X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2FEncoding.cpp;h=d0046ecc068ee674d84776b6e672276de79140a0;hb=c7d29be153debac82e3d2e8865fcc849f0a5f40d;hp=9283e55fc8e8a1b9d6ea7637604a39376fe82638;hpb=3175a288938438e4239b5c9231894613c214260d;p=lyx.git diff --git a/src/Encoding.cpp b/src/Encoding.cpp index 9283e55fc8..d0046ecc06 100644 --- a/src/Encoding.cpp +++ b/src/Encoding.cpp @@ -14,21 +14,19 @@ #include "Encoding.h" -#include "Buffer.h" -#include "BufferList.h" -#include "InsetIterator.h" -#include "LaTeXFeatures.h" #include "Lexer.h" -#include "LyXRC.h" #include "support/debug.h" -#include "support/FileName.h" +#include "support/gettext.h" #include "support/lstrings.h" +#include "support/mutex.h" #include "support/textutils.h" #include "support/unicode.h" #include +#include +#include #include using namespace std; @@ -36,6 +34,8 @@ using namespace lyx::support; namespace lyx { +int const Encoding::any = -1; + Encodings encodings; Encodings::MathCommandSet Encodings::mathcmd; @@ -44,228 +44,22 @@ Encodings::MathSymbolSet Encodings::mathsym; namespace { -char_type arabic_table[172][4] = { - {0xfe80, 0xfe80, 0xfe80, 0xfe80}, // 0x0621 = hamza - {0xfe81, 0xfe82, 0xfe81, 0xfe82}, // 0x0622 = ligature madda on alef - {0xfe83, 0xfe84, 0xfe83, 0xfe84}, // 0x0623 = ligature hamza on alef - {0xfe85, 0xfe86, 0xfe85, 0xfe86}, // 0x0624 = ligature hamza on waw - {0xfe87, 0xfe88, 0xfe87, 0xfe88}, // 0x0625 = ligature hamza under alef - {0xfe89, 0xfe8a, 0xfe8b, 0xfe8c}, // 0x0626 = ligature hamza on ya - {0xfe8d, 0xfe8e, 0xfe8d, 0xfe8e}, // 0x0627 = alef - {0xfe8f, 0xfe90, 0xfe91, 0xfe92}, // 0x0628 = baa - {0xfe93, 0xfe94, 0xfe93, 0xfe94}, // 0x0629 = taa marbuta - {0xfe95, 0xfe96, 0xfe97, 0xfe98}, // 0x062a = taa - {0xfe99, 0xfe9a, 0xfe9b, 0xfe9c}, // 0x062b = thaa - {0xfe9d, 0xfe9e, 0xfe9f, 0xfea0}, // 0x062c = jeem - {0xfea1, 0xfea2, 0xfea3, 0xfea4}, // 0x062d = haa - {0xfea5, 0xfea6, 0xfea7, 0xfea8}, // 0x062e = khaa - {0xfea9, 0xfeaa, 0xfea9, 0xfeaa}, // 0x062f = dal - - {0xfeab, 0xfeac, 0xfeab, 0xfeac}, // 0x0630 = thal - {0xfead, 0xfeae, 0xfead, 0xfeae}, // 0x0631 = ra - {0xfeaf, 0xfeb0, 0xfeaf, 0xfeb0}, // 0x0632 = zain - {0xfeb1, 0xfeb2, 0xfeb3, 0xfeb4}, // 0x0633 = seen - {0xfeb5, 0xfeb6, 0xfeb7, 0xfeb8}, // 0x0634 = sheen - {0xfeb9, 0xfeba, 0xfebb, 0xfebc}, // 0x0635 = sad - {0xfebd, 0xfebe, 0xfebf, 0xfec0}, // 0x0636 = dad - {0xfec1, 0xfec2, 0xfec3, 0xfec4}, // 0x0637 = tah - {0xfec5, 0xfec6, 0xfec7, 0xfec8}, // 0x0638 = zah - {0xfec9, 0xfeca, 0xfecb, 0xfecc}, // 0x0639 = ain - {0xfecd, 0xfece, 0xfecf, 0xfed0}, // 0x063a = ghain - {0, 0, 0, 0}, // 0x063b - {0, 0, 0, 0}, // 0x063c - {0, 0, 0, 0}, // 0x063d - {0, 0, 0, 0}, // 0x063e - {0, 0, 0, 0}, // 0x063f - - {0, 0, 0, 0}, // 0x0640 - {0xfed1, 0xfed2, 0xfed3, 0xfed4}, // 0x0641 = fa - {0xfed5, 0xfed6, 0xfed7, 0xfed8}, // 0x0642 = qaf - {0xfed9, 0xfeda, 0xfedb, 0xfedc}, // 0x0643 = kaf - {0xfedd, 0xfede, 0xfedf, 0xfee0}, // 0x0644 = lam - {0xfee1, 0xfee2, 0xfee3, 0xfee4}, // 0x0645 = meem - {0xfee5, 0xfee6, 0xfee7, 0xfee8}, // 0x0646 = noon - {0xfee9, 0xfeea, 0xfeeb, 0xfeec}, // 0x0647 = ha - {0xfeed, 0xfeee, 0xfeed, 0xfeee}, // 0x0648 = waw - {0xfeef, 0xfef0, 0xfeef, 0xfef0}, // 0x0649 = alef maksura - {0xfef1, 0xfef2, 0xfef3, 0xfef4}, // 0x064a = ya - {0x065b, 0x065b, 0x065b, 0x065b}, // 0x064b = fathatan - {0x065c, 0x065c, 0x065c, 0x065c}, // 0x064c = dammatan - {0x064d, 0x064d, 0x064d, 0x064d}, // 0x064d = kasratan - {0x064e, 0x064e, 0x064e, 0x064e}, // 0x064e = fatha - {0x064f, 0x064f, 0x064f, 0x064f}, // 0x064f = damma - - {0x0650, 0x0650, 0x0650, 0x0650}, // 0x0650 = kasra - {0x0651, 0x0651, 0x0651, 0x0651}, // 0x0651 = shadda - {0x0652, 0x0652, 0x0652, 0x0652}, // 0x0652 = sukun - - {0, 0, 0, 0}, // 0x0653 - {0, 0, 0, 0}, // 0x0654 - {0, 0, 0, 0}, // 0x0655 - {0, 0, 0, 0}, // 0x0656 - {0, 0, 0, 