[AST] Add generator for source location introspection
authorStephen Kelly <steveire@gmail.com>
Sat, 12 Dec 2020 13:17:49 +0000 (13:17 +0000)
committerStephen Kelly <steveire@gmail.com>
Wed, 10 Mar 2021 22:38:39 +0000 (22:38 +0000)
Generate a json file containing descriptions of AST classes and their
public accessors which return SourceLocation or SourceRange.

Use the JSON file to generate a C++ API and implementation for accessing
the source locations and method names for accessing them for a given AST
node.

This new API can be used to implement 'srcloc' output in clang-query:

  http://ce.steveire.com/z/m_kTIo

In this first version of this feature, only the accessors for Stmt
classes are generated, not Decls, TypeLocs etc.  Those can be added
after this change is reviewed, as this change is mostly about
infrastructure of these code generators.

Differential Revision: https://reviews.llvm.org/D93164

12 files changed:
clang/include/clang/Tooling/NodeIntrospection.h [new file with mode: 0644]
clang/lib/Tooling/CMakeLists.txt
clang/lib/Tooling/DumpTool/APIData.h [new file with mode: 0644]
clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.cpp [new file with mode: 0644]
clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.h [new file with mode: 0644]
clang/lib/Tooling/DumpTool/CMakeLists.txt [new file with mode: 0644]
clang/lib/Tooling/DumpTool/ClangSrcLocDump.cpp [new file with mode: 0644]
clang/lib/Tooling/DumpTool/generate_cxx_src_locs.py [new file with mode: 0755]
clang/lib/Tooling/NodeIntrospection.cpp [new file with mode: 0644]
clang/unittests/CMakeLists.txt
clang/unittests/Introspection/CMakeLists.txt [new file with mode: 0644]
clang/unittests/Introspection/IntrospectionTest.cpp [new file with mode: 0644]

diff --git a/clang/include/clang/Tooling/NodeIntrospection.h b/clang/include/clang/Tooling/NodeIntrospection.h
new file mode 100644 (file)
index 0000000..abaa58b
--- /dev/null
@@ -0,0 +1,85 @@
+//===- NodeIntrospection.h ------------------------------------*- C++ -*---===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+//  This file contains the implementation of the NodeIntrospection.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_TOOLING_NODEINTROSPECTION_H
+#define LLVM_CLANG_TOOLING_NODEINTROSPECTION_H
+
+#include "clang/AST/ASTTypeTraits.h"
+#include "clang/AST/DeclarationName.h"
+
+#include <memory>
+#include <set>
+
+namespace clang {
+
+class Stmt;
+
+namespace tooling {
+
+class LocationCall {
+public:
+  enum LocationCallFlags { NoFlags, ReturnsPointer, IsCast };
+  LocationCall(std::shared_ptr<LocationCall> on, std::string name,
+               LocationCallFlags flags = NoFlags)
+      : m_on(on), m_name(name), m_flags(flags) {}
+  LocationCall(std::shared_ptr<LocationCall> on, std::string name,
+               std::vector<std::string> const &args,
+               LocationCallFlags flags = NoFlags)
+      : m_on(on), m_name(name), m_flags(flags) {}
+
+  LocationCall *on() const { return m_on.get(); }
+  StringRef name() const { return m_name; }
+  std::vector<std::string> const &args() const { return m_args; }
+  bool returnsPointer() const { return m_flags & ReturnsPointer; }
+  bool isCast() const { return m_flags & IsCast; }
+
+private:
+  std::shared_ptr<LocationCall> m_on;
+  std::string m_name;
+  std::vector<std::string> m_args;
+  LocationCallFlags m_flags;
+};
+
+class LocationCallFormatterCpp {
+public:
+  static std::string format(LocationCall *Call);
+};
+
+namespace internal {
+struct RangeLessThan {
+  bool operator()(
+      std::pair<SourceRange, std::shared_ptr<LocationCall>> const &LHS,
+      std::pair<SourceRange, std::shared_ptr<LocationCall>> const &RHS) const;
+};
+} // namespace internal
+
+template <typename T, typename U, typename Comp = std::less<std::pair<T, U>>>
+using UniqueMultiMap = std::set<std::pair<T, U>, Comp>;
+
+using SourceLocationMap =
+    UniqueMultiMap<SourceLocation, std::shared_ptr<LocationCall>>;
+using SourceRangeMap =
+    UniqueMultiMap<SourceRange, std::shared_ptr<LocationCall>,
+                   internal::RangeLessThan>;
+
+struct NodeLocationAccessors {
+  SourceLocationMap LocationAccessors;
+  SourceRangeMap RangeAccessors;
+};
+
+namespace NodeIntrospection {
+NodeLocationAccessors GetLocations(clang::Stmt const *Object);
+NodeLocationAccessors GetLocations(clang::DynTypedNode const &Node);
+} // namespace NodeIntrospection
+} // namespace tooling
+} // namespace clang
+#endif
index 7a58af5..9600bf7 100644 (file)
@@ -8,10 +8,78 @@ add_subdirectory(Core)
 add_subdirectory(Inclusions)
 add_subdirectory(Refactoring)
 add_subdirectory(ASTDiff)
+add_subdirectory(DumpTool)
 add_subdirectory(Syntax)
 add_subdirectory(DependencyScanning)
 add_subdirectory(Transformer)
 
+find_package(Python3 COMPONENTS Interpreter)
+
+# The generation of ASTNodeAPI.json takes a long time in a
+# Debug build due to parsing AST.h. Disable the processing
+# but setting CLANG_TOOLING_BUILD_AST_INTROSPECTION as an
+# internal hidden setting to override.
