X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2FEncoding.cpp;h=ff83ab8c139a972f2392b399043e08a717947447;hb=4db3e641ed6765e005343010cb90ee8af26f8f99;hp=48e9271395fc43dcdf5bafca902b8067e14dbdec;hpb=ee3ce572a0b4fc2b71fcf7da0f1136174c2f4928;p=lyx.git diff --git a/src/Encoding.cpp b/src/Encoding.cpp index 48e9271395..ff83ab8c13 100644 --- a/src/Encoding.cpp +++ b/src/Encoding.cpp @@ -37,6 +37,8 @@ using namespace lyx::support; namespace lyx { +int const Encoding::any = -1; + Encodings encodings; Encodings::MathCommandSet Encodings::mathcmd; @@ -143,7 +145,7 @@ char_type arabic_table[172][4] = { {0, 0, 0, 0}, // 0x067b {0, 0, 0, 0}, // 0x067c {0, 0, 0, 0}, // 0x067d - {0xfb56, 0xfb57, 0xfb58, 0xfb59}, // 0x067e = peh + {0xfb56, 0xfb57, 0xfb58, 0xfb59}, // 0x067e = peh {0, 0, 0, 0}, // 0x067f {0, 0, 0, 0}, // 0x0680 {0, 0, 0, 0}, // 0x0681 @@ -151,7 +153,7 @@ char_type arabic_table[172][4] = { {0, 0, 0, 0}, // 0x0683 {0, 0, 0, 0}, // 0x0684 {0, 0, 0, 0}, // 0x0685 - {0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d}, // 0x0686 = tcheh + {0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d}, // 0x0686 = tcheh {0, 0, 0, 0}, // 0x0687 {0, 0, 0, 0}, // 0x0688 {0, 0, 0, 0}, // 0x0689 @@ -186,13 +188,13 @@ char_type arabic_table[172][4] = { {0, 0, 0, 0}, // 0x06a6 {0, 0, 0, 0}, // 0x06a7 {0, 0, 0, 0}, // 0x06a8 - {0xfb8e, 0xfb8f, 0xfb90, 0xfb91}, // 0x06a9 = farsi kaf + {0xfb8e, 0xfb8f, 0xfb90, 0xfb91}, // 0x06a9 = farsi kaf {0, 0, 0, 0}, // 0x06aa {0, 0, 0, 0}, // 0x06ab {0, 0, 0, 0}, // 0x06ac {0, 0, 0, 0}, // 0x06ad {0, 0, 0, 0}, // 0x06ae - {0xfb92, 0xfb93, 0xfb94, 0xfb95}, // 0x06af = gaf + {0xfb92, 0xfb93, 0xfb94, 0xfb95}, // 0x06af = gaf {0, 0, 0, 0}, // 0x06b0 {0, 0, 0, 0}, // 0x06b1 {0, 0, 0, 0}, // 0x06b2 @@ -221,7 +223,7 @@ char_type arabic_table[172][4] = { {0, 0, 0, 0}, // 0x06c9 {0, 0, 0, 0}, // 0x06ca {0, 0, 0, 0}, // 0x06cb - {0xfbfc, 0xfbfd, 0xfbfe, 0xfbff} // 0x06cc = farsi yeh + {0xfbfc, 0xfbfd, 0xfbfe, 0xfbff} // 0x06cc = farsi yeh }; @@ -242,6 +244,8 @@ enum CharInfoFlags { CharInfoTextNoTermination = 16, /// CharInfoMathNoTermination = 32, + /// + CharInfoForceSelected = 64, }; /// Information about a single UCS4 character @@ -265,6 +269,8 @@ struct CharInfo { /// Always force the LaTeX command, even if the encoding contains /// this character? bool force() const { return flags & CharInfoForce ? true : false; } + /// Force the LaTeX command for some encodings? + bool forceselected() const { return flags & CharInfoForceSelected ? true : false; } /// TIPA shortcut string tipashortcut; /// \c textcommand needs no termination (such as {} or space). @@ -280,7 +286,9 @@ typedef map CharInfoMap; CharInfoMap unicodesymbols; typedef std::set CharSet; +typedef std::map CharSetMap; CharSet forced; +CharSetMap forcedselected; typedef std::set MathAlphaSet; MathAlphaSet mathalpha; @@ -307,7 +315,7 @@ const char * EncodingException::what() const throw() Encoding::Encoding(string const & n, string const & l, string const & g, string const & i, bool f, bool u, Encoding::Package p) : name_(n), latexName_(l), guiName_(g), iconvName_(i), fixedwidth_(f), - unsafe_(u), package_(p) + unsafe_(u), forced_(&forcedselected[n]), package_(p) { if (n == "ascii") { // ASCII can encode 128 code points and nothing else @@ -318,6 +326,7 @@ Encoding::Encoding(string const & n, string const & l, string const & g, start_encodable_ = max_ucs4; complete_ = true; } else { + start_encodable_ = 0; complete_ = false; } } @@ -341,8 +350,12 @@ void Encoding::init() const continue; char_type const uc = ucs4[0]; CharInfoMap::const_iterator const it = unicodesymbols.find(uc); - if (it == unicodesymbols.end() || !it->second.force()) + if (it == unicodesymbols.end()) encodable_.insert(uc); + else if (!it->second.force()) { + if (forced_->empty() || forced_->find(uc) == forced_->end()) + encodable_.insert(uc); + } } } else { // We do not know how many code points this encoding has, and @@ -353,8 +366,12 @@ void Encoding::init() const vector const eightbit = ucs4_to_eightbit(&c, 1, iconvName_); if (!eightbit.empty()) { CharInfoMap::const_iterator const it = unicodesymbols.find(c); - if (it == unicodesymbols.end() || !it->second.force()) + if (it == unicodesymbols.end()) encodable_.insert(c); + else if (!it->second.force()) { + if (forced_->empty() || forced_->find(c) == forced_->end()) + encodable_.insert(c); + } } } } @@ -369,6 +386,14 @@ void Encoding::init() const } +bool Encoding::isForced(char_type