[InstallAPI] Add --extra* and --exclude* cli options for header input (…

…#86522)

InstallAPI takes a json list of headers that is typically generated from
a build system like Xcode based on a project's attributes. Sometimes,
maintainers may want to alter this for tapi input. Using e.g.
`--extra-public-headers`, users can manipulate what headers will be used
for TBD file generation.

clang/include/clang/Basic/DiagnosticInstallAPIKinds.td

@@ -15,6 +15,9 @@ let CategoryName = "Command line" in {
 def err_cannot_write_file : Error<"cannot write file '%0': %1">;
 def err_no_install_name : Error<"no install name specified: add -install_name <path>">;
 def err_no_output_file: Error<"no output file specified">;
+def err_no_such_header_file : Error<"no such %select{public|private|project}1 header file: '%0'">;
+def warn_no_such_excluded_header_file : Warning<"no such excluded %select{public|private}0 header file: '%1'">, InGroup<InstallAPIViolation>;
+def warn_glob_did_not_match: Warning<"glob '%0' did not match any header file">, InGroup<InstallAPIViolation>;
 } // end of command line category.
 
 let CategoryName = "Verification" in {

clang/include/clang/InstallAPI/HeaderFile.h

@@ -13,7 +13,9 @@
 #ifndef LLVM_CLANG_INSTALLAPI_HEADERFILE_H
 #define LLVM_CLANG_INSTALLAPI_HEADERFILE_H
 
+#include "clang/Basic/FileManager.h"
 #include "clang/Basic/LangStandard.h"
+#include "clang/InstallAPI/MachO.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/Regex.h"
@@ -56,6 +58,10 @@ class HeaderFile {
   std::string IncludeName;
   /// Supported language mode for header.
   std::optional<clang::Language> Language;
+  /// Exclude header file from processing.
+  bool Excluded{false};
+  /// Add header file to processing.
+  bool Extra{false};
 
 public:
   HeaderFile() = delete;
@@ -71,17 +77,48 @@ class HeaderFile {
   StringRef getIncludeName() const { return IncludeName; }
   StringRef getPath() const { return FullPath; }
 
+  void setExtra(bool V = true) { Extra = V; }
+  void setExcluded(bool V = true) { Excluded = V; }
+  bool isExtra() const { return Extra; }
+  bool isExcluded() const { return Excluded; }
+
   bool useIncludeName() const {
     return Type != HeaderType::Project && !IncludeName.empty();
   }
 
   bool operator==(const HeaderFile &Other) const {
-    return std::tie(Type, FullPath, IncludeName, Language) ==
+    return std::tie(Type, FullPath, IncludeName, Language, Excluded, Extra) ==
            std::tie(Other.Type, Other.FullPath, Other.IncludeName,
-                    Other.Language);
+                    Other.Language, Other.Excluded, Other.Extra);
   }
 };
 
+/// Glob that represents a pattern of header files to retreive.
+class HeaderGlob {
+private:
+  std::string GlobString;
+  llvm::Regex Rule;
+  HeaderType Type;
+  bool FoundMatch{false};
+
+public:
+  HeaderGlob(StringRef GlobString, llvm::Regex &&, HeaderType Type);
+
+  /// Create a header glob from string for the header access level.
+  static llvm::Expected<std::unique_ptr<HeaderGlob>>
+  create(StringRef GlobString, HeaderType Type);
+
+  /// Query if provided header matches glob.
+  bool match(const HeaderFile &Header);
+
+  /// Query if a header was matched in the glob, used primarily for error
+  /// reporting.
+  bool didMatch() { return FoundMatch; }
+
+  /// Provide back input glob string.
+  StringRef str() { return GlobString; }
+};
+
 /// Assemble expected way header will be included by clients.
 /// As in what maps inside the brackets of `#include <IncludeName.h>`
 /// For example,
@@ -93,6 +130,19 @@ class HeaderFile {
 std::optional<std::string> createIncludeHeaderName(const StringRef FullPath);
 using HeaderSeq = std::vector<HeaderFile>;
 
+/// Determine if Path is a header file.
+/// It does not touch the file system.
+///
+/// \param  Path File path to file.
+bool isHeaderFile(StringRef Path);
+
+/// Given input directory, collect all header files.
+///
+/// \param FM FileManager for finding input files.
+/// \param Directory Path to directory file.
+llvm::Expected<PathSeq> enumerateFiles(clang::FileManager &FM,
+                                       StringRef Directory);
+
 } // namespace clang::installapi
 