+# When the processing is disabled, a trivial/empty JSON
+# file is generated by clang-ast-dump and generate_cxx_src_locs.py
+# generates the same API, but with a trivial implementation.
+
+option(CLANG_TOOLING_BUILD_AST_INTROSPECTION "Enable AST introspection" TRUE)
+set(skip_expensive_processing $<OR:$<CONFIG:Debug>,$<NOT:$<BOOL:${CLANG_TOOLING_BUILD_AST_INTROSPECTION}>>>)
+
+add_custom_command(
+    COMMENT Generate ASTNodeAPI.json
+    OUTPUT ${CMAKE_BINARY_DIR}/ASTNodeAPI.json
+    DEPENDS clang-ast-dump clang-headers
+    COMMAND
+    $<TARGET_FILE:clang-ast-dump>
+      # Skip this in debug mode because parsing AST.h is too slow
+      --skip-processing=${skip_expensive_processing}
+      --astheader=${CMAKE_SOURCE_DIR}/../clang/include/clang/AST/AST.h
+      -I ${CMAKE_BINARY_DIR}/lib/clang/${LLVM_VERSION_MAJOR}.${LLVM_VERSION_MINOR}.${LLVM_VERSION_PATCH}/include
+      -I ${CMAKE_SOURCE_DIR}/../clang/include
+      -I ${CMAKE_BINARY_DIR}/tools/clang/include/
+      -I ${CMAKE_BINARY_DIR}/include
+      -I ${CMAKE_SOURCE_DIR}/include
+      --json-output-path ${CMAKE_BINARY_DIR}/ASTNodeAPI.json
+)
+
+add_custom_target(run-ast-api-dump-tool
+    DEPENDS ${CMAKE_BINARY_DIR}/ASTNodeAPI.json
+)
+
+# Replace the last lib component of the current binary directory with include
+string(FIND ${CMAKE_CURRENT_BINARY_DIR} "/lib/" PATH_LIB_START REVERSE)
+if(PATH_LIB_START EQUAL -1)
+  message(FATAL_ERROR "Couldn't find lib component in binary directory")
+endif()
+math(EXPR PATH_LIB_END "${PATH_LIB_START}+5")
+string(SUBSTRING ${CMAKE_CURRENT_BINARY_DIR} 0 ${PATH_LIB_START} PATH_HEAD)
+string(SUBSTRING ${CMAKE_CURRENT_BINARY_DIR} ${PATH_LIB_END} -1 PATH_TAIL)
+string(CONCAT BINARY_INCLUDE_DIR ${PATH_HEAD} "/include/clang/" ${PATH_TAIL})
+
+add_custom_command(
+    COMMENT Generate NodeIntrospection.inc
+    OUTPUT ${BINARY_INCLUDE_DIR}/NodeIntrospection.inc
+    DEPENDS ${CMAKE_BINARY_DIR}/ASTNodeAPI.json ${CMAKE_CURRENT_SOURCE_DIR}/DumpTool/generate_cxx_src_locs.py
+    COMMAND
+    ${CMAKE_COMMAND} -E make_directory
+      ${CMAKE_CURRENT_BINARY_DIR}/generated/
+    COMMAND
+    ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/DumpTool/generate_cxx_src_locs.py
+      --json-input-path ${CMAKE_BINARY_DIR}/ASTNodeAPI.json
+      --output-file generated/NodeIntrospection.inc
+      --empty-implementation ${skip_expensive_processing}
+    COMMAND
+    ${CMAKE_COMMAND} -E copy_if_different
+      ${CMAKE_CURRENT_BINARY_DIR}/generated/NodeIntrospection.inc
+      ${BINARY_INCLUDE_DIR}/NodeIntrospection.inc
+)
+
+add_custom_target(run-ast-api-generate-tool
+    DEPENDS
+    ${BINARY_INCLUDE_DIR}/NodeIntrospection.inc
+)
+
 add_clang_library(clangTooling
   AllTUsExecution.cpp
   ArgumentsAdjusters.cpp
@@ -27,6 +95,8 @@ add_clang_library(clangTooling
   Refactoring.cpp
   RefactoringCallbacks.cpp
   StandaloneExecution.cpp
+  NodeIntrospection.cpp
+  ${BINARY_INCLUDE_DIR}/NodeIntrospection.inc
   Tooling.cpp
 
   DEPENDS
diff --git a/clang/lib/Tooling/DumpTool/APIData.h b/clang/lib/Tooling/DumpTool/APIData.h
new file mode 100644 (file)
index 0000000..bf68a3b
--- /dev/null
@@ -0,0 +1,33 @@
+//===- APIData.h ---------------------------------------------*- C++ -*----===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_LIB_TOOLING_DUMPTOOL_APIDATA_H
+#define LLVM_CLANG_LIB_TOOLING_DUMPTOOL_APIDATA_H
+
+#include <string>
+#include <vector>
+
+namespace clang {
+namespace tooling {
+
+struct ClassData {
+
+  bool isEmpty() const {
+    return ASTClassLocations.empty() && ASTClassRanges.empty();
+  }
+
+  std::vector<std::string> ASTClassLocations;
+  std::vector<std::string> ASTClassRanges;
+  // TODO: Extend this with locations available via typelocs etc.