0}, // 0x0657 - {0, 0, 0, 0}, // 0x0658 - {0, 0, 0, 0}, // 0x0659 - {0, 0, 0, 0}, // 0x065a - {0, 0, 0, 0}, // 0x065b - {0, 0, 0, 0}, // 0x065c - {0, 0, 0, 0}, // 0x065d - {0, 0, 0, 0}, // 0x065e - {0, 0, 0, 0}, // 0x065f - {0, 0, 0, 0}, // 0x0660 - {0, 0, 0, 0}, // 0x0661 - {0, 0, 0, 0}, // 0x0662 - {0, 0, 0, 0}, // 0x0663 - {0, 0, 0, 0}, // 0x0664 - {0, 0, 0, 0}, // 0x0665 - {0, 0, 0, 0}, // 0x0666 - {0, 0, 0, 0}, // 0x0667 - {0, 0, 0, 0}, // 0x0668 - {0, 0, 0, 0}, // 0x0669 - {0, 0, 0, 0}, // 0x066a - {0, 0, 0, 0}, // 0x066b - {0, 0, 0, 0}, // 0x066c - {0, 0, 0, 0}, // 0x066d - {0, 0, 0, 0}, // 0x066e - {0, 0, 0, 0}, // 0x066f - {0, 0, 0, 0}, // 0x0670 - {0, 0, 0, 0}, // 0x0671 - {0, 0, 0, 0}, // 0x0672 - {0, 0, 0, 0}, // 0x0673 - {0, 0, 0, 0}, // 0x0674 - {0, 0, 0, 0}, // 0x0675 - {0, 0, 0, 0}, // 0x0676 - {0, 0, 0, 0}, // 0x0677 - {0, 0, 0, 0}, // 0x0678 - {0, 0, 0, 0}, // 0x0679 - {0, 0, 0, 0}, // 0x067a - {0, 0, 0, 0}, // 0x067b - {0, 0, 0, 0}, // 0x067c - {0, 0, 0, 0}, // 0x067d - {0xfb56, 0xfb57, 0xfb58, 0xfb59}, // 0x067e = peh - {0, 0, 0, 0}, // 0x067f - {0, 0, 0, 0}, // 0x0680 - {0, 0, 0, 0}, // 0x0681 - {0, 0, 0, 0}, // 0x0682 - {0, 0, 0, 0}, // 0x0683 - {0, 0, 0, 0}, // 0x0684 - {0, 0, 0, 0}, // 0x0685 - {0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d}, // 0x0686 = tcheh - {0, 0, 0, 0}, // 0x0687 - {0, 0, 0, 0}, // 0x0688 - {0, 0, 0, 0}, // 0x0689 - {0, 0, 0, 0}, // 0x068a - {0, 0, 0, 0}, // 0x068b - {0, 0, 0, 0}, // 0x068c - {0, 0, 0, 0}, // 0x068d - {0, 0, 0, 0}, // 0x068e - {0, 0, 0, 0}, // 0x068f - {0, 0, 0, 0}, // 0x0690 - {0, 0, 0, 0}, // 0x0691 - {0, 0, 0, 0}, // 0x0692 - {0, 0, 0, 0}, // 0x0693 - {0, 0, 0, 0}, // 0x0694 - {0, 0, 0, 0}, // 0x0695 - {0, 0, 0, 0}, // 0x0696 - {0, 0, 0, 0}, // 0x0697 - {0xfb8a, 0xfb8b, 0xfb8a, 0xfb8b}, // 0x0698 = jeh - {0, 0, 0, 0}, // 0x0699 - {0, 0, 0, 0}, // 0x069a - {0, 0, 0, 0}, // 0x069b - {0, 0, 0, 0}, // 0x069c - {0, 0, 0, 0}, // 0x069d - {0, 0, 0, 0}, // 0x069e - {0, 0, 0, 0}, // 0x069f - {0, 0, 0, 0}, // 0x06a0 - {0, 0, 0, 0}, // 0x06a1 - {0, 0, 0, 0}, // 0x06a2 - {0, 0, 0, 0}, // 0x06a3 - {0, 0, 0, 0}, // 0x06a4 - {0, 0, 0, 0}, // 0x06a5 - {0, 0, 0, 0}, // 0x06a6 - {0, 0, 0, 0}, // 0x06a7 - {0, 0, 0, 0}, // 0x06a8 - {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 - {0, 0, 0, 0}, // 0x06b0 - {0, 0, 0, 0}, // 0x06b1 - {0, 0, 0, 0}, // 0x06b2 - {0, 0, 0, 0}, // 0x06b3 - {0, 0, 0, 0}, // 0x06b4 - {0, 0, 0, 0}, // 0x06b5 - {0, 0, 0, 0}, // 0x06b6 - {0, 0, 0, 0}, // 0x06b7 - {0, 0, 0, 0}, // 0x06b8 - {0, 0, 0, 0}, // 0x06b9 - {0, 0, 0, 0}, // 0x06ba - {0, 0, 0, 0}, // 0x06bb - {0, 0, 0, 0}, // 0x06bc - {0, 0, 0, 0}, // 0x06bd - {0, 0, 0, 0}, // 0x06be - {0, 0, 0, 0}, // 0x06bf - {0, 0, 0, 0}, // 0x06c0 - {0, 0, 0, 0}, // 0x06c1 - {0, 0, 0, 0}, // 0x06c2 - {0, 0, 0, 0}, // 0x06c3 - {0, 0, 0, 0}, // 0x06c4 - {0, 0, 0, 0}, // 0x06c5 - {0, 0, 0, 0}, // 0x06c6 - {0, 0, 0, 0}, // 0x06c7 - {0, 0, 0, 0}, // 0x06c8 - {0, 0, 0, 0}, // 0x06c9 - {0, 0, 0, 0}, // 0x06ca - {0, 0, 0, 0}, // 0x06cb - {0xfbfc, 0xfbfd, 0xfbfe, 0xfbff} // 0x06cc = farsi yeh -}; - - -char_type const arabic_start = 0x0621; -char_type const arabic_end = 0x06cc; - - -/// Information about a single UCS4 character -struct CharInfo { - /// LaTeX command (text mode) for this character - docstring textcommand; - /// LaTeX command (math mode) for this character - docstring mathcommand; - /// Needed LaTeX preamble (or feature) for text mode - string textpreamble; - /// Needed LaTeX preamble (or feature) for math mode - string mathpreamble; - /// Is this a combining character? - bool combining; - /// Is \c textpreamble a feature known by LaTeXFeatures, or a raw LaTeX - /// command? - bool textfeature; - /// Is \c mathpreamble a feature known by LaTeXFeatures, or a raw LaTeX - /// command? - bool mathfeature; - /// Always force the