[clang][InstallAPI] Add input file support to library

This patch adds support for expected InstallAPI inputs.
InstallAPI  accepts a well defined filelist of headers and how those headers
represent a single library.

InstallAPI captures header files to determine linkable symbols
to then compare against what was compiled in a binary dylib and
generate TBD files.

clang/include/clang/InstallAPI/FileList.h

@@ -0,0 +1,80 @@
+//===- InstallAPI/FileList.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
+//
+//===----------------------------------------------------------------------===//
+///
+/// The JSON file list parser is used to communicate input to InstallAPI.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_INSTALLAPI_FILELIST_H
+#define LLVM_CLANG_INSTALLAPI_FILELIST_H
+
+#include "clang/Basic/Diagnostic.h"
+#include "clang/Basic/FileManager.h"
+#include "clang/InstallAPI/HeaderFile.h"
+#include "llvm/Support/Error.h"
+#include "llvm/Support/MemoryBuffer.h"
+
+namespace clang {
+namespace installapi {
+
+/// Abstract Interface for reading FileList JSON Input.
+class FileListReader {
+  class Implementation;
+
+  Implementation &Impl;
+
+  FileListReader(std::unique_ptr<llvm::MemoryBuffer> InputBuffer,
+                 llvm::Error &Err);
+
+public:
+  static llvm::Expected<std::unique_ptr<FileListReader>>
+  get(std::unique_ptr<llvm::MemoryBuffer> InputBuffer);
+
+  ~FileListReader();
+
+  FileListReader(const FileListReader &) = delete;
+  FileListReader &operator=(const FileListReader &) = delete;
+
+  int getVersion() const;
+
+  struct HeaderInfo {
+    HeaderType Type;
+    std::string Path;
+    std::optional<clang::Language> Language;
+  };
+
+  /// Visitor used when walking the contents of the file list.
+  class Visitor {
+  public:
+    virtual ~Visitor();
+
+    virtual void visitHeaderFile(HeaderInfo &header) = 0;
+  };
+
+  /// Visit the contents of the header list file, passing each entity to the
+  /// given visitor. It visits in the same order as they appear in the json
+  /// file.
+  void visit(Visitor &visitor);
+};
+
+class FileListVisitor final : public FileListReader::Visitor {
+  FileManager &FM;
+  DiagnosticsEngine &Diag;
+  HeaderSeq &HeaderFiles;
+
+public:
+  FileListVisitor(FileManager &FM, DiagnosticsEngine &Diag,
+                  HeaderSeq &HeaderFiles)
+      : FM(FM), Diag(Diag), HeaderFiles(HeaderFiles) {}
+
+  void visitHeaderFile(FileListReader::HeaderInfo &Header) override;
+};
+} // namespace installapi
+} // namespace clang
+
+#endif // LLVM_CLANG_INSTALLAPI_FILELIST_H

clang/include/clang/InstallAPI/HeaderFile.h

@@ -0,0 +1,69 @@
+//===- InstallAPI/HeaderFile.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
+//
+//===----------------------------------------------------------------------===//
+///
+/// Representations of a library's headers for InstallAPI.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_INSTALLAPI_HEADERFILE_H
+#define LLVM_CLANG_INSTALLAPI_HEADERFILE_H
+
+#include "clang/Basic/LangStandard.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/Regex.h"
+#include <optional>
+#include <string>
+
+namespace clang::installapi {
+
+const llvm::Regex DarwinFwkHeaderRule("/(.+)\\.framework/(.+)?Headers/(.+)");
+
+enum class HeaderType {
+  /// Represents declarations accessible to all clients.
+  Public,
+  /// Represents declarations accessible to a disclosed set of clients.
+  Private,
+  /// Represents declarations only accessible as implementation details to the
+  /// input library.
+  Project,
+};
+
+class HeaderFile {
+  /// Full input path to header.
+  std::string FullPath;
+  /// Access level of header.
+  HeaderType Type;
+  /// Expected way header will be included by clients.
+  std::string IncludeName;
+  /// Supported language mode for header.
+  std::optional<clang::Language> Language;
+
+public:
+  HeaderFile(StringRef FullPath, HeaderType Type,
+             StringRef IncludeName = StringRef(),
+             std::optional<clang::Language> Language = std::nullopt)
+      : FullPath(FullPath), Type(Type), IncludeName(IncludeName),
+        Language(Language) {}
+
+  HeaderType getType() const { return Type; }
+};
+
+/// Assemble expected way header will be included by clients.
+/// As in what maps inside the brackets of `#include <IncludeName.h>`
+/// For example,
+/// "/System/Library/Frameworks/Foo.framework/Headers/Foo.h" returns
+/// "Foo/Foo.h"
+///
+/// \param FullPath Path to the header file which includes the library
+/// structure.
+std::optional<std::string> createIncludeHeaderName(const StringRef FullPath);
+using HeaderSeq = std::vector<HeaderFile>;
+
+} // namespace clang::installapi
+
+#endif // LLVM_CLANG_INSTALLAPI_HEADERFILE_H

clang/lib/ExtractAPI/ExtractAPIConsumer.cpp

@@ -30,6 +30,7 @@
 #include "clang/Frontend/CompilerInstance.h"
 #include "clang/Frontend/FrontendOptions.h"
 #include "clang/Frontend/MultiplexConsumer.h"
+#include "clang/InstallAPI/HeaderFile.h"
 #include "clang/Lex/MacroInfo.h"
 #include "clang/Lex/PPCallbacks.h"
 #include "clang/Lex/Preprocessor.h"
@@ -61,9 +62,6 @@ std::optional<std::string> getRelativeIncludeName(const CompilerInstance &CI,
          "CompilerInstance does not have a FileNamager!");
 
   using namespace llvm::sys;
-  // Matches framework include patterns
-  const llvm::Regex Rule("/(.+)\\.framework/(.+)?Headers/(.+)");
-
   const auto &FS = CI.getVirtualFileSystem();
 
   SmallString<128> FilePath(File.begin(), File.end());
@@ -147,7 +145,7 @@ std::optional<std::string> getRelativeIncludeName(const CompilerInstance &CI,
       // include name `<Framework/Header.h>`
       if (Entry.IsFramework) {
         SmallVector<StringRef, 4> Matches;
-        Rule.match(File, &Matches);
+        clang::installapi::DarwinFwkHeaderRule.match(File, &Matches);
         // Returned matches are always in stable order.
         if (Matches.size() != 4)
           return std::nullopt;

clang/lib/InstallAPI/CMakeLists.txt

@@ -5,6 +5,8 @@ set(LLVM_LINK_COMPONENTS
 
 add_clang_library(clangInstallAPI
   Context.cpp
+  FileList.cpp
+  HeaderFile.cpp
 
   LINK_LIBS
   clangAST