c) const +{ + if (!forced.empty() && forced.find(c) != forced.end()) + return true; + return !forced_->empty() && forced_->find(c) != forced_->end(); +} + + bool Encoding::encodable(char_type c) const { // assure the used encoding is properly initialized @@ -376,7 +401,7 @@ bool Encoding::encodable(char_type c) const if (iconvName_ == "UTF-8" && package_ == none) return true; - if (c < start_encodable_ && !encodings.isForced(c)) + if (c < start_encodable_ && !isForced(c)) return true; if (encodable_.find(c) != encodable_.end()) return true; @@ -534,14 +559,28 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, bool const mathmode = cmdtype & MATH_CMD; bool const textmode = cmdtype & TEXT_CMD; docstring symbols; - size_t i = 0; size_t const cmdend = cmd.size(); + size_t prefix = 0; CharInfoMap::const_iterator const uniend = unicodesymbols.end(); - for (size_t j = 0; j < cmdend; ++j) { + for (size_t i = 0, j = 0; j < cmdend; ++j) { // Also get the char after a backslash - if (j + 1 < cmdend && cmd[j] == '\\') + if (j + 1 < cmdend && cmd[j] == '\\') { ++j; + prefix = 1; + // Detect things like \=*{e} as well + if (j + 3 < cmdend && cmd[j+1] == '*' && + cmd[j+2] == '{') { + ++j; + prefix = 2; + } + } + // position of the last character before a possible macro + // argument + size_t m = j; // If a macro argument follows, get it, too + // Do it here only for single character commands. Other + // combining commands need this too, but they are handled in + // the loop below for performance reasons. if (j + 1 < cmdend && cmd[j + 1] == '{') { size_t k = j + 1; int count = 1; @@ -554,12 +593,19 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, } if (k != docstring::npos) j = k; + } else if (m + 1 < cmdend && isAlphaASCII(cmd[m])) { + while (m + 2 < cmdend && isAlphaASCII(cmd[m+1])) + m++; } // Start with this substring and try augmenting it when it is // the prefix of some command in the unicodesymbols file - docstring const subcmd = cmd.substr(i, j - i + 1); + docstring subcmd = cmd.substr(i, j - i + 1); CharInfoMap::const_iterator it = unicodesymbols.begin(); + // First part of subcmd which might be a combining character + docstring combcmd = (m == j) ? docstring() : cmd.substr(i, m - i + 1); + // The combining character of combcmd if it exists + CharInfoMap::const_iterator combining = uniend; size_t unicmd_size = 0; char_type c = 0; for (; it != uniend; ++it) { @@ -567,6 +613,9 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, : docstring(); docstring const text = textmode ? it->second.textcommand : docstring(); + if (!combcmd.empty() && it->second.combining() && + (math == combcmd || text == combcmd)) + combining = it; size_t cur_size = max(math.size(), text.size()); // The current math or text unicode command cannot // match, or we already matched a longer one @@ -593,6 +642,26 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, // If this is an exact match, we found a (longer) // matching entry in the unicodesymbols file. + if (math != tmp && text != tmp) + continue; + // If we found a combining command, we need to append + // the macro argument if this has not been done above. + if (tmp == combcmd && combining != uniend && + k < cmdend && cmd[k] == '{') { + size_t l = k; + int count = 1; + while (l < cmdend && count && l != docstring::npos) { + l = cmd.find_first_of(from_ascii("{}"), l + 1); + if (cmd[l] == '{') + ++count; + else + --count; + } + if (l != docstring::npos) { + j = l; + subcmd = cmd.substr(i, j - i + 1); + } + } // If the entry doesn't start with '\', we take note // of the match and continue (this is not a ultimate // acceptance, as some other entry may match a longer @@ -602,12 +671,13 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, // (nonletter) char macro, or nothing else follows, // or what follows is a nonletter char, or the last // character is a }. - if ((math == tmp || text == tmp) - && (tmp[0] != '\\' - || (tmp.size() == 2 && !isAlphaASCII(tmp[1])) + else if (tmp[0] != '\\' + || (tmp.size() == prefix + 1 && + !isAlphaASCII(tmp[1]) && + (prefix == 1 || !isAlphaASCII(tmp[2]))) || k == cmdend || !isAlphaASCII(cmd[k]) - || tmp[tmp.size() - 1] == '}') + || tmp[tmp.size() - 1] == '}' ) { c = it->first; j = k - 1; @@ -629,7 +699,40 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, } if (unicmd_size) symbols += c; - else if (j + 1 == cmdend) { + else if (combining != uniend && + prefixIs(subcmd, combcmd + '{')) { + // We know that subcmd