X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2FParagraph.cpp;h=d69fa55bb539459b0760adc24d19724a6da75ec3;hb=2b02b73f4a59bb73f9e819a960ea98c86ef8716f;hp=e066d2b0df654d9cc1f3c15d008754651cb6b16c;hpb=ef11fdc77dc617ec8059216f745fd51ad4c527f2;p=lyx.git diff --git a/src/Paragraph.cpp b/src/Paragraph.cpp index e066d2b0df..d69fa55bb5 100644 --- a/src/Paragraph.cpp +++ b/src/Paragraph.cpp @@ -66,6 +66,7 @@ #include "support/textutils.h" #include "output_docbook.h" +#include #include #include #include @@ -342,7 +343,10 @@ public: Change & running_change, Layout const & style, pos_type & i, - unsigned int & column); + unsigned int & column, + bool const fontswitch_inset, + bool const closeLanguage, + bool const lang_switched_at_inset); /// void latexSpecialChar( @@ -957,7 +961,10 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, Change & running_change, Layout const & style, pos_type & i, - unsigned int & column) + unsigned int & column, + bool const fontswitch_inset, + bool const closeLanguage, + bool const lang_switched_at_inset) { Inset * inset = owner_->getInset(i); LBUFERR(inset); @@ -1027,20 +1034,10 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, close = true; } - if (open_font && (!inset->inheritFont() || inset->allowMultiPar())) { - // Some insets cannot be inside a font change command. - // However, even such insets *can* be placed in \L or \R - // or their equivalents (for RTL language switches), so we don't - // close the language in those cases. - // ArabTeX, though, cannot handle this special behavior, it seems. - bool arabtex = basefont.language()->lang() == "arabic_arabtex" - || running_font.language()->lang() == "arabic_arabtex"; - bool closeLanguage = arabtex - || basefont.isRightToLeft() == running_font.isRightToLeft(); + if (open_font && fontswitch_inset) { bool lang_closed = false; - bool lang_switched_at_inset = false; // Close language if needed - if (closeLanguage) { + if (closeLanguage && !lang_switched_at_inset) { // We need prev_font here as language changes directly at inset // will only be started inside the inset. Font const prev_font = (i > 0) ? @@ -1054,20 +1051,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, needPar, closeLanguage); column += count; lang_closed = count > 0; - lang_switched_at_inset = prev_font.language() != running_font.language(); } - // Now re-do font changes in a way needed here - // (using switches with multi-par insets) - InsetText const * textinset = inset->asInsetText(); - bool const cprotect = textinset - ? textinset->hasCProtectContent(runparams.moving_arg) - && !textinset->text().isMainText() - : false; - unsigned int count2 = running_font.latexWriteStartChanges(os, bparams, - runparams, basefont, - running_font, true, - cprotect); - column += count2; // Update the running_font, making sure, however, // to leave the language as it was. // FIXME: probably a better way to keep track of the old @@ -1077,11 +1061,26 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, running_font = basefont; if (!closeLanguage) running_font.setLanguage(copy_font.language()); + OutputParams rp = runparams; + rp.encoding = basefont.language()->encoding(); // For these, we use switches, so they should be taken as // base inside the inset. basefont.fontInfo().setSize(copy_font.fontInfo().size()); basefont.fontInfo().setFamily(copy_font.fontInfo().family()); basefont.fontInfo().setSeries(copy_font.fontInfo().series()); + // Now re-do font changes in a way needed here + // (using switches with multi-par insets) + InsetText const * textinset = inset->asInsetText(); + bool const cprotect = textinset + ? textinset->hasCProtectContent(runparams.moving_arg) + && !textinset->text().isMainText() + : false; + unsigned int count2 = basefont.latexWriteStartChanges(os, bparams, + rp, running_font, + basefont, true, + cprotect); + open_font = true; + column += count2; if (count2 == 0 && (lang_closed || lang_switched_at_inset)) // All fonts closed open_font = false; @@ -1089,6 +1088,12 