X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2FParagraph.cpp;h=545ab0baee24f6796670443b2e3108989c166cf7;hb=7950ace3d96543e3099aa3bde823f970a1ce6a86;hp=be955097dce86aa5468e71463ef051e0197dfdea;hpb=361995459068a5e53b597f1195f7f29522714ec4;p=lyx.git diff --git a/src/Paragraph.cpp b/src/Paragraph.cpp index be955097dc..545ab0baee 100644 --- a/src/Paragraph.cpp +++ b/src/Paragraph.cpp @@ -25,7 +25,7 @@ #include "BufferParams.h" #include "Changes.h" #include "Counters.h" -#include "Encoding.h" +#include "BufferEncodings.h" #include "InsetList.h" #include "Language.h" #include "LaTeXFeatures.h" @@ -40,6 +40,7 @@ #include "ParagraphParameters.h" #include "SpellChecker.h" #include "sgml.h" +#include "texstream.h" #include "TextClass.h" #include "TexRow.h" #include "Text.h" @@ -51,6 +52,9 @@ #include "insets/InsetBibitem.h" #include "insets/InsetLabel.h" #include "insets/InsetSpecialChar.h" +#include "insets/InsetText.h" + +#include "mathed/InsetMathHull.h" #include "support/debug.h" #include "support/docstring_list.h" @@ -60,12 +64,22 @@ #include "support/lstrings.h" #include "support/textutils.h" +#include #include #include using namespace std; using namespace lyx::support; +// OSX clang, gcc < 4.8.0, and msvc < 2015 do not support C++11 thread_local +#if defined(__APPLE__) || (defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ < 8) +#define THREAD_LOCAL_STATIC static __thread +#elif defined(_MSC_VER) && (_MSC_VER < 1900) +#define THREAD_LOCAL_STATIC static __declspec(thread) +#else +#define THREAD_LOCAL_STATIC thread_local static +#endif + namespace lyx { namespace { @@ -73,7 +87,7 @@ namespace { /// Inset identifier (above 0x10ffff, for ucs-4) char_type const META_INSET = 0x200001; -} +} // namespace ///////////////////////////////////////////////////////////////////// @@ -96,7 +110,7 @@ public: /// void result(SpellChecker::Result r) { result_ = r; } /// - bool inside(pos_type pos) const { return range_.inside(pos); } + bool contains(pos_type pos) const { return range_.contains(pos); } /// bool covered(FontSpan const & r) const { @@ -104,8 +118,9 @@ public: // 2. last of new range inside current range or // 3. first of current range inside new range or // 4. last of current range inside new range - return range_.inside(r.first) || range_.inside(r.last) || - r.inside(range_.first) || r.inside(range_.last); + //FIXME: is this the same as !range_.intersect(r).empty() ? + return range_.contains(r.first) || range_.contains(r.last) || + r.contains(range_.first) || r.contains(range_.last); } /// void shift(pos_type pos, int offset) @@ -191,7 +206,7 @@ public: RangesIterator et = ranges_.end(); RangesIterator it = ranges_.begin(); for (; it != et; ++it) { - if(it->inside(pos)) { + if(it->contains(pos)) { return it->result(); } } @@ -205,7 +220,7 @@ public: RangesIterator et = ranges_.end(); RangesIterator it = ranges_.begin(); for (; it != et; ++it) { - if(it->inside(pos)) { + if(it->contains(pos)) { return it->range(); } } @@ -281,6 +296,11 @@ private: class Paragraph::Private { + // Enforce our own "copy" constructor + Private(Private const &) = delete; + Private & operator=(Private const &) = delete; + // Unique ID generator + static int make_id(); public: /// Private(Paragraph * owner, Layout const & layout); @@ -353,24 +373,17 @@ public: pos_type i, unsigned int & column); /// - bool latexSpecialT3( + bool latexSpecialTU( char_type const c, otexstream & os, pos_type i, unsigned int & column); /// - bool latexSpecialTypewriter( + bool latexSpecialT3( char_type const c, otexstream & os, pos_type i, unsigned int & column); - /// - bool latexSpecialPhrase( - otexstream & os, - pos_type & i, - pos_type end_pos, - unsigned int & column, - OutputParams const & runparams); /// void validate(LaTeXFeatures & features) const; @@ -379,15 +392,12 @@ public: bool onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const; - /// match a string against a particular point in the paragraph - bool isTextAt(string const & str, pos_type pos) const; - /// a vector of speller skip positions typedef vector SkipPositions; typedef SkipPositions::const_iterator SkipPositionsIterator; void appendSkipPosition(SkipPositions & skips, pos_type const pos) const; - + Language * getSpellLanguage(pos_type const from) const; Language * locateSpellRange(pos_type & from, pos_type & to, @@ -402,7 +412,7 @@ public: } bool ignoreWord(docstring const & word) const ; - + void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state) { pos_type textsize = owner_->size(); @@ -499,7 +509,7 @@ public: TextContainer text_; typedef set Words; - typedef map LangWordsMap; + typedef map LangWordsMap; /// LangWordsMap words_; /// @@ -509,26 +519,6 @@ public: }; -namespace { - -struct special_phrase { - string phrase; - docstring macro; - bool builtin; -}; - -special_phrase const special_phrases[] = { - { "LyX", from_ascii("\\LyX{}"), false }, - { "TeX", from_ascii("\\TeX{}"), true }, - { "LaTeX2e", from_ascii("\\LaTeXe{}"), true }, - { "LaTeX", from_ascii("\\LaTeX{}"), true }, -}; - -size_t const phrases_nr = sizeof(special_phrases)/sizeof(special_phrase); - -} // namespace anon - - Paragraph::Private::Private(Paragraph * owner, Layout const & layout) : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout) { @@ -536,33 +526,37 @@ Paragraph::Private::Private(Paragraph * owner, Layout const & layout) } -// Initialization of the counter for the paragraph id's, -// -// FIXME: There should be a more intelligent way to generate and use the -// paragraph ids per buffer instead a global static counter for all InsetText -// in the running program. -static int paragraph_id = -1; +//static +int Paragraph::Private::make_id() +{ + // The id is unique per session across buffers because it is used in + // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the + // outliner. + // (thread-safe) + static atomic_uint next_id(0); + return next_id++; +} + Paragraph::Private::Private(Private const & p, Paragraph * owner) : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_), + id_(make_id()), params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_), begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_), layout_(p.layout_) { - id_ = ++paragraph_id; requestSpellCheck(p.text_.size()); } Paragraph::Private::Private(Private const & p, Paragraph * owner, pos_type beg, pos_type end) - : owner_(owner), inset_owner_(p.inset_owner_), + : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()), params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_, beg, end), begin_of_body_(p.begin_of_body_), words_(p.words_), layout_(p.layout_) { - id_ = ++paragraph_id; if (beg >= pos_type(p.text_.size())) return; text_ = p.text_.substr(beg, end - beg); @@ -584,10 +578,22 @@ Paragraph::Private::Private(Private const & p, Paragraph * owner, } -void Paragraph::addChangesToToc(DocIterator const & cdit, - Buffer const & buf, bool output_active) const +void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf, + bool output_active, TocBackend & backend) const +{ + d->changes_.addToToc(cdit, buf, output_active, backend); +} + + +void Paragraph::addChangesToBuffer(Buffer const & buf) const { - d->changes_.addToToc(cdit, buf, output_active); + d->changes_.updateBuffer(buf); +} + + +bool Paragraph::isChangeUpdateRequired() const +{ + return d->changes_.isUpdateRequired(); } @@ -615,7 +621,7 @@ bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const if (!trackChanges) return true; - Change const change = d->changes_.lookup(size()); + Change const & change = d->changes_.lookup(size()); return change.inserted() && change.currentAuthor(); } @@ -767,7 +773,7 @@ void Paragraph::Private::insertChar(pos_type pos, char_type c, bool Paragraph::insertInset(pos_type pos, Inset * inset, - Font const & font, Change const & change) + Font const & font, Change const & change) { LASSERT(inset, return false); LASSERT(pos >= 0 && pos <= size(), return false); @@ -785,7 +791,6 @@ bool Paragraph::insertInset(pos_type pos, Inset * inset, // Some insets require run of spell checker requestSpellCheck(pos); - setFont(pos, font); return true; } @@ -947,10 +952,13 @@ int Paragraph::Private::writeScriptChars(otexstream & os, { // FIXME: modifying i here is not very nice... - // We only arrive here when a proper language for character text_[i] has - // not been specified (i.e., it could not be translated in the current - // latex encoding) or its latex translation has been forced, and it - // belongs to a known