3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
28 #include "BufferEncodings.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "output_docbook.h"
41 #include "ParagraphParameters.h"
42 #include "SpellChecker.h"
44 #include "texstream.h"
45 #include "TextClass.h"
48 #include "WordLangTuple.h"
51 #include "frontends/alert.h"
53 #include "insets/InsetBibitem.h"
54 #include "insets/InsetLabel.h"
55 #include "insets/InsetSpecialChar.h"
56 #include "insets/InsetText.h"
58 #include "mathed/InsetMathHull.h"
60 #include "support/debug.h"
61 #include "support/docstring_list.h"
62 #include "support/ExceptionMessage.h"
63 #include "support/gettext.h"
64 #include "support/lassert.h"
65 #include "support/lstrings.h"
66 #include "support/textutils.h"
67 #include "output_docbook.h"
74 using namespace lyx::support;
76 // OSX clang, gcc < 4.8.0, and msvc < 2015 do not support C++11 thread_local
77 #if defined(__APPLE__) || (defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ < 8)
78 #define THREAD_LOCAL_STATIC static __thread
79 #elif defined(_MSC_VER) && (_MSC_VER < 1900)
80 #define THREAD_LOCAL_STATIC static __declspec(thread)
82 #define THREAD_LOCAL_STATIC thread_local static
89 /// Inset identifier (above 0x10ffff, for ucs-4)
90 char_type const META_INSET = 0x200001;
95 /////////////////////////////////////////////////////////////////////
99 /////////////////////////////////////////////////////////////////////
101 class SpellResultRange {
103 SpellResultRange(FontSpan range, SpellChecker::Result result)
104 : range_(range), result_(result)
107 FontSpan const & range() const { return range_; }
109 void range(FontSpan const & r) { range_ = r; }
111 SpellChecker::Result result() const { return result_; }
113 void result(SpellChecker::Result r) { result_ = r; }
115 bool contains(pos_type pos) const { return range_.contains(pos); }
117 bool covered(FontSpan const & r) const
119 // 1. first of new range inside current range or
120 // 2. last of new range inside current range or
121 // 3. first of current range inside new range or
122 // 4. last of current range inside new range
123 //FIXME: is this the same as !range_.intersect(r).empty() ?
124 return range_.contains(r.first) || range_.contains(r.last) ||
125 r.contains(range_.first) || r.contains(range_.last);
128 void shift(pos_type pos, int offset)
130 if (range_.first > pos) {
131 range_.first += offset;
132 range_.last += offset;
133 } else if (range_.last >= pos) {
134 range_.last += offset;
139 SpellChecker::Result result_ ;
143 /////////////////////////////////////////////////////////////////////
147 /////////////////////////////////////////////////////////////////////
149 class SpellCheckerState {
153 needs_refresh_ = true;
154 current_change_number_ = 0;
157 void setRange(FontSpan const & fp, SpellChecker::Result state)
160 RangesIterator et = ranges_.end();
161 RangesIterator it = ranges_.begin();
162 for (; it != et; ++it) {
163 if (!it->covered(fp))
164 result.push_back(SpellResultRange(it->range(), it->result()));
165 else if (state == SpellChecker::WORD_OK) {
166 // trim or split the current misspelled range
167 // store misspelled ranges only
168 FontSpan range = it->range();
169 if (fp.first > range.first) {
170 // misspelled area in front of WORD_OK
171 range.last = fp.first - 1;
172 result.push_back(SpellResultRange(range, it->result()));
175 if (fp.last < range.last) {
176 // misspelled area after WORD_OK range
177 range.first = fp.last + 1;
178 result.push_back(SpellResultRange(range, it->result()));
183 if (state != SpellChecker::WORD_OK)
184 ranges_.push_back(SpellResultRange(fp, state));
187 void increasePosAfterPos(pos_type pos)
189 correctRangesAfterPos(pos, 1);
193 void decreasePosAfterPos(pos_type pos)
195 correctRangesAfterPos(pos, -1);
199 void refreshLast(pos_type pos)
201 if (pos < refresh_.last)
205 SpellChecker::Result getState(pos_type pos) const
207 SpellChecker::Result result = SpellChecker::WORD_OK;
208 RangesIterator et = ranges_.end();
209 RangesIterator it = ranges_.begin();
210 for (; it != et; ++it) {
211 if(it->contains(pos)) {
218 FontSpan const & getRange(pos_type pos) const
220 /// empty span to indicate mismatch
221 static FontSpan empty_;
222 RangesIterator et = ranges_.end();
223 RangesIterator it = ranges_.begin();
224 for (; it != et; ++it) {
225 if(it->contains(pos)) {
232 bool needsRefresh() const
234 return needs_refresh_;
237 SpellChecker::ChangeNumber currentChangeNumber() const
239 return current_change_number_;
242 void refreshRange(pos_type & first, pos_type & last) const
244 first = refresh_.first;
245 last = refresh_.last;
248 void needsRefresh(pos_type pos)
250 if (needs_refresh_ && pos != -1) {
251 if (pos < refresh_.first)
252 refresh_.first = pos;
253 if (pos > refresh_.last)
255 } else if (pos != -1) {
256 // init request check for neighbour positions too
257 refresh_.first = pos > 0 ? pos - 1 : 0;
258 // no need for special end of paragraph check
259 refresh_.last = pos + 1;
261 needs_refresh_ = pos != -1;
264 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
266 needs_refresh_ = true;
269 current_change_number_ = change_number;
272 typedef vector<SpellResultRange> Ranges;
273 typedef Ranges::const_iterator RangesIterator;
275 /// the area of the paragraph with pending spell check
278 /// spell state cache version number
279 SpellChecker::ChangeNumber current_change_number_;
282 void correctRangesAfterPos(pos_type pos, int offset)
284 RangesIterator et = ranges_.end();
285 Ranges::iterator it = ranges_.begin();
286 for (; it != et; ++it) {
287 it->shift(pos, offset);
293 /////////////////////////////////////////////////////////////////////
295 // Paragraph::Private
297 /////////////////////////////////////////////////////////////////////
299 class Paragraph::Private
301 // Enforce our own "copy" constructor
302 Private(Private const &) = delete;
303 Private & operator=(Private const &) = delete;
304 // Unique ID generator
305 static int make_id();
308 Private(Paragraph * owner, Layout const & layout);
309 /// "Copy constructor"
310 Private(Private const &, Paragraph * owner);
311 /// Copy constructor from \p beg to \p end
312 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
315 void insertChar(pos_type pos, char_type c, Change const & change);
317 /// Output the surrogate pair formed by \p c and \p next to \p os.
318 /// \return the number of characters written.
319 int latexSurrogatePair(BufferParams const &, otexstream & os,
320 char_type c, char_type next,
321 OutputParams const &);
323 /// Output a space in appropriate formatting (or a surrogate pair
324 /// if the next character is a combining character).
325 /// \return whether a surrogate pair was output.
326 bool simpleTeXBlanks(BufferParams const &,
327 OutputParams const &,
330 unsigned int & column,
332 Layout const & style);
334 /// This could go to ParagraphParameters if we want to.
335 int startTeXParParams(BufferParams const &, otexstream &,
336 OutputParams const &) const;
338 /// This could go to ParagraphParameters if we want to.
339 bool endTeXParParams(BufferParams const &, otexstream &,
340 OutputParams const &) const;
343 void latexInset(BufferParams const &,
348 Font const & outerfont,
350 Change & running_change,
351 Layout const & style,
353 unsigned int & column);
356 void latexSpecialChar(
358 BufferParams const & bparams,
359 OutputParams const & runparams,
360 Font const & running_font,
361 string & alien_script,
362 Layout const & style,
365 unsigned int & column);
372 unsigned int & column);
378 unsigned int & column);
384 unsigned int & column);
387 void validate(LaTeXFeatures & features) const;
389 /// Checks if the paragraph contains only text and no inset or font change.
390 bool onlyText(Buffer const & buf, Font const & outerfont,
391 pos_type initial) const;
393 /// a vector of speller skip positions
394 typedef vector<FontSpan> SkipPositions;
395 typedef SkipPositions::const_iterator SkipPositionsIterator;
397 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
399 Language * getSpellLanguage(pos_type const from) const;
401 Language * locateSpellRange(pos_type & from, pos_type & to,
402 SkipPositions & skips) const;
404 bool hasSpellerChange() const
406 SpellChecker::ChangeNumber speller_change_number = 0;
407 if (theSpellChecker())
408 speller_change_number = theSpellChecker()->changeNumber();
409 return speller_change_number > speller_state_.currentChangeNumber();
412 bool ignoreWord(docstring const & word) const ;
414 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
416 pos_type textsize = owner_->size();
417 // check for sane arguments
418 if (to <= from || from >= textsize)
420 FontSpan fp = FontSpan(from, to - 1);
421 speller_state_.setRange(fp, state);
424 void requestSpellCheck(pos_type pos)
427 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
429 speller_state_.needsRefresh(pos);
432 void readySpellCheck()
434 speller_state_.needsRefresh(-1);
437 bool needsSpellCheck() const
439 return speller_state_.needsRefresh();
442 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
444 speller_state_.refreshRange(first, last);
446 last = owner_->size();
449 pos_type endpos = last;
450 owner_->locateWord(first, endpos, WHOLE_WORD, true);
453 owner_->locateWord(last, endpos, WHOLE_WORD, true);
458 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
462 while (it != et && it->first < start) {
463 int skip = it->last - it->first + 1;
471 void markMisspelledWords(pos_type const & first, pos_type const & last,
472 SpellChecker::Result result,
473 docstring const & word,
474 SkipPositions const & skips);
476 InsetCode ownerCode() const
478 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
481 /// Which Paragraph owns us?
