[clang-tidy] support query based custom check (llvm#131804)

## summary

### Design for Compatibility

For new field design
1. we must make sure the required fields do not change in quiet long
time.
2. we should tolerant the unknown optional field (do not exist now) in
the future.

For large project integration (3rd party)
1. For config itself, since we can tolerant the unknown optional fields
and required fields should not change, the user can decide whether to
use custom check from third-party.
2.For clang-query, if there are some break change, since the query will
be parsed at check time, the user can disable the check and it will not
damage the whole clang-tidy

----

Inherit from llvm#123734
RFC:
https://discourse.llvm.org/t/support-query-based-clang-tidy-external-check/85331

this patch introduce query based custom check by `CustomChecks` options.

The major improvement compared with llvm#123734:
1. don't need to define config yaml file in command line
4. all behavior including `InheritFromParantConfig` is same as other
config.
6. change configuration schema from KV structured to List structured to
make extend new function easier.
7. Split bind string and diag message to two field to give more freedom
to design query.

example:
```yaml
Checks: -*,custom-call-main-function
CustomChecks:
  - Name: call-main-function
    Query: |
        match callExpr(
          callee(
            functionDecl(isMain()).bind("fn")
          )
        ).bind("callee")
    Diagnostic:
      - BindName: fn
        Message: main function.
        Level: Note
      - BindName: callee
        Message: call to main function.
        Level: Warning
```
```c++
int main(); // note: main function.

void bar() {
  main(); // warning: call to main function. [custom-call-main-function]
}
```

---

To make this PR don't do too much things that hard to review, here are
some possible features not included in this PR
- support language
- support traverse
- easier used template string in diagnostics message

---------

Co-authored-by: DeNiCoN <[email protected]>
Co-authored-by: Baranov Victor <[email protected]>

Author: 3 people

clang-tools-extra/CMakeLists.txt

@@ -5,6 +5,8 @@ include(GNUInstallDirs)
 
 option(CLANG_TIDY_ENABLE_STATIC_ANALYZER
   "Include static analyzer checks in clang-tidy" ON)
+option(CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS
+  "Enable query-based custom checks in clang-tidy" ON)
 
 if(CLANG_INCLUDE_TESTS)
   umbrella_lit_testsuite_begin(check-clang-tools)

clang-tools-extra/clang-tidy/CMakeLists.txt

@@ -58,6 +58,7 @@ add_subdirectory(bugprone)
 add_subdirectory(cert)
 add_subdirectory(concurrency)
 add_subdirectory(cppcoreguidelines)
+add_subdirectory(custom)
 add_subdirectory(darwin)
 add_subdirectory(fuchsia)
 add_subdirectory(google)
@@ -101,6 +102,10 @@ set(ALL_CLANG_TIDY_CHECKS
   clangTidyReadabilityModule
   clangTidyZirconModule
   )
+
+if(CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS)
+  list(APPEND ALL_CLANG_TIDY_CHECKS clangTidyCustomModule)
+endif()
 if(CLANG_TIDY_ENABLE_STATIC_ANALYZER)
   list(APPEND ALL_CLANG_TIDY_CHECKS clangTidyMPIModule)
 endif()

clang-tools-extra/clang-tidy/ClangTidy.cpp

@@ -53,6 +53,11 @@ LLVM_INSTANTIATE_REGISTRY(clang::tidy::ClangTidyModuleRegistry)
 
 namespace clang::tidy {
 
+namespace custom {
+extern void registerCustomChecks(const ClangTidyOptions &O,
+                                 ClangTidyCheckFactories &Factories);
+} // namespace custom
+
 namespace {
 #if CLANG_TIDY_ENABLE_STATIC_ANALYZER
 #define ANALYZER_CHECK_NAME_PREFIX "clang-analyzer-"
@@ -342,6 +347,10 @@ ClangTidyASTConsumerFactory::ClangTidyASTConsumerFactory(
     IntrusiveRefCntPtr<llvm::vfs::OverlayFileSystem> OverlayFS)
     : Context(Context), OverlayFS(std::move(OverlayFS)),
       CheckFactories(new ClangTidyCheckFactories) {
+#if CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS
+  if (Context.canExperimentalCustomChecks())
+    custom::registerCustomChecks(Context.getOptions(), *CheckFactories);
+#endif
   for (ClangTidyModuleRegistry::entry E : ClangTidyModuleRegistry::entries()) {
     std::unique_ptr<ClangTidyModule> Module = E.instantiate();
     Module->addCheckFactories(*CheckFactories);
@@ -411,7 +420,10 @@ ClangTidyASTConsumerFactory::createASTConsumer(
                         .getCurrentWorkingDirectory();
   if (WorkingDir)
     Context.setCurrentBuildDirectory(WorkingDir.get());
-
+#if CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS
+  if (Context.canExperimentalCustomChecks())
+    custom::registerCustomChecks(Context.getOptions(), *CheckFactories);
+#endif
   std::vector<std::unique_ptr<ClangTidyCheck>> Checks =
       CheckFactories->createChecksForLanguage(&Context);
 
