aboutsummaryrefslogtreecommitdiffhomepage
path: root/src/UniConversion.cxx
diff options
context:
space:
mode:
authormitchell <unknown>2018-05-05 12:08:22 -0400
committermitchell <unknown>2018-05-05 12:08:22 -0400
commitad5951840a7e9d19c3652151e57466fa2c94e6a7 (patch)
treea77144132bd8fe505c8e0340e1e4e7d66d44bf07 /src/UniConversion.cxx
parent93462d87c3c8f398d5900be84349f29cb088d849 (diff)
downloadscintilla-mirror-ad5951840a7e9d19c3652151e57466fa2c94e6a7.tar.gz
Backport: Feature [feature-requests:#1212]. Move Unicode conversions into UniConversion.
Move Unicode conversion functions UnicodeFromUTF8 and UTF8FromUTF32Character into UniConversion. Backport of changeset 6645:463fa6965d9a.
Diffstat (limited to 'src/UniConversion.cxx')
-rw-r--r--src/UniConversion.cxx22
1 files changed, 21 insertions, 1 deletions
diff --git a/src/UniConversion.cxx b/src/UniConversion.cxx
index c4025c403..f2bf8d30f 100644
--- a/src/UniConversion.cxx
+++ b/src/UniConversion.cxx
@@ -65,6 +65,26 @@ void UTF8FromUTF16(const wchar_t *uptr, size_t tlen, char *putf, size_t len) {
putf[k] = '\0';
}
+void UTF8FromUTF32Character(int uch, char *putf) {
+ size_t k = 0;
+ if (uch < 0x80) {
+ putf[k++] = static_cast<char>(uch);
+ } else if (uch < 0x800) {
+ putf[k++] = static_cast<char>(0xC0 | (uch >> 6));
+ putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
+ } else if (uch < 0x10000) {
+ putf[k++] = static_cast<char>(0xE0 | (uch >> 12));
+ putf[k++] = static_cast<char>(0x80 | ((uch >> 6) & 0x3f));
+ putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
+ } else {
+ putf[k++] = static_cast<char>(0xF0 | (uch >> 18));
+ putf[k++] = static_cast<char>(0x80 | ((uch >> 12) & 0x3f));
+ putf[k++] = static_cast<char>(0x80 | ((uch >> 6) & 0x3f));
+ putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
+ }
+ putf[k] = '\0';
+}
+
size_t UTF16Length(const char *s, size_t len) {
size_t ulen = 0;
const unsigned char *us = reinterpret_cast<const unsigned char *>(s);
@@ -101,7 +121,7 @@ size_t UTF16FromUTF8(const char *s, size_t len, wchar_t *tbuf, size_t tlen) {
break;
}
- const size_t outLen = (byteCount < 4) ? 1 : 2;
+ const size_t outLen = UTF16LengthFromUTF8ByteCount(byteCount);
if (ui + outLen > tlen) {
throw std::runtime_error("UTF16FromUTF8: attempted write beyond end");
}