485 Inset const * inset_owner_;
494 ParagraphParameters params_;
496 /// for recording and looking up changes
500 InsetList insetlist_;
503 pos_type begin_of_body_;
505 typedef docstring TextContainer;
509 typedef set<docstring> Words;
510 typedef map<string, Words> LangWordsMap;
514 Layout const * layout_;
516 SpellCheckerState speller_state_;
520 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
521 : owner_(owner), inset_owner_(nullptr), id_(-1), begin_of_body_(0), layout_(&layout)
528 int Paragraph::Private::make_id()
530 // The id is unique per session across buffers because it is used in
531 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
534 static atomic_uint next_id(0);
539 Paragraph::Private::Private(Private const & p, Paragraph * owner)
540 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
542 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
543 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
546 requestSpellCheck(p.text_.size());
550 Paragraph::Private::Private(Private const & p, Paragraph * owner,
551 pos_type beg, pos_type end)
552 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
553 params_(p.params_), changes_(p.changes_),
554 insetlist_(p.insetlist_, beg, end),
555 begin_of_body_(p.begin_of_body_), words_(p.words_),
558 if (beg >= pos_type(p.text_.size()))
560 text_ = p.text_.substr(beg, end - beg);
562 FontList::const_iterator fcit = fontlist_.begin();
563 FontList::const_iterator fend = fontlist_.end();
564 for (; fcit != fend; ++fcit) {
565 if (fcit->pos() < beg)
567 if (fcit->pos() >= end) {
568 // Add last entry in the fontlist_.
569 fontlist_.set(text_.size() - 1, fcit->font());
572 // Add a new entry in the fontlist_.
573 fontlist_.set(fcit->pos() - beg, fcit->font());
575 requestSpellCheck(p.text_.size());
579 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
580 bool output_active, TocBackend & backend) const
582 d->changes_.addToToc(cdit, buf, output_active, backend);
586 bool Paragraph::isDeleted(pos_type start, pos_type end) const
588 LASSERT(start >= 0 && start <= size(), return false);
589 LASSERT(end > start && end <= size() + 1, return false);
591 return d->changes_.isDeleted(start, end);
595 bool Paragraph::isChanged(pos_type start, pos_type end) const
597 LASSERT(start >= 0 && start <= size(), return false);
598 LASSERT(end > start && end <= size() + 1, return false);
600 return d->changes_.isChanged(start, end);
603 // FIXME: Ideally the diverse isChanged() methods should account for that!
604 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
606 LASSERT(start >= 0 && start <= size(), return false);
607 LASSERT(end > start && end <= size() + 1, return false);
609 for (auto const & icit : d->insetlist_) {
610 if (icit.pos < start)
614 if (icit.inset && icit.inset->isChanged())
620 bool Paragraph::isChanged() const
622 return d->changes_.isChanged();
626 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
628 // keep the logic here in sync with the logic of eraseChars()
632 Change const & change = d->changes_.lookup(size());
633 return change.inserted() && change.currentAuthor();
636 Change Paragraph::parEndChange() const
638 return d->changes_.lookup(size());
642 void Paragraph::setChange(Change const & change)
644 // beware of the imaginary end-of-par character!
645 d->changes_.set(change, 0, size() + 1);
648 * Propagate the change recursively - but not in case of DELETED!
650 * Imagine that your co-author makes changes in an existing inset. He
651 * sends your document to you and you come to the conclusion that the
652 * inset should go completely. If you erase it, LyX must not delete all
653 * text within the inset. Otherwise, the change tracked insertions of
654 * your co-author get lost and there is no way to restore them later.
656 * Conclusion: An inset's content should remain untouched if you delete it
659 if (!change.deleted()) {
660 for (pos_type pos = 0; pos < size(); ++pos) {
661 if (Inset * inset = getInset(pos))
662 inset->setChange(change);
668 void Paragraph::setChange(pos_type pos, Change const & change)
670 LASSERT(pos >= 0 && pos <= size(), return);
671 d->changes_.set(change, pos);
673 // see comment in setChange(Change const &) above
674 if (!change.deleted() && pos < size())
675 if (Inset * inset = getInset(pos))
676 inset->setChange(change);
680 Change const & Paragraph::lookupChange(pos_type pos) const
682 LBUFERR(pos >= 0 && pos <= size());
683 return d->changes_.lookup(pos);
687 void Paragraph::acceptChanges(pos_type start, pos_type end)
689 LASSERT(start >= 0 && start <= size(), return);
690 LASSERT(end > start && end <= size() + 1, return);
692 for (pos_type pos = start; pos < end; ++pos) {
693 switch (lookupChange(pos).type) {
694 case Change::UNCHANGED:
695 // accept changes in nested inset
696 if (Inset * inset = getInset(pos))
697 inset->acceptChanges();
700 case Change::INSERTED:
701 d->changes_.set(Change(Change::UNCHANGED), pos);
702 // also accept changes in nested inset
703 if (Inset * inset = getInset(pos))
704 inset->acceptChanges();
707 case Change::DELETED:
708 // Suppress access to non-existent
709 // "end-of-paragraph char"
711 eraseChar(pos, false);
722 void Paragraph::rejectChanges(pos_type start, pos_type end)
724 LASSERT(start >= 0 && start <= size(), return);
725 LASSERT(end > start && end <= size() + 1, return);
727 for (pos_type pos = start; pos < end; ++pos) {
728 switch (lookupChange(pos).type) {
729 case Change::UNCHANGED:
730 // reject changes in nested inset
731 if (Inset * inset = getInset(pos))
732 inset->rejectChanges();
735 case Change::INSERTED:
736 // Suppress access to non-existent
737 // "end-of-paragraph char"
739 eraseChar(pos, false);
745 case Change::DELETED:
746 d->changes_.set(Change(Change::UNCHANGED), pos);
748 // Do NOT reject changes within a deleted inset!
749 // There may be insertions of a co-author inside of it!
757 void Paragraph::Private::insertChar(pos_type pos, char_type c,
758 Change const & change)
760 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
763 changes_.insert(change, pos);
765 // This is actually very common when parsing buffers (and
766 // maybe inserting ascii text)
767 if (pos == pos_type(text_.size())) {
768 // when appending characters, no need to update tables
770 // but we want spell checking
771 requestSpellCheck(pos);
775 text_.insert(text_.begin() + pos, c);
777 // Update the font table.
778 fontlist_.increasePosAfterPos(pos);
781 insetlist_.increasePosAfterPos(pos);
783 // Update list of misspelled positions
784 speller_state_.increasePosAfterPos(pos);
788 bool Paragraph::insertInset(pos_type pos, Inset * inset,
789 Font const & font, Change const & change)
791 LASSERT(inset, return false);
792 LASSERT(pos >= 0 && pos <= size(), return false);
794 // Paragraph::insertInset() can be used in cut/copy/paste operation where
795 // d->inset_owner_ is not set yet.
796 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
799 d->insertChar(pos, META_INSET, change);
800 LASSERT(d->text_[pos] == META_INSET, return false);
802 // Add a new entry in the insetlist_.
803 d->insetlist_.insert(inset, pos);
805 // Some insets require run of spell checker
806 requestSpellCheck(pos);
812 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
814 LASSERT(pos >= 0 && pos <= size(), return false);
816 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
819 Change change = d->changes_.lookup(pos);
821 // set the character to DELETED if
822 // a) it was previously unchanged or
823 // b) it was inserted by a co-author
825 if (!change.changed() ||
826 (change.inserted() && !change.currentAuthor())) {
827 setChange(pos, Change(Change::DELETED));
828 // request run of spell checker
829 requestSpellCheck(pos);
833 if (change.deleted())
837 // Don't physically access the imaginary end-of-paragraph character.
838 // eraseChar() can only mark it as DELETED. A physical deletion of
839 // end-of-par must be handled externally.
845 d->changes_.erase(pos);
847 // if it is an inset, delete the inset entry
848 if (d->text_[pos] == META_INSET)
849 d->insetlist_.erase(pos);
851 d->text_.erase(d->text_.begin() + pos);
853 // Update the fontlist_
854 d->fontlist_.erase(pos);
856 // Update the insetlist_
857 d->insetlist_.decreasePosAfterPos(pos);
859 // Update list of misspelled positions
860 d->speller_state_.decreasePosAfterPos(pos);
861 d->speller_state_.refreshLast(size());
867 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
869 LASSERT(start >= 0 && start <= size(), return 0);
870 LASSERT(end >= start && end <= size() + 1, return 0);
873 for (pos_type count = end - start; count; --count) {
874 if (!eraseChar(i, trackChanges))
880 // Handle combining characters
881 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
882 otexstream & os, char_type c, char_type next,
883 OutputParams const & runparams)
885 // Writing next here may circumvent a possible font change between
886 // c and next. Since next is only output if it forms a surrogate pair
887 // with c we can ignore this:
888 // A font change inside a surrogate pair does not make sense and is
889 // hopefully impossible to input.
890 // FIXME: change tracking
891 // Is this correct WRT change tracking?
892 Encoding const & encoding = *(runparams.encoding);
893 docstring latex1 = encoding.latexChar(next).first;
894 if (runparams.inIPA) {
895 string const tipashortcut = Encodings::TIPAShortcut(next);
896 if (!tipashortcut.empty()) {
897 latex1 = from_ascii(tipashortcut);
900 docstring latex2 = encoding.latexChar(c).first;
902 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
903 // Encoding supports the combination:
904 // output as is (combining char after base char).
905 os << latex2 << latex1;
906 return latex1.length() + latex2.length();
909 os << latex1 << "{" << latex2 << "}";
910 return latex1.length() + latex2.length() + 2;
914 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
915 OutputParams const & runparams,
918 unsigned int & column,
920 Layout const & style)
922 if (style.pass_thru || runparams.pass_thru)
925 if (i + 1 < int(text_.size())) {
926 char_type next = text_[i + 1];
927 if (Encodings::isCombiningChar(next)) {
928 // This space has an accent, so we must always output it.
929 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
934 if (runparams.linelen > 0
935 && column > runparams.linelen
937 && text_[i - 1] != ' '
938 && (i + 1 < int(text_.size()))
939 // same in FreeSpacing mode
940 && !owner_->isFreeSpacing()
941 // In typewriter mode, we want to avoid
942 // ! . ? : at the end of a line
943 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
944 && (text_[i - 1] == '.'
945 || text_[i - 1] == '?'
946 || text_[i - 1] == ':'
947 || text_[i - 1] == '!'))) {
949 os.texrow().start(owner_->id(), i + 1);
951 } else if (style.free_spacing) {
960 void Paragraph::Private::latexInset(BufferParams const & bparams,
962 OutputParams & runparams,
965 Font const & outerfont,
967 Change & running_change,
968 Layout const & style,
970 unsigned int & column)
972 Inset * inset = owner_->getInset(i);
975 if (style.pass_thru) {
976 odocstringstream ods;
977 inset->plaintext(ods, runparams);
982 // FIXME: move this to InsetNewline::latex
983 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
984 // newlines are handled differently here than
985 // the default in simpleTeXSpecialChars().
