Merge pull request #5277 from diffblue/unicode

unicode.h: introduce utf8_to_utf32

Author: kroening

src/util/unicode.cpp

@@ -18,124 +18,117 @@ Author: Daniel Kroening, kroening@kroening.com
 #include "invariant.h"
 
 #ifdef _WIN32
-#include <util/pragma_push.def>
-#ifdef _MSC_VER
-#pragma warning(disable:4668)
-  // using #if/#elif on undefined macro
-#pragma warning(disable : 5039)
+#  include <util/pragma_push.def>
+#  ifdef _MSC_VER
+#    pragma warning(disable : 4668)
+// using #if/#elif on undefined macro
+#    pragma warning(disable : 5039)
 // pointer or reference to potentially throwing function passed to extern C
-#endif
-#include <windows.h>
-#include <util/pragma_pop.def>
+#  endif
+#  include <util/pragma_pop.def>
+#  include <windows.h>
 #endif
 
+static void utf8_append_code(unsigned int c, std::string &);
+
 std::string narrow(const wchar_t *s)
 {
-  #ifdef _WIN32
+#ifdef _WIN32
 
-  int slength=static_cast<int>(wcslen(s));
-  int rlength=
+  int slength = static_cast<int>(wcslen(s));
+  int rlength =
     WideCharToMultiByte(CP_UTF8, 0, s, slength, NULL, 0, NULL, NULL);
   std::string r(rlength, 0);
   WideCharToMultiByte(CP_UTF8, 0, s, slength, &r[0], rlength, NULL, NULL);
   return r;
 
-  #else
-  // dummy conversion
-  std::string r;
-  r.reserve(wcslen(s));
-  while(*s!=0)
-  {
-    r+=static_cast<char>(*s);
-    s++;
-  }
-
-  return r;
-  #endif
+#else
+  return narrow(std::wstring(s));
+#endif
 }
 
 std::wstring widen(const char *s)
 {
-  #ifdef _WIN32
+#ifdef _WIN32
 
-  int slength=static_cast<int>(strlen(s));
-  int rlength=
-    MultiByteToWideChar(CP_UTF8, 0, s, slength, NULL, 0);
+  int slength = static_cast<int>(strlen(s));
+  int rlength = MultiByteToWideChar(CP_UTF8, 0, s, slength, NULL, 0);
   std::wstring r(rlength, 0);
   MultiByteToWideChar(CP_UTF8, 0, s, slength, &r[0], rlength);
   return r;
 
-  #else
-  // dummy conversion
-  std::wstring r;
-  r.reserve(strlen(s));
-  while(*s!=0)
-  {
-    r+=wchar_t(*s);
-    s++;
-  }
-
-  return r;
-  #endif
+#else
+  return widen(std::string(s));
+#endif
 }
 
 std::string narrow(const std::wstring &s)
 {
-  #ifdef _WIN32
+#ifdef _WIN32
 
-  int slength=static_cast<int>(s.size());
-  int rlength=
+  int slength = static_cast<int>(s.size());
+  int rlength =
     WideCharToMultiByte(CP_UTF8, 0, &s[0], slength, NULL, 0, NULL, NULL);
   std::string r(rlength, 0);
   WideCharToMultiByte(CP_UTF8, 0, &s[0], slength, &r[0], rlength, NULL, NULL);
   return r;
 
-  #else
-  // dummy conversion
-  return std::string(s.begin(), s.end());
-  #endif
+#else
+  std::string result;
+
+  result.reserve(s.size()); // at least that long
+
+  for(const auto codepoint : s)
+    utf8_append_code(codepoint, result);
+
+  return result;
+#endif
 }
 
 std::wstring widen(const std::string &s)
 {
-  #ifdef _WIN32
+#ifdef _WIN32
 
-  int slength=static_cast<int>(s.size());
-  int rlength=
-    MultiByteToWideChar(CP_UTF8, 0, &s[0], slength, NULL, 0);
+  int slength = static_cast<int>(s.size());
+  int rlength = MultiByteToWideChar(CP_UTF8, 0, &s[0], slength, NULL, 0);
   std::wstring r(rlength, 0);
   MultiByteToWideChar(CP_UTF8, 0, &s[0], slength, &r[0], rlength);
   return r;
 
-  #else
-  // dummy conversion
-  return std::wstring(s.begin(), s.end());
-  #endif
+#else
+  auto utf32 = utf8_to_utf32(std::string(s));
+
+  std::wstring r;
+  r.reserve(utf32.size());
+  for(auto codepoint : utf32)
+    r += codepoint;
+  return r;
+#endif
 }
 