LaTeX command, even if the encoding contains - /// this character? - bool force; -}; - - typedef map CharInfoMap; CharInfoMap unicodesymbols; -typedef std::set CharSet; +typedef set CharSet; +typedef map CharSetMap; CharSet forced; +CharSetMap forcedselected; -typedef std::set MathAlphaSet; +typedef set MathAlphaSet; MathAlphaSet mathalpha; /// The highest code point in UCS4 encoding (1<<20 + 1<<16) char_type const max_ucs4 = 0x110000; -} // namespace anon +} // namespace EncodingException::EncodingException(char_type c) @@ -280,9 +74,21 @@ const char * EncodingException::what() const throw() } +CharInfo::CharInfo( + docstring const & textcommand, docstring const & mathcommand, + std::string const & textpreamble, std::string const & mathpreamble, + std::string const & tipashortcut, unsigned int flags) + : textcommand_(textcommand), mathcommand_(mathcommand), + textpreamble_(textpreamble), mathpreamble_(mathpreamble), + tipashortcut_(tipashortcut), flags_(flags) +{ +} + + Encoding::Encoding(string const & n, string const & l, string const & g, - string const & i, bool f, Encoding::Package p) - : name_(n), latexName_(l), guiName_(g), iconvName_(i), fixedwidth_(f), package_(p) + string const & i, bool f, bool u, Encoding::Package p) + : name_(n), latexName_(l), guiName_(g), iconvName_(i), fixedwidth_(f), + unsafe_(u), forced_(&forcedselected[n]), package_(p) { if (n == "ascii") { // ASCII can encode 128 code points and nothing else @@ -293,6 +99,7 @@ Encoding::Encoding(string const & n, string const & l, string const & g, start_encodable_ = max_ucs4; complete_ = true; } else { + start_encodable_ = 0; complete_ = false; } } @@ -300,10 +107,28 @@ Encoding::Encoding(string const & n, string const & l, string const & g, void Encoding::init() const { + // Since the the constructor is the only method which sets complete_ + // to false the test for complete_ is thread-safe without mutex. if (complete_) return; - start_encodable_ = 0; + static Mutex mutex; + Mutex::Locker lock(&mutex); + + // We need to test again for complete_, since another thread could + // have set it to true while we were waiting for the lock and we must + // not modify an encoding which is already complete. + if (complete_) + return; + + // We do not make any member mutable so that it can be easily verified + // that all const methods are thread-safe: init() is the only const + // method which changes complete_, encodable_ and start_encodable_, and + // it uses a mutex to ensure thread-safety. + CharSet & encodable = const_cast(this)->encodable_; + char_type & start_encodable = const_cast(this)->start_encodable_; + + start_encodable = 0; // temporarily switch off lyxerr, since we will generate iconv errors lyxerr.disable(); if (fixedwidth_) { @@ -316,8 +141,12 @@ 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) - encodable_.insert(uc); + 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 @@ -328,44 +157,101 @@ 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) - encodable_.insert(c); + if (it == unicodesymbols.end()) + encodable.insert(c); + else if (!it->second.force()) { + if (forced_->empty() || forced_->find(c) == forced_->end()) + encodable.insert(c); + } } } } lyxerr.enable(); - CharSet::iterator it = encodable_.find(start_encodable_); - while (it != encodable_.end()) { - encodable_.erase(it); - ++start_encodable_; - it = encodable_.find(start_encodable_); + CharSet::iterator it = encodable.find(start_encodable); + while (it != encodable.end()) { + encodable.erase(it); + ++start_encodable; + it = encodable.find(start_encodable); } - complete_ = true; + const_cast(this)->complete_ = true; } -docstring Encoding::latexChar(char_type c, bool for_mathed) 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 init(); if (iconvName_ == "UTF-8" && package_ == none) - return docstring(1, c); - if (c < start_encodable_ && !encodings.isForced(c)) - return docstring(1, c); + return true; + if (c < start_encodable_ && !isForced(c)) + return true; if (encodable_.find(c) != encodable_.end()) - return docstring(1, c); - if (for_mathed) - return docstring(); + return true; + return false; +} + + +pair Encoding::latexChar(char_type c) const +{ + if (encodable(c)) + return make_pair(docstring(1, c), false); // c cannot (or should not) be encoded in this encoding CharInfoMap::const_iterator const it = unicodesymbols.find(c); if (it == unicodesymbols.end()) throw EncodingException(c); // at least one of mathcommand and textcommand is nonempty - if (it->second.textcommand.empty()) - return "\\ensuremath{" + it->second.mathcommand + '}'; - return it->second.textcommand; + if (it->second.textcommand().empty()) + return make_pair( + "\\ensuremath{" + it->second.mathcommand() + '}', false); + return make_pair(it->second.textcommand(), !it->second.textnotermination()); +} + + +pair Encoding::latexString(docstring const & input, bool dryrun) const +{ + docstring result; + docstring uncodable; + bool terminate = false; + for (size_t n = 0; n < input.size(); ++n) { + try { + char_type const c = input[n]; + pair latex_char = latexChar(c); + docstring const latex = latex_char.first; + if (terminate && !prefixIs(latex, '\\') + && !prefixIs(latex, '{') + && !prefixIs(latex, '}')) { + // Prevent eating of a following + // space or command corruption by + // following characters + if (latex == " ") + result += "{}"; + else + result += " "; + } + result += latex; + terminate = latex_char.second; + } catch (EncodingException & /* e */) { + LYXERR0("Uncodable character in latexString!"); + if (dryrun) { + result += "<" + _("LyX Warning: ") + + _("uncodable character") + " '"; + result += docstring(1, input[n]); + result += "'>"; + } else + uncodable += input[n]; + } + } + return make_pair(result, uncodable); } @@ -374,25 +260,32 @@ vector Encoding::symbolsList() const // assure the used encoding is properly initialized init(); - // first all encodable characters - vector symbols(encodable_.begin(), encodable_.end()); - // add those below start_encodable_ + // first all those below start_encodable_ + vector symbols; for (char_type c = 0; c < start_encodable_; ++c) symbols.push_back(c); - // now the ones from the unicodesymbols file - CharInfoMap::const_iterator const end = unicodesymbols.end(); - CharInfoMap::const_iterator it = unicodesymbols.begin(); - for (; it != end; ++it) - symbols.push_back(it->first); + // add all encodable characters + copy(encodable_.begin(), encodable_.end(), back_inserter(symbols)); + // now the ones from the unicodesymbols file that are not already there + for (pair const & elem : unicodesymbols) { + if (find(symbols.begin(), symbols.end(), elem.first) == symbols.end()) + symbols.push_back(elem.first); + } + // finally, sort the vector + sort(symbols.begin(), symbols.end()); return symbols; } bool Encodings::latexMathChar(char_type c, bool mathmode, - Encoding const * encoding, docstring & command) + Encoding const * encoding, docstring & command, + bool & needsTermination) { + command = empty_docstring(); if (encoding) - command = encoding->latexChar(c, true); + if (encoding->encodable(c)) + command = docstring(1, c); + needsTermination = false; CharInfoMap::const_iterator const it = unicodesymbols.find(c); if (it == unicodesymbols.end()) { @@ -403,55 +296,96 @@ bool Encodings::latexMathChar(char_type c, bool mathmode, return false; } // at least one of mathcommand and textcommand is nonempty - bool use_math = (mathmode && !it->second.mathcommand.empty()) || - (!mathmode && it->second.textcommand.empty()); + bool use_math = (mathmode && !it->second.mathcommand().empty()) || + (!mathmode && it->second.textcommand().empty()); if (use_math) { - command = it->second.mathcommand; + command = it->second.mathcommand(); + needsTermination = !it->second.mathnotermination(); addMathCmd(c); } else { if (!encoding || command.empty()) { - command = it->second.textcommand; - addTextCmd(c); + command = it->second.textcommand(); + needsTermination = !it->second.textnotermination(); } if (mathmode) addMathSym(c); + else + addTextCmd(c); } return use_math; } -char_type Encodings::fromLaTeXCommand(docstring const & cmd, bool & combining) +char_type Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, + bool & combining, bool & needsTermination, set * req) { CharInfoMap::const_iterator const end = unicodesymbols.end(); CharInfoMap::const_iterator it = unicodesymbols.begin(); for (combining = false; it != end; ++it) { - docstring const math = it->second.mathcommand; - docstring const text = it->second.textcommand; - if (math == cmd || text == cmd) { - combining = it->second.combining; + if (it->second.deprecated()) + continue; + docstring const math = it->second.mathcommand(); + docstring const text = it->second.textcommand(); + if ((cmdtype & MATH_CMD) && math == cmd) { + combining = it->second.combining(); + needsTermination = !it->second.mathnotermination(); + if (req && it->second.mathfeature() && + !it->second.mathpreamble().empty()) + req->insert(it->second.mathpreamble()); + return it->first; + } + if ((cmdtype & TEXT_CMD) && text == cmd) { + combining = it->second.combining(); + needsTermination = !it->second.textnotermination(); + if (req && it->second.textfeature() && + !it->second.textpreamble().empty()) + req->insert(it->second.textpreamble()); return it->first; } } + needsTermination = false; return 0; } -docstring Encodings::fromLaTeXCommand(docstring const & cmd, docstring & rem) +docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype, + bool & needsTermination, docstring & rem, set * req) { + needsTermination = false; + rem = empty_docstring(); + 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; - while (k < cmdend && count && k != docstring::npos) { + while (k < cmdend && count) { k = cmd.find_first_of(from_ascii("{}"), k + 1); + // braces may not be balanced + if (k == docstring::npos) + break; if (cmd[k] == '{') ++count; else @@ -459,17 +393,31 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, docstring & rem) } 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) { - docstring const math = it->second.mathcommand; - docstring const text = it->second.textcommand; + if (it->second.deprecated()) + continue; + docstring const math = mathmode ? it->second.mathcommand() + : 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 @@ -496,6 +444,29 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, docstring & rem) // 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 = cmd.find_first_of(from_ascii("{}"), l + 1); + // braces may not be balanced + if (l == docstring::npos) + break; + 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 @@ -505,152 +476,109 @@ docstring Encodings::fromLaTeXCommand(docstring const & cmd, docstring & rem) // (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])) - || k == cmdend + 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; i = j + 1; unicmd_size = cur_size; + if (math == tmp) + needsTermination = !it->second.mathnotermination(); + else + needsTermination = !it->second.textnotermination(); + if (req) { + if (math == tmp && it->second.mathfeature() && + !it->second.mathpreamble().empty()) + req->insert(it->second.mathpreamble()); + if (text == tmp && it->second.textfeature() && + !it->second.textpreamble().empty()) + req->insert(it->second.textpreamble()); + } } } if (unicmd_size) symbols += c; - else if (j + 1 == cmdend) - // No luck. Return what remains - rem = cmd.substr(i); - } - return symbols; -} - - -void Encodings::initUnicodeMath(Buffer