986 if (!style.newline_allowed) {
990 bool needPar = false;
991 column += running_font.latexWriteEndChanges(
992 os, bparams, runparams,
993 basefont, basefont, needPar);
997 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1000 basefont = owner_->getLayoutFont(bparams, outerfont);
1001 running_font = basefont;
1003 if (runparams.moving_arg)
1007 os.texrow().start(owner_->id(), i + 1);
1011 if (owner_->isDeleted(i)) {
1012 if( ++runparams.inDeletedInset == 1)
1013 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1016 if (inset->canTrackChanges()) {
1017 column += Changes::latexMarkChange(os, bparams, running_change,
1018 Change(Change::UNCHANGED), runparams);
1019 running_change = Change(Change::UNCHANGED);
1023 odocstream::pos_type const len = os.os().tellp();
1025 if (inset->forceLTR(runparams)
1026 && running_font.isRightToLeft()
1027 // ERT is an exception, it should be output with no
1028 // decorations at all
1029 && inset->lyxCode() != ERT_CODE) {
1030 if (runparams.use_polyglossia) {
1032 } else if (running_font.language()->lang() == "farsi"
1033 || running_font.language()->lang() == "arabic_arabi")
1034 os << "\\textLR{" << termcmd;
1040 // FIXME: Bug: we can have an empty font change here!
1041 // if there has just been a font change, we are going to close it
1042 // right now, which means stupid latex code like \textsf{}. AFAIK,
1043 // this does not harm dvi output. A minor bug, thus (JMarc)
1045 // Some insets cannot be inside a font change command.
1046 // However, even such insets *can* be placed in \L or \R
1047 // or their equivalents (for RTL language switches), so we don't
1048 // close the language in those cases.
1049 // ArabTeX, though, cannot handle this special behavior, it seems.
1050 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1051 || running_font.language()->lang() == "arabic_arabtex";
1052 if (open_font && !inset->inheritFont()) {
1053 bool needPar = false;
1054 bool closeLanguage = arabtex
1055 || basefont.isRightToLeft() == running_font.isRightToLeft();
1056 // We pass non_inherit_inset = true here since size switches
1057 // ought not to be terminated here (#8384).
1058 unsigned int count = running_font.latexWriteEndChanges(os,
1059 bparams, runparams, basefont, basefont,
1060 needPar, closeLanguage, true);
1062 // if any font properties were closed, update the running_font,
1063 // making sure, however, to leave the language as it was
1065 // FIXME: probably a better way to keep track of the old
1066 // language, than copying the entire font?
1067 Font const copy_font(running_font);
1068 basefont = owner_->getLayoutFont(bparams, outerfont);
1069 running_font = basefont;
1071 running_font.setLanguage(copy_font.language());
1072 // For these, we use switches, so no need to close
1073 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1074 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1075 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1076 // leave font open if language or any of the switches is still open
1077 open_font = (running_font.language() == basefont.language()
1078 || running_font.fontInfo().size() == basefont.fontInfo().size()
1079 || running_font.fontInfo().family() == basefont.fontInfo().family()
1080 || running_font.fontInfo().series() == basefont.fontInfo().series());
1082 runparams.local_font = &basefont;
1086 size_t const previous_row_count = os.texrow().rows();
1089 runparams.lastid = id_;
1090 runparams.lastpos = i;
1091 inset->latex(os, runparams);
1092 } catch (EncodingException & e) {
1093 // add location information and throw again.
1102 if (os.texrow().rows() > previous_row_count) {
1103 os.texrow().start(owner_->id(), i + 1);
1106 column += (unsigned int)(os.os().tellp() - len);
1109 if (owner_->isDeleted(i))
1110 --runparams.inDeletedInset;
1114 void Paragraph::Private::latexSpecialChar(otexstream & os,
1115 BufferParams const & bparams,
1116 OutputParams const & runparams,
1117 Font const & running_font,
1118 string & alien_script,
1119 Layout const & style,
1122 unsigned int & column)
1124 char_type const c = owner_->getUChar(bparams, runparams, i);
1126 if (style.pass_thru || runparams.pass_thru
1127 || contains(style.pass_thru_chars, c)
1128 || contains(runparams.pass_thru_chars, c)) {
1130 Encoding const * const enc = runparams.encoding;
1131 if (enc && !enc->encodable(c))
1132 throw EncodingException(c);
1138 // TIPA uses its own T3 encoding
1139 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1141 // If T1 font encoding is used, use the special
1142 // characters it provides.
1143 // NOTE: Some languages reset the font encoding internally to a
1144 // non-standard font encoding. If we are using such a language,
1145 // we do not output special T1 chars.
1146 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1147 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1148 && latexSpecialT1(c, os, i, column))
1150 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1151 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1152 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1155 // Otherwise, we use what LaTeX provides us.
1158 os << "\\textbackslash" << termcmd;
1162 os << "\\textless" << termcmd;
1166 os << "\\textgreater" << termcmd;
1170 os << "\\textbar" << termcmd;
1175 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1176 (end_pos == -1 || i + 1 < end_pos) &&
1177 text_[i+1] == '-') {
1178 // Prevent "--" becoming an en dash and "---" an em dash.
1179 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1185 os << "\\textquotedbl" << termcmd;
1190 case '%': case '#': case '{':
1198 os << "\\textasciitilde" << termcmd;
1203 os << "\\textasciicircum" << termcmd;
1210 // avoid being mistaken for optional arguments
1218 // Blanks are printed before font switching.
1219 // Sure? I am not! (try nice-latex)
1220 // I am sure it's correct. LyX might be smarter
1221 // in the future, but for now, nothing wrong is
1227 // XeTeX's dash behaviour is determined via a global setting
1228 if (bparams.use_dash_ligatures
1229 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1231 // TODO #10961: && not in inset Flex Code
1232 // TODO #10961: && not in layout LyXCode
1233 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1250 Encoding const & encoding = *(runparams.encoding);
1251 char_type next = '\0';
1252 if (i + 1 < int(text_.size())) {
1253 next = text_[i + 1];
1254 if (Encodings::isCombiningChar(next)) {
1255 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1260 pair<docstring, bool> latex = encoding.latexChar(c);
1261 docstring nextlatex;
1262 bool nexttipas = false;
1263 string nexttipashortcut;
1264 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1265 nextlatex = encoding.latexChar(next).first;
1266 if (runparams.inIPA) {
1267 nexttipashortcut = Encodings::TIPAShortcut(next);
1268 nexttipas = !nexttipashortcut.empty();
1272 if (runparams.inIPA) {
1273 string const tipashortcut = Encodings::TIPAShortcut(c);
1274 if (!tipashortcut.empty()) {
1275 latex.first = from_ascii(tipashortcut);
1276 latex.second = false;
1280 // eventually close "script wrapper" command (see `Paragraph::latex`)
1281 if (!alien_script.empty()
1282 && alien_script != Encodings::isKnownScriptChar(next)) {
1283 column += latex.first.length();
1284 alien_script.clear();
1285 os << latex.first << "}";
1289 && ((!prefixIs(nextlatex, '\\')
1290 && !prefixIs(nextlatex, '{')
1291 && !prefixIs(nextlatex, '}'))
1293 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1295 // Prevent eating of a following space or command corruption by
1296 // following characters
1297 if (next == ' ' || next == '\0') {
1298 column += latex.first.length() + 1;
1299 os << latex.first << "{}";
1301 column += latex.first.length();
1302 os << latex.first << " ";
1305 column += latex.first.length() - 1;
1313 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1314 pos_type i, unsigned int & column)
1320 // In T1 encoding, these characters exist
1321 // but we should avoid ligatures
1322 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1324 os << "\\textcompwordmark" << termcmd;
1331 // soul.sty breaks with \char`\"
1332 os << "\\textquotedbl" << termcmd;
1341 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1342 pos_type i, unsigned int & column)
1344 // TU encoding is currently on par with T1.
1345 return latexSpecialT1(c, os, i, column);
1349 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1350 pos_type /*i*/, unsigned int & column)
1360 os << "\\textvertline" << termcmd;
1369 void Paragraph::Private::validate(LaTeXFeatures & features) const
1371 if (layout_->inpreamble && inset_owner_) {
1372 // FIXME: Using a string stream here circumvents the encoding
1373 // switching machinery of odocstream. Therefore the
1374 // output is wrong if this paragraph contains content
1375 // that needs to switch encoding.
1376 Buffer const & buf = inset_owner_->buffer();
1377 otexstringstream os;
1378 os << layout_->preamble();
1379 size_t const length = os.length();
1380 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1381 buf.getParFromID(owner_->id()).pit(), os,
1382 features.runparams(), string(), 0, -1, true);
1383 if (os.length() > length)
1384 features.addPreambleSnippet(os.release(), true);
1387 if (features.runparams().flavor == OutputParams::HTML
1388 && layout_->htmltitle()) {
1389 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1392 // check the params.
1393 if (!params_.spacing().isDefault())
1394 features.require("setspace");
1397 features.useLayout(layout_->name());
1400 fontlist_.validate(features);
1402 // then the indentation
1403 if (!params_.leftIndent().zero())
1404 features.require("ParagraphLeftIndent");
1407 InsetList::const_iterator icit = insetlist_.begin();
1408 InsetList::const_iterator iend = insetlist_.end();
1409 for (; icit != iend; ++icit) {
1411 features.inDeletedInset(owner_->isDeleted(icit->pos));
1412 if (icit->inset->lyxCode() == FOOT_CODE) {
1413 // FIXME: an item inset would make things much easier.