+};
+
+} // namespace tooling
+} // namespace clang
+
+#endif
diff --git a/clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.cpp b/clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.cpp
new file mode 100644 (file)
index 0000000..3f97b7d
--- /dev/null
@@ -0,0 +1,171 @@
+//===- ASTSrcLocProcessor.cpp --------------------------------*- C++ -*----===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "ASTSrcLocProcessor.h"
+
+#include "clang/Frontend/CompilerInstance.h"
+#include "llvm/Support/JSON.h"
+
+using namespace clang::tooling;
+using namespace llvm;
+using namespace clang::ast_matchers;
+
+ASTSrcLocProcessor::ASTSrcLocProcessor(StringRef JsonPath)
+    : JsonPath(JsonPath) {
+
+  MatchFinder::MatchFinderOptions FinderOptions;
+
+  Finder = std::make_unique<MatchFinder>(std::move(FinderOptions));
+  Finder->addMatcher(
+      cxxRecordDecl(
+          isDefinition(),
+          isSameOrDerivedFrom(
+              // TODO: Extend this with other clades
+              namedDecl(hasName("clang::Stmt")).bind("nodeClade")),
+          optionally(isDerivedFrom(cxxRecordDecl().bind("derivedFrom"))))
+          .bind("className"),
+      this);
+}
+
+std::unique_ptr<clang::ASTConsumer>
+ASTSrcLocProcessor::createASTConsumer(clang::CompilerInstance &Compiler,
+                                      StringRef File) {
+  return Finder->newASTConsumer();
+}
+
+llvm::json::Object toJSON(llvm::StringMap<std::vector<StringRef>> const &Obj) {
+  using llvm::json::toJSON;
+
+  llvm::json::Object JsonObj;
+  for (const auto &Item : Obj) {
+    JsonObj[Item.first()] = Item.second;
+  }
+  return JsonObj;
+}
+
+llvm::json::Object toJSON(llvm::StringMap<StringRef> const &Obj) {
+  using llvm::json::toJSON;
+
+  llvm::json::Object JsonObj;
+  for (const auto &Item : Obj) {
+    JsonObj[Item.first()] = Item.second;
+  }
+  return JsonObj;
+}
+
+llvm::json::Object toJSON(ClassData const &Obj) {
+  llvm::json::Object JsonObj;
+
+  if (!Obj.ASTClassLocations.empty())
+    JsonObj["sourceLocations"] = Obj.ASTClassLocations;
+  if (!Obj.ASTClassRanges.empty())
+    JsonObj["sourceRanges"] = Obj.ASTClassRanges;
+  return JsonObj;
+}
+
+llvm::json::Object toJSON(llvm::StringMap<ClassData> const &Obj) {
+  using llvm::json::toJSON;
+
+  llvm::json::Object JsonObj;
+  for (const auto &Item : Obj) {
+    if (!Item.second.isEmpty())
+      JsonObj[Item.first()] = ::toJSON(Item.second);
+  }
+  return JsonObj;
+}
+
+void WriteJSON(std::string JsonPath,
+               llvm::StringMap<StringRef> const &ClassInheritance,
+               llvm::StringMap<std::vector<StringRef>> const &ClassesInClade,
+               llvm::StringMap<ClassData> const &ClassEntries) {
+  llvm::json::Object JsonObj;
+
+  using llvm::json::toJSON;
+
+  JsonObj["classInheritance"] = ::toJSON(ClassInheritance);
+  JsonObj["classesInClade"] = ::toJSON(ClassesInClade);
+  JsonObj["classEntries"] = ::toJSON(ClassEntries);
+
+  std::error_code EC;
+  llvm::raw_fd_ostream JsonOut(JsonPath, EC, llvm::sys::fs::F_Text);
+  if (EC)
+    return;
+
+  llvm::json::Value JsonVal(std::move(JsonObj));
+  JsonOut << formatv("{0:2}", JsonVal);
+}
+
+void ASTSrcLocProcessor::generate() {
+  WriteJSON(JsonPath, ClassInheritance, ClassesInClade, ClassEntries);
+}
+
+std::vector<std::string>
+CaptureMethods(std::string TypeString, const clang::CXXRecordDecl *ASTClass,
+               const MatchFinder::MatchResult &Result) {
+
+  auto publicAccessor = [](auto... InnerMatcher) {
+    return cxxMethodDecl(isPublic(), parameterCountIs(0), isConst(),
+                         InnerMatcher...);
+  };
+
+  auto BoundNodesVec =
+      match(findAll(publicAccessor(ofClass(equalsNode(ASTClass)),
+                                   returns(asString(TypeString)))
+                        .bind("classMethod")),
+            *ASTClass, *Result.Context);
+
+  std::vector<std::string> Methods;
+  for (const auto &BN : BoundNodesVec) {
+    if (const auto *Node = BN.getNodeAs<clang::NamedDecl>("classMethod")) {
+      // Only record the getBeginLoc etc on Stmt etc, because it will call
+      // more-derived implementations pseudo-virtually.
