aboutsummaryrefslogtreecommitdiffstats
path: root/libdevcore/UTF8.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'libdevcore/UTF8.cpp')
-rw-r--r--libdevcore/UTF8.cpp84
1 files changed, 73 insertions, 11 deletions
diff --git a/libdevcore/UTF8.cpp b/libdevcore/UTF8.cpp
index 449ccc5d..2ae720ec 100644
--- a/libdevcore/UTF8.cpp
+++ b/libdevcore/UTF8.cpp
@@ -27,25 +27,74 @@
namespace dev
{
+namespace
+{
-bool validateUTF8(std::string const& _input, size_t& _invalidPosition)
+/// Validate byte sequence against Unicode chapter 3 Table 3-7.
+bool isWellFormed(unsigned char byte1, unsigned char byte2)
+{
+ if (byte1 == 0xc0 || byte1 == 0xc1)
+ return false;
+ else if (byte1 >= 0xc2 && byte1 <= 0xdf)
+ return true;
+ else if (byte1 == 0xe0)
+ {
+ if (byte2 < 0xa0)
+ return false;
+ else
+ return true;
+ }
+ else if (byte1 >= 0xe1 && byte1 <= 0xec)
+ return true;
+ else if (byte1 == 0xed)
+ {
+ if (byte2 > 0x9f)
+ return false;
+ else
+ return true;
+ }
+ else if (byte1 == 0xee || byte1 == 0xef)
+ return true;
+ else if (byte1 == 0xf0)
+ {
+ if (byte2 < 0x90)
+ return false;
+ else
+ return true;
+ }
+ else if (byte1 >= 0xf1 && byte1 <= 0xf3)
+ return true;
+ else if (byte1 == 0xf4)
+ {
+ if (byte2 > 0x8f)
+ return false;
+ else
+ return true;
+ }
+ /// 0xf5 .. 0xf7 is disallowed
+ /// Technically anything below 0xc0 or above 0xf7 is
+ /// not possible to encode using Table 3-6 anyway.
+ return false;
+}
+
+bool validateUTF8(const unsigned char *_input, size_t _length, size_t& _invalidPosition)
{
- const size_t length = _input.length();
bool valid = true;
size_t i = 0;
- for (; i < length; i++)
+ for (; i < _length; i++)
{
- if ((unsigned char)_input[i] < 0x80)
+ // Check for Unicode Chapter 3 Table 3-6 conformity.
+ if (_input[i] < 0x80)
continue;
size_t count = 0;
- switch(_input[i] & 0xf0) {
- case 0xc0: count = 1; break;
- case 0xe0: count = 2; break;
- case 0xf0: count = 3; break;
- default: break;
- }
+ if (_input[i] >= 0xc0 && _input[i] <= 0xdf)
+ count = 1;
+ else if (_input[i] >= 0xe0 && _input[i] <= 0xef)
+ count = 2;
+ else if (_input[i] >= 0xf0 && _input[i] <= 0xf7)
+ count = 3;
if (count == 0)
{
@@ -53,7 +102,7 @@ bool validateUTF8(std::string const& _input, size_t& _invalidPosition)
break;
}
- if ((i + count) >= length)
+ if ((i + count) >= _length)
{
valid = false;
break;
@@ -67,6 +116,13 @@ bool validateUTF8(std::string const& _input, size_t& _invalidPosition)
valid = false;
break;
}
+
+ // Check for Unicode Chapter 3 Table 3-7 conformity.
+ if ((j == 0) && !isWellFormed(_input[i - 1], _input[i]))
+ {
+ valid = false;
+ break;
+ }
}
}
@@ -77,5 +133,11 @@ bool validateUTF8(std::string const& _input, size_t& _invalidPosition)
return false;
}
+}
+
+bool validateUTF8(std::string const& _input, size_t& _invalidPosition)
+{
+ return validateUTF8(reinterpret_cast<unsigned char const*>(_input.c_str()), _input.length(), _invalidPosition);
+}
}