1414 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1415 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1416 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1417 && (icit->pos < begin_of_body_
1418 || (icit->pos == begin_of_body_
1419 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1420 features.saveNoteEnv("description");
1422 icit->inset->validate(features);
1423 features.inDeletedInset(false);
1424 if (layout_->needprotect &&
1425 icit->inset->lyxCode() == FOOT_CODE)
1426 features.require("NeedLyXFootnoteCode");
1430 // then the contents
1431 BufferParams const bp = features.runparams().is_child
1432 ? features.buffer().masterParams() : features.buffer().params();
1433 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1434 char_type c = text_[i];
1435 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1437 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1438 features.require("textquotedblp");
1439 else if (bp.main_font_encoding() != "T1"
1440 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1441 features.require("textquotedbl");
1442 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1443 // features that depend on the font or input encoding
1444 string feats = ci.textpreamble();
1445 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1446 if (fontenc.empty())
1447 fontenc = features.runparams().main_fontenc;
1448 while (!feats.empty()) {
1450 feats = split(feats, feat, ',');
1451 if (contains(feat, "!=")) {
1452 // a feature that is required except for the spcified
1453 // font or input encodings
1455 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1456 // multiple encodings are separated by semicolon
1457 vector<string> context = getVectorFromString(contexts, ";");
1458 // require feature if the context matches neither current font
1459 // nor input encoding
1460 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1461 && std::find(context.begin(), context.end(),
1462 features.runparams().encoding->name()) == context.end())
1463 features.require(realfeature);
1464 } else if (contains(feat, '=')) {
1465 // a feature that is required only for the spcified
1466 // font or input encodings
1468 string const contexts = split(feat, realfeature, '=');
1469 // multiple encodings are separated by semicolon
1470 vector<string> context = getVectorFromString(contexts, ";");
1471 // require feature if the context matches either current font
1472 // or input encoding
1473 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1474 || std::find(context.begin(), context.end(),
1475 features.runparams().encoding->name()) != context.end())
1476 features.require(realfeature);
1479 } else if (!bp.use_dash_ligatures
1480 && (c == 0x2013 || c == 0x2014)
1481 && bp.useNonTeXFonts
1482 && features.runparams().flavor == OutputParams::XETEX)
1483 // XeTeX's dash behaviour is determined via a global setting
1484 features.require("xetexdashbreakstate");
1485 BufferEncodings::validate(c, features);
1489 /////////////////////////////////////////////////////////////////////
1493 /////////////////////////////////////////////////////////////////////
1496 Layout const emptyParagraphLayout;
1499 Paragraph::Paragraph()
1500 : d(new Paragraph::Private(this, emptyParagraphLayout))
1507 Paragraph::Paragraph(Paragraph const & par)
1508 : itemdepth(par.itemdepth),
1509 d(new Paragraph::Private(*par.d, this))
1515 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1516 : itemdepth(par.itemdepth),
1517 d(new Paragraph::Private(*par.d, this, beg, end))
1523 Paragraph & Paragraph::operator=(Paragraph const & par)
1525 // needed as we will destroy the private part before copying it
1527 itemdepth = par.itemdepth;
1531 d = new Private(*par.d, this);
1538 Paragraph::~Paragraph()
1547 // this shall be called just before every "os << ..." action.
1548 void flushString(ostream & os, docstring & s)
1557 void Paragraph::write(ostream & os, BufferParams const & bparams,
1558 depth_type & dth) const
1560 // The beginning or end of a deeper (i.e. nested) area?
1561 if (dth != d->params_.depth()) {
1562 if (d->params_.depth() > dth) {
1563 while (d->params_.depth() > dth) {
1564 os << "\n\\begin_deeper";
1568 while (d->params_.depth() < dth) {
1569 os << "\n\\end_deeper";
1575 // First write the layout
1576 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1578 d->params_.write(os);
1580 Font font1(inherit_font, bparams.language);
1582 Change running_change = Change(Change::UNCHANGED);
1584 // this string is used as a buffer to avoid repetitive calls
1585 // to to_utf8(), which turn out to be expensive (JMarc)
1586 docstring write_buffer;
1589 for (pos_type i = 0; i <= size(); ++i) {
1591 Change const & change = lookupChange(i);
1592 if (change != running_change)
1593 flushString(os, write_buffer);
1594 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1595 running_change = change;
1600 // Write font changes
1601 Font font2 = getFontSettings(bparams, i);
1602 if (font2 != font1) {
1603 flushString(os, write_buffer);
1604 font2.lyxWriteChanges(font1, os);
1609 char_type const c = d->text_[i];
1612 if (Inset const * inset = getInset(i)) {
1613 flushString(os, write_buffer);
1614 if (inset->directWrite()) {
1615 // international char, let it write
1616 // code directly so it's shorter in
1622 os << "\\begin_inset ";
1624 os << "\n\\end_inset\n\n";
1627 // FIXME This can be removed again once the mystery
1628 // crash has been resolved.
1633 flushString(os, write_buffer);
1634 os << "\n\\backslash\n";
1638 flushString(os, write_buffer);
1639 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1646 if ((column > 70 && c == ' ')
1648 flushString(os, write_buffer);
1652 // this check is to amend a bug. LyX sometimes
1653 // inserts '\0' this could cause problems.
1655 write_buffer.push_back(c);
1657 LYXERR0("NUL char in structure.");
1663 flushString(os, write_buffer);
1664 os << "\n\\end_layout\n";
1665 // FIXME This can be removed again once the mystery
1666 // crash has been resolved.
1671 void Paragraph::validate(LaTeXFeatures & features) const
1673 d->validate(features);
1674 bool fragile = features.runparams().moving_arg;
1675 fragile |= layout().needprotect;
1676 if (inInset().getLayout().isNeedProtect())
1678 if (needsCProtection(fragile))
1679 features.require("cprotect");
1683 void Paragraph::insert(pos_type start, docstring const & str,
1684 Font const & font, Change const & change)
1686 for (size_t i = 0, n = str.size(); i != n ; ++i)
1687 insertChar(start + i, str[i], font, change);
1691 void Paragraph::appendChar(char_type c, Font const & font,
1692 Change const & change)
1695 d->changes_.insert(change, d->text_.size());
1696 // when appending characters, no need to update tables
1697 d->text_.push_back(c);
1698 setFont(d->text_.size() - 1, font);
1699 d->requestSpellCheck(d->text_.size() - 1);
1703 void Paragraph::appendString(docstring const & s, Font const & font,
1704 Change const & change)
1706 pos_type end = s.size();
1707 size_t oldsize = d->text_.size();
1708 size_t newsize = oldsize + end;
1709 size_t capacity = d->text_.capacity();
1710 if (newsize >= capacity)
1711 d->text_.reserve(max(capacity + 100, newsize));
1713 // when appending characters, no need to update tables
1716 // FIXME: Optimize this!
1717 for (size_t i = oldsize; i != newsize; ++i) {
1719 d->changes_.insert(change, i);
1720 d->requestSpellCheck(i);
1722 d->fontlist_.set(oldsize, font);
1723 d->fontlist_.set(newsize - 1, font);
1727 void Paragraph::insertChar(pos_type pos, char_type c,
1730 d->insertChar(pos, c, Change(trackChanges ?
1731 Change::INSERTED : Change::UNCHANGED));
1735 void Paragraph::insertChar(pos_type pos, char_type c,
1736 Font const & font, bool trackChanges)
1738 d->insertChar(pos, c, Change(trackChanges ?
1739 Change::INSERTED : Change::UNCHANGED));
1744 void Paragraph::insertChar(pos_type pos, char_type c,
1745 Font const & font, Change const & change)
1747 d->insertChar(pos, c, change);
1752 void Paragraph::resetFonts(Font const & font)
1754 d->fontlist_.clear();
1755 d->fontlist_.set(0, font);
1756 d->fontlist_.set(d->text_.size() - 1, font);
1759 // Gets uninstantiated font setting at position.
1760 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1764 LYXERR0("pos: " << pos << " size: " << size());
1768 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1769 if (cit != d->fontlist_.end())
1772 if (pos == size() && !empty())
1773 return getFontSettings(bparams, pos - 1);
1775 // Optimisation: avoid a full font instantiation if there is no
1776 // language change from previous call.
1777 static Font previous_font;
1778 static Language const * previous_lang = nullptr;
1779 Language const * lang = getParLanguage(bparams);
1780 if (lang != previous_lang) {
1781 previous_lang = lang;
1782 previous_font = Font(inherit_font, lang);
1784 return previous_font;
1788 FontSpan Paragraph::fontSpan(pos_type pos) const
1790 LBUFERR(pos <= size());
1793 return FontSpan(pos, pos);
1796 FontList::const_iterator cit = d->fontlist_.begin();
1797 FontList::const_iterator end = d->fontlist_.end();
1798 for (; cit != end; ++cit) {
1799 if (cit->pos() >= pos) {
1800 if (pos >= beginOfBody())
1801 return FontSpan(max(start, beginOfBody()),
1804 return FontSpan(start,
1805 min(beginOfBody() - 1,
1808 start = cit->pos() + 1;
1811 // This should not happen, but if so, we take no chances.
1812 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1813 LASSERT(false, return FontSpan(pos, pos));
1817 // Gets uninstantiated font setting at position 0
1818 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1820 if (!empty() && !d->fontlist_.empty())
1821 return d->fontlist_.begin()->font();
1823 // Optimisation: avoid a full font instantiation if there is no
1824 // language change from previous call.
1825 static Font previous_font;
1826 static Language const * previous_lang = nullptr;
1827 if (bparams.language != previous_lang) {
1828 previous_lang = bparams.language;
1829 previous_font = Font(inherit_font, bparams.language);
1832 return previous_font;
1836 // Gets the fully instantiated font at a given position in a paragraph
1837 // This is basically the same function as Text::GetFont() in text2.cpp.
1838 // The difference is that this one is used for generating the LaTeX file,
1839 // and thus cosmetic "improvements" are disallowed: This has to deliver
1840 // the true picture of the buffer. (Asger)
1841 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1842 Font const & outerfont) const
1846 Font font = getFontSettings(bparams, pos);
1848 pos_type const body_pos = beginOfBody();
1849 FontInfo & fi = font.fontInfo();
1851 fi.realize(d->layout_->labelfont);
1853 fi.realize(d->layout_->font);
1855 fi.realize(outerfont.fontInfo());
1856 fi.realize(bparams.getFont().fontInfo());
1862 Font const Paragraph::getLabelFont
1863 (BufferParams const & bparams, Font const & outerfont) const
1865 FontInfo tmpfont = d->layout_->labelfont;
1866 tmpfont.realize(outerfont.fontInfo());
1867 tmpfont.realize(bparams.getFont().fontInfo());
1868 return Font(tmpfont, getParLanguage(bparams));
1872 Font const Paragraph::getLayoutFont
1873 (BufferParams const & bparams, Font const & outerfont) const
1875 FontInfo tmpfont = d->layout_->font;
1876 tmpfont.realize(outerfont.fontInfo());
1877 tmpfont.realize(bparams.getFont().fontInfo());
1878 return Font(tmpfont, getParLanguage(bparams));
1882 char_type Paragraph::getUChar(BufferParams const & bparams,
1883 OutputParams const & rp,
1886 char_type c = d->text_[pos];
1888 // Return unchanged character in LTR languages
1889 // or if we use poylglossia/bidi (XeTeX).