+      if ((ASTClass->getName() != "Stmt" && ASTClass->getName() != "Decl") &&
+          (Node->getName() == "getBeginLoc" || Node->getName() == "getEndLoc" ||
+           Node->getName() == "getSourceRange")) {
+        continue;
+      }
+      // Only record the getExprLoc on Expr, because it will call
+      // more-derived implementations pseudo-virtually.
+      if (ASTClass->getName() != "Expr" && Node->getName() == "getExprLoc") {
+        continue;
+      }
+      Methods.push_back(Node->getName().str());
+    }
+  }
+  return Methods;
+}
+
+void ASTSrcLocProcessor::run(const MatchFinder::MatchResult &Result) {
+
+  if (const auto *ASTClass =
+          Result.Nodes.getNodeAs<clang::CXXRecordDecl>("className")) {
+
+    StringRef ClassName = ASTClass->getName();
+
+    ClassData CD;
+
+    const auto *NodeClade =
+        Result.Nodes.getNodeAs<clang::CXXRecordDecl>("nodeClade");
+    StringRef CladeName = NodeClade->getName();
+
+    if (const auto *DerivedFrom =
+            Result.Nodes.getNodeAs<clang::CXXRecordDecl>("derivedFrom"))
+      ClassInheritance[ClassName] = DerivedFrom->getName();
+
+    CD.ASTClassLocations =
+        CaptureMethods("class clang::SourceLocation", ASTClass, Result);
+    CD.ASTClassRanges =
+        CaptureMethods("class clang::SourceRange", ASTClass, Result);
+
+    if (!CD.isEmpty()) {
+      ClassEntries[ClassName] = CD;
+      ClassesInClade[CladeName].push_back(ClassName);
+    }
+  }
+}
diff --git a/clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.h b/clang/lib/Tooling/DumpTool/ASTSrcLocProcessor.h
new file mode 100644 (file)
index 0000000..a4a9856
--- /dev/null
@@ -0,0 +1,49 @@
+//===- ASTSrcLocProcessor.h ---------------------------------*- C++ -*-----===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_TOOLING_DUMPTOOL_ASTSRCLOCPROCESSOR_H
+#define LLVM_CLANG_TOOLING_DUMPTOOL_ASTSRCLOCPROCESSOR_H
+
+#include "APIData.h"
+#include "clang/ASTMatchers/ASTMatchFinder.h"
+#include "llvm/ADT/StringRef.h"
+#include <memory>
+#include <string>
+#include <vector>
+
+namespace clang {
+
+class CompilerInstance;
+
+namespace tooling {
+
+class ASTSrcLocProcessor : public ast_matchers::MatchFinder::MatchCallback {
+public:
+  explicit ASTSrcLocProcessor(StringRef JsonPath);
+
+  std::unique_ptr<ASTConsumer> createASTConsumer(CompilerInstance &Compiler,
+                                                 StringRef File);
+
+  void generate();
+
+private:
+  void run(const ast_matchers::MatchFinder::MatchResult &Result) override;
+
+  llvm::StringMap<StringRef> ClassInheritance;
+  llvm::StringMap<std::vector<StringRef>> ClassesInClade;
+  llvm::StringMap<ClassData> ClassEntries;
+
+  std::string JsonPath;
+  std::unique_ptr<clang::ast_matchers::MatchFinder> Finder;
+};
+
+} // namespace tooling
+} // namespace clang
+
+#endif
diff --git a/clang/lib/Tooling/DumpTool/CMakeLists.txt b/clang/lib/Tooling/DumpTool/CMakeLists.txt
new file mode 100644 (file)
index 0000000..aad9d7b
--- /dev/null
@@ -0,0 +1,19 @@
+
+add_clang_executable(clang-ast-dump
+  ASTSrcLocProcessor.cpp
+  ClangSrcLocDump.cpp
+)
+
+target_link_libraries(clang-ast-dump
+  PRIVATE
+  clangAST
+  clangASTMatchers
+  clangBasic
+  clangDriver
+  clangFormat
+  clangFrontend
+  clangLex
+  clangRewrite
+  clangSerialization
+  clangToolingCore
+)
diff --git a/clang/lib/Tooling/DumpTool/ClangSrcLocDump.cpp b/clang/lib/Tooling/DumpTool/ClangSrcLocDump.cpp
new file mode 100644 (file)
index 0000000..68e28a9
--- /dev/null
@@ -0,0 +1,139 @@
+//===- ClangSrcLocDump.cpp ------------------------------------*- C++ -*---===//\r
+//\r
+//                     The LLVM Compiler Infrastructure\r
+//\r
+// This file is distributed under the University of Illinois Open Source\r
+// License. See LICENSE.TXT for details.\r
+//\r
+//===----------------------------------------------------------------------===//\r
+\r
+#include "clang/Basic/Diagnostic.h"\r
+#include "clang/Driver/Compilation.h"\r
+#include "clang/Driver/Driver.h"\r
+#include "clang/Driver/Job.h"\r
+#include "clang/Driver/Options.h"\r
+#include "clang/Driver/Tool.h"\r
