X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2FParagraph.cpp;h=6410d570853afec660e81ff245edbe423ec354ee;hb=4cb209b121430b38606932e31d2659bd7ed57447;hp=574a830690452919cb210b1510210c81ba5fa967;hpb=02d37932d0e3000e31d726593d953aa5a506a474;p=lyx.git diff --git a/src/Paragraph.cpp b/src/Paragraph.cpp index 574a830690..6410d57085 100644 --- a/src/Paragraph.cpp +++ b/src/Paragraph.cpp @@ -314,13 +314,15 @@ public: /// Output the surrogate pair formed by \p c and \p next to \p os. /// \return the number of characters written. - int latexSurrogatePair(otexstream & os, char_type c, char_type next, + int latexSurrogatePair(BufferParams const &, otexstream & os, + char_type c, char_type next, OutputParams const &); /// Output a space in appropriate formatting (or a surrogate pair /// if the next character is a combining character). /// \return whether a surrogate pair was output. - bool simpleTeXBlanks(OutputParams const &, + bool simpleTeXBlanks(BufferParams const &, + OutputParams const &, otexstream &, pos_type i, unsigned int & column, @@ -330,7 +332,8 @@ public: /// Output consecutive unicode chars, belonging to the same script as /// specified by the latex macro \p ltx, to \p os starting from \p i. /// \return the number of characters written. - int writeScriptChars(OutputParams const &, otexstream & os, + int writeScriptChars(BufferParams const &, OutputParams const &, + otexstream & os, docstring const & ltx, Change const &, Encoding const &, std::string const, pos_type & i); @@ -867,8 +870,9 @@ int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges) } -int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c, - char_type next, OutputParams const & runparams) +int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams, + otexstream & os, char_type c, char_type next, + OutputParams const & runparams) { // Writing next here may circumvent a possible font change between // c and next. Since next is only output if it forms a surrogate pair @@ -907,9 +911,9 @@ int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c, int length = brace2; string fontenc; if (runparams.local_font) - fontenc = runparams.local_font->language()->fontenc(); + fontenc = runparams.local_font->language()->fontenc(bparams); else - fontenc = runparams.main_fontenc; + fontenc = bparams.language->fontenc(bparams); docstring scriptmacro; docstring cb; if (script == "textgreek" || script == "textcyrillic") { @@ -940,7 +944,8 @@ int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c, } -bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams, +bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams, + OutputParams const & runparams, otexstream & os, pos_type i, unsigned int & column, @@ -954,7 +959,7 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams, char_type next = text_[i + 1]; if (Encodings::isCombiningChar(next)) { // This space has an accent, so we must always output it. - column += latexSurrogatePair(os, ' ', next, runparams) - 1; + column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1; return true; } } @@ -985,7 +990,8 @@ bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams, } -int Paragraph::Private::writeScriptChars(OutputParams const & runparams, +int Paragraph::Private::writeScriptChars(BufferParams const & bparams, + OutputParams const & runparams, otexstream & os, docstring const & ltx, Change const & runningChange, @@ -1047,7 +1053,7 @@ int Paragraph::Private::writeScriptChars(OutputParams const & runparams, if (i + 2 < size) { next_next = text_[i + 2]; if (Encodings::isCombiningChar(next_next)) { - length += latexSurrogatePair(os, next, next_next, runparams) - 1; + length += latexSurrogatePair(bparams, os, next, next_next, runparams) - 1; i += 2; continue; } @@ -1228,10 +1234,7 @@ 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 || contains(style.pass_thru_chars, c) @@ -1362,7 +1365,7 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, if (i + 1 < int(text_.size())) { next = text_[i + 1]; if (Encodings::isCombiningChar(next)) { - column += latexSurrogatePair(os, c, next, runparams) - 1; + column += latexSurrogatePair(bparams, os, c, next, runparams) - 1; ++i; break; } @@ -1389,10 +1392,7 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, } } string fontenc; - if (running_font.language()->lang() == bparams.language->lang()) - fontenc = runparams.main_fontenc; - else - fontenc = running_font.language()->fontenc(); + fontenc = running_font.language()->fontenc(bparams); // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding // whether they are encodable or not (it only depends on the font encoding) if (!runparams.isFullUnicode() && Encodings::isKnownScriptChar(c, script)) { @@ -1400,7 +1400,7 @@ void Paragraph::Private::latexSpecialChar(otexstream & os, docstring ltx = latex.first; if (!prefixIs(ltx, wrapper)) ltx = wrapper + latex.first + from_ascii("}"); - column += writeScriptChars(runparams, os, ltx, running_change, + column += writeScriptChars(bparams, runparams, os, ltx, running_change, encoding, fontenc, i) - 1; } else if (latex.second && ((!prefixIs(nextlatex, '\\') @@ -1549,17 +1549,54 @@ void Paragraph::Private::validate(LaTeXFeatures & features) const ? features.buffer().masterParams() : features.buffer().params(); for (pos_type i = 0; i < int(text_.size()) ; ++i) { char_type c = text_[i]; + CharInfo const & ci = Encodings::unicodeCharInfo(c); 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"); - } - if (!bp.use_dash_ligatures - && (c == 0x2013 || c == 0x2014) - && bp.useNonTeXFonts - && features.runparams().flavor == OutputParams::XETEX) + } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) { + // features that depend on the font or input encoding + string feats = ci.textpreamble(); + string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp); + if (fontenc.empty()) + fontenc = features.runparams().main_fontenc; + while (!feats.empty()) { + string feat; + feats = split(feats, feat, ','); + if (contains(feat, "!