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, runparams.local_font = &basefont; } + if (fontswitch_inset && !closeLanguage) { + // The directionality has been switched at inset. + // Force markup inside. + runparams.local_font = &basefont; + } + size_t const previous_row_count = os.texrow().rows(); try { @@ -1099,7 +1104,7 @@ void Paragraph::Private::latexInset(BufferParams const & bparams, // add location information and throw again. e.par_id = id_; e.pos = i; - throw(e); + throw; } if (close) @@ -2575,7 +2580,7 @@ void Paragraph::latex(BufferParams const & bparams, ++column; // Fully instantiated font - Font const current_font = getFont(bparams, i, outerfont); + Font current_font = getFont(bparams, i, outerfont); // Previous font Font const prev_font = (i > 0) ? getFont(bparams, i - 1, outerfont) @@ -2586,16 +2591,37 @@ void Paragraph::latex(BufferParams const & bparams, && runningChange == change && change.type == Change::DELETED && !os.afterParbreak()); - bool const multipar_inset = - (c == META_INSET && getInset(i) && getInset(i)->allowMultiPar()); + // Insets where font switches are used (rather than font commands) + bool const fontswitch_inset = + c == META_INSET + && getInset(i) + && getInset(i)->allowMultiPar() + && getInset(i)->lyxCode() != ERT_CODE; + + bool closeLanguage = false; + bool lang_switched_at_inset = false; + if (fontswitch_inset) { + // Some insets cannot be inside a font change command. + // However, even such insets *can* be placed in \L or \R + // or their equivalents (for RTL language switches), + // so we don't close the language in those cases + // (= differing isRightToLeft()). + // ArabTeX, though, doesn't seem to handle this special behavior. + closeLanguage = basefont.isRightToLeft() == current_font.isRightToLeft() + || basefont.language()->lang() == "arabic_arabtex" + || current_font.language()->lang() == "arabic_arabtex"; + // We need to check prev_font as language changes directly at inset + // will only be started inside the inset. + lang_switched_at_inset = prev_font.language() != current_font.language(); + } // Do we need to close the previous font? + bool langClosed = false; if (open_font && ((current_font != running_font || current_font.language() != running_font.language()) - || (multipar_inset - && (current_font == prev_font - || current_font.language() == prev_font.language())))) + || (fontswitch_inset + && (current_font == prev_font)))) { // ensure there is no open script-wrapper if (!alien_script.empty()) { @@ -2610,10 +2636,13 @@ void Paragraph::latex(BufferParams const & bparams, os << '}'; column += 1; } + if (closeLanguage) + // Force language closing + current_font.setLanguage(basefont.language()); + Font const nextfont = (i == body_pos-1) ? basefont : current_font; column += running_font.latexWriteEndChanges( os, bparams, runparams, basefont, - (i == body_pos-1) ? basefont : current_font, - needPar); + nextfont, needPar); if (in_ct_deletion) { // We have to close and then reopen \lyxdeleted, // as strikeout needs to be on lowest level. @@ -2621,8 +2650,12 @@ void Paragraph::latex(BufferParams const & bparams, column += Changes::latexMarkChange(os, bparams, Change(Change::UNCHANGED), Change(Change::DELETED), rp); } - running_font = basefont; open_font = false; + // Has the language been closed in the latexWriteEndChanges() call above? + langClosed = running_font.language() != basefont.language() + && running_font.language() != nextfont.language() + && (running_font.language()->encoding()->package() != Encoding::CJK); + running_font = basefont; } // if necessary, close language environment before opening CJK @@ -2639,7 +2672,8 @@ void Paragraph::latex(BufferParams const & bparams, } // Switch file encoding if necessary (and allowed) - if (!runparams.pass_thru && !style.pass_thru && + if ((!fontswitch_inset || closeLanguage) + && !runparams.pass_thru && !style.pass_thru && runparams.encoding->package() != Encoding::none && current_font.language()->encoding()->package() != Encoding::none) { pair const