script. + // We only arrive here when character text_[i] could not be translated + // into the current latex encoding (or its latex translation has been forced,) + // and it belongs to a known script. + // TODO: We need \textcyr and \textgreek wrappers also for characters + // that can be encoded in the "LaTeX encoding" but not in the + // current *font encoding*. + // (See #9681 for details and test) // Parameter ltx contains the latex translation of text_[i] as specified // in the unicodesymbols file and is something like "\textXXX{}". // The latex macro name "textXXX" specifies the script to which text_[i] @@ -966,6 +974,10 @@ int Paragraph::Private::writeScriptChars(otexstream & os, bool closing_brace = true; if (script == "textgreek" && encoding.latexName() == "iso-8859-7") { // Correct encoding is being used, so we can avoid \textgreek. + // TODO: wrong test: we need to check the *font encoding* + // (i.e. the active language and its FontEncoding tag) + // instead of the LaTeX *input encoding*! + // See #9637 for details and test-cases. pos = brace1 + 1; length -= pos; closing_brace = false; @@ -1012,26 +1024,6 @@ int Paragraph::Private::writeScriptChars(otexstream & os, } -bool Paragraph::Private::isTextAt(string const & str, pos_type pos) const -{ - pos_type const len = str.length(); - - // is the paragraph large enough? - if (pos + len > int(text_.size())) - return false; - - // does the wanted text start at point? - for (string::size_type i = 0; i < str.length(); ++i) { - // Caution: direct comparison of characters works only - // because str is pure ASCII. - if (str[i] != text_[pos + i]) - return false; - } - - return fontlist_.hasChangeInRange(pos, len); -} - - void Paragraph::Private::latexInset(BufferParams const & bparams, otexstream & os, OutputParams & runparams, @@ -1055,16 +1047,17 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, } // FIXME: move this to InsetNewline::latex - if (inset->lyxCode() == NEWLINE_CODE) { + if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) { // newlines are handled differently here than // the default in simpleTeXSpecialChars(). if (!style.newline_allowed) { os << '\n'; } else { if (open_font) { + bool needPar = false; column += running_font.latexWriteEndChanges( os, bparams, runparams, - basefont, basefont); + basefont, basefont, needPar); open_font = false; } @@ -1097,13 +1090,15 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, odocstream::pos_type const len = os.os().tellp(); if (inset->forceLTR() - && !runparams.use_polyglossia && running_font.isRightToLeft() // ERT is an exception, it should be output with no // decorations at all && inset->lyxCode() != ERT_CODE) { - if (running_font.language()->lang() == "farsi") - os << "\\beginL{}"; + if (runparams.use_polyglossia) { + os << "\\LRE{"; + } else if (running_font.language()->lang() == "farsi" + || running_font.language()->lang() == "arabic_arabi") + os << "\\textLR{" << termcmd; else os << "\\L{"; close = true; @@ -1122,10 +1117,12 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, bool arabtex = basefont.language()->lang() == "arabic_arabtex" || running_font.language()->lang() == "arabic_arabtex"; if (open_font && !inset->inheritFont()) { + bool needPar = false; bool closeLanguage = arabtex || basefont.isRightToLeft() == running_font.isRightToLeft(); unsigned int count = running_font.latexWriteEndChanges(os, - bparams, runparams, basefont, basefont, closeLanguage); + bparams, runparams, basefont, basefont, + needPar, closeLanguage); column += count; // if any font properties were closed, update the running_font, // making sure, however, to leave the language as it was @@ -1144,7 +1141,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, } } - int prev_rows = os.texrow().rows(); + size_t const previous_row_count = os.texrow().rows(); try { runparams.lastid = id_; @@ -1157,14 +1154,10 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, throw(e); } - if (close) { - if (running_font.language()->lang() == "farsi") - os << "\\endL{}"; - else - os << '}'; - } + if (close) + os << '}'; - if (os.texrow().rows() > prev_rows) { + if (os.texrow().rows() > previous_row_count) { os.texrow().start(owner_->id(), i + 1); column = 0; } else { @@ -1186,12 +1179,11 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, pos_type end_pos, unsigned int & column) { - // With polyglossia, brackets and stuff need not be reversed - // in RTL scripts (see bug #8251) - char_type const c = (runparams.use_polyglossia) ? - owner_->getUChar(bparams, i) : text_[i]; + char_type const c = owner_->getUChar(bparams, runparams, i); - if (style.pass_thru || runparams.pass_thru) { + if (style.pass_thru || runparams.pass_thru + || contains(style.pass_thru_chars, c) + || contains(runparams.pass_thru_chars, c)) { if (c != '\0') { Encoding const * const enc = runparams.encoding; if (enc && !enc->encodable(c)) @@ -1206,42 +1198,52 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, return; // If T1 font encoding is used, use the special // characters it provides. - // NOTE: some languages reset the font encoding - // internally + // NOTE: Some languages reset the font encoding internally to a + // non-standard font encoding. If we are using such a language, + // we do not output special T1 chars. if (!runparams.inIPA && !running_font.language()->internalFontEncoding() - && lyxrc.fontenc == "T1" && latexSpecialT1(c, os, i, column)) - return; - - // \tt font needs special treatment - if (!runparams.inIPA - && running_font.fontInfo().family() == TYPEWRITER_FAMILY - && latexSpecialTypewriter(c, os, i, column)) + && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1" + && latexSpecialT1(c, os, i, column)) return; + // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding + else if (!runparams.inIPA && !running_font.language()->internalFontEncoding() + && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column)) + return; // Otherwise, we use what LaTeX provides us. switch (c) { case '\\': - os << "\\textbackslash{}"; + os << "\\textbackslash" << termcmd; column += 15; break; case '<': - os << "\\textless{}"; + os << "\\textless" << termcmd; column += 10; break; case '>': - os << "\\textgreater{}"; + os << "\\textgreater" << termcmd; column += 13; break; case '|': - os << "\\textbar{}"; + os << "\\textbar" << termcmd; column += 9; break; case '-': os << '-'; + if (i + 1 < static_cast(text_.size()) && + (end_pos == -1 || i + 1 < end_pos) && + text_[i+1] == '-') { + // Prevent "--" becoming an endash and "---" becoming + // an emdash. + // Within \ttfamily, "--" is merged to "-" (no endash) + // so we avoid this rather irritating ligature as well + os << "{}"; + column += 2; + } break; case '\"': - os << "\\char`\\\"{}"; - column += 9; + os << "\\textquotedbl" << termcmd; + column += 14; break; case '$': case '&': @@ -1253,12 +1255,12 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, break; case '~': - os << "\\textasciitilde{}"; + os << "\\textasciitilde" << termcmd; column += 16; break; case '^': - os << "\\textasciicircum{}"; + os << "\\textasciicircum" << termcmd; column += 17; break; @@ -1280,11 +1282,25 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, // written. (Asger) break; + case 0x2013: + case 0x2014: + // XeTeX's dash behaviour is determined via a global setting + if (bparams.use_dash_ligatures + && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY + && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) { + if (c == 0x2013) { + // en-dash + os << "--"; + column +=2; + } else { + // em-dash + os << "---"; + column +=3; + } + break; + } + // fall through default: - // LyX, LaTeX etc. - if (latexSpecialPhrase(os, i, end_pos, column, runparams)) - return; - if (c == '\0') return; @@ -1303,7 +1319,7 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, docstring nextlatex; bool nexttipas = false; string nexttipashortcut; - if (next != '\0' && next != META_INSET) { + if (next != '\0' && next != META_INSET && encoding.encodable(next)) { nextlatex = encoding.latexChar(next).first; if (runparams.inIPA) { nexttipashortcut = Encodings::TIPAShortcut(next); @@ -1360,7 +1376,7 @@ bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os, // but we should avoid ligatures if (i + 1 >= int(text_.size()) || text_[i + 1] != c) return true; - os << "\\textcompwordmark{}"; + os << "\\textcompwordmark" << termcmd; column += 19; return true; case '|': @@ -1368,7 +1384,7 @@ bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os, return