+#include "clang/Frontend/CompilerInstance.h"\r
+#include "clang/Frontend/TextDiagnosticPrinter.h"\r
+#include "clang/Tooling/Tooling.h"\r
+#include "llvm/Option/ArgList.h"\r
+#include "llvm/Support/CommandLine.h"\r
+#include "llvm/Support/Host.h"\r
+#include "llvm/Support/JSON.h"\r
+\r
+#include "ASTSrcLocProcessor.h"\r
+\r
+using namespace clang::tooling;\r
+using namespace clang;\r
+using namespace llvm;\r
+\r
+static cl::list<std::string> IncludeDirectories(\r
+    "I", cl::desc("Include directories to use while compiling"),\r
+    cl::value_desc("directory"), cl::Required, cl::OneOrMore, cl::Prefix);\r
+\r
+static cl::opt<std::string>\r
+    AstHeaderFile("astheader", cl::desc("AST header to parse API from"),\r
+                  cl::Required, cl::value_desc("AST header file"));\r
+\r
+static cl::opt<bool>\r
+    SkipProcessing("skip-processing",\r
+                   cl::desc("Avoid processing the AST header file"),\r
+                   cl::Required, cl::value_desc("bool"));\r
+\r
+static cl::opt<std::string> JsonOutputPath("json-output-path",\r
+                                           cl::desc("json output path"),\r
+                                           cl::Required,\r
+                                           cl::value_desc("path"));\r
+\r
+class ASTSrcLocGenerationAction : public clang::ASTFrontendAction {\r
+public:\r
+  ASTSrcLocGenerationAction() : Processor(JsonOutputPath) {}\r
+\r
+  ~ASTSrcLocGenerationAction() { Processor.generate(); }\r
+\r
+  std::unique_ptr<clang::ASTConsumer>\r
+  CreateASTConsumer(clang::CompilerInstance &Compiler,\r
+                    llvm::StringRef File) override {\r
+    return Processor.createASTConsumer(Compiler, File);\r
+  }\r
+\r
+private:\r
+  ASTSrcLocProcessor Processor;\r
+};\r
+\r
+int main(int argc, const char **argv) {\r
+\r
+  cl::ParseCommandLineOptions(argc, argv);\r
+\r
+  if (SkipProcessing) {\r
+    std::error_code EC;\r
+    llvm::raw_fd_ostream JsonOut(JsonOutputPath, EC, llvm::sys::fs::F_Text);\r
+    if (EC)\r
+      return 1;\r
+    JsonOut << formatv("{0:2}", llvm::json::Value(llvm::json::Object()));\r
+    return 0;\r
+  }\r
+\r
+  std::vector<std::string> Args;\r
+  Args.push_back("-xc++-header");\r
+\r
+  llvm::transform(IncludeDirectories, std::back_inserter(Args),\r
+                  [](const std::string &IncDir) { return "-I" + IncDir; });\r
+\r
+  Args.push_back(AstHeaderFile);\r
+\r
+  std::vector<const char *> Argv(Args.size(), nullptr);\r
+  llvm::transform(Args, Argv.begin(),\r
+                  [](const std::string &Arg) { return Arg.c_str(); });\r
+\r
+  IntrusiveRefCntPtr<DiagnosticOptions> DiagOpts = new DiagnosticOptions();\r
+  unsigned MissingArgIndex, MissingArgCount;\r
+  auto Opts = driver::getDriverOptTable();\r
+  auto ParsedArgs = Opts.ParseArgs(llvm::makeArrayRef(Argv).slice(1),\r
+                                   MissingArgIndex, MissingArgCount);\r
+  ParseDiagnosticArgs(*DiagOpts, ParsedArgs);\r
+  TextDiagnosticPrinter DiagnosticPrinter(llvm::errs(), &*DiagOpts);\r
+  DiagnosticsEngine Diagnostics(\r
+      IntrusiveRefCntPtr<DiagnosticIDs>(new DiagnosticIDs()), &*DiagOpts,\r
+      &DiagnosticPrinter, false);\r
+\r
+  FileManager Files(FileSystemOptions(), vfs::getRealFileSystem());\r
+\r
+  auto Driver = std::make_unique<driver::Driver>(\r
+      "clang++", llvm::sys::getDefaultTargetTriple(), Diagnostics,\r
+      "ast-api-dump-tool", &Files.getVirtualFileSystem());\r
+\r
+  auto Comp = Driver->BuildCompilation(llvm::makeArrayRef(Argv));\r
+  if (!Comp)\r
+    return 1;\r
+\r
+  const auto &Jobs = Comp->getJobs();\r
+  if (Jobs.size() != 1 || !isa<driver::Command>(*Jobs.begin())) {\r
+    SmallString<256> error_msg;\r
+    llvm::raw_svector_ostream error_stream(error_msg);\r
+    Jobs.Print(error_stream, "; ", true);\r
+    return 1;\r
+  }\r
+\r
+  const auto &Cmd = cast<driver::Command>(*Jobs.begin());\r
+  const llvm::opt::ArgStringList &CC1Args = Cmd.getArguments();\r
+\r
+  auto Invocation = std::make_unique<CompilerInvocation>();\r
+  CompilerInvocation::CreateFromArgs(*Invocation, CC1Args, Diagnostics);\r
+\r