enc_switch = @@ -2669,19 +2703,19 @@ void Paragraph::latex(BufferParams const & bparams, current_font.language() != running_font.language()) && i != body_pos - 1) { - if (in_ct_deletion) { - // We have to close and then reopen \lyxdeleted, - // as strikeout needs to be on lowest level. - bool needPar = false; - OutputParams rp = runparams; - column += running_font.latexWriteEndChanges( - os, bparams, rp, basefont, - basefont, needPar); - os << '}'; - column += 1; - } - otexstringstream ots; - if (!multipar_inset) { + if (!fontswitch_inset) { + if (in_ct_deletion) { + // We have to close and then reopen \lyxdeleted, + // as strikeout needs to be on lowest level. + bool needPar = false; + OutputParams rp = runparams; + column += running_font.latexWriteEndChanges( + os, bparams, rp, basefont, + basefont, needPar); + os << '}'; + column += 1; + } + otexstringstream ots; InsetText const * textinset = inInset().asInsetText(); bool const cprotect = textinset ? textinset->hasCProtectContent(runparams.moving_arg) @@ -2690,35 +2724,38 @@ void Paragraph::latex(BufferParams const & bparams, column += current_font.latexWriteStartChanges(ots, bparams, runparams, basefont, last_font, false, cprotect); - } - // 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 = ots.str(); - os << fontchange; - // check whether the fontchange ends with a \\textcolor - // modifier and the text starts with a space. If so we - // need to add } in order to prevent \\textcolor from gobbling - // the space (bug 4473). - docstring const last_modifier = rsplit(fontchange, '\\'); - if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ') - os << from_ascii("{}"); - else if (ots.terminateCommand()) - os << termcmd; - if (in_ct_deletion) { - // We have to close and then reopen \lyxdeleted, - // as strikeout needs to be on lowest level. - OutputParams rp = runparams; - column += Changes::latexMarkChange(os, bparams, - Change(Change::UNCHANGED), change, rp); + // 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 = ots.str(); + os << fontchange; + // check whether the fontchange ends with a \\textcolor + // modifier and the text starts with a space. If so we + // need to add } in order to prevent \\textcolor from gobbling + // the space (bug 4473). + docstring const last_modifier = rsplit(fontchange, '\\'); + if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ') + os << from_ascii("{}"); + else if (ots.terminateCommand()) + os << termcmd; + if (in_ct_deletion) { + // We have to close and then reopen \lyxdeleted, + // as strikeout needs to be on lowest level. + OutputParams rp = runparams; + column += Changes::latexMarkChange(os, bparams, + Change(Change::UNCHANGED), change, rp); + } + } else { + running_font = current_font; + open_font = !langClosed; } } @@ -2781,16 +2818,24 @@ void Paragraph::latex(BufferParams const & bparams, incremented = true; } } - // We need to restore these after insets with + // We need to restore parts of this after insets with // allowMultiPar() true - Font const save_running_font = running_font; Font const save_basefont = basefont; d->latexInset(bparams, os, rp, running_font, basefont, real_outerfont, open_font, - runningChange, style, i, column); - if (multipar_inset) { - running_font = save_running_font; - basefont = save_basefont; + runningChange, style, i, column, fontswitch_inset, + closeLanguage, lang_switched_at_inset); + if (fontswitch_inset) { + if (open_font) { + bool needPar = false; + column += running_font.latexWriteEndChanges( + os, bparams, runparams, + basefont, basefont, needPar); + open_font = false; + } + basefont.fontInfo().setSize(save_basefont.fontInfo().size()); + basefont.fontInfo().setFamily(save_basefont.fontInfo().family()); + basefont.fontInfo().setSeries(save_basefont.fontInfo().series()); } if (incremented) --parInline; @@ -2834,7 +2879,7 @@ void Paragraph::latex(BufferParams const & bparams, // add location information and throw again. e.par_id = id(); e.pos = i; - throw(e); + throw; } } } @@ -3288,19 +3333,16 @@ std::tuple, vector> computeDocBookFontSwit } // anonymous namespace -void Paragraph::simpleDocBookOnePar(Buffer const & buf, - XMLStream & xs, - OutputParams const & runparams, - Font const & outerfont, - bool start_paragraph, bool close_paragraph, - pos_type initial) const +std::vector Paragraph::simpleDocBookOnePar(Buffer const & buf, + OutputParams const & runparams, + Font const & outerfont, + pos_type initial, + bool is_last_par, + bool ignore_fonts) const { - // track whether we have opened these tags + // Track whether we have opened these tags DocBookFontState fs; - if (start_paragraph) - xs.startDivision(allowEmpty()); - Layout const & style = *d->layout_; FontInfo font_old = style.labeltype == LABEL_MANUAL ? style.labelfont : style.font; @@ -3311,48 +3353,88 @@ void Paragraph::simpleDocBookOnePar(Buffer const & buf, vector tagsToOpen; vector tagsToClose; - // parsing main loop + std::vector generatedParagraphs; + DocBookFontState old_fs = fs; + odocstringstream os; + auto * xs = new XMLStream(os); // XMLStream has no copy constructor: to create a new object, the only solution + // is to hold a pointer to the XMLStream (xs = XMLStream(os) is not allowed once the first object is built). + + // When a font tag ends with a space, output it after the closing font tag. This requires to store delayed + // characters at some point. + std::vector delayedChars; + + // Parsing main loop. for (pos_type i = initial; i < size(); ++i) { - // let's not show deleted material in the output + // Don't show deleted material in the output. if (isDeleted(i)) continue; - Font const font = getFont(buf.masterBuffer()->params(), i, outerfont); - - // Determine which tags should be opened or closed. - tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs); - - // FIXME XHTML - // Other such tags? What about the other text ranges? + // If this is an InsetNewline, generate a new paragraph. Also reset the fonts, so that tags are closed in + // this paragraph. + if (getInset(i) != nullptr && getInset(i)->lyxCode() == NEWLINE_CODE) { + if (!ignore_fonts) + xs->closeFontTags(); + + // Output one paragraph (i.e. one string entry in generatedParagraphs). + generatedParagraphs.push_back(os.str()); + + // Create a new XMLStream for the new paragraph, completely independent from the previous one. This implies + // that the string stream must be reset. + os.str(from_ascii("")); + delete xs; + xs = new XMLStream(os); + + // Restore the fonts for the new paragraph, so that the right tags are opened for the new entry. + if (!ignore_fonts) { + font_old = outerfont.fontInfo(); + fs = old_fs; + } + } - vector::const_iterator cit = tagsToClose.begin(); - vector::const_iterator cen = tagsToClose.end(); - for (; cit != cen; ++cit) - xs << *cit; + // Determine which tags should be opened or closed regarding fonts. + Font const font = getFont(buf.masterBuffer()->params(), i, outerfont); + if (!ignore_fonts) { + tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs); + + // 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; + + // Deal with the delayed characters *after* closing font tags. + if (!delayedChars.empty()) { + for (char_type c: delayedChars) + *xs << c; + delayedChars.clear(); + } - vector::const_iterator sit = tagsToOpen.begin(); - vector::const_iterator sen = tagsToOpen.end(); - for (; sit != sen; ++sit) - xs << *sit; + vector::const_iterator sit = tagsToOpen.begin(); + vector::const_iterator sen = tagsToOpen.end(); + for (; sit != sen; ++sit) + *xs << *sit; - tagsToClose.clear(); - tagsToOpen.clear(); + tagsToClose.clear(); + tagsToOpen.clear(); + } if (Inset const * inset = getInset(i)) { if (!runparams.for_toc || inset->isInToc()) { OutputParams np = runparams; np.local_font = &font; - // If the paragraph has size 1, then we are in the "special - // case" where we do not output the containing paragraph info. - // This "special case" is defined in more details in output_docbook.cpp, makeParagraphs. The results - // of that brittle logic is passed to this function