=")) { + // a feature that is required except for the spcified + // font or input encodings + string realfeature; + string const contexts = ltrim(split(feat, realfeature, '!'), "="); + // multiple encodings are separated by semicolon + vector context = getVectorFromString(contexts, ";"); + // require feature if the context matches neither current font + // nor input encoding + if (std::find(context.begin(), context.end(), fontenc) == context.end() + && std::find(context.begin(), context.end(), + features.runparams().encoding->name()) == context.end()) + features.require(realfeature); + } else if (contains(feat, '=')) { + // a feature that is required only for the spcified + // font or input encodings + string realfeature; + string const contexts = split(feat, realfeature, '='); + // multiple encodings are separated by semicolon + vector context = getVectorFromString(contexts, ";"); + // require feature if the context matches either current font + // or input encoding + if (std::find(context.begin(), context.end(), fontenc) != context.end() + || std::find(context.begin(), context.end(), + features.runparams().encoding->name()) != context.end()) + features.require(realfeature); + } + } + } else 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); @@ -1751,7 +1788,9 @@ void Paragraph::write(ostream & os, BufferParams const & bparams, void Paragraph::validate(LaTeXFeatures & features) const { d->validate(features); - if (needsCProtection()) + bool fragile = features.runparams().moving_arg; + fragile |= layout().needprotect; + if (needsCProtection(fragile)) features.require("cprotect"); } @@ -1955,32 +1994,60 @@ Font const Paragraph::getLayoutFont } -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]; + + // 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 = '}'; @@ -2316,13 +2383,13 @@ int Paragraph::Private::startTeXParParams(BufferParams const & bparams, case LYX_ALIGN_DECIMAL: break; case LYX_ALIGN_LEFT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") + if (!owner_->getParLanguage(bparams)->rightToLeft()) corrected_env(os, begin_tag, "flushleft", code, lastpar, column); else corrected_env(os, begin_tag, "flushright", code, lastpar, column); break; } case LYX_ALIGN_RIGHT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") + if (!owner_->getParLanguage(bparams)->rightToLeft()) corrected_env(os, begin_tag, "flushright", code, lastpar, column); else corrected_env(os, begin_tag, "flushleft", code, lastpar, column); @@ -2374,13 +2441,13 @@ bool Paragraph::Private::endTeXParParams(BufferParams const & bparams, case LYX_ALIGN_DECIMAL: break; case LYX_ALIGN_LEFT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") + if (!owner_->getParLanguage(bparams)->rightToLeft()) output = corrected_env(os, end_tag, "flushleft", code, lastpar, col); else output = corrected_env(os, end_tag, "flushright", code, lastpar, col); break; } case LYX_ALIGN_RIGHT: { - if (owner_->getParLanguage(bparams)->babel() != "hebrew") + if (!owner_->getParLanguage(bparams)->rightToLeft()) output = corrected_env(os, end_tag, "flushright", code, lastpar, col); else output = corrected_env(os, end_tag, "flushleft", code, lastpar, col); @@ -2683,7 +2750,7 @@ void Paragraph::latex(BufferParams const & bparams, // latexSpecialChar ignores spaces if // style.pass_thru is false. if (i != body_pos - 1) { - if (d->simpleTeXBlanks(runparams, os, + if (d->simpleTeXBlanks(bparams, runparams, os, i, column, current_font, style)) { // A surrogate pair was output. We // must not call latexSpecialChar @@ -3345,7 +3412,8 @@ docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf, retval += inset->xhtml(xs, np); } } else { - char_type c = getUChar(buf.masterBuffer()->params(), i); + char_type c = getUChar(buf.masterBuffer()->params(), + runparams, i); xs << c; } font_old = font.fontInfo(); @@ -3435,10 +3503,15 @@ bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const } -bool Paragraph::needsCProtection() const +bool Paragraph::needsCProtection(bool const fragile) const { - // first check the layout of the paragraph - if (layout().needcprotect) { + // first check the layout of the paragraph, but only in insets + InsetText const * textinset = inInset().asInsetText(); + bool const maintext = textinset + ? textinset->text().isMainText() + : false; + + if (!maintext && layout().needcprotect) { // Environments need cprotection regardless the content if (layout().latextype == LATEX_ENVIRONMENT) return true; @@ -3458,7 +3531,7 @@ bool Paragraph::needsCProtection() const // now check whether we have insets that need cprotection pos_type size = d->text_.size(); for (pos_type i = 0; i < size; ++i) - if (isInset(i) && getInset(i)->needsCProtection()) + if (isInset(i) && getInset(i)->needsCProtection(maintext, fragile)) return true; return false; @@ -4184,6 +4257,9 @@ SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to, docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE); Language * lang = d->getSpellLanguage(from); + if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON) + return result; + wl = WordLangTuple(word, lang); if (word.empty())