From fcc654699dd59d08226a1efc46ea5f66178ce274 Mon Sep 17 00:00:00 2001
From: Jason Volk <jason@zemos.net>
Date: Fri, 4 Nov 2016 14:46:42 -0700
Subject: [PATCH] ircd::locale: Add char16_t conversion overloads for
 unterminated strings. ircd::js: Use sized conversions when calling
 ircd::locale.

---
 include/ircd/js/native.h |   2 +-
 include/ircd/locale.h    |   8 +++
 ircd/locale.cc           | 102 ++++++++++++++++++++++++++++++++++++++-
 3 files changed, 109 insertions(+), 3 deletions(-)

diff --git a/include/ircd/js/native.h b/include/ircd/js/native.h
index 553e7aa4d..0dd0b48b5 100644
--- a/include/ircd/js/native.h
+++ b/include/ircd/js/native.h
@@ -56,6 +56,6 @@ ircd::js::native_external_copy(const char *const &s,
                                const size_t &len)
 {
 	auto buf(std::make_unique<char16_t[]>(len + 1));
-	ircd::locale::char16::conv(s, buf.get(), len + 1);
+	ircd::locale::char16::conv(s, len, buf.get(), len + 1);
 	return buf;
 }
diff --git a/include/ircd/locale.h b/include/ircd/locale.h
index a717600bf..a87f0cde0 100644
--- a/include/ircd/locale.h
+++ b/include/ircd/locale.h
@@ -35,10 +35,18 @@ namespace char16
 {
 	char conv(const char16_t &);
 	char16_t conv(const char &);
+
+	size_t conv(const char16_t *const &, const size_t &len, char *const &buf, const size_t &max);
+	size_t conv(const char *const &, const size_t &len, char16_t *const &buf, const size_t &max);  // uint8_t = max*2
+
 	size_t conv(const char16_t *const &, char *const &buf, const size_t &max);
 	size_t conv(const char *const &, char16_t *const &buf, const size_t &max);  // uint8_t = max*2
+
+	std::string conv(const char16_t *const &, const size_t &len);
 	std::string conv(const char16_t *const &);
 	std::string conv(const std::u16string &);
+
+	std::u16string conv(const char *const &, const size_t &len);
 	std::u16string conv(const char *const &);
 	std::u16string conv(const std::string &);
 }
diff --git a/ircd/locale.cc b/ircd/locale.cc
index 70c7e5417..dee2eed90 100644
--- a/ircd/locale.cc
+++ b/ircd/locale.cc
@@ -59,6 +59,23 @@ ircd::locale::char16::conv(const char *const &s)
 }
 #endif
 
+#ifdef HAVE_CODECVT
+std::u16string
+ircd::locale::char16::conv(const char *const &s,
+                           const size_t &len)
+{
+	static std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> converter;
+	return s && len? converter.from_bytes(s, s + len) : std::u16string{};
+}
+#else
+std::u16string
+ircd::locale::char16::conv(const char *const &s,
+                           const size_t &len)
+{
+	return boost::locale::conv::utf_to_utf<char16_t>(s, s + len);
+}
+#endif
+
 #ifdef HAVE_CODECVT
 std::string
 ircd::locale::char16::conv(const std::u16string &s)
@@ -89,6 +106,23 @@ ircd::locale::char16::conv(const char16_t *const &s)
 }
 #endif
 
+#ifdef HAVE_CODECVT
+std::string
+ircd::locale::char16::conv(const char16_t *const &s,
+                           const size_t &len)
+{
+	static std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> converter;
+	return s && len? converter.to_bytes(s, s + len) : std::string{};
+}
+#else
+std::string
+ircd::locale::char16::conv(const char16_t *const &s,
+                           const size_t &len)
+{
+	return boost::locale::conv::utf_to_utf<char>(s, s + len);
+}
+#endif
+
 #ifdef HAVE_CODECVT
 size_t
 ircd::locale::char16::conv(const char16_t *const &str,
@@ -104,8 +138,8 @@ ircd::locale::char16::conv(const char16_t *const &str,
 #else
 size_t
 ircd::locale::char16::conv(const char16_t *const &str,
-                      char *const &buf,
-                      const size_t &max)
+                           char *const &buf,
+                           const size_t &max)
 {
 	//TODO: optimize
 	const auto s(boost::locale::conv::utf_to_utf<char>(str));
@@ -113,6 +147,32 @@ ircd::locale::char16::conv(const char16_t *const &str,
 }
 #endif
 
+#ifdef HAVE_CODECVT
+size_t
+ircd::locale::char16::conv(const char16_t *const &str,
+                           const size_t &len,
+                           char *const &buf,
+                           const size_t &max)
+{
+	static std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> converter;
+
+	//TODO: optimize
+	const auto s(converter.to_bytes(str, str + len));
+	return rb_strlcpy(buf, s.c_str(), max);
+}
+#else
+size_t
+ircd::locale::char16::conv(const char16_t *const &str,
+                           const size_t &len,
+                           char *const &buf,
+                           const size_t &max)
+{
+	//TODO: optimize
+	const auto s(boost::locale::conv::utf_to_utf<char>(str, str + len));
+	return rb_strlcpy(buf, s.c_str(), max);
+}
+#endif
+
 #ifdef HAVE_CODECVT
 size_t
 ircd::locale::char16::conv(const char *const &str,
@@ -149,6 +209,44 @@ ircd::locale::char16::conv(const char *const &str,
 }
 #endif
 
+#ifdef HAVE_CODECVT
+size_t
+ircd::locale::char16::conv(const char *const &str,
+                           const size_t &len,
+                           char16_t *const &buf,
+                           const size_t &max)
+{
+	static std::wstring_convert<std::codecvt_utf8<char16_t>, char16_t> converter;
+
+	if(unlikely(!max))
+		return 0;
+
+	//TODO: optimize
+	const auto s(converter.from_bytes(str, str + len));
+	const auto cpsz(std::min(s.size(), size_t(max - 1)));
+	memcpy(buf, s.data(), cpsz * 2);
+	buf[cpsz] = char16_t(0);
+	return cpsz;
+}
+#else
+size_t
+ircd::locale::char16::conv(const char *const &str,
+                           const size_t &len,
+                           char16_t *const &buf,
+                           const size_t &max)
+{
+	if(unlikely(!max))
+		return 0;
+
+	//TODO: optimize
+	const auto s(boost::locale::conv::utf_to_utf<char16_t>(str, str + len));
+	const auto cpsz(std::min(s.size(), size_t(max - 1)));
+	memcpy(buf, s.data(), cpsz * 2);
+	buf[cpsz] = char16_t(0);
+	return cpsz;
+}
+#endif
+
 char16_t
 ircd::locale::char16::conv(const char &c)
 {