Boost logo

Boost-Commit :

Subject: [Boost-commit] svn:boost r86613 - trunk/boost/detail
From: ramey_at_[hidden]
Date: 2013-11-10 21:19:46


Author: ramey
Date: 2013-11-10 21:19:46 EST (Sun, 10 Nov 2013)
New Revision: 86613
URL: http://svn.boost.org/trac/boost/changeset/86613

Log:
changed codecvt_utf8 so that serialization uses the standard one if available. If there is no standard one available, it uses the one in boost/utility.

Added:
   trunk/boost/detail/utf8_codecvt_facet.ipp (contents, props changed)

Added: trunk/boost/detail/utf8_codecvt_facet.ipp
==============================================================================
--- /dev/null 00:00:00 1970 (empty, because file is newly added)
+++ trunk/boost/detail/utf8_codecvt_facet.ipp 2013-11-10 21:19:46 EST (Sun, 10 Nov 2013) (r86613)
@@ -0,0 +1,284 @@
+/////////1/////////2/////////3/////////4/////////5/////////6/////////7/////////8
+// utf8_codecvt_facet.ipp
+
+// Copyright (c) 2001 Ronald Garcia, Indiana University (garcia_at_[hidden])
+// Andrew Lumsdaine, Indiana University (lums_at_[hidden]).
+// Use, modification and distribution is subject to the Boost Software
+// License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at
+// http://www.boost.org/LICENSE_1_0.txt)
+
+// Please see the comments in <boost/detail/utf8_codecvt_facet.hpp> to
+// learn how this file should be used.
+
+#include <boost/detail/utf8_codecvt_facet.hpp>
+
+#include <cstdlib> // for multi-byte converson routines
+#include <cassert>
+
+#include <boost/limits.hpp>
+#include <boost/config.hpp>
+
+// If we don't have wstring, then Unicode support
+// is not available anyway, so we don't need to even
+// compiler this file. This also fixes the problem
+// with mingw, which can compile this file, but will
+// generate link error when building DLL.
+#ifndef BOOST_NO_STD_WSTRING
+
+BOOST_UTF8_BEGIN_NAMESPACE
+
+/////////1/////////2/////////3/////////4/////////5/////////6/////////7/////////8
+// implementation for wchar_t
+
+// Translate incoming UTF-8 into UCS-4
+std::codecvt_base::result utf8_codecvt_facet::do_in(
+ std::mbstate_t& /*state*/,
+ const char * from,
+ const char * from_end,
+ const char * & from_next,
+ wchar_t * to,
+ wchar_t * to_end,
+ wchar_t * & to_next
+) const {
+ // Basic algorithm: The first octet determines how many
+ // octets total make up the UCS-4 character. The remaining
+ // "continuing octets" all begin with "10". To convert, subtract
+ // the amount that specifies the number of octets from the first
+ // octet. Subtract 0x80 (1000 0000) from each continuing octet,
+ // then mash the whole lot together. Note that each continuing
+ // octet only uses 6 bits as unique values, so only shift by
+ // multiples of 6 to combine.
+ while (from != from_end && to != to_end) {
+
+ // Error checking on the first octet
+ if (invalid_leading_octet(*from)){
+ from_next = from;
+ to_next = to;
+ return std::codecvt_base::error;
+ }
+
+ // The first octet is adjusted by a value dependent upon
+ // the number of "continuing octets" encoding the character
+ const int cont_octet_count = get_cont_octet_count(*from);
+ const wchar_t octet1_modifier_table[] = {
+ 0x00, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc
+ };
+
+ // The unsigned char conversion is necessary in case char is
+ // signed (I learned this the hard way)
+ wchar_t ucs_result =
+ (unsigned char)(*from++) - octet1_modifier_table[cont_octet_count];
+
+ // Invariants :
+ // 1) At the start of the loop, 'i' continuing characters have been
+ // processed
+ // 2) *from points to the next continuing character to be processed.
+ int i = 0;
+ while(i != cont_octet_count && from != from_end) {
+
+ // Error checking on continuing characters
+ if (invalid_continuing_octet(*from)) {
+ from_next = from;
+ to_next = to;
+ return std::codecvt_base::error;
+ }
+
+ ucs_result *= (1 << 6);
+
+ // each continuing character has an extra (10xxxxxx)b attached to
+ // it that must be removed.
+ ucs_result += (unsigned char)(*from++) - 0x80;
+ ++i;
+ }
+
+ // If the buffer ends with an incomplete unicode character...
+ if (from == from_end && i != cont_octet_count) {
+ // rewind "from" to before the current character translation
+ from_next = from - (i+1);
+ to_next = to;
+ return std::codecvt_base::partial;
+ }
+ *to++ = ucs_result;
+ }
+ from_next = from;
+ to_next = to;
+
+ // Were we done converting or did we run out of destination space?
+ if(from == from_end) return std::codecvt_base::ok;
+ else return std::codecvt_base::partial;
+}
+
+std::codecvt_base::result utf8_codecvt_facet::do_out(
+ std::mbstate_t& /*state*/,
+ const wchar_t * from,
+ const wchar_t * from_end,
+ const wchar_t * & from_next,
+ char * to,
+ char * to_end,
+ char * & to_next
+) const
+{