starts with combcmd and + // contains an argument in braces. + docstring const arg = subcmd.substr( + combcmd.length() + 1, + subcmd.length() - combcmd.length() - 2); + // If arg is a single character we can construct a + // combining sequence. + char_type a; + bool argcomb = false; + if (arg.size() == 1 && isAlnumASCII(arg[0])) + a = arg[0]; + else { + // Use the version of fromLaTeXCommand() that + // parses only one command, since we cannot + // use more than one character. + bool dummy = false; + set r; + a = fromLaTeXCommand(arg, cmdtype, argcomb, + dummy, &r); + if (a && req && !argcomb) + req->insert(r.begin(), r.end()); + } + if (a && !argcomb) { + // In unicode the combining character comes + // after its base + symbols += a; + symbols += combining->first; + i = j + 1; + unicmd_size = 2; + } + } + if (j + 1 == cmdend && !unicmd_size) { // No luck. Return what remains rem = cmd.substr(i); if (needsTermination && !rem.empty()) { @@ -812,12 +915,6 @@ bool Encodings::isKnownScriptChar(char_type const c, string & preamble) } -bool Encodings::isForced(char_type c) -{ - return (!forced.empty() && forced.find(c) != forced.end()); -} - - bool Encodings::isMathAlpha(char_type c) { return mathalpha.count(c); @@ -835,7 +932,7 @@ Encodings::fromLyXName(string const & name, bool allowUnsafe) const Encoding const * -Encodings::fromLaTeXName(string const & n, bool allowUnsafe) const +Encodings::fromLaTeXName(string const & n, int const & p, bool allowUnsafe) const { string name = n; // FIXME: if we have to test for too many of these synonyms, @@ -850,11 +947,21 @@ Encodings::fromLaTeXName(string const & n, bool allowUnsafe) const // most at the top of lib/encodings. EncodingList::const_iterator const end = encodinglist.end(); for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it) - if (it->second.latexName() == name) { - if (!allowUnsafe && it->second.unsafe()) - return 0; + if ((it->second.latexName() == name) && (it->second.package() & p) + && (!it->second.unsafe() || allowUnsafe)) + return &it->second; + return 0; +} + + +Encoding const * +Encodings::fromIconvName(string const & n, int const & p, bool allowUnsafe) const +{ + EncodingList::const_iterator const end = encodinglist.end(); + for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it) + if ((it->second.iconvName() == n) && (it->second.package() & p) + && (!it->second.unsafe() || allowUnsafe)) return &it->second; - } return 0; } @@ -867,6 +974,7 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) { // We must read the symbolsfile first, because the Encoding // constructor depends on it. + CharSetMap forcednotselected; Lexer symbolslex; symbolslex.setFile(symbolsfile); bool getNextToken = true; @@ -912,6 +1020,18 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) } else if (flag == "force") { info.flags |= CharInfoForce; forced.insert(symbol); + } else if (prefixIs(flag, "force=")) { + vector encodings = + getVectorFromString(flag.substr(6), ";"); + for (size_t i = 0; i < encodings.size(); ++i) + forcedselected[encodings[i]].insert(symbol); + info.flags |= CharInfoForceSelected; + } else if (prefixIs(flag, "force!=")) { + vector encodings = + getVectorFromString(flag.substr(7), ";"); + for (size_t i = 0; i < encodings.size(); ++i) + forcednotselected[encodings[i]].insert(symbol); + info.flags |= CharInfoForceSelected; } else if (flag == "mathalpha") { mathalpha.insert(symbol); } else if (flag == "notermination=text") { @@ -974,7 +1094,8 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) << " '" << info.textfeature() << ' ' << info.textnotermination() << ' ' << to_utf8(info.mathcommand) << "' '" << info.mathpreamble << "' " << info.mathfeature() << ' ' << info.mathnotermination() - << ' ' << info.combining() << ' ' << info.force()); + << ' ' << info.combining() << ' ' << info.force() + << ' ' << info.forceselected()); // we assume that at least one command is nonempty when using unicodesymbols if (!info.textcommand.empty() || !info.mathcommand.empty()) @@ -1058,6 +1179,15 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) break; } } + + // Move all information from forcednotselected to forcedselected + for (CharSetMap::const_iterator it1 = forcednotselected.begin(); it1 != forcednotselected.end(); ++it1) { + for (CharSetMap::iterator it2 = forcedselected.begin(); it2 != forcedselected.end(); ++it2) { + if (it2->first != it1->first) + it2->second.insert(it1->second.begin(), it1->second.end()); + } + } + }