diff options
Diffstat (limited to 'Swiften/IDN/ICUConverter.cpp')
-rw-r--r-- | Swiften/IDN/ICUConverter.cpp | 157 |
1 files changed, 157 insertions, 0 deletions
diff --git a/Swiften/IDN/ICUConverter.cpp b/Swiften/IDN/ICUConverter.cpp new file mode 100644 index 0000000..18ff231 --- /dev/null +++ b/Swiften/IDN/ICUConverter.cpp @@ -0,0 +1,157 @@ +/* + * Copyright (c) 2012-2013 Remko Tronçon + * Licensed under the GNU General Public License v3. + * See Documentation/Licenses/GPLv3.txt for more information. + */ + +#include <Swiften/IDN/ICUConverter.h> + +#pragma GCC diagnostic ignored "-Wold-style-cast" +#pragma clang diagnostic ignored "-Wheader-hygiene" +#include <unicode/uidna.h> +#include <unicode/usprep.h> +#include <unicode/ucnv.h> +#include <unicode/ustring.h> + + #include <boost/numeric/conversion/cast.hpp> + +using namespace Swift; +using boost::numeric_cast; + +namespace { + typedef std::vector<UChar, SafeAllocator<UChar> > ICUString; + + const char* toConstCharArray(const std::string& input) { + return input.c_str(); + } + + const char* toConstCharArray(const std::vector<unsigned char, SafeAllocator<unsigned char> >& input) { + return reinterpret_cast<const char*>(vecptr(input)); + } + + template<typename T> + ICUString convertToICUString(const T& s) { + ICUString result; + result.resize(s.size()); + UErrorCode status = U_ZERO_ERROR; + int32_t icuResultLength = numeric_cast<int32_t>(result.size()); + u_strFromUTF8Lenient(vecptr(result), numeric_cast<int32_t>(result.size()), &icuResultLength, toConstCharArray(s), numeric_cast<int32_t>(s.size()), &status); + if (status == U_BUFFER_OVERFLOW_ERROR) { + status = U_ZERO_ERROR; + result.resize(numeric_cast<size_t>(icuResultLength)); + u_strFromUTF8Lenient(vecptr(result), numeric_cast<int32_t>(result.size()), &icuResultLength, toConstCharArray(s), numeric_cast<int32_t>(s.size()), &status); + } + if (U_FAILURE(status)) { + return ICUString(); + } + result.resize(numeric_cast<size_t>(icuResultLength)); + return result; + } + + std::vector<char, SafeAllocator<char> > convertToArray(const ICUString& input) { + std::vector<char, SafeAllocator<char> > result; + result.resize(input.size()); + UErrorCode status = U_ZERO_ERROR; + int32_t inputLength = numeric_cast<int32_t>(result.size()); + u_strToUTF8(vecptr(result), numeric_cast<int32_t>(result.size()), &inputLength, vecptr(input), numeric_cast<int32_t>(input.size()), &status); + if (status == U_BUFFER_OVERFLOW_ERROR) { + status = U_ZERO_ERROR; + result.resize(numeric_cast<size_t>(inputLength)); + u_strToUTF8(vecptr(result), numeric_cast<int32_t>(result.size()), &inputLength, vecptr(input), numeric_cast<int32_t>(input.size()), &status); + } + if (U_FAILURE(status)) { + return std::vector<char, SafeAllocator<char> >(); + } + + result.resize(numeric_cast<size_t>(inputLength) + 1); + result[result.size() - 1] = '\0'; + return result; + } + + std::string convertToString(const ICUString& input) { + return std::string(vecptr(convertToArray(input))); + } + + UStringPrepProfileType getICUProfileType(IDNConverter::StringPrepProfile profile) { + switch(profile) { + case IDNConverter::NamePrep: return USPREP_RFC3491_NAMEPREP; + case IDNConverter::XMPPNodePrep: return USPREP_RFC3920_NODEPREP; + case IDNConverter::XMPPResourcePrep: return USPREP_RFC3920_RESOURCEPREP; + case IDNConverter::SASLPrep: return USPREP_RFC4013_SASLPREP; + } + assert(false); + return USPREP_RFC3491_NAMEPREP; + } + + template<typename StringType> + std::vector<char, SafeAllocator<char> > getStringPreparedDetail(const StringType& s, IDNConverter::StringPrepProfile profile) { + UErrorCode status = U_ZERO_ERROR; + + boost::shared_ptr<UStringPrepProfile> icuProfile(usprep_openByType(getICUProfileType(profile), &status), usprep_close); + assert(U_SUCCESS(status)); + + ICUString icuInput = convertToICUString(s); + ICUString icuResult; + UParseError parseError; + icuResult.resize(icuInput.size()); + int32_t icuResultLength = usprep_prepare(icuProfile.get(), vecptr(icuInput), numeric_cast<int32_t>(icuInput.size()), vecptr(icuResult), numeric_cast<int32_t>(icuResult.size()), USPREP_ALLOW_UNASSIGNED, &parseError, &status); + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + if (status == U_BUFFER_OVERFLOW_ERROR) { + status = U_ZERO_ERROR; + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + icuResultLength = usprep_prepare(icuProfile.get(), vecptr(icuInput), numeric_cast<int32_t>(icuInput.size()), vecptr(icuResult), numeric_cast<int32_t>(icuResult.size()), USPREP_ALLOW_UNASSIGNED, &parseError, &status); + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + } + if (U_FAILURE(status)) { + return std::vector<char, SafeAllocator<char> >(); + } + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + + return convertToArray(icuResult); + } +} + +namespace Swift { + +std::string ICUConverter::getStringPrepared(const std::string& s, StringPrepProfile profile) { + if (s.empty()) { + return ""; + } + std::vector<char, SafeAllocator<char> > preparedData = getStringPreparedDetail(s, profile); + if (preparedData.empty()) { + throw std::exception(); + } + return std::string(vecptr(preparedData)); +} + +SafeByteArray ICUConverter::getStringPrepared(const SafeByteArray& s, StringPrepProfile profile) { + if (s.empty()) { + return SafeByteArray(); + } + std::vector<char, SafeAllocator<char> > preparedData = getStringPreparedDetail(s, profile); + if (preparedData.empty()) { + throw std::exception(); + } + return createSafeByteArray(reinterpret_cast<const char*>(vecptr(preparedData))); +} + +std::string ICUConverter::getIDNAEncoded(const std::string& domain) { + UErrorCode status = U_ZERO_ERROR; + ICUString icuInput = convertToICUString(domain); + ICUString icuResult; + icuResult.resize(icuInput.size()); + UParseError parseError; + int32_t icuResultLength = uidna_IDNToASCII(vecptr(icuInput), numeric_cast<int32_t>(icuInput.size()), vecptr(icuResult), numeric_cast<int32_t>(icuResult.size()), UIDNA_DEFAULT, &parseError, &status); + if (status == U_BUFFER_OVERFLOW_ERROR) { + status = U_ZERO_ERROR; + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + icuResultLength = uidna_IDNToASCII(vecptr(icuInput), numeric_cast<int32_t>(icuInput.size()), vecptr(icuResult), numeric_cast<int32_t>(icuResult.size()), UIDNA_DEFAULT, &parseError, &status); + } + if (U_FAILURE(status)) { + return domain; + } + icuResult.resize(numeric_cast<size_t>(icuResultLength)); + return convertToString(icuResult); +} + +} |