 #endif // LLVM_CLANG_INSTALLAPI_HEADERFILE_H

clang/include/clang/InstallAPI/MachO.h

@@ -40,6 +40,7 @@ using SymbolSet = llvm::MachO::SymbolSet;
 using SimpleSymbol = llvm::MachO::SimpleSymbol;
 using FileType = llvm::MachO::FileType;
 using PackedVersion = llvm::MachO::PackedVersion;
+using PathSeq = llvm::MachO::PathSeq;
 using Target = llvm::MachO::Target;
 using TargetList = llvm::MachO::TargetList;
 

clang/lib/InstallAPI/Frontend.cpp

@@ -138,6 +138,8 @@ std::unique_ptr<MemoryBuffer> createInputBuffer(InstallAPIContext &Ctx) {
   SmallString<4096> Contents;
   raw_svector_ostream OS(Contents);
   for (const HeaderFile &H : Ctx.InputHeaders) {
+    if (H.isExcluded())
+      continue;
     if (H.getType() != Ctx.Type)
       continue;
     if (Ctx.LangMode == Language::C || Ctx.LangMode == Language::CXX)

clang/lib/InstallAPI/HeaderFile.cpp

@@ -7,6 +7,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "clang/InstallAPI/HeaderFile.h"
+#include "llvm/TextAPI/Utils.h"
 
 using namespace llvm;
 namespace clang::installapi {
@@ -34,4 +35,54 @@ std::optional<std::string> createIncludeHeaderName(const StringRef FullPath) {
   return Matches[1].drop_front(Matches[1].rfind('/') + 1).str() + "/" +
          Matches[3].str();
 }
+
+bool isHeaderFile(StringRef Path) {
+  return StringSwitch<bool>(sys::path::extension(Path))
+      .Cases(".h", ".H", ".hh", ".hpp", ".hxx", true)
+      .Default(false);
+}
+
+llvm::Expected<PathSeq> enumerateFiles(FileManager &FM, StringRef Directory) {
+  PathSeq Files;
+  std::error_code EC;
+  auto &FS = FM.getVirtualFileSystem();
+  for (llvm::vfs::recursive_directory_iterator i(FS, Directory, EC), ie;
+       i != ie; i.increment(EC)) {
+    if (EC)
+      return errorCodeToError(EC);
+
+    // Skip files that do not exist. This usually happens for broken symlinks.
+    if (FS.status(i->path()) == std::errc::no_such_file_or_directory)
+      continue;
+
+    StringRef Path = i->path();
+    if (isHeaderFile(Path))
+      Files.emplace_back(Path);
+  }
+
+  return Files;
+}
+
+HeaderGlob::HeaderGlob(StringRef GlobString, Regex &&Rule, HeaderType Type)
+    : GlobString(GlobString), Rule(std::move(Rule)), Type(Type) {}
+
+bool HeaderGlob::match(const HeaderFile &Header) {
+  if (Header.getType() != Type)
+    return false;
+
+  bool Match = Rule.match(Header.getPath());
+  if (Match)
+    FoundMatch = true;
+  return Match;
+}
+
+Expected<std::unique_ptr<HeaderGlob>> HeaderGlob::create(StringRef GlobString,
+                                                         HeaderType Type) {
+  auto Rule = MachO::createRegexFromGlob(GlobString);
+  if (!Rule)
+    return Rule.takeError();
+
+  return std::make_unique<HeaderGlob>(GlobString, std::move(*Rule), Type);
+}
+
 } // namespace clang::installapi

clang/test/InstallAPI/Inputs/Simple/Extra/SimpleExtraAPI1.h

@@ -0,0 +1 @@
+extern int extraGlobalAPI1;

clang/test/InstallAPI/Inputs/Simple/Extra/SimpleExtraAPI2.h

@@ -0,0 +1 @@
+extern int extraGlobalAPI2;