1890 if (rp.useBidiPackage()
1891 || !getFontSettings(bparams, pos).isRightToLeft())
1894 // Without polyglossia/bidi, we need to account for some special cases.
1895 // FIXME This needs to be audited!
1897 // * The input is as expected for all delimiters
1898 // => checked for Hebrew!
1899 // * The output matches the display in the LyX workarea
1900 // => checked for Hebrew!
1901 // * The special cases below are really necessary
1902 // => checked for Hebrew!
1903 // * In arabic_arabi, brackets are transformed to Arabic
1904 // Ornate Parentheses. Is this is really wanted?
1906 string const & lang = getFontSettings(bparams, pos).language()->lang();
1909 // 1. In the following languages, parentheses need to be reversed.
1910 // Also with polyglodia/luabidi
1911 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1913 // 2. In the following languages, brackets don't need to be reversed.
1914 bool const reversebrackets = lang != "arabic_arabtex"
1915 && lang != "arabic_arabi"
1918 // Now swap delimiters if needed.
1929 if (reversebrackets)
1933 if (reversebrackets)
1954 void Paragraph::setFont(pos_type pos, Font const & font)
1956 LASSERT(pos <= size(), return);
1958 // First, reduce font against layout/label font
1959 // Update: The setCharFont() routine in text2.cpp already
1960 // reduces font, so we don't need to do that here. (Asger)
1962 d->fontlist_.set(pos, font);
1966 void Paragraph::makeSameLayout(Paragraph const & par)
1968 d->layout_ = par.d->layout_;
1969 d->params_ = par.d->params_;
1973 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1975 if (isFreeSpacing())
1981 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1982 if (eraseChar(pos, trackChanges))
1988 return count > 0 || pos > 0;
1992 bool Paragraph::hasSameLayout(Paragraph const & par) const
1994 return par.d->layout_ == d->layout_
1995 && d->params_.sameLayout(par.d->params_);
1999 depth_type Paragraph::getDepth() const
2001 return d->params_.depth();
2005 depth_type Paragraph::getMaxDepthAfter() const
2007 if (d->layout_->isEnvironment())
2008 return d->params_.depth() + 1;
2010 return d->params_.depth();
2014 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2016 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2017 return getDefaultAlign(bparams);
2019 return d->params_.align();
2023 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2025 LyXAlignment res = layout().align;
2026 if (isRTL(bparams)) {
2028 if (res == LYX_ALIGN_LEFT)
2029 res = LYX_ALIGN_RIGHT;
2030 else if (res == LYX_ALIGN_RIGHT)
2031 res = LYX_ALIGN_LEFT;
2037 docstring const & Paragraph::labelString() const
2039 return d->params_.labelString();
2043 // the next two functions are for the manual labels
2044 docstring const Paragraph::getLabelWidthString() const
2046 if (d->layout_->margintype == MARGIN_MANUAL
2047 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2048 return d->params_.labelWidthString();
2050 return _("Senseless with this layout!");
2054 void Paragraph::setLabelWidthString(docstring const & s)
2056 d->params_.labelWidthString(s);
2060 docstring Paragraph::expandLabel(Layout const & layout,
2061 BufferParams const & bparams) const
2063 return expandParagraphLabel(layout, bparams, true);
2067 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2068 BufferParams const & bparams, bool process_appendix) const
2070 DocumentClass const & tclass = bparams.documentClass();
2071 string const & lang = getParLanguage(bparams)->code();
2072 bool const in_appendix = process_appendix && d->params_.appendix();
2073 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2075 if (fmt.empty() && !layout.counter.empty())
2076 return tclass.counters().theCounter(layout.counter, lang);
2078 // handle 'inherited level parts' in 'fmt',
2079 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2080 size_t const i = fmt.find('@', 0);
2081 if (i != docstring::npos) {
2082 size_t const j = fmt.find('@', i + 1);
2083 if (j != docstring::npos) {
2084 docstring parent(fmt, i + 1, j - i - 1);
2085 docstring label = from_ascii("??");
2086 if (tclass.hasLayout(parent))
2087 label = expandParagraphLabel(tclass[parent], bparams,
2089 fmt = docstring(fmt, 0, i) + label
2090 + docstring(fmt, j + 1, docstring::npos);
2094 return tclass.counters().counterLabel(fmt, lang);
2098 void Paragraph::applyLayout(Layout const & new_layout)
2100 d->layout_ = &new_layout;
2101 LyXAlignment const oldAlign = d->params_.align();
2103 if (!(oldAlign & d->layout_->alignpossible)) {
2104 frontend::Alert::warning(_("Alignment not permitted"),
2105 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2106 d->params_.align(LYX_ALIGN_LAYOUT);
2111 pos_type Paragraph::beginOfBody() const
2113 return d->begin_of_body_;
2117 void Paragraph::setBeginOfBody()
2119 if (d->layout_->labeltype != LABEL_MANUAL) {
2120 d->begin_of_body_ = 0;
2124 // Unroll the first two cycles of the loop
2125 // and remember the previous character to
2126 // remove unnecessary getChar() calls
2128 pos_type end = size();
2129 bool prev_char_deleted = false;
2130 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2133 char_type previous_char = d->text_[i];
2134 if (!(isNewline(i) || isEnvSeparator(i))) {
2136 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2137 char_type temp = d->text_[i];
2138 prev_char_deleted = isDeleted(i);
2139 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2142 previous_char = temp;
2148 d->begin_of_body_ = i;
2152 bool Paragraph::allowParagraphCustomization() const
2154 return inInset().allowParagraphCustomization();
2158 bool Paragraph::usePlainLayout() const
2160 return inInset().usePlainLayout();
2164 bool Paragraph::isPassThru() const
2166 return inInset().isPassThru() || d->layout_->pass_thru;
2171 // paragraphs inside floats need different alignment tags to avoid
2174 bool noTrivlistCentering(InsetCode code)
2176 return code == FLOAT_CODE
2177 || code == WRAP_CODE
2178 || code == CELL_CODE;
2182 string correction(string const & orig)
2184 if (orig == "flushleft")
2185 return "raggedright";
2186 if (orig == "flushright")
2187 return "raggedleft";
2188 if (orig == "center")
2194 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2195 InsetCode code, bool const lastpar, int & col)
2197 string macro = suffix + "{";
2198 if (noTrivlistCentering(code)) {
2200 // the last paragraph in non-trivlist-aligned
2201 // context is special (to avoid unwanted whitespace)
2202 if (suffix == "\\begin") {
2203 macro = "\\" + correction(env) + "{}";
2204 os << from_ascii(macro);
2205 col += macro.size();
2210 macro += correction(env);
2214 if (suffix == "\\par\\end") {
2218 os << from_ascii(macro);
2219 col += macro.size();
2220 if (suffix == "\\begin") {
2230 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2231 otexstream & os, OutputParams const & runparams) const
2236 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2237 (layout_->toggle_indent != ITOGGLE_NEVER) :
2238 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2240 if (canindent && params_.noindent() && !layout_->pass_thru) {
2241 os << "\\noindent ";
2245 LyXAlignment const curAlign = params_.align();
2247 if (curAlign == layout_->align)
2251 case LYX_ALIGN_NONE:
2252 case LYX_ALIGN_BLOCK:
2253 case LYX_ALIGN_LAYOUT:
2254 case LYX_ALIGN_SPECIAL:
2255 case LYX_ALIGN_DECIMAL:
2257 case LYX_ALIGN_LEFT:
2258 case LYX_ALIGN_RIGHT:
2259 case LYX_ALIGN_CENTER:
2260 if (runparams.moving_arg) {
2267 string const begin_tag = "\\begin";
2268 InsetCode code = ownerCode();
2269 bool const lastpar = runparams.isLastPar;
2270 // RTL in classic (PDF)LaTeX (without the Bidi package)
2271 // Luabibdi (used by LuaTeX) behaves like classic
2272 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2273 && !runparams.useBidiPackage();
2276 case LYX_ALIGN_NONE:
2277 case LYX_ALIGN_BLOCK:
2278 case LYX_ALIGN_LAYOUT:
2279 case LYX_ALIGN_SPECIAL:
2280 case LYX_ALIGN_DECIMAL:
2282 case LYX_ALIGN_LEFT: {
2284 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2285 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2287 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2289 } case LYX_ALIGN_RIGHT: {
2291 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2292 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2294 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2296 } case LYX_ALIGN_CENTER: {
2297 corrected_env(os, begin_tag, "center", code, lastpar, column);
2306 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2307 otexstream & os, OutputParams const & runparams) const
2309 LyXAlignment const curAlign = params_.align();
2311 if (curAlign == layout_->align)
2315 case LYX_ALIGN_NONE:
2316 case LYX_ALIGN_BLOCK:
2317 case LYX_ALIGN_LAYOUT:
2318 case LYX_ALIGN_SPECIAL:
2319 case LYX_ALIGN_DECIMAL:
2321 case LYX_ALIGN_LEFT:
2322 case LYX_ALIGN_RIGHT:
2323 case LYX_ALIGN_CENTER:
2324 if (runparams.moving_arg)
2329 bool output = false;
2331 string const end_tag = "\\par\\end";
2332 InsetCode code = ownerCode();
2333 bool const lastpar = runparams.isLastPar;
2334 // RTL in classic (PDF)LaTeX (without the Bidi package)
2335 // Luabibdi (used by LuaTeX) behaves like classic
2336 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2337 && !runparams.useBidiPackage();
2340 case LYX_ALIGN_NONE:
2341 case LYX_ALIGN_BLOCK:
2342 case LYX_ALIGN_LAYOUT:
2343 case LYX_ALIGN_SPECIAL:
2344 case LYX_ALIGN_DECIMAL:
2346 case LYX_ALIGN_LEFT: {
2348 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2349 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2351 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2353 } case LYX_ALIGN_RIGHT: {
2355 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2356 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2358 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2360 } case LYX_ALIGN_CENTER: {
2361 corrected_env(os, end_tag, "center", code, lastpar, col);
2366 return output || lastpar;
2370 // This one spits out the text of the paragraph
2371 void Paragraph::latex(BufferParams const & bparams,
2372 Font const & outerfont,
2374 OutputParams const & runparams,
2375 int start_pos, int end_pos, bool force) const
2377 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2379 // FIXME This check should not be needed. Perhaps issue an
2380 // error if it triggers.