@@ -497,13 +509,13 @@ ClangTidyOptions::OptionMap ClangTidyASTConsumerFactory::getCheckOptions() {
   return Options;
 }
 
-std::vector<std::string>
-getCheckNames(const ClangTidyOptions &Options,
-              bool AllowEnablingAnalyzerAlphaCheckers) {
+std::vector<std::string> getCheckNames(const ClangTidyOptions &Options,
+                                       bool AllowEnablingAnalyzerAlphaCheckers,
+                                       bool ExperimentalCustomChecks) {
   clang::tidy::ClangTidyContext Context(
       std::make_unique<DefaultOptionsProvider>(ClangTidyGlobalOptions(),
                                                Options),
-      AllowEnablingAnalyzerAlphaCheckers);
+      AllowEnablingAnalyzerAlphaCheckers, false, ExperimentalCustomChecks);
   ClangTidyASTConsumerFactory Factory(Context);
   return Factory.getCheckNames();
 }
@@ -524,11 +536,12 @@ void filterCheckOptions(ClangTidyOptions &Options,
 
 ClangTidyOptions::OptionMap
 getCheckOptions(const ClangTidyOptions &Options,
-                bool AllowEnablingAnalyzerAlphaCheckers) {
+                bool AllowEnablingAnalyzerAlphaCheckers,
+                bool ExperimentalCustomChecks) {
   clang::tidy::ClangTidyContext Context(
       std::make_unique<DefaultOptionsProvider>(ClangTidyGlobalOptions(),
                                                Options),
-      AllowEnablingAnalyzerAlphaCheckers);
+      AllowEnablingAnalyzerAlphaCheckers, false, ExperimentalCustomChecks);
   ClangTidyDiagnosticConsumer DiagConsumer(Context);
   auto DiagOpts = std::make_unique<DiagnosticOptions>();
   DiagnosticsEngine DE(llvm::makeIntrusiveRefCnt<DiagnosticIDs>(), *DiagOpts,
@@ -665,15 +678,19 @@ void exportReplacements(const llvm::StringRef MainFilePath,
   YAML << TUD;
 }
 
-ChecksAndOptions
-getAllChecksAndOptions(bool AllowEnablingAnalyzerAlphaCheckers) {
+ChecksAndOptions getAllChecksAndOptions(bool AllowEnablingAnalyzerAlphaCheckers,
+                                        bool ExperimentalCustomChecks) {
   ChecksAndOptions Result;
   ClangTidyOptions Opts;
   Opts.Checks = "*";
   clang::tidy::ClangTidyContext Context(
       std::make_unique<DefaultOptionsProvider>(ClangTidyGlobalOptions(), Opts),
-      AllowEnablingAnalyzerAlphaCheckers);
+      AllowEnablingAnalyzerAlphaCheckers, false, ExperimentalCustomChecks);
   ClangTidyCheckFactories Factories;
+#if CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS
+  if (ExperimentalCustomChecks)
+    custom::registerCustomChecks(Context.getOptions(), Factories);
+#endif
   for (const ClangTidyModuleRegistry::entry &Module :
        ClangTidyModuleRegistry::entries()) {
     Module.instantiate()->addCheckFactories(Factories);

clang-tools-extra/clang-tidy/ClangTidy.h

@@ -56,15 +56,16 @@ class ClangTidyASTConsumerFactory {
 /// Fills the list of check names that are enabled when the provided
 /// filters are applied.
 std::vector<std::string> getCheckNames(const ClangTidyOptions &Options,
-                                       bool AllowEnablingAnalyzerAlphaCheckers);
+                                       bool AllowEnablingAnalyzerAlphaCheckers,
+                                       bool ExperimentalCustomChecks);
 
 struct ChecksAndOptions {
   llvm::StringSet<> Checks;
   llvm::StringSet<> Options;
 };
 
-ChecksAndOptions
-getAllChecksAndOptions(bool AllowEnablingAnalyzerAlphaCheckers = true);
+ChecksAndOptions getAllChecksAndOptions(bool AllowEnablingAnalyzerAlphaCheckers,
+                                        bool ExperimentalCustomChecks);
 
 /// Returns the effective check-specific options.
 ///
@@ -74,7 +75,8 @@ getAllChecksAndOptions(bool AllowEnablingAnalyzerAlphaCheckers = true);
 /// Options.
 ClangTidyOptions::OptionMap
 getCheckOptions(const ClangTidyOptions &Options,
-                bool AllowEnablingAnalyzerAlphaCheckers);
+                bool AllowEnablingAnalyzerAlphaCheckers,
+                bool ExperimentalCustomChecks);
 