 /// Appends a unicode character to a utf8-encoded string
 /// \par parameters: character to append, string to append to
 static void utf8_append_code(unsigned int c, std::string &result)
 {
-  if(c<=0x7f)
-    result+=static_cast<char>(c);
-  else if(c<=0x7ff)
+  if(c <= 0x7f)
+    result += static_cast<char>(c);
+  else if(c <= 0x7ff)
   {
-    result+=static_cast<char>((c >> 6)   | 0xc0);
-    result+=static_cast<char>((c &0x3f) | 0x80);
+    result += static_cast<char>((c >> 6) | 0xc0);
+    result += static_cast<char>((c & 0x3f) | 0x80);
   }
-  else if(c<=0xffff)
+  else if(c <= 0xffff)
   {
-    result+=static_cast<char>((c >> 12)         | 0xe0);
-    result+=static_cast<char>(((c >> 6) &0x3f) | 0x80);
-    result+=static_cast<char>((c &0x3f)        | 0x80);
+    result += static_cast<char>((c >> 12) | 0xe0);
+    result += static_cast<char>(((c >> 6) & 0x3f) | 0x80);
+    result += static_cast<char>((c & 0x3f) | 0x80);
   }
   else
   {
-    result+=static_cast<char>((c >> 18)         | 0xf0);
-    result+=static_cast<char>(((c >> 12) &0x3f)| 0x80);
-    result+=static_cast<char>(((c >> 6) &0x3f) | 0x80);
-    result+=static_cast<char>((c &0x3f)        | 0x80);
+    result += static_cast<char>((c >> 18) | 0xf0);
+    result += static_cast<char>(((c >> 12) & 0x3f) | 0x80);
+    result += static_cast<char>(((c >> 6) & 0x3f) | 0x80);
+    result += static_cast<char>((c & 0x3f) | 0x80);
   }
 }
 
@@ -156,13 +149,13 @@ utf32_native_endian_to_utf8(const std::basic_string<unsigned int> &s)
 
 std::vector<std::string> narrow_argv(int argc, const wchar_t **argv_wide)
 {
-  if(argv_wide==nullptr)
+  if(argv_wide == nullptr)
     return std::vector<std::string>();
 
   std::vector<std::string> argv_narrow;
   argv_narrow.reserve(argc);
 
-  for(int i=0; i!=argc; ++i)
+  for(int i = 0; i != argc; ++i)
     argv_narrow.push_back(narrow(argv_wide[i]));
 
   return argv_narrow;
@@ -173,7 +166,7 @@ static void utf16_append_code(unsigned int code, std::wstring &result)
   // we do not treat 0xD800 to 0xDFFF, although
   // they are not valid unicode symbols
 
-  if(code<0xFFFF)
+  if(code < 0xFFFF)
   {
     // code is encoded as one UTF16 character
     result += static_cast<wchar_t>(code);
@@ -185,76 +178,89 @@ static void utf16_append_code(unsigned int code, std::wstring &result)
     // but let's not check it programmatically
 
     // encode the code in UTF16
-    code=code-0x10000;
+    code = code - 0x10000;
     const uint16_t i1 = static_cast<uint16_t>(((code >> 10) & 0x3ff) | 0xD800);
     result += static_cast<wchar_t>(i1);
     const uint16_t i2 = static_cast<uint16_t>((code & 0x3ff) | 0xDC00);
     result += static_cast<wchar_t>(i2);
   }
 }
 