+  CompilerInstance Compiler(std::make_shared<clang::PCHContainerOperations>());\r
+  Compiler.setInvocation(std::move(Invocation));\r
+\r
+  Compiler.createDiagnostics(&DiagnosticPrinter, false);\r
+  if (!Compiler.hasDiagnostics())\r
+    return 1;\r
+\r
+  Compiler.createSourceManager(Files);\r
+\r
+  ASTSrcLocGenerationAction ScopedToolAction;\r
+  Compiler.ExecuteAction(ScopedToolAction);\r
+\r
+  Files.clearStatCache();\r
+\r
+  return 0;\r
+}\r
diff --git a/clang/lib/Tooling/DumpTool/generate_cxx_src_locs.py b/clang/lib/Tooling/DumpTool/generate_cxx_src_locs.py
new file mode 100755 (executable)
index 0000000..dee7a41
--- /dev/null
@@ -0,0 +1,208 @@
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+
+import os
+import sys
+import json
+
+from optparse import OptionParser
+
+parser = OptionParser()
+parser.add_option('--json-input-path',
+                  help='Read API description from FILE', metavar='FILE')
+parser.add_option('--output-file', help='Generate output in FILEPATH',
+                  metavar='FILEPATH')
+parser.add_option('--empty-implementation', help='Generate empty implementation',
+                  action="store", type="int", metavar='FILEPATH')
+
+(options, args) = parser.parse_args()
+
+if options.empty_implementation:
+    with open(os.path.join(os.getcwd(),
+              options.output_file), 'w') as f:
+        f.write("""
+namespace clang {
+namespace tooling {
+
+NodeLocationAccessors NodeIntrospection::GetLocations(clang::Stmt const *) {
+  return {};
+}
+NodeLocationAccessors
+NodeIntrospection::GetLocations(clang::DynTypedNode const &) {
+  return {};
+}
+} // namespace tooling
+} // namespace clang
+""")
+    sys.exit(0)
+
+with open(options.json_input_path) as f:
+    jsonData = json.load(f)
+
+
+class Generator(object):
+
+    implementationContent = ''
+
+    def GeneratePrologue(self):
+
+        self.implementationContent += \
+            """
+/*===- Generated file -------------------------------------------*- C++ -*-===*\
+|*                                                                            *|
+|* Introspection of available AST node SourceLocations                        *|
+|*                                                                            *|
+|* Automatically generated file, do not edit!                                 *|
+|*                                                                            *|
+\*===----------------------------------------------------------------------===*/
+
+namespace clang {
+namespace tooling {
+
+using LocationAndString = SourceLocationMap::value_type;
+using RangeAndString = SourceRangeMap::value_type;
+"""
+
+    def GenerateBaseGetLocationsDeclaration(self, CladeName):
+        self.implementationContent += \
+            """
+void GetLocationsImpl(std::shared_ptr<LocationCall> const& Prefix, clang::{0} const *Object, SourceLocationMap &Locs,
+    SourceRangeMap &Rngs);
+""".format(CladeName)
+
+    def GenerateSrcLocMethod(self, ClassName, ClassData):
+
+        self.implementationContent += \
+            """
+static void GetLocations{0}(std::shared_ptr<LocationCall> const& Prefix,
+    clang::{0} const &Object,
+    SourceLocationMap &Locs, SourceRangeMap &Rngs)
+{{
+""".format(ClassName)
+
+        if 'sourceLocations' in ClassData:
+            for locName in ClassData['sourceLocations']:
+                self.implementationContent += \
+                    """
+  Locs.insert(LocationAndString(Object.{0}(), std::make_shared<LocationCall>(Prefix, "{0}")));
+""".format(locName)
+
+            self.implementationContent += '\n'
+
+        if 'sourceRanges' in ClassData:
+            for rngName in ClassData['sourceRanges']:
+                self.implementationContent += \
+                    """
+  Rngs.insert(RangeAndString(Object.{0}(), std::make_shared<LocationCall>(Prefix, "{0}")));
+""".format(rngName)
+
+            self.implementationContent += '\n'
+
+        self.implementationContent += '}\n'
+
+    def GenerateFiles(self, OutputFile):
+        with open(os.path.join(os.getcwd(),
+                  OutputFile), 'w') as f:
+            f.write(self.implementationContent)