const & buffer, bool clear_sets) -{ -#ifdef TEX2LYX - // The code below is not needed in tex2lyx and requires additional stuff - (void)buffer; - (void)clear_sets; -#else - if (clear_sets) { - mathcmd.clear(); - textcmd.clear(); - mathsym.clear(); - } - - // Check master - Inset & inset = buffer.inset(); - InsetIterator it = inset_iterator_begin(inset); - InsetIterator const end = inset_iterator_end(inset); - for (; it != end; ++it) - it->initUnicodeMath(); - - // Check children - BufferList::iterator bit = theBufferList().begin(); - BufferList::iterator const bend = theBufferList().end(); - for (; bit != bend; ++bit) - if (buffer.isChild(*bit)) - initUnicodeMath(**bit, false); -#endif -} - - -void Encodings::validate(char_type c, LaTeXFeatures & features, bool for_mathed) -{ -#ifdef TEX2LYX - // The code below is not needed in tex2lyx and requires additional stuff - (void)c; - (void)features; - (void)for_mathed; -#else - CharInfoMap::const_iterator const it = unicodesymbols.find(c); - if (it != unicodesymbols.end()) { - // In mathed, c could be used both in textmode and mathmode - bool const use_math = (for_mathed && isMathCmd(c)) || - (!for_mathed && it->second.textcommand.empty()); - bool const use_text = (for_mathed && isTextCmd(c)) || - (!for_mathed && !it->second.textcommand.empty()); - if (use_math) { - if (!it->second.mathpreamble.empty()) { - if (it->second.mathfeature) { - string feats = it->second.mathpreamble; - while (!feats.empty()) { - string feat; - feats = split(feats, feat, ','); - features.require(feat); - } - } else - features.addPreambleSnippet(it->second.mathpreamble); + 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 (use_text) { - if (!it->second.textpreamble.empty()) { - if (it->second.textfeature) { - string feats = it->second.textpreamble; - while (!feats.empty()) { - string feat; - feats = split(feats, feat, ','); - features.require(feat); - } - } else - features.addPreambleSnippet(it->second.textpreamble); + if (j + 1 == cmdend && !unicmd_size) { + // No luck. Return what remains + rem = cmd.substr(i); + if (needsTermination && !rem.empty()) { + if (rem.substr(0, 2) == "{}") { + rem = rem.substr(2); + needsTermination = false; + } else if (rem[0] == ' ') { + needsTermination = false; + // LaTeX would swallow all spaces + rem = ltrim(rem); + } } } } - if (for_mathed && isMathSym(c)) { - features.require("amstext"); - features.require("lyxmathsym"); - } -#endif -} - - -bool Encodings::isHebrewComposeChar(char_type c) -{ - return c <= 0x05c2 && c >= 0x05b0 && c != 0x05be && c != 0x05c0; -} - - -// Special Arabic letters are ones that do not get connected from left -// they are hamza, alef_madda, alef_hamza, waw_hamza, alef_hamza_under, -// alef, tah_marbota, dal, thal, rah, zai, wow, alef_maksoura - -bool Encodings::isArabicSpecialChar(char_type c) -{ - return (c >= 0x0621 && c <= 0x0625) || (c >= 0x0630 && c <= 0x0632) - || c == 0x0627 || c == 0x0629 || c == 0x062f || c == 0x0648 - || c == 0x0649 || c == 0x0698; -} - - -bool Encodings::isArabicComposeChar(char_type c) -{ - return c >= 0x064b && c <= 0x0652; + return symbols; } -bool Encodings::isArabicChar(char_type c) +CharInfo const & Encodings::unicodeCharInfo(char_type c) { - return c >= arabic_start && c <= arabic_end - && arabic_table[c-arabic_start][0]; + static CharInfo empty; + CharInfoMap::const_iterator const it = unicodesymbols.find(c); + return it != unicodesymbols.end() ? it->second : empty; } -char_type Encodings::transformChar(char_type c, Encodings::LetterForm form) +bool Encodings::isCombiningChar(char_type c) { - return isArabicChar(c) ? arabic_table[c-arabic_start][form] : c; + CharInfoMap::const_iterator const it = unicodesymbols.find(c); + if (it != unicodesymbols.end()) + return it->second.combining(); + return false; } -bool Encodings::isCombiningChar(char_type