From fc4547b412e28164af1bf8981234c6af959ccc0b Mon Sep 17 00:00:00 2001 From: Joel Klinghed Date: Tue, 13 Jun 2023 10:07:16 +0200 Subject: WIP --- utf/src/utf32.cc | 43 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) create mode 100644 utf/src/utf32.cc (limited to 'utf/src/utf32.cc') diff --git a/utf/src/utf32.cc b/utf/src/utf32.cc new file mode 100644 index 0000000..cfa29b6 --- /dev/null +++ b/utf/src/utf32.cc @@ -0,0 +1,43 @@ +#include "utf32.hh" + +#include "utf_error.hh" + +namespace utf { + +namespace { + +inline bool valid_codepoint(uint32_t c) { + return (c < 0xd800) || (c > 0xdfff && c <= 0x10ffff); +} + +} // namespace + +uint32_t read32be(std::string_view data, std::size_t& offset) { + if (offset > data.size() || data.size() - offset < 4) + return NEED_MORE; + uint32_t c = static_cast(data[offset]) << 24 + | static_cast(data[offset + 1] & 0xff) << 16 + | static_cast(data[offset + 2] & 0xff) << 8 + | static_cast(data[offset + 3] & 0xff); + if (valid_codepoint(c)) { + offset += 4; + return c; + } + return INVALID; +} + +uint32_t read32le(std::string_view data, std::size_t& offset) { + if (offset > data.size() || data.size() - offset < 4) + return NEED_MORE; + uint32_t c = static_cast(data[offset + 3]) << 24 + | static_cast(data[offset + 2] & 0xff) << 16 + | static_cast(data[offset + 1] & 0xff) << 8 + | static_cast(data[offset] & 0xff); + if (valid_codepoint(c)) { + offset += 4; + return c; + } + return INVALID; +} + +} // namespace utf -- cgit v1.2.3-70-g09d2