2381 Layout const & style = inInset().forcePlainLayout() ?
2382 bparams.documentClass().plainLayout() : *d->layout_;
2384 if (!force && style.inpreamble)
2387 bool const allowcust = allowParagraphCustomization();
2389 // Current base font for all inherited font changes, without any
2390 // change caused by an individual character, except for the language:
2391 // It is set to the language of the first character.
2392 // As long as we are in the label, this font is the base font of the
2393 // label. Before the first body character it is set to the base font
2397 // If there is an open font-encoding changing command (script wrapper),
2398 // alien_script is set to its name
2399 string alien_script;
2402 // Maybe we have to create a optional argument.
2403 pos_type body_pos = beginOfBody();
2404 unsigned int column = 0;
2406 // If we are inside an inset, the real outerfont is local_font
2407 Font const real_outerfont = (runparams.local_font != nullptr)
2408 ? Font(runparams.local_font->fontInfo()) : outerfont;
2411 // the optional argument is kept in curly brackets in
2412 // case it contains a ']'
2413 // This is not strictly needed, but if this is changed it
2414 // would be a file format change, and tex2lyx would need
2415 // to be adjusted, since it unconditionally removes the
2416 // braces when it parses \item.
2419 basefont = getLabelFont(bparams, real_outerfont);
2421 basefont = getLayoutFont(bparams, real_outerfont);
2424 // Which font is currently active?
2425 Font running_font(basefont);
2426 // Do we have an open font change?
2427 bool open_font = false;
2429 Change runningChange = Change(Change::UNCHANGED);
2431 Encoding const * const prev_encoding = runparams.encoding;
2433 os.texrow().start(id(), 0);
2435 // if the paragraph is empty, the loop will not be entered at all
2437 // For InTitle commands, we have already opened a group
2438 // in output_latex::TeXOnePar.
2439 if (style.isCommand() && !style.intitle) {
2443 if (!style.leftdelim().empty()) {
2444 os << style.leftdelim();
2445 column += style.leftdelim().size();
2448 column += d->startTeXParParams(bparams, os, runparams);
2451 // Whether a \par can be issued for insets typeset inline with text.
2452 // Yes if greater than 0. This has to be static.
2453 THREAD_LOCAL_STATIC int parInline = 0;
2455 for (pos_type i = 0; i < size(); ++i) {
2456 // First char in paragraph or after label?
2457 if (i == body_pos) {
2460 bool needPar = false;
2461 column += running_font.latexWriteEndChanges(
2462 os, bparams, runparams,
2463 basefont, basefont, needPar);
2466 basefont = getLayoutFont(bparams, real_outerfont);
2467 running_font = basefont;
2469 column += Changes::latexMarkChange(os, bparams,
2470 runningChange, Change(Change::UNCHANGED),
2472 runningChange = Change(Change::UNCHANGED);
2477 // For InTitle commands, we have already opened a group
2478 // in output_latex::TeXOnePar.
2479 if (style.isCommand() && !style.intitle) {
2484 if (!style.leftdelim().empty()) {
2485 os << style.leftdelim();
2486 column += style.leftdelim().size();
2490 column += d->startTeXParParams(bparams, os,
2494 runparams.wasDisplayMath = runparams.inDisplayMath;
2495 runparams.inDisplayMath = false;
2496 bool deleted_display_math = false;
2497 Change const & change = runparams.inDeletedInset
2498 ? runparams.changeOfDeletedInset : lookupChange(i);
2500 char_type const c = d->text_[i];
2502 // Check whether a display math inset follows
2504 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2506 runparams.ctObject = getInset(i)->CtObject(runparams);
2508 InsetMath const * im = getInset(i)->asInsetMath();
2509 if (im && im->asHullInset()
2510 && im->asHullInset()->outerDisplay()) {
2511 runparams.inDisplayMath = true;
2512 // runparams.inDeletedInset will be set by
2513 // latexInset later, but we need this info
2514 // before it is called. On the other hand, we
2515 // cannot set it here because it is a counter.
2516 deleted_display_math = isDeleted(i);
2518 if (bparams.output_changes && deleted_display_math
2519 && runningChange == change
2520 && change.type == Change::DELETED
2521 && !os.afterParbreak()) {
2522 // A display math in the same paragraph follows.
2523 // We have to close and then reopen \lyxdeleted,
2524 // otherwise the math will be shifted up.
2525 OutputParams rp = runparams;
2527 bool needPar = false;
2528 column += running_font.latexWriteEndChanges(
2529 os, bparams, rp, basefont,
2533 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2534 : getLayoutFont(bparams, real_outerfont);
2535 running_font = basefont;
2536 column += Changes::latexMarkChange(os, bparams,
2537 Change(Change::INSERTED), change, rp);
2541 if (bparams.output_changes && runningChange != change) {
2542 if (!alien_script.empty()) {
2545 alien_script.clear();
2548 bool needPar = false;
2549 column += running_font.latexWriteEndChanges(
2550 os, bparams, runparams,
2551 basefont, basefont, needPar);
2554 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2555 : getLayoutFont(bparams, real_outerfont);
2556 running_font = basefont;
2557 column += Changes::latexMarkChange(os, bparams, runningChange,
2559 runningChange = change;
2562 // do not output text which is marked deleted
2563 // if change tracking output is disabled
2564 if (!bparams.output_changes && change.deleted()) {
2570 // Fully instantiated font
2571 Font const current_font = getFont(bparams, i, real_outerfont);
2573 Font const last_font = running_font;
2574 bool const in_ct_deletion = (bparams.output_changes
2575 && runningChange == change
2576 && change.type == Change::DELETED
2577 && !os.afterParbreak());
2579 // Do we need to close the previous font?
2581 (current_font != running_font ||
2582 current_font.language() != running_font.language()))
2584 // ensure there is no open script-wrapper
2585 if (!alien_script.empty()) {
2588 alien_script.clear();
2590 bool needPar = false;
2591 if (in_ct_deletion) {
2592 // We have to close and then reopen \lyxdeleted,
2593 // as strikeout needs to be on lowest level.
2597 column += running_font.latexWriteEndChanges(
2598 os, bparams, runparams, basefont,
2599 (i == body_pos-1) ? basefont : current_font,
2601 if (in_ct_deletion) {
2602 // We have to close and then reopen \lyxdeleted,
2603 // as strikeout needs to be on lowest level.
2604 OutputParams rp = runparams;
2605 column += Changes::latexMarkChange(os, bparams,
2606 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2608 running_font = basefont;
2612 // if necessary, close language environment before opening CJK
2613 string const running_lang = running_font.language()->babel();
2614 string const lang_end_command = lyxrc.language_command_end;
2615 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2616 && !running_lang.empty()
2617 && running_lang == openLanguageName()
2618 && current_font.language()->encoding()->package() == Encoding::CJK) {
2619 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2620 os << from_ascii(end_tag);
2621 column += end_tag.length();
2625 // Switch file encoding if necessary (and allowed)
2626 if (!runparams.pass_thru && !style.pass_thru &&
2627 runparams.encoding->package() != Encoding::none &&
2628 current_font.language()->encoding()->package() != Encoding::none) {
2629 pair<bool, int> const enc_switch =
2630 switchEncoding(os.os(), bparams, runparams,
2631 *(current_font.language()->encoding()));
2632 if (enc_switch.first) {
2633 column += enc_switch.second;
2634 runparams.encoding = current_font.language()->encoding();
2638 // A display math inset inside an ulem command will be output
2639 // as a box of width \linewidth, so we have to either disable
2640 // indentation if the inset starts a paragraph, or start a new
2641 // line to accommodate such box. This has to be done before
2642 // writing any font changing commands.
2643 if (runparams.inDisplayMath && !deleted_display_math
2644 && runparams.inulemcmd) {
2645 if (os.afterParbreak())
2651 // Do we need to change font?
2652 if ((current_font != running_font ||
2653 current_font.language() != running_font.language())
2654 && i != body_pos - 1)
2656 if (in_ct_deletion) {
2657 // We have to close and then reopen \lyxdeleted,
2658 // as strikeout needs to be on lowest level.
2659 bool needPar = false;
2660 OutputParams rp = runparams;
2661 column += running_font.latexWriteEndChanges(
2662 os, bparams, rp, basefont,
2667 otexstringstream ots;
2668 bool const non_inherit_inset = (c == META_INSET && getInset(i) && !getInset(i)->inheritFont());
2669 column += current_font.latexWriteStartChanges(ots, bparams,
2670 runparams, basefont,
2671 last_font, non_inherit_inset);
2672 // Check again for display math in ulem commands as a
2673 // font change may also occur just before a math inset.
2674 if (runparams.inDisplayMath && !deleted_display_math
2675 && runparams.inulemcmd) {
2676 if (os.afterParbreak())
2681 running_font = current_font;
2683 docstring fontchange = ots.str();
2685 // check whether the fontchange ends with a \\textcolor
2686 // modifier and the text starts with a space. If so we
2687 // need to add } in order to prevent \\textcolor from gobbling
2688 // the space (bug 4473).
2689 docstring const last_modifier = rsplit(fontchange, '\\');
2690 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2691 os << from_ascii("{}");
2692 else if (ots.terminateCommand())
2694 if (in_ct_deletion) {
2695 // We have to close and then reopen \lyxdeleted,
2696 // as strikeout needs to be on lowest level.
2697 OutputParams rp = runparams;
2698 column += Changes::latexMarkChange(os, bparams,
2699 Change(Change::UNCHANGED), change, rp);
2703 // FIXME: think about end_pos implementation...
2704 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2705 // FIXME: integrate this case in latexSpecialChar
2706 // Do not print the separation of the optional argument
2707 // if style.pass_thru is false. This works because
2708 // latexSpecialChar ignores spaces if
2709 // style.pass_thru is false.
2710 if (i != body_pos - 1) {
2711 if (d->simpleTeXBlanks(bparams, runparams, os,
2712 i, column, current_font, style)) {
2713 // A surrogate pair was output. We
2714 // must not call latexSpecialChar
2715 // in this iteration, since it would output
2716 // the combining character again.
