[clang-fuzzer] Add a tiny tool to generate a fuzzing dictionary for clang

[sammccall]

Summary

It should be useful clang-fuzzer itself, though my own motivation is
to use this in fuzzing clang-pseudo. (clang-tools-extra/pseudo/fuzzer).

[hokein]

clang/tools/clang-fuzzer/dictionary/CMakeLists.txt

3

nit: remove the extra blank line.

clang/tools/clang-fuzzer/dictionary/dictionary.c

58

nit: remove the extra blank line.

clang/test/CMakeLists.txt

 if(CLANG_TEST_USE_VG)
   set(CLANG_TEST_EXTRA_ARGS ${CLANG_TEST_EXTRA_ARGS} "--vg")
 endif ()
 
 list(APPEND CLANG_TEST_DEPS
   apinotes-test
   c-index-test
   clang
+  clang-fuzzer-dictionary
   clang-resource-headers
   clang-format
   clang-tblgen
   clang-offload-bundler
   clang-import-test
   clang-rename
   clang-refactor
   clang-repl

clang/test/Misc/fuzzer-dictionary.test

+RUN: clang-fuzzer-dictionary | FileCheck %s
+CHECK-DAG: or="or"
+CHECK-DAG: catch="catch"
+CHECK-DAG: rawstart="R\x22("

clang/tools/clang-fuzzer/CMakeLists.txt

target_link_libraries(clang-llvm-proto-fuzzer
     clangCXXLoopProto
     clangLoopProtoToLLVM
     )
 
 endif()
 
 add_clang_subdirectory(handle-cxx)
 add_clang_subdirectory(handle-llvm)
+add_clang_subdirectory(dictionary)
 
 add_clang_executable(clang-fuzzer
   EXCLUDE_FROM_ALL
   ${DUMMY_MAIN}
   ClangFuzzer.cpp
   )
 
 target_link_libraries(clang-fuzzer

clang/tools/clang-fuzzer/dictionary/CMakeLists.txt

+add_clang_executable(clang-fuzzer-dictionary dictionary.c)
+

clang/tools/clang-fuzzer/dictionary/dictionary.c

+//===-- dictionary.c - Generate fuzzing dictionary for clang --------------===//
+//
+// 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 binary emits a fuzzing dictionary describing strings that are
+// significant to the clang parser: keywords and other tokens.
+//
+// The dictionary can be used by a fuzzer to reach interesting parser states
+// much more quickly.
+//
+// The output is a single-file dictionary supported by libFuzzer and AFL:
+// https://llvm.org/docs/LibFuzzer.html#dictionaries
+//
+//===----------------------------------------------------------------------===//
+
+#include <stdio.h>
+
+static void emit(const char *Name, const char *Spelling) {
+  static char Hex[] = "0123456789abcdef";
+
+  printf("%s=\"", Name);
+  unsigned char C;
+  while ((C = *Spelling++)) {
+    if (C < 32 || C == '"' || C == '\\')
+      printf("\\x%c%c", Hex[C>>4], Hex[C%16]);
+    else
+      printf("%c", C);
+  }
+  printf("\"\n");
+}
+
+int main(int argc, char **argv) {
+#define PUNCTUATOR(Name, Spelling) emit(#Name, Spelling);
+#define KEYWORD(Name, Criteria) emit(#Name, #Name);
+#define PPKEYWORD(Name) emit(#Name, #Name);
+#define CXX_KEYWORD_OPERATOR(Name, Equivalent) emit(#Name, #Name);
+#define OBJC_AT_KEYWORD(Name) emit(#Name, #Name);
+#define ALIAS(Spelling, Equivalent, Criteria) emit(Spelling, Spelling);
+#include "clang/Basic/TokenKinds.def"
+  // Some other sub-token chunks significant to the lexer.
+  emit("ucn16", "\\u0000");
+  emit("ucn32", "\\U00000000");
+  emit("rawstart", "R\"(");
+  emit("rawend", ")\"");
+  emit("quote", "\"");
+  emit("squote", "'");
+  emit("u8quote", "u8\"");
+  emit("u16quote", "u\"");
+  emit("u32quote", "U\"");
+  emit("esc_nl", "\\\n");
+  emit("hex", "0x");
+}
+