-
 /// Convert UTF8-encoded string to UTF-16 with architecture-native endianness.
 /// \par parameters: String in UTF-8 format
 /// \return String in UTF-16 format. The encoding follows the endianness of the
 ///   architecture iff swap_bytes is true.
 std::wstring utf8_to_utf16_native_endian(const std::string &in)
 {
-    std::wstring result;
-    result.reserve(in.size());
-    std::string::size_type i=0;
-    while(i<in.size())
+  std::wstring result;
+  result.reserve(in.size());
+
+  for(auto codepoint : utf8_to_utf32(in))
+    utf16_append_code(codepoint, result);
+
+  return result;
+}
+
+/// Convert UTF8-encoded string to UTF-32 with architecture-native endianness.
+/// \par parameters: String in UTF-8 format
+/// \return String in UTF-32 format.
+std::u32string utf8_to_utf32(const std::string &utf8_str)
+{
+  std::u32string result;
+  result.reserve(utf8_str.size());
+  std::string::size_type i = 0;
+  while(i < utf8_str.size())
+  {
+    unsigned char c = utf8_str[i++];
+    char32_t code = 0;
+    // the ifs that follow find out how many UTF8 characters (1-4) store the
+    // next unicode character. This is determined by the few most
+    // significant bits.
+    if(c <= 0x7F)
     {
-      unsigned char c=in[i++];
-      unsigned int code=0;
-      // the ifs that follow find out how many UTF8 characters (1-4) store the
-      // next unicode character. This is determined by the few most
-      // significant bits.
-      if(c<=0x7F)
-      {
-        // if it's one character, then code is exactly the value
-        code=c;
-      }
-      else if(c<=0xDF && i<in.size())
-      { // in other cases, we need to read the right number of chars and decode
-        // note: if we wanted to make sure that we capture incorrect strings,
-        // we should check that whatever follows first character starts with
-        // bits 10.
-        code = (c & 0x1Fu) << 6;
-        c=in[i++];
-        code += c & 0x3Fu;
-      }
-      else if(c<=0xEF && i+1<in.size())
-      {
-        code = (c & 0xFu) << 12;
-        c=in[i++];
-        code += (c & 0x3Fu) << 6;
-        c=in[i++];
-        code += c & 0x3Fu;
-      }
-      else if(c<=0xF7 && i+2<in.size())
-      {
-        code = (c & 0x7u) << 18;
-        c=in[i++];
-        code += (c & 0x3Fu) << 12;
-        c=in[i++];
-        code += (c & 0x3Fu) << 6;
-        c=in[i++];
-        code += c & 0x3Fu;
-      }
-      else
-      {
-        // The string is not a valid UTF8 string! Either it has some characters
-        // missing from a multi-character unicode symbol, or it has a char with
-        // too high value.
-        // For now, let's replace the character with a space
-        code=32;
-      }
-
-      utf16_append_code(code, result);
+      // if it's one character, then code is exactly the value
+      code = c;
+    }
+    else if(c <= 0xDF && i < utf8_str.size())
+    { // in other cases, we need to read the right number of chars and decode
+      // note: if we wanted to make sure that we capture incorrect strings,
+      // we should check that whatever follows first character starts with
+      // bits 10.
+      code = (c & 0x1Fu) << 6;
+      c = utf8_str[i++];
+      code += c & 0x3Fu;
+    }
+    else if(c <= 0xEF && i + 1 < utf8_str.size())
+    {
+      code = (c & 0xFu) << 12;
+      c = utf8_str[i++];
+      code += (c & 0x3Fu) << 6;
+      c = utf8_str[i++];
+      code += c & 0x3Fu;
+    }
+    else if(c <= 0xF7 && i + 2 < utf8_str.size())
+    {
+      code = (c & 0x7u) << 18;
+      c = utf8_str[i++];
+      code += (c & 0x3Fu) << 12;
+      c = utf8_str[i++];
+      code += (c & 0x3Fu) << 6;
+      c = utf8_str[i++];
+      code += c & 0x3Fu;
+    }
+    else
+    {
+      // The string is not a valid UTF8 string! Either it has some characters
+      // missing from a multi-character unicode symbol, or it has a char with
+      // too high value.
+      // For now, let's replace the character with a space
+      code = 32;
     }
 
-    return result;
+    result.append(1, code);
+  }
+
+  return result;
 }
 
 /// Escapes non-printable characters, whitespace except for spaces, double

src/util/unicode.h

@@ -6,7 +6,6 @@ Author: Daniel Kroening, kroening@kroening.com
 
 \*******************************************************************/
 
-
 #ifndef CPROVER_UTIL_UNICODE_H
 #define CPROVER_UTIL_UNICODE_H
 
@@ -25,6 +24,10 @@ std::wstring widen(const std::string &s);
 std::string
 utf32_native_endian_to_utf8(const std::basic_string<unsigned int> &s);
 
+/// \param utf8_str: UTF-8 string
+/// \return UTF-32 encoding of the string
+std::u32string utf8_to_utf32(const std::string &utf8_str);
+
 std::wstring utf8_to_utf16_native_endian(const std::string &in);
 std::string utf16_native_endian_to_java(const char16_t ch);
 std::string utf16_native_endian_to_java(const std::wstring &in);
@@ -57,10 +60,8 @@ std::vector<const char *> to_c_str_array(It b, It e)
 {
   // Assumes that walking the range will be faster than repeated allocation
   std::vector<const char *> ret(std::distance(b, e) + 1, nullptr);
-  std::transform(b, e, std::begin(ret), [] (const std::string & s)
-    {
-      return s.c_str();
-    });
+  std::transform(
+    b, e, std::begin(ret), [](const std::string &s) { return s.c_str(); });
   return ret;
 }