2723 OutputParams rp = runparams;
2724 rp.free_spacing = style.free_spacing;
2725 rp.local_font = ¤t_font;
2726 rp.intitle = style.intitle;
2728 // Two major modes: LaTeX or plain
2729 // Handle here those cases common to both modes
2730 // and then split to handle the two modes separately.
2731 if (c == META_INSET) {
2732 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2733 // Greyedout notes and, in general, all insets
2734 // with InsetLayout::isDisplay() == false,
2735 // are typeset inline with the text. So, we
2736 // can add a \par to the last paragraph of
2737 // such insets only if nothing else follows.
2738 bool incremented = false;
2739 Inset const * inset = getInset(i);
2740 InsetText const * textinset = inset
2741 ? inset->asInsetText()
2743 if (i + 1 == size() && textinset
2744 && !inset->getLayout().isDisplay()) {
2745 ParagraphList const & pars =
2746 textinset->text().paragraphs();
2747 pit_type const pit = pars.size() - 1;
2748 Font const lastfont =
2749 pit < 0 || pars[pit].empty()
2750 ? pars[pit].getLayoutFont(
2753 : pars[pit].getFont(bparams,
2754 pars[pit].size() - 1,
2756 if (lastfont.fontInfo().size() !=
2757 basefont.fontInfo().size()) {
2762 d->latexInset(bparams, os, rp, running_font,
2763 basefont, real_outerfont, open_font,
2764 runningChange, style, i, column);
2768 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2769 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2770 // Close \lyx*deleted and force its
2771 // reopening (if needed)
2774 runningChange = Change(Change::UNCHANGED);
2775 runparams.ctObject = OutputParams::CT_NORMAL;
2778 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2779 if (!bparams.useNonTeXFonts)
2780 script = Encodings::isKnownScriptChar(c);
2781 if (script != alien_script) {
2782 if (!alien_script.empty()) {
2784 alien_script.clear();
2786 string fontenc = running_font.language()->fontenc(bparams);
2788 && !Encodings::fontencSupportsScript(fontenc, script)) {
2789 column += script.length() + 2;
2790 os << "\\" << script << "{";
2791 alien_script = script;
2795 d->latexSpecialChar(os, bparams, rp, running_font,
2796 alien_script, style, i, end_pos, column);
2797 } catch (EncodingException & e) {
2798 if (runparams.dryrun) {
2799 os << "<" << _("LyX Warning: ")
2800 << _("uncodable character") << " '";
2804 // add location information and throw again.
2812 // Set the encoding to that returned from latexSpecialChar (see
2813 // comment for encoding member in OutputParams.h)
2814 runparams.encoding = rp.encoding;
2816 // Also carry on the info on a closed ulem command for insets
2817 // such as Note that do not produce any output, so that no
2818 // command is ever executed but its opening was recorded.
2819 runparams.inulemcmd = rp.inulemcmd;
2821 // These need to be passed upstream as well
2822 runparams.need_maketitle = rp.need_maketitle;
2823 runparams.have_maketitle = rp.have_maketitle;
2825 // And finally, pass the post_macros upstream
2826 runparams.post_macro = rp.post_macro;
2829 // Close wrapper for alien script
2830 if (!alien_script.empty()) {
2832 alien_script.clear();
2835 // If we have an open font definition, we have to close it
2837 // Make sure that \\par is done with the font of the last
2838 // character if this has another size as the default.
2839 // This is necessary because LaTeX (and LyX on the screen)
2840 // calculates the space between the baselines according
2841 // to this font. (Matthias)
2843 // We must not change the font for the last paragraph
2844 // of non-multipar insets, tabular cells or commands,
2845 // since this produces unwanted whitespace.
2847 Font const font = empty()
2848 ? getLayoutFont(bparams, real_outerfont)
2849 : getFont(bparams, size() - 1, real_outerfont);
2851 InsetText const * textinset = inInset().asInsetText();
2853 bool const maintext = textinset
2854 ? textinset->text().isMainText()
2857 size_t const numpars = textinset
2858 ? textinset->text().paragraphs().size()
2861 bool needPar = false;
2863 if (style.resfont.size() != font.fontInfo().size()
2864 && (!runparams.isLastPar || maintext
2865 || (numpars > 1 && d->ownerCode() != CELL_CODE
2866 && (inInset().getLayout().isDisplay()
2868 && !style.isCommand()) {
2871 #ifdef FIXED_LANGUAGE_END_DETECTION
2873 running_font.latexWriteEndChanges(os, bparams,
2874 runparams, basefont,
2875 next_->getFont(bparams, 0, outerfont),
2878 running_font.latexWriteEndChanges(os, bparams,
2879 runparams, basefont, basefont, needPar);
2882 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2883 //FIXME: there as we start another \selectlanguage with the next paragraph if
2884 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2885 running_font.latexWriteEndChanges(os, bparams, runparams,
2886 basefont, basefont, needPar);
2889 // The \par could not be inserted at the same nesting
2890 // level of the font size change, so do it now.
2891 os << "{\\" << font.latexSize() << "\\par}";
2895 column += Changes::latexMarkChange(os, bparams, runningChange,
2896 Change(Change::UNCHANGED), runparams);
2898 // Needed if there is an optional argument but no contents.
2899 if (body_pos > 0 && body_pos == size()) {
2903 if (!style.rightdelim().empty()) {
2904 os << style.rightdelim();
2905 column += style.rightdelim().size();
2908 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2909 && runparams.encoding != prev_encoding) {
2910 runparams.encoding = prev_encoding;
2911 os << setEncoding(prev_encoding->iconvName());
2914 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2918 bool Paragraph::emptyTag() const
2920 for (pos_type i = 0; i < size(); ++i) {
2921 if (Inset const * inset = getInset(i)) {
2922 InsetCode lyx_code = inset->lyxCode();
2923 // FIXME testing like that is wrong. What is
2925 if (lyx_code != TOC_CODE &&
2926 lyx_code != INCLUDE_CODE &&
2927 lyx_code != GRAPHICS_CODE &&
2928 lyx_code != ERT_CODE &&
2929 lyx_code != LISTINGS_CODE &&
2930 lyx_code != FLOAT_CODE &&
2931 lyx_code != TABULAR_CODE) {
2935 char_type c = d->text_[i];
2936 if (c != ' ' && c != '\t')
2944 string Paragraph::getID(Buffer const &, OutputParams const &)
2947 for (pos_type i = 0; i < size(); ++i) {
2948 if (Inset const * inset = getInset(i)) {
2949 InsetCode lyx_code = inset->lyxCode();
2950 if (lyx_code == LABEL_CODE) {
2951 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2952 docstring const & id = il->getParam("name");
2953 return "id='" + to_utf8(xml::cleanID(id)) + "'";
2961 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
2964 for (i = 0; i < size(); ++i) {
2965 if (Inset const * inset = getInset(i)) {
2966 inset->docbook(xs, runparams);
2968 char_type c = d->text_[i];
2978 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
2982 for (i = 0; i < size(); ++i) {
2983 if (Inset const * inset = getInset(i)) {
2984 inset->xhtml(xs, runparams);
2986 char_type c = d->text_[i];
2996 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2999 pos_type size = text_.size();
3000 for (pos_type i = initial; i < size; ++i) {
3001 Font font = owner_->getFont(buf.params(), i, outerfont);
3002 if (text_[i] == META_INSET)
3004 if (i != initial && font != font_old)
3015 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3016 vector<xml::EndFontTag> & tagsToClose,
3017 bool & flag, FontState curstate, xml::FontTypes type)
3019 if (curstate == FONT_ON) {
3020 tagsToOpen.push_back(docbookStartFontTag(type));
3023 tagsToClose.push_back(docbookEndFontTag(type));
3028 class OptionalFontType {
3033 OptionalFontType(): has_value(false), ft(xml::FT_EMPH) {} // A possible value at random for ft.
3034 OptionalFontType(xml::FontTypes ft): has_value(true), ft(ft) {}
3037 OptionalFontType fontShapeToXml(FontShape fs)
3041 return {xml::FT_ITALIC};
3043 return {xml::FT_SLANTED};
3044 case SMALLCAPS_SHAPE:
3045 return {xml::FT_SMALLCAPS};
3050 // the other tags are for internal use
3056 OptionalFontType fontFamilyToXml(FontFamily fm)
3060 return {xml::FT_ROMAN};
3063 return {xml::FT_SANS};
3065 case TYPEWRITER_FAMILY:
3066 return {xml::FT_TYPE};
3068 case INHERIT_FAMILY:
3071 // the other tags are for internal use
3077 OptionalFontType fontSizeToXml(FontSize fs)
3081 return {xml::FT_SIZE_TINY};
3083 return {xml::FT_SIZE_SCRIPT};
3085 return {xml::FT_SIZE_FOOTNOTE};
3087 return {xml::FT_SIZE_SMALL};
3089 return {xml::FT_SIZE_LARGE};
3091 return {xml::FT_SIZE_LARGER};
3093 return {xml::FT_SIZE_LARGEST};
3095 return {xml::FT_SIZE_HUGE};
3097 return {xml::FT_SIZE_HUGER};
3099 return {xml::FT_SIZE_INCREASE};
3101 return {xml::FT_SIZE_DECREASE};
3106 // the other tags are for internal use
3112 }// anonymous namespace
3115 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3117 OutputParams const & runparams,
3118 Font const & outerfont,
3119 bool start_paragraph, bool close_paragraph,
3120 pos_type initial) const
3122 // track whether we have opened these tags
3123 bool emph_flag = false;
3124 bool bold_flag = false;
3125 bool noun_flag = false;
3126 bool ubar_flag = false;
3127 bool dbar_flag = false;
3128 bool sout_flag = false;
3129 bool wave_flag = false;
3131 bool shap_flag = false;
3133 bool faml_flag = false;
3135 bool size_flag = false;
3137 Layout const & style = *d->layout_;
3139 if (start_paragraph)
3140 xs.startDivision(allowEmpty());
3143 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3145 FontShape curr_fs = INHERIT_SHAPE;
3146 FontFamily curr_fam = INHERIT_FAMILY;
3147 FontSize curr_size = INHERIT_SIZE;
3149 string const default_family =
3150 buf.masterBuffer()->params().fonts_default_family;
3152 vector<xml::FontTag> tagsToOpen;
3153 vector<xml::EndFontTag> tagsToClose;
3155 // parsing main loop
3156 for (pos_type i = initial; i < size(); ++i) {
3157 // let's not show deleted material in the output
3161 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3163 if (start_paragraph) {
3165 FontState curstate = font.fontInfo().emph();
3166 if (font_old.emph() != curstate)
3167 doFontSwitchDocBook(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3170 curstate = font.fontInfo().noun();
3171 if (font_old.noun() != curstate)
3172 doFontSwitchDocBook(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3175 curstate = font.fontInfo().underbar();
3176 if (font_old.underbar() != curstate)
3177 doFontSwitchDocBook(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3180 curstate = font.fontInfo().strikeout();
3181 if (font_old.strikeout() != curstate)
3182 doFontSwitchDocBook(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3185 curstate = font.fontInfo().uuline();
3186 if (font_old.uuline() != curstate)
3187 doFontSwitchDocBook(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3190 curstate = font.fontInfo().uwave();
3191 if (font_old.uwave() != curstate)
3192 doFontSwitchDocBook(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3195 // a little hackish, but allows us to reuse what we have.