 /// Filters CheckOptions in \p Options to only include options specified in
 /// the \p EnabledChecks which is a sorted vector.

clang-tools-extra/clang-tidy/ClangTidyDiagnosticConsumer.cpp

@@ -160,11 +160,12 @@ ClangTidyError::ClangTidyError(StringRef CheckName,
 
 ClangTidyContext::ClangTidyContext(
     std::unique_ptr<ClangTidyOptionsProvider> OptionsProvider,
-    bool AllowEnablingAnalyzerAlphaCheckers, bool EnableModuleHeadersParsing)
+    bool AllowEnablingAnalyzerAlphaCheckers, bool EnableModuleHeadersParsing,
+    bool ExperimentalCustomChecks)
     : OptionsProvider(std::move(OptionsProvider)),
-
       AllowEnablingAnalyzerAlphaCheckers(AllowEnablingAnalyzerAlphaCheckers),
-      EnableModuleHeadersParsing(EnableModuleHeadersParsing) {
+      EnableModuleHeadersParsing(EnableModuleHeadersParsing),
+      ExperimentalCustomChecks(ExperimentalCustomChecks) {
   // Before the first translation unit we can get errors related to command-line
   // parsing, use dummy string for the file name in this case.
   setCurrentFile("dummy");

clang-tools-extra/clang-tidy/ClangTidyDiagnosticConsumer.h

@@ -19,6 +19,7 @@
 #include "llvm/ADT/StringSet.h"
 #include "llvm/Support/Regex.h"
 #include <optional>
+#include <utility>
 
 namespace clang {
 
@@ -68,10 +69,13 @@ struct ClangTidyStats {
 /// \endcode
 class ClangTidyContext {
 public:
+  ClangTidyContext(std::unique_ptr<ClangTidyOptionsProvider> OptionsProvider)
+      : ClangTidyContext(std::move(OptionsProvider), false, false, false) {}
   /// Initializes \c ClangTidyContext instance.
   ClangTidyContext(std::unique_ptr<ClangTidyOptionsProvider> OptionsProvider,
-                   bool AllowEnablingAnalyzerAlphaCheckers = false,
-                   bool EnableModuleHeadersParsing = false);
+                   bool AllowEnablingAnalyzerAlphaCheckers,
+                   bool EnableModuleHeadersParsing,
+                   bool ExperimentalCustomChecks);
   /// Sets the DiagnosticsEngine that diag() will emit diagnostics to.
   // FIXME: this is required initialization, and should be a constructor param.
   // Fix the context -> diag engine -> consumer -> context initialization cycle.
@@ -210,6 +214,10 @@ class ClangTidyContext {
     return EnableModuleHeadersParsing;
   }
 
+  // whether experimental custom checks can be enabled.
+  // enabled with `--experimental-custom-checks`
+  bool canExperimentalCustomChecks() const { return ExperimentalCustomChecks; }
+
   void setSelfContainedDiags(bool Value) { SelfContainedDiags = Value; }
 
   bool areDiagsSelfContained() const { return SelfContainedDiags; }
@@ -258,6 +266,7 @@ class ClangTidyContext {
 
   bool AllowEnablingAnalyzerAlphaCheckers;
   bool EnableModuleHeadersParsing;
+  bool ExperimentalCustomChecks;
 
   bool SelfContainedDiags = false;
 

clang-tools-extra/clang-tidy/ClangTidyForceLinker.h

@@ -54,6 +54,13 @@ extern volatile int CppCoreGuidelinesModuleAnchorSource;
 static int LLVM_ATTRIBUTE_UNUSED CppCoreGuidelinesModuleAnchorDestination =
     CppCoreGuidelinesModuleAnchorSource;
 
+#if CLANG_TIDY_ENABLE_QUERY_BASED_CUSTOM_CHECKS
+// This anchor is used to force the linker to link the CustomModule.
+extern volatile int CustomModuleAnchorSource;
+static int LLVM_ATTRIBUTE_UNUSED CustomModuleAnchorDestination =
+    CustomModuleAnchorSource;
+#endif
+
 // This anchor is used to force the linker to link the DarwinModule.
 extern volatile int DarwinModuleAnchorSource;
 static int LLVM_ATTRIBUTE_UNUSED DarwinModuleAnchorDestination =

clang-tools-extra/clang-tidy/ClangTidyModule.h

@@ -62,6 +62,8 @@ class ClangTidyCheckFactories {
                          });
   }
 