through open_par. - if (!inset->getLayout().htmlisblock() && size() != 1) // TODO: htmlisblock here too! - np.docbook_in_par = true; - inset->docbook(xs, np); + + // TODO: special case will bite here. + np.docbook_in_par = true; + inset->docbook(*xs, np); } } else { char_type c = getUChar(buf.masterBuffer()->params(), runparams, i); - xs << c; + if (lyx::isSpace(c) && !ignore_fonts) + delayedChars.push_back(c); + else + *xs << c; } font_old = font.fontInfo(); } @@ -3360,11 +3442,24 @@ void Paragraph::simpleDocBookOnePar(Buffer const & buf, // FIXME, this code is just imported from 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(); - if (runparams.docbook_in_listing) - xs << xml::CR(); - if (close_paragraph) - xs.endDivision(); + if (!ignore_fonts) + xs->closeFontTags(); + + // Deal with the delayed characters *after* closing font tags. + if (!delayedChars.empty()) + for (char_type c: delayedChars) + *xs << c; + + // In listings, new lines (i.e. \n characters in the output) are very important. Avoid generating one for the + // last line to get a clean output. + if (runparams.docbook_in_listing && !is_last_par) + *xs << xml::CR(); + + // Finalise the last (and most likely only) paragraph. + generatedParagraphs.push_back(os.str()); + delete xs; + + return generatedParagraphs; } @@ -3799,8 +3894,8 @@ bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const char_type const c = d->text_[pos]; if (c != '-' && c != '\'') return false; - int nextpos = pos + 1; - int prevpos = pos > 0 ? pos - 1 : 0; + pos_type nextpos = pos + 1; + pos_type prevpos = pos > 0 ? pos - 1 : 0; if ((nextpos == psize || isSpace(nextpos)) && (pos == 0 || isSpace(prevpos))) return false; @@ -3841,9 +3936,6 @@ bool Paragraph::needsCProtection(bool const fragile) const Inset const * ins = getInset(i); if (ins->needsCProtection(maintext, fragile)) return true; - if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT) - // Environments need cprotection regardless the content - return true; // Now check math environments InsetMath const * im = getInset(i)->asInsetMath(); if (!im || im->cell(0).empty()) @@ -4277,10 +4369,8 @@ 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].first, - changes[i].second, - trackChanges); + for (auto const & change : changes) { + insertChar(pos, change.first, change.second, trackChanges); if (!eraseChar(erasePos, trackChanges)) { ++erasePos; ++pos; // advance @@ -4303,9 +4393,12 @@ int Paragraph::find(docstring const & str, bool cs, bool mw, // Ignore "invisible" letters such as ligature breaks // and hyphenation chars while searching while (pos < parsize - 1 && isInset(pos)) { + Inset const * inset = getInset(pos); + if (!inset->isLetter()) + break; odocstringstream os; - getInset(pos)->toString(os); - if (!getInset(pos)->isLetter() || !os.str().empty()) + inset->toString(os); + if (!os.str().empty()) break; pos++; } @@ -4499,7 +4592,18 @@ Language * Paragraph::Private::locateSpellRange( while (last < to && samelang && sameinset) { // hop to end of word while (last < to && !owner_->isWordSeparator(last)) { - if (owner_->getInset(last)) { + Inset const * inset = owner_->getInset(last); + if (inset && dynamic_cast(inset)) { + // check for "invisible" letters such as ligature breaks + odocstringstream os; + inset->toString(os); + if (os.str().length() != 0) { + // avoid spell check of visible special char insets + // stop the loop in front of the special char inset + sameinset = false; + break; + } + } else if (inset) { appendSkipPosition(skips, last); } else if (owner_->isDeleted(last)) { appendSkipPosition(skips, last); @@ -4725,7 +4829,7 @@ void Paragraph::spellCheck() const // start the spell checker on the unit of meaning docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE); WordLangTuple wl = WordLangTuple(word, lang); - SpellChecker::Result result = word.size() ? + SpellChecker::Result result = !word.empty() ? speller->check(wl) : SpellChecker::WORD_OK; d->markMisspelledWords(first, last, result, word, skips); first = ++last;