true; case '\"': // soul.sty breaks with \char`\" - os << "\\textquotedbl{}"; + os << "\\textquotedbl" << termcmd; column += 14; return true; default: @@ -1377,6 +1393,14 @@ bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os, } +bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os, + pos_type i, unsigned int & column) +{ + // TU encoding is currently on par with T1. + return latexSpecialT1(c, os, i, column); +} + + bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os, pos_type /*i*/, unsigned int & column) { @@ -1388,105 +1412,38 @@ bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os, os.put(c); return true; case '|': - os << "\\textvertline{}"; + os << "\\textvertline" << termcmd; column += 14; return true; - default: - return false; - } -} - - -bool Paragraph::Private::latexSpecialTypewriter(char_type const c, otexstream & os, - pos_type i, unsigned int & column) -{ - switch (c) { - case '-': - // within \ttfamily, "--" is merged to "-" (no endash) - // so we avoid this rather irritating ligature - if (i + 1 < int(text_.size()) && text_[i + 1] == '-') { - os << "-{}"; - column += 2; - } else - os << '-'; + case 0x2013: + os << "\\textendash" << termcmd; + column += 12; + return true; + case 0x2014: + os << "\\textemdash" << termcmd; + column += 12; return true; - - // everything else has to be checked separately - // (depending on the encoding) default: return false; } } -/// \param end_pos -/// If [start_pos, end_pos) does not include entirely the special phrase, then -/// do not apply the macro transformation. -bool Paragraph::Private::latexSpecialPhrase(otexstream & os, pos_type & i, pos_type end_pos, - unsigned int & column, OutputParams const & runparams) -{ - // FIXME: if we have "LaTeX" with a font - // change in the middle (before the 'T', then - // the "TeX" part is still special cased. - // Really we should only operate this on - // "words" for some definition of word - - for (size_t pnr = 0; pnr < phrases_nr; ++pnr) { - if (!isTextAt(special_phrases[pnr].phrase, i) - || (end_pos != -1 && i + int(special_phrases[pnr].phrase.size()) > end_pos)) - continue; - if (runparams.moving_arg) - os << "\\protect"; - os << special_phrases[pnr].macro; - i += special_phrases[pnr].phrase.length() - 1; - column += special_phrases[pnr].macro.length() - 1; - return true; - } - return false; -} - - void Paragraph::Private::validate(LaTeXFeatures & features) const { if (layout_->inpreamble && inset_owner_) { - bool const is_command = layout_->latextype == LATEX_COMMAND; - Buffer const & buf = inset_owner_->buffer(); - BufferParams const & bp = buf.params(); - Font f; - TexRow texrow; - // Using a string stream here circumvents the encoding + // FIXME: Using a string stream here circumvents the encoding // switching machinery of odocstream. Therefore the // output is wrong if this paragraph contains content // that needs to switch encoding. - odocstringstream ods; - otexstream os(ods, texrow); - if (is_command) { - os << '\\' << from_ascii(layout_->latexname()); - // we have to provide all the optional arguments here, even though - // the last one is the only one we care about. - // Separate handling of optional argument inset. - if (!layout_->latexargs().empty()) { - OutputParams rp = features.runparams(); - rp.local_font = &owner_->getFirstFontSettings(bp); - latexArgInsets(*owner_, os, rp, layout_->latexargs()); - } - os << from_ascii(layout_->latexparam()); - } - docstring::size_type const length = ods.str().length(); - // this will output "{" at the beginning, but not at the end - owner_->latex(bp, f, os, features.runparams(), 0, -1, true); - if (ods.str().length() > length) { - if (is_command) { - ods << '}'; - if (!layout_->postcommandargs().empty()) { - OutputParams rp = features.runparams(); - rp.local_font = &owner_->getFirstFontSettings(bp); - latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:"); - } - } - string const snippet = to_utf8(ods.str()); - features.addPreambleSnippet(snippet); - } + Buffer const & buf = inset_owner_->buffer(); + otexstringstream os; + os << layout_->preamble(); + size_t const length = os.length(); + TeXOnePar(buf, buf.text(), buf.getParFromID(owner_->id()).pit(), os, + features.runparams(), string(), 0, -1, true); + if (os.length() > length) + features.addPreambleSnippet(os.release(), true); } if (features.runparams().flavor == OutputParams::HTML @@ -1513,7 +1470,19 @@ void Paragraph::Private::validate(LaTeXFeatures & features) const InsetList::const_iterator iend = insetlist_.end(); for (; icit != iend; ++icit) { if (icit->inset) { + features.inDeletedInset(owner_->isDeleted(icit->pos)); + if (icit->inset->lyxCode() == FOOT_CODE) { + // FIXME: an item inset would make things much easier. + if ((layout_->latextype == LATEX_LIST_ENVIRONMENT + || (layout_->latextype == LATEX_ITEM_ENVIRONMENT + && layout_->margintype == MARGIN_FIRST_DYNAMIC)) + && (icit->pos < begin_of_body_ + || (icit->pos == begin_of_body_ + && (icit->pos == 0 || text_[icit->pos - 1] != ' ')))) + features.saveNoteEnv("description"); + } icit->inset->validate(features); + features.inDeletedInset(false); if (layout_->needprotect && icit->inset->lyxCode() == FOOT_CODE) features.require("NeedLyXFootnoteCode"); @@ -1521,15 +1490,24 @@ void Paragraph::Private::validate(LaTeXFeatures & features) const } // then the contents + BufferParams const bp = features.runparams().is_child + ? features.buffer().masterParams() : features.buffer().params(); for (pos_type i = 0; i < int(text_.size()) ; ++i) { - for (size_t pnr = 0; pnr < phrases_nr; ++pnr) { - if (!special_phrases[pnr].builtin - && isTextAt(special_phrases[pnr].phrase, i)) { - features.require(special_phrases[pnr].phrase); - break; - } + char_type c = text_[i]; + if (c == 0x0022) { + if (features.runparams().isFullUnicode() && bp.useNonTeXFonts) + features.require("textquotedblp"); + else if (bp.main_font_encoding() != "T1" + || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding())) + features.require("textquotedbl"); } - Encodings::validate(text_[i], features); + if (!bp.use_dash_ligatures + && (c == 0x2013 || c == 0x2014) + && bp.useNonTeXFonts + && features.runparams().flavor == OutputParams::XETEX) + // XeTeX's dash behaviour is determined via a global setting + features.require("xetexdashbreakstate"); + BufferEncodings::validate(c, features); } } @@ -1598,7 +1576,7 @@ void flushString(ostream & os, docstring & s) s.erase(); } -} +} // namespace void Paragraph::write(ostream & os, BufferParams const & bparams, @@ -1635,7 +1613,7 @@ void Paragraph::write(ostream & os, BufferParams const & bparams, int column = 0; for (pos_type i = 0; i <= size(); ++i) { - Change const change = lookupChange(i); + Change const & change = lookupChange(i); if (change != running_change) flushString(os, write_buffer); Changes::lyxMarkChange(os, bparams, column, running_change, change); @@ -1671,6 +1649,9 @@ void Paragraph::write(ostream & os, BufferParams const & bparams, os << "\n\\end_inset\n\n"; column = 0; } + // FIXME This can be removed again once the mystery + // crash has been resolved. + os << flush; } break; case '\\': @@ -1706,6 +1687,9 @@ void Paragraph::write(ostream & os, BufferParams const & bparams, flushString(os, write_buffer); os << "\n\\end_layout\n"; + // FIXME This can be removed again once the mystery + // crash has been resolved. + os << flush; } @@ -1784,14 +1768,6 @@ void Paragraph::insertChar(pos_type pos, char_type c, } -bool Paragraph::insertInset(pos_type pos, Inset * inset, - Change const & change) -{ - Font no_font; - return insertInset(pos, inset, no_font, change); -} - - void Paragraph::resetFonts(Font const & font) { d->fontlist_.clear(); @@ -1830,7 +1806,10 @@ Font const & Paragraph::getFontSettings(BufferParams const & bparams, FontSpan Paragraph::fontSpan(pos_type pos) const { - LBUFERR(pos < size()); + LBUFERR(pos <= size()); + + if (pos == size()) + return FontSpan(pos, pos); pos_type start = 0; FontList::const_iterator cit = d->fontlist_.begin(); @@ -1919,40 +1898,60 @@ Font const Paragraph::getLayoutFont } -/// Returns the height of the highest font in range -FontSize Paragraph::highestFontInRange - (pos_type startpos, pos_type endpos, FontSize def_size) const -{ - return d->fontlist_.highestInRange(startpos, endpos, def_size); -} - - -char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const +char_type Paragraph::getUChar(BufferParams const & bparams, + OutputParams const & rp, + pos_type