+ // RG - consider merging this table with the other one
+ const wchar_t octet1_modifier_table[] = {
+ 0x00, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc
+ };
+
+ wchar_t max_wchar = (std::numeric_limits<wchar_t>::max)();
+ while (from != from_end && to != to_end) {
+
+ // Check for invalid UCS-4 character
+ if (*from > max_wchar) {
+ from_next = from;
+ to_next = to;
+ return std::codecvt_base::error;
+ }
+
+ int cont_octet_count = get_cont_octet_out_count(*from);
+
+ // RG - comment this formula better
+ int shift_exponent = (cont_octet_count) * 6;
+
+ // Process the first character
+ *to++ = static_cast<char>(octet1_modifier_table[cont_octet_count] +
+ (unsigned char)(*from / (1 << shift_exponent)));
+
+ // Process the continuation characters
+ // Invariants: At the start of the loop:
+ // 1) 'i' continuing octets have been generated
+ // 2) '*to' points to the next location to place an octet
+ // 3) shift_exponent is 6 more than needed for the next octet
+ int i = 0;
+ while (i != cont_octet_count && to != to_end) {
+ shift_exponent -= 6;
+ *to++ = static_cast<char>(0x80 + ((*from / (1 << shift_exponent)) % (1 << 6)));
+ ++i;
+ }
+ // If we filled up the out buffer before encoding the character
+ if(to == to_end && i != cont_octet_count) {
+ from_next = from;
+ to_next = to - (i+1);
+ return std::codecvt_base::partial;
+ }
+ ++from;
+ }
+ from_next = from;
+ to_next = to;
+ // Were we done or did we run out of destination space
+ if(from == from_end) return std::codecvt_base::ok;
+ else return std::codecvt_base::partial;
+}
+
+// How many char objects can I process to get <= max_limit
+// wchar_t objects?
+int utf8_codecvt_facet::do_length(
+ const std::mbstate_t &,
+ const char * from,
+ const char * from_end,
+ std::size_t max_limit
+) const
+#if BOOST_WORKAROUND(__IBMCPP__, BOOST_TESTED_AT(600))
+ throw()
+#endif
+{
+ // RG - this code is confusing! I need a better way to express it.
+ // and test cases.
+
+ // Invariants:
+ // 1) last_octet_count has the size of the last measured character
+ // 2) char_count holds the number of characters shown to fit
+ // within the bounds so far (no greater than max_limit)
+ // 3) from_next points to the octet 'last_octet_count' before the
+ // last measured character.
+ int last_octet_count=0;
+ std::size_t char_count = 0;
+ const char* from_next = from;
+ // Use "<" because the buffer may represent incomplete characters
+ while (from_next+last_octet_count <= from_end && char_count <= max_limit) {
+ from_next += last_octet_count;
+ last_octet_count = (get_octet_count(*from_next));
+ ++char_count;
+ }
+ return static_cast<int>(from_next-from_end);
+}
+
+unsigned int utf8_codecvt_facet::get_octet_count(
+ unsigned char lead_octet
+){
+ // if the 0-bit (MSB) is 0, then 1 character
+ if (lead_octet <= 0x7f) return 1;
+
+ // Otherwise the count number of consecutive 1 bits starting at MSB
+// assert(0xc0 <= lead_octet && lead_octet <= 0xfd);
+
+ if (0xc0 <= lead_octet && lead_octet <= 0xdf) return 2;
+ else if (0xe0 <= lead_octet && lead_octet <= 0xef) return 3;
+ else if (0xf0 <= lead_octet && lead_octet <= 0xf7) return 4;
+ else if (0xf8 <= lead_octet && lead_octet <= 0xfb) return 5;
+ else return 6;
+}
+BOOST_UTF8_END_NAMESPACE
+
+namespace {
+template<std::size_t s>
+int get_cont_octet_out_count_impl(wchar_t word){
+ if (word < 0x80) {
+ return 0;
+ }
+ if (word < 0x800) {
+ return 1;
+ }
+ return 2;
+}
+
+template<>
+int get_cont_octet_out_count_impl<4>(wchar_t word){
+ if (word < 0x80) {
+ return 0;
+ }
+ if (word < 0x800) {
+ return 1;
+ }
+
+ // Note that the following code will generate warnings on some platforms
+ // where wchar_t is defined as UCS2. The warnings are superfluous as the
+ // specialization is never instantitiated with such compilers, but this
+ // can cause problems if warnings are being treated as errors, so we guard
+ // against that. Including <boost/detail/utf8_codecvt_facet.hpp> as we do
+ // should be enough to get WCHAR_MAX defined.
+#if !defined(WCHAR_MAX)
+# error WCHAR_MAX not defined!
+#endif
+ // cope with VC++ 7.1 or earlier having invalid WCHAR_MAX
+#if defined(_MSC_VER) && _MSC_VER <= 1310 // 7.1 or earlier
+ return 2;
+#elif WCHAR_MAX > 0x10000
+
+ if (word < 0x10000) {
+ return 2;
+ }
+ if (word < 0x200000) {
+ return 3;
+ }
+ if (word < 0x4000000) {
+ return 4;
+ }
+ return 5;
+
+#else
+ return 2;
+#endif
+}
+
+} // namespace anonymous
+
+BOOST_UTF8_BEGIN_NAMESPACE
+// How many "continuing octets" will be needed for this word
+// == total octets - 1.
+int utf8_codecvt_facet::get_cont_octet_out_count(
+ wchar_t word
+) const {
+ return get_cont_octet_out_count_impl<sizeof(wchar_t)>(word);
+}
+BOOST_UTF8_END_NAMESPACE
+
+#endif


Boost-Commit list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk