n";
+
+      SmallString<32> DirectoryName;
+      if (llvm::sys::fs::createUniqueDirectory(
+              "SourceKit-IncrementalParsing-Inconsistency", DirectoryName)) {
+        Log->getOS() << "Failed to create log directory\n";
+      }
+
+      std::error_code ErrorCode;
+
+      // Write the incremental syntax tree
+      auto IncrTreeFilename = DirectoryName + "/incrementalTree.json";
+      llvm::raw_fd_ostream IncrementalFilestream(
+          IncrTreeFilename.str(), ErrorCode, llvm::sys::fs::F_RW);
+      IncrementalFilestream << IncrTreeStream.str();
+      if (ErrorCode) {
+        Log->getOS() << "Failed to write incremental syntax tree to "
+                     << IncrTreeFilename << "(error code " << ErrorCode.value()
+                     << ": " << ErrorCode.message() << ")\n";
+      } else {
+        Log->getOS() << "Incremental syntax tree written to "
+                     << IncrTreeFilename << '\n';
+      }
+
+      // Write from-scratch syntax tree
+      auto ScratchTreeFilename = DirectoryName + "/fromScratchTree.json";
+      llvm::raw_fd_ostream ScratchTreeFilestream(
+          ScratchTreeFilename.str(), ErrorCode, llvm::sys::fs::F_RW);
+      ScratchTreeFilestream << ScratchTreeStream.str();
+      if (ErrorCode) {
+        Log->getOS() << "Failed to write from-scratch syntax tree to "
+                     << ScratchTreeFilename << "(error code "
+                     << ErrorCode.value() << ": " << ErrorCode.message()
+                     << ")\n";
+      } else {
+        Log->getOS() << "From-scratch syntax tree written to "
+                     << ScratchTreeFilename << '\n';
+      }
+
+      // Write source file
+      auto SourceFilename = DirectoryName + "/postEditSource.swift";
+      llvm::raw_fd_ostream SourceFilestream(SourceFilename.str(), ErrorCode,
+                                            llvm::sys::fs::F_RW);
+      auto FileBuffer = EditorDoc->getLatestSnapshot()->getBuffer();
+      SourceFilestream << FileBuffer->getText();
+    }
+  }
+}
+
 void SwiftLangSupport::editorReplaceText(StringRef Name,
                                          llvm::MemoryBuffer *Buf,
                                          unsigned Offset, unsigned Length,
                                          EditorConsumer &Consumer) {
   bool LogReuseRegions = ::getenv("SOURCEKIT_LOG_INCREMENTAL_REUSE_REGIONS");
+  bool ValidateSyntaxTree = ::getenv("SOURCEKIT_INCREMENTAL_PARSE_VALIDATION");
 
   auto EditorDoc = EditorDocuments.getByUnresolvedName(Name);
   if (!EditorDoc) {
@@ -2248,6 +2341,13 @@ void SwiftLangSupport::editorReplaceText(StringRef Name,
 
   ImmutableTextSnapshotRef Snapshot;
   if (Length != 0 || Buf->getBufferSize() != 0) {
+    std::string PreEditText;
+    if (ValidateSyntaxTree) {
+      auto CurBuffer = EditorDoc->getLatestSnapshot()->getBuffer();
+      auto BufferStart = CurBuffer->getInternalBuffer()->getBufferStart();
+      StringRef PreEditTextRef(BufferStart + Offset, Length);
+      PreEditText = PreEditTextRef.str();
+    }
     Snapshot = EditorDoc->replaceText(Offset, Length, Buf,
                                       Consumer.needsSemanticInfo());
     assert(Snapshot);
@@ -2310,6 +2410,11 @@ void SwiftLangSupport::editorReplaceText(StringRef Name,
     } else {
       Consumer.handleSyntaxReuseRegions({});
     }
+
+    if (ValidateSyntaxTree) {
+      verifyIncrementalParse(EditorDoc, Offset, Length, PreEditText,
+                             Buf->getBuffer());
+    }
   } else {
     Snapshot = EditorDoc->getLatestSnapshot();
   }

tools/SourceKit/lib/SwiftLang/SwiftLangSupport.h

@@ -111,9 +111,12 @@ public:
   const llvm::Optional<swift::SourceFileSyntax> &getSyntaxTree() const;
 
   const swift::SourceManager &getSourceManager() const;
+  swift::SourceManager &getSourceManager();
 
   /// Get the buffer ID of this file in its source manager
   unsigned getBufferID() const;
+
+  std::string getFilePath() const;
 };
 
 typedef IntrusiveRefCntPtr<SwiftEditorDocument> SwiftEditorDocumentRef;