+
+    def GenerateTrivialBaseGetLocationsFunction(self, CladeName):
+        MethodReturnType = 'NodeLocationAccessors'
+
+        Signature = \
+            'GetLocations(clang::{0} const *Object)'.format(CladeName)
+
+        self.implementationContent += \
+            '{0} NodeIntrospection::{1} {{ return {{}}; }}'.format(MethodReturnType,
+                Signature)
+
+    def GenerateBaseGetLocationsFunction(self, ASTClassNames, CladeName):
+
+        MethodReturnType = 'NodeLocationAccessors'
+
+        Signature = \
+            'GetLocations(clang::{0} const *Object)'.format(CladeName)
+        ImplSignature = \
+            """
+GetLocationsImpl(std::shared_ptr<LocationCall> const& Prefix,
+    clang::{0} const *Object, SourceLocationMap &Locs,
+    SourceRangeMap &Rngs)
+""".format(CladeName)
+
+        self.implementationContent += \
+            'void {0} {{ GetLocations{1}(Prefix, *Object, Locs, Rngs);'.format(ImplSignature,
+                CladeName)
+
+        for ASTClassName in ASTClassNames:
+            if ASTClassName != CladeName:
+                self.implementationContent += \
+                    """
+if (auto Derived = llvm::dyn_cast<clang::{0}>(Object)) {{
+  GetLocations{0}(Prefix, *Derived, Locs, Rngs);
+}}
+""".format(ASTClassName)
+
+        self.implementationContent += '}'
+
+        self.implementationContent += \
+            """
+{0} NodeIntrospection::{1} {{
+  NodeLocationAccessors Result;
+  std::shared_ptr<LocationCall> Prefix;
+
+  GetLocationsImpl(Prefix, Object, Result.LocationAccessors,
+                   Result.RangeAccessors);
+""".format(MethodReturnType,
+                Signature)
+
+        self.implementationContent += 'return Result; }'
+
+    def GenerateDynNodeVisitor(self, CladeNames):
+        MethodReturnType = 'NodeLocationAccessors'
+
+        Signature = \
+            'GetLocations(clang::DynTypedNode const &Node)'
+
+        self.implementationContent += MethodReturnType \
+            + ' NodeIntrospection::' + Signature + '{'
+
+        for CladeName in CladeNames:
+            self.implementationContent += \
+                """
+    if (const auto *N = Node.get<{0}>())
+      return GetLocations(const_cast<{0} *>(N));""".format(CladeName)
+
+        self.implementationContent += 'return {}; }'
+
+    def GenerateEpilogue(self):
+
+        self.implementationContent += '''
+  }
+}
+'''
+
+
+g = Generator()
+
+g.GeneratePrologue()
+
+if 'classesInClade' in jsonData:
+    for (CladeName, ClassNameData) in jsonData['classesInClade'].items():
+        g.GenerateBaseGetLocationsDeclaration(CladeName)
+
+    for (ClassName, ClassAccessors) in jsonData['classEntries'].items():
+        if ClassAccessors:
+            g.GenerateSrcLocMethod(ClassName, ClassAccessors)
+
+    for (CladeName, ClassNameData) in jsonData['classesInClade'].items():
+        g.GenerateBaseGetLocationsFunction(ClassNameData, CladeName)
+
+    g.GenerateDynNodeVisitor(jsonData['classesInClade'].keys())
+else:
+    for CladeName in ['Stmt']:
+        g.GenerateTrivialBaseGetLocationsFunction(CladeName)
+
+    g.GenerateDynNodeVisitor([])
+
+g.GenerateEpilogue()
+
+g.GenerateFiles(options.output_file)
diff --git a/clang/lib/Tooling/NodeIntrospection.cpp b/clang/lib/Tooling/NodeIntrospection.cpp
new file mode 100644 (file)
index 0000000..89e8c19
--- /dev/null
@@ -0,0 +1,61 @@
+//===- NodeIntrospection.h -----------------------------------*- C++ -*----===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+//  This file contains the implementation of the NodeIntrospection.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/Tooling/NodeIntrospection.h"
+
+#include "clang/AST/AST.h"
+
+namespace clang {
+
+namespace tooling {
+
+std::string LocationCallFormatterCpp::format(LocationCall *Call) {
+  SmallVector<LocationCall *> vec;
+  while (Call) {
+    vec.push_back(Call);
+    Call = Call->on();
+  }
+  std::string result;
+  for (auto *VecCall : llvm::reverse(llvm::makeArrayRef(vec).drop_front())) {
+    result +=
+        (VecCall->name() + "()" + (VecCall->returnsPointer() ? "->" : "."))
