QUtil::toUTF16

Jay Berkenbilt
1 parent 44674a3e
Showing 4 changed files with 75 additions and 0 deletions
include/qpdf/QUtil.hh
libqpdf/QUtil.cc
libtests/qtest/qutil/qutil.out
libtests/qutil.cc
@@ -140,6 +140,12 @@ namespace QUtil
     QPDF_DLL
     std::string toUTF8(unsigned long uval);
  
+    // Return a string containing the byte representation of the
+    // UTF-16 BE encoding for the unicode value passed in.
+    // Unrepresentable code points are converted to U+FFFD.
+    QPDF_DLL
+    std::string toUTF16(unsigned long uval);
+
     // If secure random number generation is supported on your
     // platform and qpdf was not compiled with insecure random number
     // generation, this returns a cryptographically secure random
@@ -506,6 +506,41 @@ QUtil::toUTF8(unsigned long uval)
     return result;
 }
  
+std::string
+QUtil::toUTF16(unsigned long uval)
+{
+    std::string result;
+    if ((uval >= 0xd800) && (uval <= 0xdfff))
+    {
+        result = "\xff\xfd";
+    }
+    else if (uval <= 0xffff)
+    {
+        char out[2];
+        out[0] = (uval & 0xff00) >> 8;
+        out[1] = (uval & 0xff);
+        result = std::string(out, 2);
+    }
+    else if (uval <= 0x10ffff)
+    {
+        char out[4];
+        uval -= 0x10000;
+        unsigned short high = ((uval & 0xffc00) >> 10) + 0xd800;
+        unsigned short low = (uval & 0x3ff) + 0xdc00;
+        out[0] = (high & 0xff00) >> 8;
+        out[1] = (high & 0xff);
+        out[2] = (low & 0xff00) >> 8;
+        out[3] = (low & 0xff);
+        result = std::string(out, 4);
+    }
+    else
+    {
+        result = "\xff\xfd";
+    }
+
+    return result;
+}
+
 // Random data support
  
 long
@@ -39,6 +39,14 @@ HAGOOGAMAGOOGLE: 0
 0x16059 -> f0 96 81 99
 0x7fffffff -> fd bf bf bf bf bf
 0x80000000: bounds error in QUtil::toUTF8
+---- utf16
+0x41 -> 00 41
+0xf7 -> 00 f7
+0x3c0 -> 03 c0
+0x16059 -> d8 18 dc 59
+0xdead -> ff fd
+0x7fffffff -> ff fd
+0x80000000 -> ff fd
 ---- whoami
 quack1
 quack2
@@ -193,6 +193,30 @@ void to_utf8_test()
     }
 }
  
+static void print_utf16(unsigned long val)
+{
+    std::string result = QUtil::toUTF16(val);
+    std::cout << "0x" << QUtil::int_to_string_base(val, 16) << " ->";
+    for (std::string::iterator iter = result.begin();
+         iter != result.end(); ++iter)
+    {
+        std::cout << " " << QUtil::int_to_string_base(
+            static_cast<int>(static_cast<unsigned char>(*iter)), 16, 2);
+    }
+    std::cout << std::endl;
+}
+
+void to_utf16_test()
+{
+    print_utf16(0x41UL);
+    print_utf16(0xF7UL);
+    print_utf16(0x3c0UL);
+    print_utf16(0x16059UL);
+    print_utf16(0xdeadUL);
+    print_utf16(0x7fffffffUL);
+    print_utf16(0x80000000UL);
+}
+
 void print_whoami(char const* str)
 {
     PointerHolder<char> dup(true, QUtil::copy_string(str));
@@ -299,6 +323,8 @@ int main(int argc, char* argv[])
 	getenv_test();
 	std::cout << "---- utf8" << std::endl;
 	to_utf8_test();
+	std::cout << "---- utf16" << std::endl;
+	to_utf16_test();
 	std::cout << "---- whoami" << std::endl;
 	get_whoami_test();
 	std::cout << "---- file" << std::endl;