clang/test/InstallAPI/Inputs/Simple/Simple.framework/Headers/Basic.h

@@ -0,0 +1,103 @@
+#import <Foundation/Foundation.h>
+
+// Basic class with no super class
+@interface Basic1
+@end
+
+@interface Basic2 : NSObject
+@end
+
+@interface Basic3 : NSObject
+@property BOOL property1;
+@property(readonly) BOOL property2;
+@property(getter=isProperty3) BOOL property3;
+@property BOOL dynamicProp;
+@end
+
+@interface Basic4 : NSObject {
+@public
+  BOOL ivar1;
+@protected
+  BOOL ivar2;
+@package
+  BOOL ivar3;
+@private
+  BOOL ivar4;
+}
+@end
+
+__attribute__((visibility("hidden"))) @interface Basic4_1 : NSObject {
+@public
+  BOOL ivar1;
+@protected
+  BOOL ivar2;
+@package
+  BOOL ivar3;
+@private
+  BOOL ivar4;
+}
+@end
+
+@interface Basic4_2 : NSObject {
+@private
+  BOOL ivar4;
+@package
+  BOOL ivar3;
+@protected
+  BOOL ivar2;
+@public
+  BOOL ivar1;
+}
+@end
+
+@interface Basic5 : NSObject
++ (void)aClassMethod;
+- (void)anInstanceMethod;
+@end
+
+@interface Basic6 : NSObject
+@end
+
+@interface Basic6 () {
+@public
+  BOOL ivar1;
+}
+@property BOOL property1;
+- (void)anInstanceMethodFromAnExtension;
+@end
+
+@interface Basic6 (Foo)
+@property BOOL property2;
+- (void)anInstanceMethodFromACategory;
+@end
+
+__attribute__((visibility("hidden")))
+@interface Basic7 : NSObject
+@end
+
+@interface Basic7 ()
+- (void) anInstanceMethodFromAnHiddenExtension;
+@end
+
+@interface Basic8 : NSObject
++ (void)useSameName;
+@end
+
+// Classes and protocols can have the same name. For now they would only clash
+// in the selector map if the protocl starts with '_'.
+@protocol _A
+- (void)aMethod;
+@end
+
+@interface A : NSObject
+- (void)aMethod NS_AVAILABLE(10_11, 9_0);
+- (void)bMethod NS_UNAVAILABLE;
+@end
+
+@interface Basic9 : NSObject
+@property(readonly) BOOL aProperty NS_AVAILABLE(10_10, 8_0);
+@end
+
+@interface Basic9 (deprecated)
+@property(readwrite) BOOL aProperty NS_DEPRECATED_MAC(10_8, 10_10);
+@end

clang/test/InstallAPI/Inputs/Simple/Simple.framework/Headers/External.h

@@ -0,0 +1,19 @@
+#import <Foundation/Foundation.h>
+
+// Sub-class an external defined ObjC Class.
+@interface ExternalManagedObject : NSManagedObject
+- (void)foo;
+@end
+
+// Add category to external defined ObjC Class.
+@interface NSManagedObject (Simple)
+- (int)supportsSimple;
+@end
+
+// CoreData Accessors are dynamically generated and have no implementation.
+@interface ExternalManagedObject (CoreDataGeneratedAccessors)
+- (void)addChildObject:(ExternalManagedObject *)value;
+- (void)removeChildObject:(ExternalManagedObject *)value;
+- (void)addChild:(NSSet *)values;
+- (void)removeChild:(NSSet *)values;
+@end

clang/test/InstallAPI/Inputs/Simple/Simple.framework/Headers/Simple.h

@@ -0,0 +1,45 @@
+#import <Foundation/Foundation.h>
+
+// Useless forward declaration. This is used for testing.
+@class FooBar;
+@protocol FooProtocol;
+
+@protocol ForwardProcotol;
+
+// Test public global.
+extern int publicGlobalVariable;
+
+// Test weak public global.
+extern int weakPublicGlobalVariable __attribute__((weak));
+
+// Test public ObjC class
+@interface Simple : NSObject
+@end
+
+__attribute__((objc_exception))
+@interface Base : NSObject
+@end
+
+@interface SubClass : Base
+@end
+
+@protocol BaseProtocol
+- (void) baseMethod;
+@end
+
+NS_AVAILABLE(10_11, 9_0)
+@protocol FooProtocol <BaseProtocol>
+- (void) protocolMethod;
+@end
+
+@protocol BarProtocol
+- (void) barMethod;
+@end
+
+@interface FooClass <FooProtocol, BarProtocol>
+@end
+
+// Create an empty category conforms to a forward declared protocol.
+// <rdar://problem/35605892>
+@interface FooClass (Test) <ForwardProcotol>
+@end

clang/test/InstallAPI/Inputs/Simple/Simple.framework/Headers/SimpleAPI.h

@@ -0,0 +1 @@
+extern int otherFrameworkAPI;

clang/test/InstallAPI/Inputs/Simple/Simple.framework/PrivateHeaders/SimplePrivate.h

@@ -0,0 +1,5 @@
+// Test private global variable.
+extern int privateGlobalVariable;
+
+// Test weak private global.
+extern int weakPrivateGlobalVariable __attribute__((weak));

clang/test/InstallAPI/Inputs/Simple/Simple.framework/PrivateHeaders/SimplePrivateSPI.h

@@ -0,0 +1,2 @@
+// Test private global variable.
+extern int otherFrameworkSPI;