c) +string const Encodings::TIPAShortcut(char_type c) { CharInfoMap::const_iterator const it = unicodesymbols.find(c); if (it != unicodesymbols.end()) - return it->second.combining; - return false; + return it->second.tipashortcut(); + return string(); } @@ -661,38 +589,54 @@ bool Encodings::isKnownScriptChar(char_type const c, string & preamble) if (it == unicodesymbols.end()) return false; - if (it->second.textpreamble != "textgreek" && it->second.textpreamble != "textcyr") + if (it->second.textpreamble() != "textgreek" && it->second.textpreamble() != "textcyr") return false; if (preamble.empty()) { - preamble = it->second.textpreamble; + preamble = it->second.textpreamble(); return true; } - return it->second.textpreamble == preamble; + return it->second.textpreamble() == preamble; } -bool Encodings::isForced(char_type c) +bool Encodings::isMathAlpha(char_type c) { - return (!forced.empty() && forced.find(c) != forced.end()); + return mathalpha.count(c); } -bool Encodings::isMathAlpha(char_type c) +bool Encodings::isUnicodeTextOnly(char_type c) { - return mathalpha.count(c); + if (isASCII(c) || isMathAlpha(c)) + return false; + + CharInfoMap::const_iterator const it = unicodesymbols.find(c); + return it == unicodesymbols.end() || it->second.mathcommand().empty(); } -Encoding const * Encodings::fromLyXName(string const & name) const +Encoding const * +Encodings::fromLyXName(string const & name, bool allowUnsafe) const { EncodingList::const_iterator const it = encodinglist.find(name); - return it != encodinglist.end() ? &it->second : 0; + if (it == encodinglist.end()) + return 0; + if (!allowUnsafe && it->second.unsafe()) + return 0; + return &it->second; } -Encoding const * Encodings::fromLaTeXName(string const & name) const +Encoding 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, + // we should instead extend the format of lib/encodings + if (n == "ansinew") + name = "cp1252"; + // We don't use find_if because it makes copies of the pairs in // the map. // This linear search is OK since we don't have many encodings. @@ -700,28 +644,39 @@ Encoding const * Encodings::fromLaTeXName(string const & name) 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 ((it->second.latexName() == name) && (it->second.package() & p) + && (!it->second.unsafe() || allowUnsafe)) return &it->second; return 0; } -Encodings::Encodings() +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; } +Encodings::Encodings() +{} + + 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; while (symbolslex.isOK()) { char_type symbol; - CharInfo info; - string flags; if (getNextToken) { if (!symbolslex.next(true)) @@ -739,27 +694,55 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) if (!symbolslex.next(true)) break; - info.textcommand = symbolslex.getDocString(); + docstring textcommand = symbolslex.getDocString(); if (!symbolslex.next(true)) break; - info.textpreamble = symbolslex.getString(); + string textpreamble = symbolslex.getString(); if (!symbolslex.next(true)) break; - flags = symbolslex.getString(); + string sflags = symbolslex.getString(); + + string tipashortcut; + int flags = 0; - info.combining = false; - info.textfeature = false; - info.force = false; - while (!flags.empty()) { + if (suffixIs(textcommand, '}')) + flags |= CharInfoTextNoTermination; + while (!sflags.empty()) { string flag; - flags = split(flags, flag, ','); + sflags = split(sflags, flag, ','); if (flag == "combining") { - info.combining = true; + flags |= CharInfoCombining; } else if (flag == "force") { - info.force = true; + 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); + 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); + flags |= CharInfoForceSelected; } else if (flag == "mathalpha") { mathalpha.insert(symbol); + } else if (flag == "notermination=text") { + flags |= CharInfoTextNoTermination; + } else if (flag == "notermination=math") { + flags |= CharInfoMathNoTermination; + } else if (flag == "notermination=both") { + flags |= CharInfoTextNoTermination; + flags |= CharInfoMathNoTermination; + } else if (flag == "notermination=none") { + flags &= ~CharInfoTextNoTermination; + flags &= ~CharInfoMathNoTermination; + } else if (contains(flag, "tipashortcut=")) { + tipashortcut = split(flag, '='); + } else if (flag == "deprecated") { + flags |= CharInfoDeprecated; } else { lyxerr << "Ignoring unknown flag `" << flag << "' for symbol `0x" @@ -771,19 +754,23 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) // make them optional so that old files still work. int const lineno = symbolslex.lineNumber(); bool breakout = false; + docstring mathcommand; + string mathpreamble; if (symbolslex.next(true)) { if (symbolslex.lineNumber() != lineno) { // line in old format without mathcommand and mathpreamble getNextToken = false; } else { - info.mathcommand = symbolslex.getDocString(); + mathcommand = symbolslex.getDocString(); + if (suffixIs(mathcommand, '}')) + flags |= CharInfoMathNoTermination; if (symbolslex.next(true)) { if (symbolslex.lineNumber() != lineno) { // line in new format with mathcommand only getNextToken = false; } else { // line in new format with mathcommand and mathpreamble - info.mathpreamble = symbolslex.getString(); + mathpreamble = symbolslex.getString(); } } else breakout = true; @@ -792,20 +779,33 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) breakout = true; } - if (!info.textpreamble.empty()) - info.textfeature = info.textpreamble[0] != '\\'; - if (!info.mathpreamble.empty()) - info.mathfeature = info.mathpreamble[0] != '\\'; - + // backward compatibility + if (mathpreamble == "esintoramsmath") + mathpreamble = "esint|amsmath"; + + if (!textpreamble.empty()) + if (textpreamble[0] != '\\') + flags |= CharInfoTextFeature; + if (!mathpreamble.empty()) + if (mathpreamble[0] != '\\') + flags |= CharInfoMathFeature; + + CharInfo info = CharInfo( + textcommand, mathcommand, + textpreamble, mathpreamble, + tipashortcut, flags); LYXERR(Debug::INFO, "Read unicode symbol " << symbol << " '" - << to_utf8(info.textcommand) << "' '" << info.textpreamble - << "' " << info.combining << ' ' << info.textfeature - << " '" << to_utf8(info.mathcommand) << "' '" - << info.mathpreamble << "' " << info.mathfeature); + << to_utf8(info.textcommand()) << "' '" << info.textpreamble() + << " '" << info.textfeature() << ' ' << info.textnotermination() + << ' ' << to_utf8(info.mathcommand()) << "' '" << info.mathpreamble() + << "' " << info.mathfeature() << ' ' << info.mathnotermination() + << ' ' << 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()) + if (info.isUnicodeSymbol()) { unicodesymbols[symbol] = info; + } if (breakout) break; @@ -814,7 +814,7 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) // Now read the encodings enum { et_encoding = 1, - et_end, + et_end }; LexerKeyword encodingtags[] = { @@ -840,10 +840,15 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) lex.next(); string const width = lex.getString(); bool fixedwidth = false; + bool unsafe = false; if (width == "fixed") fixedwidth = true; else if (width == "variable") fixedwidth = false; + else if (width == "variableunsafe") { + fixedwidth = false; + unsafe = true; + } else lex.printError("Unknown width"); @@ -863,7 +868,8 @@ void Encodings::read(FileName const & encfile, FileName const & symbolsfile) LYXERR(Debug::INFO, "Reading encoding " << name); encodinglist[name] = Encoding(name, latexname, - guiname, iconvname, fixedwidth, package); + guiname, iconvname, fixedwidth, unsafe, + package); if (lex.lex() != et_end) lex.printError("Missing end"); @@ -879,6 +885,15 @@ 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()); + } + } + }