pos) const { char_type c = d->text_[pos]; - if (!lyxrc.rtl_support || !getFontSettings(bparams, pos).isRightToLeft()) + + // Return unchanged character in LTR languages. + if (!getFontSettings(bparams, pos).isRightToLeft()) return c; - // FIXME: The arabic special casing is due to the difference of arabic - // round brackets input introduced in r18599. Check if this should be - // unified with Hebrew or at least if all bracket types should be - // handled the same (file format change in either case). + // FIXME This is a complete mess due to all the language-specific + // special cases. We need to unify this eventually, but this + // requires a file format change and some thought. + // We also need to unify the input of parentheses in different RTL + // languages. Currently, some have their own methods (Arabic: + // 18599/lyxsvn, Hebrew: e5f42f67d/lyxgit), some don't (Urdu, Syriac). + // Also note that the representation in the LyX file is probably wrong + // (see FIXME in TextMetrics::breakRow). + // Most likely, we should simply rely on Qt's unicode handling here. string const & lang = getFontSettings(bparams, pos).language()->lang(); - bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi" - || lang == "farsi"; + + // With polyglossia, brackets and stuff need not be reversed in RTL scripts + // FIXME: The special casing for Hebrew parens is due to the special + // handling on input (for Hebrew in e5f42f67d/lyxgit); see #8251. char_type uc = c; + if (rp.use_polyglossia) { + switch (c) { + case '(': + if (lang == "hebrew") + uc = ')'; + break; + case ')': + if (lang == "hebrew") + uc = '('; + break; + } + return uc; + } + + // In the following languages, brackets don't need to be reversed. + // Furthermore, in arabic_arabi, they are transformed to Arabic + // Ornate Parentheses (dunno if this is really wanted) + bool const reversebrackets = lang != "arabic_arabtex" + && lang != "arabic_arabi" + && lang != "farsi"; + switch (c) { - case '(': - uc = arabic ? c : ')'; - break; - case ')': - uc = arabic ? c : '('; - break; case '[': - uc = ']'; + if (reversebrackets) + uc = ']'; break; case ']': - uc = '['; + if (reversebrackets) + uc = '['; break; case '{': uc = '}'; @@ -2032,7 +2031,7 @@ depth_type Paragraph::getMaxDepthAfter() const } -char Paragraph::getAlign() const +LyXAlignment Paragraph::getAlign() const { if (d->params_.align() == LYX_ALIGN_LAYOUT) return d->layout_->align; @@ -2098,7 +2097,7 @@ docstring Paragraph::expandParagraphLabel(Layout const & layout, docstring parent(fmt, i + 1, j - i - 1); docstring label = from_ascii("??"); if (tclass.hasLayout(parent)) - docstring label = expandParagraphLabel(tclass[parent], bparams, + label = expandParagraphLabel(tclass[parent], bparams, process_appendix); fmt = docstring(fmt, 0, i) + label + docstring(fmt, j + 1, docstring::npos); @@ -2140,17 +2139,15 @@ void Paragraph::setBeginOfBody() // remove unnecessary getChar() calls pos_type i = 0; pos_type end = size(); - if (i < end && !isNewline(i)) { + if (i < end && !(isNewline(i) || isEnvSeparator(i))) { ++i; - char_type previous_char = 0; - char_type temp = 0; if (i < end) { - previous_char = d->text_[i]; - if (!isNewline(i)) { + char_type previous_char = d->text_[i]; + if (!(isNewline(i) || isEnvSeparator(i))) { ++i; while (i < end && previous_char != ' ') { - temp = d->text_[i]; - if (isNewline(i)) + char_type temp = d->text_[i]; + if (isNewline(i) || isEnvSeparator(i)) break; ++i; previous_char = temp; @@ -2205,39 +2202,40 @@ string correction(string const & orig) } -string const corrected_env(string const & suffix, string const & env, - InsetCode code, bool const lastpar) +bool corrected_env(otexstream & os, string const & suffix, string const & env, + InsetCode code, bool const lastpar, int & col) { - string output = suffix + "{"; + string macro = suffix + "{"; if (noTrivlistCentering(code)) { if (lastpar) { // the last paragraph in non-trivlist-aligned // context is special (to avoid unwanted whitespace) - if (suffix == "\\begin") - return "\\" + correction(env) + "{}"; - return string(); + if (suffix == "\\begin") { + macro = "\\" + correction(env) + "{}"; + os << from_ascii(macro); + col += macro.size(); + return true; + } + return false; } - output += correction(env); + macro += correction(env); } else - output += env; - output += "}"; - if (suffix == "\\begin") - output += "\n"; - return output; -} - - -void adjust_column(string const & str, int & column) -{ - if (!contains(str, "\n")) - column += str.size(); - else { - string tmp; - column = rsplit(str, tmp, '\n').size(); + macro += env; + macro += "}"; + if (suffix == "\\par\\end") { + os << breakln; + col = 0; } + os << from_ascii(macro); + col += macro.size(); + if (suffix == "\\begin") { + os << breakln; + col = 0; + } + return true; } -} // namespace anon +} // namespace int Paragraph::Private::startTeXParParams(BufferParams const & bparams, @@ -2245,7 +2243,12 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams, { int column = 0; - if (params_.noindent() && !layout_->pass_thru) { + bool canindent = + (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ? + (layout_->toggle_indent != ITOGGLE_NEVER) : + (layout_->toggle_indent == ITOGGLE_ALWAYS); + + if (canindent && params_.noindent() && !layout_->pass_thru) { os << "\\noindent "; column += 10; } @@ -2284,28 +2287,19 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams, case LYX_ALIGN_DECIMAL: break; case LYX_ALIGN_LEFT: { - string output; - if (owner_->getParLanguage(bparams)->babel() != "hebrew") - output = corrected_env(begin_tag, "flushleft", code, lastpar); + if (!owner_->getParLanguage(bparams)->rightToLeft()) + corrected_env(os, begin_tag, "flushleft", code, lastpar, column); else - output = corrected_env(begin_tag, "flushright", code, lastpar); - os << from_ascii(output); - adjust_column(output, column); + corrected_env(os, begin_tag, "flushright", code, lastpar, column); break; } case LYX_ALIGN_RIGHT: { - string output; - if (owner_->getParLanguage(bparams)->babel() != "hebrew") - output = corrected_env(begin_tag, "flushright", code, lastpar); + if (!owner_->getParLanguage(bparams)->rightToLeft()) + corrected_env(os, begin_tag, "flushright", code, lastpar, column); else - output = corrected_env(begin_tag, "flushleft", code, lastpar); - os << from_ascii(output); - adjust_column(output, column); + corrected_env(os, begin_tag, "flushleft", code, lastpar, column); break; } case LYX_ALIGN_CENTER: { - string output; - output = corrected_env(begin_tag, "center", code, lastpar); - os << from_ascii(output); - adjust_column(output, column); + corrected_env(os, begin_tag, "center", code, lastpar, column); break; } } @@ -2337,8 +2331,9 @@ bool Paragraph::Private::endTeXParParams(BufferParams const & bparams, break; } - string output; - string const end_tag = "\n\\par\\end"; + bool output = false; + int col = 0; + string const end_tag = "\\par\\end"; InsetCode code = ownerCode(); bool const lastpar = runparams.isLastPar; @@ -2350,27 +2345,24 @@ bool Paragraph::Private::endTeXParParams(BufferParams const & bparams, case LYX_ALIGN_DECIMAL: break; case LYX_ALIGN_LEFT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") - output = corrected_env(end_tag, "flushleft", code, lastpar); + if (!owner_->getParLanguage(bparams)->rightToLeft()) + output = corrected_env(os, end_tag, "flushleft", code, lastpar, col); else - output = corrected_env(end_tag, "flushright", code, lastpar); - os << from_ascii(output); + output = corrected_env(os, end_tag, "flushright", code, lastpar, col); break; } case LYX_ALIGN_RIGHT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") - output = corrected_env(end_tag, "flushright", code, lastpar); + if (!owner_->getParLanguage(bparams)->rightToLeft()) + output = corrected_env(os, end_tag, "flushright", code, lastpar, col); else - output = corrected_env(end_tag, "flushleft", code, lastpar); - os << from_ascii(output); + output = corrected_env(os, end_tag, "flushleft", code, lastpar, col); break; } case LYX_ALIGN_CENTER: { - output = corrected_env(end_tag, "center", code, lastpar); - os << from_ascii(output); + corrected_env(os, end_tag, "center", code, lastpar, col); break; } } - return !output.empty() || lastpar; + return output || lastpar; } @@ -2432,7 +2424,9 @@ void Paragraph::latex(BufferParams const & bparams, // if the paragraph is empty, the loop will not be entered at all if (empty()) { - if (style.isCommand()) { + // For InTitle commands, we have already