+            .str();
+  }
+  result += (vec.back()->name() + "()").str();
+  return result;
+}
+
+namespace internal {
+bool RangeLessThan::operator()(
+    std::pair<SourceRange, std::shared_ptr<LocationCall>> const &LHS,
+    std::pair<SourceRange, std::shared_ptr<LocationCall>> const &RHS) const {
+  if (!LHS.first.isValid() || !RHS.first.isValid())
+    return false;
+
+  if (LHS.first.getBegin() < RHS.first.getBegin())
+    return true;
+  else if (LHS.first.getBegin() != RHS.first.getBegin())
+    return false;
+
+  if (LHS.first.getEnd() < RHS.first.getEnd())
+    return true;
+  else if (LHS.first.getEnd() != RHS.first.getEnd())
+    return false;
+
+  return LHS.second->name() < RHS.second->name();
+}
+} // namespace internal
+
+} // namespace tooling
+} // namespace clang
+
+#include "clang/Tooling/NodeIntrospection.inc"
index bb635df..4e08739 100644 (file)
@@ -29,6 +29,7 @@ add_subdirectory(ASTMatchers)
 add_subdirectory(AST)
 add_subdirectory(CrossTU)
 add_subdirectory(Tooling)
+add_subdirectory(Introspection)
 add_subdirectory(Format)
 add_subdirectory(Frontend)
 add_subdirectory(Rewrite)
diff --git a/clang/unittests/Introspection/CMakeLists.txt b/clang/unittests/Introspection/CMakeLists.txt
new file mode 100644 (file)
index 0000000..2010379
--- /dev/null
@@ -0,0 +1,26 @@
+set(LLVM_LINK_COMPONENTS
+  ${LLVM_TARGETS_TO_BUILD}
+  FrontendOpenMP
+  Support
+  )
+
+add_clang_unittest(IntrospectionTests
+  IntrospectionTest.cpp
+  )
+
+clang_target_link_libraries(IntrospectionTests
+  PRIVATE
+  clangAST
+  clangASTMatchers
+  clangTooling
+  clangBasic
+  clangSerialization
+  clangFrontend
+  )
+target_compile_definitions(IntrospectionTests PRIVATE
+  SKIP_INTROSPECTION_GENERATION=$<OR:$<CONFIG:Debug>,$<NOT:$<BOOL:${CLANG_TOOLING_BUILD_AST_INTROSPECTION}>>>
+)
+target_link_libraries(IntrospectionTests
+  PRIVATE
+  LLVMTestingSupport
+)
diff --git a/clang/unittests/Introspection/IntrospectionTest.cpp b/clang/unittests/Introspection/IntrospectionTest.cpp
new file mode 100644 (file)
index 0000000..f88363c
--- /dev/null
@@ -0,0 +1,100 @@
+//===- unittest/Introspection/IntrospectionTest.cpp ----------*- C++ -*---===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// Tests for AST location API introspection.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/AST/ASTContext.h"
+#include "clang/ASTMatchers/ASTMatchFinder.h"
+#include "clang/ASTMatchers/ASTMatchers.h"
+#include "clang/Tooling/NodeIntrospection.h"
+#include "clang/Tooling/Tooling.h"
+#include "gmock/gmock-matchers.h"
+#include "gmock/gmock.h"
+#include "gtest/gtest.h"
+
+using namespace clang;
+using namespace clang::ast_matchers;
+using namespace clang::tooling;
+
+using ::testing::UnorderedElementsAre;
+using ::testing::Pair;
+
+#if SKIP_INTROSPECTION_GENERATION
+
+TEST(Introspection, NonFatalAPI) {
+  auto AST = buildASTFromCode("void foo() {} void bar() { foo(); }", "foo.cpp",
+                              std::make_shared<PCHContainerOperations>());
+  auto &Ctx = AST->getASTContext();
+  auto &TU = *Ctx.getTranslationUnitDecl();
+
+  auto BoundNodes = ast_matchers::match(
+      decl(hasDescendant(
+          callExpr(callee(functionDecl(hasName("foo")))).bind("fooCall"))),
+      TU, Ctx);
+
+  EXPECT_EQ(BoundNodes.size(), 1u);
+
+  auto *FooCall = BoundNodes[0].getNodeAs<CallExpr>("fooCall");
+
+  auto result = NodeIntrospection::GetLocations(FooCall);
+
+  EXPECT_EQ(result.LocationAccessors.size(), 0);
+  EXPECT_EQ(result.RangeAccessors.size(), 0);
+}
+
+#else
+
+TEST(Introspection, SourceLocations) {
+  auto AST = buildASTFromCode("void foo() {} void bar() { foo(); }", "foo.cpp",
+                              std::make_shared<PCHContainerOperations>());
+  auto &Ctx = AST->getASTContext();
+  auto &TU = *Ctx.getTranslationUnitDecl();
+
+  auto BoundNodes = ast_matchers::match(
+      decl(hasDescendant(
+          callExpr(callee(functionDecl(hasName("foo")))).bind("fooCall"))),
+      TU, Ctx);
+
+  EXPECT_EQ(BoundNodes.size(), 1u);
+
+  auto *FooCall = BoundNodes[0].getNodeAs<CallExpr>("fooCall");
+
+  auto result = NodeIntrospection::GetLocations(FooCall);
+
+  std::map<std::string, SourceLocation> ExpectedLocations;
+  llvm::transform(result.LocationAccessors,
+                  std::inserter(ExpectedLocations, ExpectedLocations.end()),
+                  [](const auto &Accessor) {
+                    return std::make_pair(
+                        LocationCallFormatterCpp::format(Accessor.second.get()),
+                        Accessor.first);
+                  });
+
+  EXPECT_THAT(ExpectedLocations,
+              UnorderedElementsAre(
+                  Pair("getBeginLoc()", FooCall->getBeginLoc()),
+                  Pair("getEndLoc()", FooCall->getEndLoc()),
+                  Pair("getExprLoc()", FooCall->getExprLoc()),
+                  Pair("getRParenLoc()", FooCall->getRParenLoc())));
+
+  std::map<std::string, SourceRange> ExpectedRanges;
+  llvm::transform(result.RangeAccessors,
+                  std::inserter(ExpectedRanges, ExpectedRanges.end()),
+                  [](const auto &Accessor) {
+                    return std::make_pair(
+                        LocationCallFormatterCpp::format(Accessor.second.get()),
+                        Accessor.first);
+                  });
+
+  EXPECT_THAT(ExpectedRanges,
+              UnorderedElementsAre(
+                  Pair("getSourceRange()", FooCall->getSourceRange())));
+}
+#endif