+  void eraseCheck(llvm::StringRef CheckName) { Factories.erase(CheckName); }
+
   /// Create instances of checks that are enabled.
   std::vector<std::unique_ptr<ClangTidyCheck>>
   createChecks(ClangTidyContext *Context) const;

clang-tools-extra/clang-tidy/ClangTidyOptions.cpp

@@ -8,8 +8,10 @@
 
 #include "ClangTidyOptions.h"
 #include "ClangTidyModuleRegistry.h"
+#include "clang/Basic/DiagnosticIDs.h"
 #include "clang/Basic/LLVM.h"
 #include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/StringExtras.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorOr.h"
 #include "llvm/Support/MemoryBufferRef.h"
@@ -129,6 +131,51 @@ void yamlize(IO &IO, ClangTidyOptions::OptionMap &Val, bool,
   }
 }
 
+namespace {
+struct MultiLineString {
+  std::string &S;
+};
+} // namespace
+
+template <> struct BlockScalarTraits<MultiLineString> {
+  static void output(const MultiLineString &S, void *Ctxt, raw_ostream &OS) {
+    OS << S.S;
+  }
+  static StringRef input(StringRef Str, void *Ctxt, MultiLineString &S) {
+    S.S = Str;
+    return "";
+  }
+};
+
+template <> struct ScalarEnumerationTraits<clang::DiagnosticIDs::Level> {
+  static void enumeration(IO &IO, clang::DiagnosticIDs::Level &Level) {
+    IO.enumCase(Level, "Warning", clang::DiagnosticIDs::Level::Warning);
+    IO.enumCase(Level, "Note", clang::DiagnosticIDs::Level::Note);
+  }
+};
+template <> struct SequenceElementTraits<ClangTidyOptions::CustomCheckDiag> {
+  static const bool flow = false;
+};
+template <> struct MappingTraits<ClangTidyOptions::CustomCheckDiag> {
+  static void mapping(IO &IO, ClangTidyOptions::CustomCheckDiag &D) {
+    IO.mapRequired("BindName", D.BindName);
+    MultiLineString MLS{D.Message};
+    IO.mapRequired("Message", MLS);
+    IO.mapOptional("Level", D.Level);
+  }
+};
+template <> struct SequenceElementTraits<ClangTidyOptions::CustomCheckValue> {
+  static const bool flow = false;
+};
+template <> struct MappingTraits<ClangTidyOptions::CustomCheckValue> {
+  static void mapping(IO &IO, ClangTidyOptions::CustomCheckValue &V) {
+    IO.mapRequired("Name", V.Name);
+    MultiLineString MLS{V.Query};
+    IO.mapRequired("Query", MLS);
+    IO.mapRequired("Diagnostic", V.Diags);
+  }
+};
+
 struct ChecksVariant {
   std::optional<std::string> AsString;
   std::optional<std::vector<std::string>> AsVector;
@@ -184,6 +231,7 @@ template <> struct MappingTraits<ClangTidyOptions> {
     IO.mapOptional("InheritParentConfig", Options.InheritParentConfig);
     IO.mapOptional("UseColor", Options.UseColor);
     IO.mapOptional("SystemHeaders", Options.SystemHeaders);
+    IO.mapOptional("CustomChecks", Options.CustomChecks);
   }
 };
 
@@ -245,7 +293,8 @@ ClangTidyOptions &ClangTidyOptions::mergeWith(const ClangTidyOptions &Other,
   overrideValue(UseColor, Other.UseColor);
   mergeVectors(ExtraArgs, Other.ExtraArgs);
   mergeVectors(ExtraArgsBefore, Other.ExtraArgsBefore);
-
+  // FIXME: how to handle duplicate names check?
+  mergeVectors(CustomChecks, Other.CustomChecks);
   for (const auto &KeyValue : Other.CheckOptions) {
     CheckOptions.insert_or_assign(
         KeyValue.getKey(),

clang-tools-extra/clang-tidy/ClangTidyOptions.h

@@ -9,6 +9,7 @@
 #ifndef LLVM_CLANG_TOOLS_EXTRA_CLANG_TIDY_CLANGTIDYOPTIONS_H
 #define LLVM_CLANG_TOOLS_EXTRA_CLANG_TIDY_CLANGTIDYOPTIONS_H
 
+#include "clang/Basic/DiagnosticIDs.h"
 #include "llvm/ADT/IntrusiveRefCntPtr.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringMap.h"
@@ -129,6 +130,19 @@ struct ClangTidyOptions {
   /// Key-value mapping used to store check-specific options.
   OptionMap CheckOptions;
 
+  struct CustomCheckDiag {
+    std::string BindName;
+    std::string Message;
+    std::optional<DiagnosticIDs::Level> Level;
+  };
+  struct CustomCheckValue {
+    std::string Name;
+    std::string Query;
+    llvm::SmallVector<CustomCheckDiag> Diags;
+  };
+  using CustomCheckValueList = llvm::SmallVector<CustomCheckValue>;
+  std::optional<CustomCheckValueList> CustomChecks;
+
   using ArgList = std::vector<std::string>;
 
   /// Add extra compilation arguments to the end of the list.