opened a group + // in output_latex::TeXOnePar. + if (style.isCommand() && !style.intitle) { os << '{'; ++column; } @@ -2444,14 +2438,19 @@ void Paragraph::latex(BufferParams const & bparams, column += d->startTeXParParams(bparams, os, runparams); } + // Whether a \par can be issued for insets typeset inline with text. + // Yes if greater than 0. This has to be static. + THREAD_LOCAL_STATIC int parInline = 0; + for (pos_type i = 0; i < size(); ++i) { // First char in paragraph or after label? if (i == body_pos) { if (body_pos > 0) { if (open_font) { + bool needPar = false; column += running_font.latexWriteEndChanges( os, bparams, runparams, - basefont, basefont); + basefont, basefont, needPar); open_font = false; } basefont = getLayoutFont(bparams, outerfont); @@ -2465,7 +2464,9 @@ void Paragraph::latex(BufferParams const & bparams, os << "}] "; column +=3; } - if (style.isCommand()) { + // For InTitle commands, we have already opened a group + // in output_latex::TeXOnePar. + if (style.isCommand() && !style.intitle) { os << '{'; ++column; } @@ -2480,18 +2481,57 @@ void Paragraph::latex(BufferParams const & bparams, runparams); } + runparams.wasDisplayMath = runparams.inDisplayMath; + runparams.inDisplayMath = false; + bool deleted_display_math = false; Change const & change = runparams.inDeletedInset ? runparams.changeOfDeletedInset : lookupChange(i); - if (bparams.outputChanges && runningChange != change) { + // Check whether a display math inset follows + if (d->text_[i] == META_INSET + && i >= start_pos && (end_pos == -1 || i < end_pos)) { + InsetMath const * im = getInset(i)->asInsetMath(); + if (im && im->asHullInset() + && im->asHullInset()->outerDisplay()) { + runparams.inDisplayMath = true; + // runparams.inDeletedInset will be set by + // latexInset later, but we need this info + // before it is called. On the other hand, we + // cannot set it here because it is a counter. + deleted_display_math = isDeleted(i); + } + if (bparams.output_changes && deleted_display_math + && runningChange == change + && change.type == Change::DELETED + && !os.afterParbreak()) { + // A display math in the same paragraph follows. + // We have to close and then reopen \lyxdeleted, + // otherwise the math will be shifted up. + OutputParams rp = runparams; + if (open_font) { + bool needPar = false; + column += running_font.latexWriteEndChanges( + os, bparams, rp, basefont, + basefont, needPar); + open_font = false; + } + basefont = getLayoutFont(bparams, outerfont); + running_font = basefont; + column += Changes::latexMarkChange(os, bparams, + Change(Change::INSERTED), change, rp); + } + } + + if (bparams.output_changes && runningChange != change) { if (open_font) { + bool needPar = false; column += running_font.latexWriteEndChanges( - os, bparams, runparams, basefont, basefont); + os, bparams, runparams, + basefont, basefont, needPar); open_font = false; } basefont = getLayoutFont(bparams, outerfont); running_font = basefont; - column += Changes::latexMarkChange(os, bparams, runningChange, change, runparams); runningChange = change; @@ -2499,25 +2539,27 @@ void Paragraph::latex(BufferParams const & bparams, // do not output text which is marked deleted // if change tracking output is disabled - if (!bparams.outputChanges && change.deleted()) { + if (!bparams.output_changes && change.deleted()) { continue; } ++column; // Fully instantiated font - Font const font = getFont(bparams, i, outerfont); + Font const current_font = getFont(bparams, i, outerfont); Font const last_font = running_font; // Do we need to close the previous font? if (open_font && - (font != running_font || - font.language() != running_font.language())) + (current_font != running_font || + current_font.language() != running_font.language())) { + bool needPar = false; column += running_font.latexWriteEndChanges( - os, bparams, runparams, basefont, - (i == body_pos-1) ? basefont : font); + os, bparams, runparams, basefont, + (i == body_pos-1) ? basefont : current_font, + needPar); running_font = basefont; open_font = false; } @@ -2527,40 +2569,67 @@ void Paragraph::latex(BufferParams const & bparams, // close babel's font environment before opening CJK. string const lang_end_command = runparams.use_polyglossia ? "\\end{$$lang}" : lyxrc.language_command_end; + bool const using_begin_end = runparams.use_polyglossia || + !lang_end_command.empty(); if (!running_lang.empty() && - font.language()->encoding()->package() == Encoding::CJK) { + (!using_begin_end || running_lang == openLanguageName()) && + current_font.language()->encoding()->package() == Encoding::CJK) { string end_tag = subst(lang_end_command, "$$lang", running_lang); os << from_ascii(end_tag); column += end_tag.length(); + if (using_begin_end) + popLanguageName(); } // Switch file encoding if necessary (and allowed) if (!runparams.pass_thru && !style.pass_thru && runparams.encoding->package() != Encoding::none && - font.language()->encoding()->package() != Encoding::none) { + current_font.language()->encoding()->package() != Encoding::none) { pair const enc_switch = switchEncoding(os.os(), bparams, runparams, - *(font.language()->encoding())); + *(current_font.language()->encoding())); if (enc_switch.first) { column += enc_switch.second; - runparams.encoding = font.language()->encoding(); + runparams.encoding = current_font.language()->encoding(); } } char_type const c = d->text_[i]; + // A display math inset inside an ulem command will be output + // as a box of width \linewidth, so we have to either disable + // indentation if the inset starts a paragraph, or start a new + // line to accommodate such box. This has to be done before + // writing any font changing commands. + if (runparams.inDisplayMath && !deleted_display_math + && runparams.inulemcmd) { + if (os.afterParbreak()) + os << "\\noindent"; + else + os << "\\\\\n"; + } + // Do we need to change font? - if ((font != running_font || - font.language() != running_font.language()) && + if ((current_font != running_font || + current_font.language() != running_font.language()) && i != body_pos - 1) { odocstringstream ods; - column += font.latexWriteStartChanges(ods, bparams, + column += current_font.latexWriteStartChanges(ods, bparams, runparams, basefont, last_font); - running_font = font; + // Check again for display math in ulem commands as a + // font change may also occur just before a math inset. + if (runparams.inDisplayMath && !deleted_display_math + && runparams.inulemcmd) { + if (os.afterParbreak()) + os << "\\noindent"; + else + os << "\\\\\n"; + } + running_font = current_font; open_font = true; docstring fontchange = ods.str(); // check whether the fontchange ends with a \\textcolor @@ -2586,7 +2655,7 @@ void Paragraph::latex(BufferParams const & bparams, // style.pass_thru is false. if (i != body_pos - 1) { if (d->simpleTeXBlanks(runparams, os, - i, column, font, style)) { + i, column, current_font, style)) { // A surrogate pair was output. We // must not call latexSpecialChar // in this iteration, since it would output @@ -2599,7 +2668,7 @@ void Paragraph::latex(BufferParams const & bparams, OutputParams rp = runparams; rp.free_spacing = style.free_spacing; - rp.local_font = &font; + rp.local_font = ¤t_font; rp.intitle = style.intitle; // Two major modes: LaTeX or plain @@ -2607,16 +2676,47 @@ void Paragraph::latex(BufferParams const & bparams, // and then split to handle the two modes separately. if (c == META_INSET) { if (i >= start_pos && (end_pos == -1 || i < end_pos)) { + // Greyedout notes and, in general, all insets + // with InsetLayout::isDisplay() == false, + // are typeset inline with the text. So, we + // can add a \par to the last paragraph of + // such insets only if nothing else follows. + bool incremented = false; + Inset const * inset = getInset(i); + InsetText const * textinset = inset + ? inset->asInsetText() + : 0; + if (i + 1 == size() && textinset + && !inset->getLayout().isDisplay()) { + ParagraphList const & pars = + textinset->text().paragraphs(); + pit_type const pit = pars.size() - 1; + Font const last_font = + pit < 0 || pars[pit].empty() + ? pars[pit].getLayoutFont( + bparams, + outerfont) + : pars[pit].getFont(bparams, + pars[pit].size() - 1, + outerfont); + if (last_font.fontInfo().size() != + basefont.fontInfo().size()) { + ++parInline; + incremented = true; + } + } d->latexInset(bparams, os, rp, running_font, basefont, outerfont, open_font, runningChange, style, i, column); + if (incremented) + --parInline; } - } else { - if (i >= start_pos && (end_pos == -1 || i < end_pos)) { - try { - d->latexSpecialChar(os, bparams, rp, running_font, runningChange, - style, i, end_pos, column); - } catch (EncodingException & e) { + } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) { + try { + d->latexSpecialChar(os, bparams, rp, + running_font, runningChange, + style, i, end_pos, column); + } catch (EncodingException & e) { if (runparams.dryrun) { os << "<" << _("LyX Warning: ") << _("uncodable character") << " '"; @@ -2630,31 +2730,75 @@ void Paragraph::latex(BufferParams const & bparams, } } } - } // Set the encoding to that returned from latexSpecialChar (see // comment for encoding member in OutputParams.h) runparams.encoding = rp.encoding; + + // Also carry on the info on a closed ulem command for insets + // such as Note that do not produce any output, so that no + // command is ever executed but its opening was recorded. + runparams.inulemcmd = rp.inulemcmd; } // If we have an open font definition, we have to close it if (open_font) { + // Make sure that \\par is done with the font of the last + // character if this has another size as the default. + // This is necessary because LaTeX (and LyX on the screen) + // calculates the space between the baselines according + // to this font. (Matthias) + // + // We must not change the font for the last paragraph + // of non-multipar insets, tabular cells or commands, + // since this produces unwanted whitespace. + + Font const font = empty() + ? getLayoutFont(bparams, outerfont) + : getFont(bparams, size() - 1, outerfont); + + InsetText const * textinset = inInset().asInsetText(); + + bool const maintext = textinset + ? textinset->text().isMainText() + : false; + + size_t const numpars = textinset + ? textinset->text().paragraphs().size() + : 0; + + bool needPar = false; + + if (style.resfont.size() != font.fontInfo().size() + && (!runparams.isLastPar || maintext + || (numpars > 1 && d->ownerCode() != CELL_CODE + && (inInset().getLayout().isDisplay() + || parInline))) + && !style.isCommand()) { + needPar = true; + } #ifdef FIXED_LANGUAGE_END_DETECTION if (next_) { running_font.latexWriteEndChanges(os, bparams, runparams, basefont, - next_->getFont(bparams, 0, outerfont)); + next_->getFont(bparams, 0, outerfont), + needPar); } else { running_font.latexWriteEndChanges(os, bparams, - runparams, basefont, basefont); + runparams, basefont, basefont, needPar); } #else //FIXME: For now we ALWAYS have to close the foreign font settings if they are //FIXME: there as we start another \selectlanguage with the next paragraph if //FIXME: we are in need of this. This should be fixed sometime (Jug) running_font.latexWriteEndChanges(os, bparams, runparams, - basefont, basefont); + basefont, basefont, needPar); #endif + if (needPar) { + // The \par could not be inserted at the same nesting + // level of the font size change, so do it now. + os << "{\\" << font.latexSize() << "\\par}"; + } } column += Changes::latexMarkChange(os, bparams, runningChange, @@ -2673,8 +2817,7 @@ void Paragraph::latex(BufferParams const & bparams, if (allowcust && d->endTeXParParams(bparams, os, runparams) && runparams.encoding != prev_encoding) { runparams.encoding = prev_encoding; - if (!runparams.isFullUnicode()) - os << setEncoding(prev_encoding->iconvName()); + os << setEncoding(prev_encoding->iconvName()); } LYXERR(Debug::LATEX, "Paragraph::latex... done " << this); @@ -2831,145 +2974,362 @@ void Paragraph::simpleDocBookOnePar(Buffer const & buf, } -void doFontSwitch(XHTMLStream & xs, bool startrange, - bool & flag, FontState curstate, std::string tag, std::string attr = "") +namespace { +void doFontSwitch(vector & tagsToOpen, + vector & tagsToClose, + bool & flag, FontState curstate, html::FontTypes type) { if (curstate == FONT_ON) { - xs << html::StartTag(tag, attr); + tagsToOpen.push_back(html::FontTag(type)); flag = true; - } else if (flag && !startrange) { - xs << html::EndTag(tag); + } else if (flag) { + tagsToClose.push_back(html::EndFontTag(type)); flag = false; } } +} // namespace docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf, XHTMLStream & xs, OutputParams const & runparams, Font const & outerfont, + bool start_paragraph, bool close_paragraph, pos_type initial) const { docstring retval; + // track whether we have opened these tags bool emph_flag = false; bool bold_flag = false; bool noun_flag = false; bool ubar_flag = false; bool dbar_flag = false; bool sout_flag = false; + bool xout_flag = false; bool wave_flag = false; + // shape tags + bool shap_flag = false; + // family tags + bool faml_flag = false; + // size tags + bool size_flag = false; Layout const & style = *d->layout_; - xs.startParagraph(allowEmpty()); - - if (!runparams.for_toc && runparams.html_make_pars) { - // generate a magic label for this paragraph - string const attr = "id='" + magicLabel() + "'"; - xs << html::CompTag("a", attr); - } + if (start_paragraph) + xs.startDivision(allowEmpty()); FontInfo font_old = style.labeltype == LABEL_MANUAL ? style.labelfont : style.font; + FontShape curr_fs = INHERIT_SHAPE; + FontFamily curr_fam = INHERIT_FAMILY; + FontSize curr_size = FONT_SIZE_INHERIT; + + string const default_family = + buf.masterBuffer()->params().fonts_default_family; + + vector tagsToOpen; + vector tagsToClose; + // parsing main loop for (pos_type i = initial; i < size(); ++i) { // let's not show deleted material in the output if (isDeleted(i)) continue; - Font font = getFont(buf.masterBuffer()->params(), i, outerfont); - bool const at_start = (i == initial); + Font const font = getFont(buf.masterBuffer()->params(), i, outerfont); // emphasis FontState curstate = font.fontInfo().emph(); if (font_old.emph() != curstate) - doFontSwitch(xs, at_start, emph_flag, curstate, "em"); + doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH); // noun curstate = font.fontInfo().noun(); if (font_old.noun() != curstate) - doFontSwitch(xs, at_start, noun_flag, curstate, "dfn", "class='lyxnoun'"); + doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN); // underbar curstate = font.fontInfo().underbar(); if (font_old.underbar() != curstate) - doFontSwitch(xs, at_start, ubar_flag, curstate, "u"); - + doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR); + // strikeout curstate = font.fontInfo().strikeout(); if (font_old.strikeout() != curstate) - doFontSwitch(xs, at_start, sout_flag, curstate, "del", "class='strikeout'"); + doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT); - // HTML does not really have an equivalent of the next two, so we will just - // output a single underscore with a class, and people can style it if they - // wish to do so + // xout + curstate = font.fontInfo().xout(); + if (font_old.xout() != curstate) + doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT); // double underbar curstate = font.fontInfo().uuline(); if (font_old.uuline() != curstate) - doFontSwitch(xs, at_start, dbar_flag, curstate, "u", "class='dline'"); + doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR); // wavy line curstate = font.fontInfo().uwave(); if (font_old.uwave() != curstate) - doFontSwitch(xs, at_start, wave_flag, curstate, "u", "class='wavyline'"); + doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE); // bold - if (font_old.series() != font.fontInfo().series()) { - if (font.fontInfo().series() == BOLD_SERIES) { - xs << html::StartTag("b"); - bold_flag = true; - } else if (bold_flag && i != initial) { - xs << html::EndTag("b"); - bold_flag = false; + // a little hackish, but allows us to reuse what we have. + curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF); + if (font_old.series() != font.fontInfo().series()) + doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD); + + // Font shape + curr_fs = font.fontInfo().shape(); + FontShape old_fs = font_old.shape(); + if (old_fs != curr_fs) { + if (shap_flag) { + switch (old_fs) { + case ITALIC_SHAPE: + tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC)); + break; + case SLANTED_SHAPE: + tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED)); + break; + case SMALLCAPS_SHAPE: + tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS)); + break; + case UP_SHAPE: + case INHERIT_SHAPE: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; + } + shap_flag = false; + } + switch (curr_fs) { + case ITALIC_SHAPE: + tagsToOpen.push_back(html::FontTag(html::FT_ITALIC)); + shap_flag = true; + break; + case SLANTED_SHAPE: + tagsToOpen.push_back(html::FontTag(html::FT_SLANTED)); + shap_flag = true; + break; + case SMALLCAPS_SHAPE: + tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS)); + shap_flag = true; + break; + case UP_SHAPE: + case INHERIT_SHAPE: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; } } + + // Font family + curr_fam = font.fontInfo().family(); + FontFamily old_fam = font_old.family(); + if (old_fam != curr_fam) { + if (faml_flag) { + switch (old_fam) { + case ROMAN_FAMILY: + tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN)); + break; + case SANS_FAMILY: + tagsToClose.push_back(html::EndFontTag(html::FT_SANS)); + break; + case TYPEWRITER_FAMILY: + tagsToClose.push_back(html::EndFontTag(html::FT_TYPE)); + break; + case INHERIT_FAMILY: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; + } + faml_flag = false; + } + switch (curr_fam) { + case ROMAN_FAMILY: + // we will treat a "default" font family as roman, since we have + // no other idea what to do. + if (default_family != "rmdefault" && default_family != "default") { + tagsToOpen.push_back(html::FontTag(html::FT_ROMAN)); + faml_flag = true; + } + break; + case SANS_FAMILY: + if (default_family != "sfdefault") { + tagsToOpen.push_back(html::FontTag(html::FT_SANS)); + faml_flag = true; + } + break; + case TYPEWRITER_FAMILY: + if (default_family != "ttdefault") { + tagsToOpen.push_back(html::FontTag(html::FT_TYPE)); + faml_flag = true; + } + break; + case INHERIT_FAMILY: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; + } + } + + // Font size + curr_size = font.fontInfo().size(); + FontSize old_size = font_old.size(); + if (old_size != curr_size) { + if (size_flag) { + switch (old_size) { + case FONT_SIZE_TINY: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY)); + break; + case FONT_SIZE_SCRIPT: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT)); + break; + case FONT_SIZE_FOOTNOTE: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE)); + break; + case FONT_SIZE_SMALL: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL)); + break; + case FONT_SIZE_LARGE: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE)); + break; + case FONT_SIZE_LARGER: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER)); + break; + case FONT_SIZE_LARGEST: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST)); + break; + case FONT_SIZE_HUGE: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE)); + break; + case FONT_SIZE_HUGER: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER)); + break; + case FONT_SIZE_INCREASE: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE)); + break; + case FONT_SIZE_DECREASE: + tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE)); + break; + case FONT_SIZE_INHERIT: + case FONT_SIZE_NORMAL: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; + } + size_flag = false; + } + switch (curr_size) { + case FONT_SIZE_TINY: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY)); + size_flag = true; + break; + case FONT_SIZE_SCRIPT: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT)); + size_flag = true; + break; + case FONT_SIZE_FOOTNOTE: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE)); + size_flag = true; + break; + case FONT_SIZE_SMALL: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL)); + size_flag = true; + break; + case FONT_SIZE_LARGE: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE)); + size_flag = true; + break; + case FONT_SIZE_LARGER: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER)); + size_flag = true; + break; + case FONT_SIZE_LARGEST: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST)); + size_flag = true; + break; + case FONT_SIZE_HUGE: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE)); + size_flag = true; + break; + case FONT_SIZE_HUGER: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER)); + size_flag = true; + break; + case FONT_SIZE_INCREASE: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE)); + size_flag = true; + break; + case FONT_SIZE_DECREASE: + tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE)); + size_flag = true; + break; + case FONT_SIZE_NORMAL: + case FONT_SIZE_INHERIT: + break; + default: + // the other tags are for internal use + LATTEST(false); + break; + } + } + // FIXME XHTML // Other such tags? What about the other text ranges? + vector::const_iterator cit = tagsToClose.begin(); + vector::const_iterator cen = tagsToClose.end(); + for (; cit != cen; ++cit) + xs << *cit; + + vector::const_iterator sit = tagsToOpen.begin(); + vector::const_iterator sen = tagsToOpen.end(); + for (; sit != sen; ++sit) + xs << *sit; + + tagsToClose.clear(); + tagsToOpen.clear(); + Inset const * inset = getInset(i); if (inset) { if (!runparams.for_toc || inset->isInToc()) { OutputParams np = runparams; np.local_font = &font; - if (!inset->getLayout().htmlisblock()) + // If the paragraph has size 1, then we are in the "special + // case" where we do not output the containing paragraph info + if (!inset->getLayout().htmlisblock() && size() != 1) np.html_in_par = true; retval += inset->xhtml(xs, np); } } else { - char_type c = getUChar(buf.masterBuffer()->params(), i); - - if (style.pass_thru || runparams.pass_thru) - xs << c; - else if (c == '-') { - docstring str; - int j = i + 1; - if (j < size() && d->text_[j] == '-') { - j += 1; - if (j < size() && d->text_[j] == '-') { - str += from_ascii("—"); - i += 2; - } else { - str += from_ascii("–"); - i += 1; - } - } - else - str += c; - // We don't want to escape the entities. Note that - // it is safe to do this, since str can otherwise - // only be "-". E.g., it can't be "<". - xs << XHTMLStream::ESCAPE_NONE << str; - } else - xs << c; + char_type c = getUChar(buf.masterBuffer()->params(), + runparams, i); + xs << c; } font_old = font.fontInfo(); } + // FIXME XHTML + // I'm worried about what happens if a branch, say, is itself + // wrapped in some font stuff. I think that will not work. xs.closeFontTags(); - xs.endParagraph(); + if (close_paragraph) + xs.endDivision(); + return retval; } @@ -2977,18 +3337,29 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf, bool Paragraph::isHfill(pos_type pos) const { Inset const * inset = getInset(pos); - return inset && (inset->lyxCode() == SPACE_CODE && - inset->isStretchableSpace()); + return inset && inset->isHfill(); } bool Paragraph::isNewline(pos_type pos) const { + // U+2028 LINE SEPARATOR + // U+2029 PARAGRAPH SEPARATOR + char_type const c = d->text_[pos]; + if (c == 0x2028 || c == 0x2029) + return true; Inset const * inset = getInset(pos); return inset && inset->lyxCode() == NEWLINE_CODE; } +bool Paragraph::isEnvSeparator(pos_type pos) const +{ + Inset const * inset = getInset(pos); + return inset && inset->lyxCode() == SEPARATOR_CODE; +} + + bool Paragraph::isLineSeparator(pos_type pos) const { char_type const c = d->text_[pos]; @@ -3032,16 +3403,13 @@ bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const if ((nextpos == psize || isSpace(nextpos)) && (pos == 0 || isSpace(prevpos))) return false; - return c == '\'' - || ((nextpos == psize || d->text_[nextpos] != '-') - && (pos == 0 || d->text_[prevpos] != '-')); + return true; } -bool Paragraph::isSameSpellRange(pos_type pos1, pos_type pos2) const +FontSpan const & Paragraph::getSpellRange(pos_type pos) const { - return pos1 == pos2 - || d->speller_state_.getRange(pos1) == d->speller_state_.getRange(pos2); + return d->speller_state_.getRange(pos); } @@ -3075,8 +3443,7 @@ Paragraph::getParLanguage(BufferParams const & bparams) const bool Paragraph::isRTL(BufferParams const & bparams) const { - return lyxrc.rtl_support - && getParLanguage(bparams)->rightToLeft() + return getParLanguage(bparams)->rightToLeft() && !inInset().getLayout().forceLTR(); } @@ -3131,7 +3498,7 @@ docstring Paragraph::asString(int options) const } -docstring Paragraph::asString(pos_type beg, pos_type end, int options) const +docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const { odocstringstream os; @@ -3148,9 +3515,12 @@ docstring Paragraph::asString(pos_type beg, pos_type end, int options) const || (c == '\n' && (options & AS_STR_NEWLINES))) os.put(c); else if (c == META_INSET && (options & AS_STR_INSETS)) { - getInset(i)->toString(os); - if (getInset(i)->asInsetMath()) - os << " "; + if (c == META_INSET && (options & AS_STR_PLAINTEXT)) { + LASSERT(runparams != 0, return docstring()); + getInset(i)->plaintext(os, *runparams); + } else { + getInset(i)->toString(os); + } } } @@ -3158,48 +3528,25 @@ docstring Paragraph::asString(pos_type beg, pos_type end, int options) const } -void Paragraph::forToc(docstring & os, size_t maxlen) const +void Paragraph::forOutliner(docstring & os, size_t const maxlen, + bool const shorten, bool const label) const { - if (!d->params_.labelString().empty()) - os += d->params_.labelString() + ' '; - for (pos_type i = 0; i < size() && os.length() < maxlen; ++i) { + size_t tmplen = shorten ? maxlen + 1 : maxlen; + if (label && !labelString().empty()) + os += labelString() + ' '; + if (!layout().isTocCaption()) + return; + for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) { if (isDeleted(i)) continue; char_type const c = d->text_[i]; if (isPrintable(c)) os += c; - else if (c == '\t' || c == '\n') - os += ' '; else if (c == META_INSET) - getInset(i)->forToc(os, maxlen); - } -} - - -docstring Paragraph::stringify(pos_type beg, pos_type end, int options, - OutputParams const & runparams) const -{ - odocstringstream os; - - if (beg == 0 - && options & AS_STR_LABEL - && !d->params_.labelString().empty()) - os << d->params_.labelString() << ' '; - - OutputParams op = runparams; - op.for_search = true; - - for (pos_type i = beg; i < end; ++i) { - char_type const c = d->text_[i]; - if (isPrintable(c) || c == '\t' - || (c == '\n' && (options & AS_STR_NEWLINES))) - os.put(c); - else if (c == META_INSET && (options & AS_STR_INSETS)) { - getInset(i)->plaintext(os, op); - } + getInset(i)->forOutliner(os, tmplen, false); } - - return os.str(); + if (shorten) + Text::shortenForOutliner(os, maxlen); } @@ -3289,46 +3636,6 @@ bool Paragraph::allowEmpty() const } -char_type Paragraph::transformChar(char_type c, pos_type pos) const -{ - if (!Encodings::isArabicChar(c)) - return c; - - char_type prev_char = ' '; - char_type next_char = ' '; - - for (pos_type i = pos - 1; i >= 0; --i) { - char_type const par_char = d->text_[i]; - if (!Encodings::isArabicComposeChar(par_char)) { - prev_char = par_char; - break; - } - } - - for (pos_type i = pos + 1, end = size(); i < end; ++i) { - char_type const par_char = d->text_[i]; - if (!Encodings::isArabicComposeChar(par_char)) { - next_char = par_char; - break; - } - } - - if (Encodings::isArabicChar(next_char)) { - if (Encodings::isArabicChar(prev_char) && - !Encodings::isArabicSpecialChar(prev_char)) - return Encodings::transformChar(c, Encodings::FORM_MEDIAL); - else - return Encodings::transformChar(c, Encodings::FORM_INITIAL); - } else { - if (Encodings::isArabicChar(prev_char) && - !Encodings::isArabicSpecialChar(prev_char)) - return Encodings::transformChar(c, Encodings::FORM_FINAL); - else - return Encodings::transformChar(c, Encodings::FORM_ISOLATED); - } -} - - bool Paragraph::brokenBiblio() const { // there is a problem if there is no bibitem at position 0 or @@ -3350,7 +3657,7 @@ int Paragraph::fixBiblio(Buffer const & buffer) if (d->layout_->labeltype != LABEL_BIBLIO) return 0; - bool const track_changes = buffer.params().trackChanges; + bool const track_changes = buffer.params().track_changes; int bibitem_pos = d->insetlist_.find(BIBITEM_CODE); bool const hasbibitem0 = bibitem_pos == 0; @@ -3365,8 +3672,7 @@ int Paragraph::fixBiblio(Buffer const & buffer) // these, which there should be. // FIXME: why does it make sense to do that rather // than keep the first? (JMarc) - Inset * inset = d->insetlist_.release(bibitem_pos); - eraseChar(bibitem_pos, track_changes); + Inset * inset = releaseInset(bibitem_pos); d->insetlist_.begin()->inset = inset; return -bibitem_pos; } @@ -3382,7 +3688,8 @@ int Paragraph::fixBiblio(Buffer const & buffer) inset = new InsetBibitem(const_cast(&buffer), InsetCommandParams(BIBITEM_CODE)); - insertInset(0, inset, Change(track_changes ? Change::INSERTED + Font font(inherit_font, buffer.params().language); + insertInset(0, inset, font, Change(track_changes ? Change::INSERTED : Change::UNCHANGED)); return 1; @@ -3425,6 +3732,12 @@ void Paragraph::setBuffer(Buffer & b) } +void Paragraph::resetBuffer() +{ + d->insetlist_.resetBuffer(); +} + + Inset * Paragraph::releaseInset(pos_type pos) { Inset * inset = d->insetlist_.release(pos); @@ -3454,9 +3767,11 @@ void Paragraph::changeCase(BufferParams const & bparams, pos_type pos, // process sequences of modified characters; in change // tracking mode, this approach results in much better // usability than changing case on a char-by-char basis - docstring changes; + // We also need to track the current font, since font + // changes within sequences can occur. + vector > changes; - bool const trackChanges = bparams.trackChanges; + bool const trackChanges = bparams.track_changes; bool capitalize = true; @@ -3488,7 +3803,7 @@ void Paragraph::changeCase(BufferParams const & bparams, pos_type pos, } if (oldChar != newChar) { - changes += newChar; + changes.push_back(make_pair(newChar, getFontSettings(bparams, pos))); if (pos != right - 1) continue; // step behind the changing area @@ -3497,9 +3812,8 @@ void Paragraph::changeCase(BufferParams const & bparams, pos_type pos, int erasePos = pos - changes.size(); for (size_t i = 0; i < changes.size(); i++) { - insertChar(pos, changes[i], - getFontSettings(bparams, - erasePos), + insertChar(pos, changes[i].first, + changes[i].second, trackChanges); if (!eraseChar(erasePos, trackChanges)) { ++erasePos; @@ -3589,11 +3903,11 @@ void Paragraph::deregisterWords() Private::LangWordsMap::const_iterator itl = d->words_.begin(); Private::LangWordsMap::const_iterator ite = d->words_.end(); for (; itl != ite; ++itl) { - WordList * wl = theWordList(itl->first); + WordList & wl = theWordList(itl->first); Private::Words::const_iterator it = (itl->second).begin(); Private::Words::const_iterator et = (itl->second).end(); for (; it != et; ++it) - wl->remove(*it); + wl.remove(*it); } d->words_.clear(); } @@ -3610,13 +3924,13 @@ void Paragraph::locateWord(pos_type & from, pos_type & to, to = from; return; } - // no break here, we go to the next + // fall through case WHOLE_WORD: // If we are already at the beginning of a word, do nothing if (!from || isWordSeparator(from - 1)) break; - // no break here, we go to the next + // fall through case PREVIOUS_WORD: // always move the cursor to the beginning of previous word @@ -3638,20 +3952,28 @@ void Paragraph::locateWord(pos_type & from, pos_type & to, void Paragraph::collectWords() { - pos_type n = size(); - for (pos_type pos = 0; pos < n; ++pos) { + for (pos_type pos = 0; pos < size(); ++pos) { if (isWordSeparator(pos)) continue; pos_type from = pos; locateWord(from, pos, WHOLE_WORD); - if ((pos - from) >= (int)lyxrc.completion_minlength) { - docstring word = asString(from, pos, AS_STR_NONE); - FontList::const_iterator cit = d->fontlist_.fontIterator(pos); - if (cit == d->fontlist_.end()) - return; - Language const * lang = cit->font().language(); - d->words_[*lang].insert(word); - } + // Work around MSVC warning: The statement + // if (pos < from + lyxrc.completion_minlength) + // triggers a signed vs. unsigned warning. + // I don't know why this happens, it could be a MSVC bug, or + // related to LLP64 (windows) vs. LP64 (unix) programming + // model, or the C++ standard might be ambigous in the section + // defining the "usual arithmetic conversions". However, using + // a temporary variable is safe and works on all compilers. + pos_type const endpos = from + lyxrc.completion_minlength; + if (pos < endpos) + continue; + FontList::const_iterator cit = d->fontlist_.fontIterator(from); + if (cit == d->fontlist_.end()) + return; + Language const * lang = cit->font().language(); + docstring const word = asString(from, pos, AS_STR_NONE); + d->words_[lang->lang()].insert(word); } } @@ -3661,11 +3983,11 @@ void Paragraph::registerWords() Private::LangWordsMap::const_iterator itl = d->words_.begin(); Private::LangWordsMap::const_iterator ite = d->words_.end(); for (; itl != ite; ++itl) { - WordList * wl = theWordList(itl->first); + WordList & wl = theWordList(itl->first); Private::Words::const_iterator it = (itl->second).begin(); Private::Words::const_iterator et = (itl->second).end(); for (; it != et; ++it) - wl->insert(*it); + wl.insert(*it); } } @@ -3859,6 +4181,15 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to, } +void Paragraph::anonymize() +{ + // This is a very crude anonymization for now + for (char_type & c : d->text_) + if (isLetterChar(c) || isNumber(c)) + c = 'a'; +} + + void Paragraph::Private::markMisspelledWords( pos_type const & first, pos_type const & last, SpellChecker::Result result,