embedd files using hex literals instead of string literals

String constants are limited to 64k, which requires splitting the file with
the compiler headers.

This switches the embedding to use an array of hex literals, which have no
size limit.

There is a prospect to use the #embed preprocessor command in C++26, moves
the conversion from file to hex literal array into the compiler's
preprocessor.

Author: kroening

src/ansi-c/ansi_c_internal_additions.cpp

@@ -18,41 +18,35 @@ Author: Daniel Kroening, kroening@kroening.com
 #include "ansi_c_parser.h"
 
 const char gcc_builtin_headers_types[] =
-  "#line 1 \"gcc_builtin_headers_types.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_types.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_generic[] =
-  "#line 1 \"gcc_builtin_headers_generic.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_generic.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_math[] =
-  "#line 1 \"gcc_builtin_headers_math.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_math.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_mem_string[] =
-  "#line 1 \"gcc_builtin_headers_mem_string.h\"\n"
 // NOLINTNEXTLINE(whitespace/line_length)
 #include "compiler_headers/gcc_builtin_headers_mem_string.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char gcc_builtin_headers_omp[] = "#line 1 \"gcc_builtin_headers_omp.h\"\n"
+const char gcc_builtin_headers_omp[] =
 #include "compiler_headers/gcc_builtin_headers_omp.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char gcc_builtin_headers_tm[] = "#line 1 \"gcc_builtin_headers_tm.h\"\n"
+const char gcc_builtin_headers_tm[] =
 #include "compiler_headers/gcc_builtin_headers_tm.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_ubsan[] =
-  "#line 1 \"gcc_builtin_headers_ubsan.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_ubsan.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_ia32[] =
-  "#line 1 \"gcc_builtin_headers_ia32.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_ia32.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 const char gcc_builtin_headers_ia32_2[] =
@@ -81,41 +75,38 @@ const char gcc_builtin_headers_ia32_9[] =
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_alpha[] =
-  "#line 1 \"gcc_builtin_headers_alpha.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_alpha.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char gcc_builtin_headers_arm[] = "#line 1 \"gcc_builtin_headers_arm.h\"\n"
+const char gcc_builtin_headers_arm[] =
 #include "compiler_headers/gcc_builtin_headers_arm.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_mips[] =
-  "#line 1 \"gcc_builtin_headers_mips.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_mips.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
 const char gcc_builtin_headers_power[] =
-  "#line 1 \"gcc_builtin_headers_power.h\"\n"
 #include "compiler_headers/gcc_builtin_headers_power.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char arm_builtin_headers[] = "#line 1 \"arm_builtin_headers.h\"\n"
+const char arm_builtin_headers[] =
 #include "compiler_headers/arm_builtin_headers.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char cw_builtin_headers[] = "#line 1 \"cw_builtin_headers.h\"\n"
+const char cw_builtin_headers[] =
 #include "compiler_headers/cw_builtin_headers.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char clang_builtin_headers[] = "#line 1 \"clang_builtin_headers.h\"\n"
+const char clang_builtin_headers[] =
 #include "compiler_headers/clang_builtin_headers.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 
-const char cprover_builtin_headers[] = "#line 1 \"cprover_builtin_headers.h\"\n"
+const char cprover_builtin_headers[] =
 #include "cprover_builtin_headers.inc" // IWYU pragma: keep
   ;                                    // NOLINT(whitespace/semicolon)
 
-const char windows_builtin_headers[] = "#line 1 \"windows_builtin_headers.h\"\n"
+const char windows_builtin_headers[] =
 #include "compiler_headers/windows_builtin_headers.inc" // IWYU pragma: keep
   ; // NOLINT(whitespace/semicolon)
 

src/ansi-c/compiler_headers/gcc_builtin_headers_mips.h

@@ -0,0 +1 @@
+// empty

src/ansi-c/compiler_headers/gcc_builtin_headers_power.h

@@ -0,0 +1 @@
+// empty

src/ansi-c/file_converter.cpp

@@ -1,46 +1,25 @@
 /*******************************************************************\
 
-Module: Convert file contents to C strings
+Module: Convert file contents to a C array
 
 Author: Daniel Kroening, kroening@kroening.com
 
 \*******************************************************************/
 
 /// \file
-/// Convert file contents to C strings
+/// Convert file contents to a C array
 
 #include <fstream> // IWYU pragma: keep
+#include <iomanip>
 #include <iostream>
 #include <string>
 
-static void convert_line(const std::string &line)
-{
-  std::cout << "\"";
-
-  for(std::size_t i = 0; i < line.size(); i++)
-  {
-    const char ch = line[i];
-    if(ch == '\\')
-      std::cout << "\\\\";
-    else if(ch == '"')
-      std::cout << "\\\"";
-    else if(ch == '\r' || ch == '\n')
-    {
-    }
-    else if((ch & 0x80) != 0)
-    {
-      std::cout << "\\x" << std::hex << (unsigned(ch) & 0xff) << std::dec;
-    }
-    else
-      std::cout << ch;
-  }
-
-  std::cout << "\\n\"\n";
-}
-
 int main(int argc, char *argv[])
 {
-  std::string line;
+  std::size_t counter = 0;
+  bool first = true;
+
+  std::cout << "{";
 
   for(int i = 1; i < argc; ++i)
   {
@@ -52,9 +31,31 @@ int main(int argc, char *argv[])
       return 1;
     }
 
-    while(getline(input_file, line))
-      convert_line(line);
+    char ch;
+    while(input_file.get(ch))
+    {
+      if(first)
+        first = false;
+      else
+        std::cout << ',';
+
+      if(counter == 15)
+      {
+        std::cout << '\n';
+        counter = 0;
+      }
+      else
+      {
+        counter++;
+        std::cout << ' ';
+      }
+
+      std::cout << "0x" << std::hex << std::setw(2) << std::setfill('0')
+                << int(ch);
+    }
   }
 
+  std::cout << " }\n";
+
   return 0;
 }