3196 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3197 if (font_old.series() != font.fontInfo().series())
3198 doFontSwitchDocBook(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3201 curr_fs = font.fontInfo().shape();
3202 FontShape old_fs = font_old.shape();
3203 if (old_fs != curr_fs) {
3205 OptionalFontType tag = fontShapeToXml(old_fs);
3206 if (tag.has_value) {
3207 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3212 OptionalFontType tag = fontShapeToXml(curr_fs);
3213 if (tag.has_value) {
3214 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3219 curr_fam = font.fontInfo().family();
3220 FontFamily old_fam = font_old.family();
3221 if (old_fam != curr_fam) {
3223 OptionalFontType tag = fontFamilyToXml(old_fam);
3224 if (tag.has_value) {
3225 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3231 // we will treat a "default" font family as roman, since we have
3232 // no other idea what to do.
3233 if (default_family != "rmdefault" && default_family != "default") {
3234 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3239 if (default_family != "sfdefault") {
3240 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3244 case TYPEWRITER_FAMILY:
3245 if (default_family != "ttdefault") {
3246 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3250 case INHERIT_FAMILY:
3253 // the other tags are for internal use
3260 curr_size = font.fontInfo().size();
3261 FontSize old_size = font_old.size();
3262 if (old_size != curr_size) {
3264 OptionalFontType tag = fontSizeToXml(old_size);
3265 if (tag.has_value) {
3266 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3271 OptionalFontType tag = fontSizeToXml(curr_size);
3272 if (tag.has_value) {
3273 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3279 // Other such tags? What about the other text ranges?
3281 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3282 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3283 for (; cit != cen; ++cit)
3286 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3287 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3288 for (; sit != sen; ++sit)
3291 tagsToClose.clear();
3295 if (Inset const * inset = getInset(i)) {
3296 if (!runparams.for_toc || inset->isInToc()) {
3297 OutputParams np = runparams;
3298 np.local_font = &font;
3299 // If the paragraph has size 1, then we are in the "special
3300 // case" where we do not output the containing paragraph info.
3301 // This "special case" is defined in more details in output_docbook.cpp, makeParagraphs. The results
3302 // of that brittle logic is passed to this function through open_par.
3303 if (!inset->getLayout().htmlisblock() && size() != 1) // TODO: htmlisblock here too!
3304 np.docbook_in_par = true;
3305 inset->docbook(xs, np);
3308 char_type c = getUChar(buf.masterBuffer()->params(), runparams, i);
3311 font_old = font.fontInfo();
3314 // FIXME, this code is just imported from XHTML
3315 // I'm worried about what happens if a branch, say, is itself
3316 // wrapped in some font stuff. I think that will not work.
3318 if (runparams.docbook_in_listing)
3320 if (close_paragraph)
3327 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3328 vector<xml::EndFontTag> & tagsToClose,
3329 bool & flag, FontState curstate, xml::FontTypes type)
3331 if (curstate == FONT_ON) {
3332 tagsToOpen.push_back(xhtmlStartFontTag(type));
3335 tagsToClose.push_back(xhtmlEndFontTag(type));
3340 } // anonymous namespace
3343 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3345 OutputParams const & runparams,
3346 Font const & outerfont,
3347 bool start_paragraph, bool close_paragraph,
3348 pos_type initial) const
3352 // track whether we have opened these tags
3353 bool emph_flag = false;
3354 bool bold_flag = false;
3355 bool noun_flag = false;
3356 bool ubar_flag = false;
3357 bool dbar_flag = false;
3358 bool sout_flag = false;
3359 bool xout_flag = false;
3360 bool wave_flag = false;
3362 bool shap_flag = false;
3364 bool faml_flag = false;
3366 bool size_flag = false;
3368 Layout const & style = *d->layout_;
3370 if (start_paragraph)
3371 xs.startDivision(allowEmpty());
3374 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3376 FontShape curr_fs = INHERIT_SHAPE;
3377 FontFamily curr_fam = INHERIT_FAMILY;
3378 FontSize curr_size = INHERIT_SIZE;
3380 string const default_family =
3381 buf.masterBuffer()->params().fonts_default_family;
3383 vector<xml::FontTag> tagsToOpen;
3384 vector<xml::EndFontTag> tagsToClose;
3386 // parsing main loop
3387 for (pos_type i = initial; i < size(); ++i) {
3388 // let's not show deleted material in the output
3392 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3395 FontState curstate = font.fontInfo().emph();
3396 if (font_old.emph() != curstate)
3397 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3400 curstate = font.fontInfo().noun();
3401 if (font_old.noun() != curstate)
3402 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3405 curstate = font.fontInfo().underbar();
3406 if (font_old.underbar() != curstate)
3407 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3410 curstate = font.fontInfo().strikeout();
3411 if (font_old.strikeout() != curstate)
3412 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3415 curstate = font.fontInfo().xout();
3416 if (font_old.xout() != curstate)
3417 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3420 curstate = font.fontInfo().uuline();
3421 if (font_old.uuline() != curstate)
3422 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3425 curstate = font.fontInfo().uwave();
3426 if (font_old.uwave() != curstate)
3427 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3430 // a little hackish, but allows us to reuse what we have.
3431 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3432 if (font_old.series() != font.fontInfo().series())
3433 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3436 curr_fs = font.fontInfo().shape();
3437 FontShape old_fs = font_old.shape();
3438 if (old_fs != curr_fs) {
3442 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3445 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3447 case SMALLCAPS_SHAPE:
3448 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3454 // the other tags are for internal use
3462 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3466 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3469 case SMALLCAPS_SHAPE:
3470 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3477 // the other tags are for internal use
3484 curr_fam = font.fontInfo().family();
3485 FontFamily old_fam = font_old.family();
3486 if (old_fam != curr_fam) {
3490 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3493 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3495 case TYPEWRITER_FAMILY:
3496 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3498 case INHERIT_FAMILY:
3501 // the other tags are for internal use
3509 // we will treat a "default" font family as roman, since we have
3510 // no other idea what to do.
3511 if (default_family != "rmdefault" && default_family != "default") {
3512 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3517 if (default_family != "sfdefault") {
3518 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3522 case TYPEWRITER_FAMILY:
3523 if (default_family != "ttdefault") {
3524 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3528 case INHERIT_FAMILY:
3531 // the other tags are for internal use
3538 curr_size = font.fontInfo().size();
3539 FontSize old_size = font_old.size();
3540 if (old_size != curr_size) {
3544 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3547 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3550 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3553 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3556 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3559 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3562 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3565 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3568 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3571 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3574 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3580 // the other tags are for internal use
3586 switch (curr_size) {
3588 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3592 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3596 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3600 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3604 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3608 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3612 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3616 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3620 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3624 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3628 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3635 // the other tags are for internal use
3642 // Other such tags? What about the other text ranges?
3644 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3645 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3646 for (; cit != cen; ++cit)
3649 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3650 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3651 for (; sit != sen; ++sit)
3654 tagsToClose.clear();
3657 Inset const * inset = getInset(i);
3659 if (!runparams.for_toc || inset->isInToc()) {
3660 OutputParams np = runparams;
3661 np.local_font = &font;
3662 // If the paragraph has size 1, then we are in the "special
3663 // case" where we do not output the containing paragraph info
3664 if (!inset->getLayout().htmlisblock() && size() != 1)
3665 np.html_in_par = true;
3666 retval += inset->xhtml(xs, np);
3669 char_type c = getUChar(buf.masterBuffer()->params(),
3671 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3672 xs << XMLStream::ESCAPE_NONE << " ";
3676 font_old = font.fontInfo();
3680 // I'm worried about what happens if a branch, say, is itself
3681 // wrapped in some font stuff. I think that will not work.
3683 if (close_paragraph)
3690 bool Paragraph::isHfill(pos_type pos) const
3692 Inset const * inset = getInset(pos);
3693 return inset && inset->isHfill();
3697 bool Paragraph::isNewline(pos_type pos) const
3699 // U+2028 LINE SEPARATOR
3700 // U+2029 PARAGRAPH SEPARATOR
3701 char_type const c = d->text_[pos];
3702 if (c == 0x2028 || c == 0x2029)
3704 Inset const * inset = getInset(pos);
3705 return inset && inset->lyxCode() == NEWLINE_CODE;
3709 bool Paragraph::isEnvSeparator(pos_type pos) const
3711 Inset const * inset = getInset(pos);
3712 return inset && inset->lyxCode() == SEPARATOR_CODE;
3716 bool Paragraph::isLineSeparator(pos_type pos) const
3718 char_type const c = d->text_[pos];
3719 if (isLineSeparatorChar(c))
3721 Inset const * inset = getInset(pos);
3722 return inset && inset->isLineSeparator();
3726 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3730 if (ignore_deleted && isDeleted(pos))
3732 if (Inset const * inset = getInset(pos))
3733 return !inset->isLetter();
3734 // if we have a hard hyphen (no en- or emdash) or apostrophe
3735 // we pass this to the spell checker
3736 // FIXME: this method is subject to change, visit
3737 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3738 // to get an impression how complex this is.
3739 if (isHardHyphenOrApostrophe(pos))
3741 char_type const c = d->text_[pos];
3742 // We want to pass the escape chars to the spellchecker
3743 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3744 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3748 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3750 pos_type const psize = size();
3753 char_type const c = d->text_[pos];
3754 if (c != '-' && c != '\'')
3756 int nextpos = pos + 1;
3757 int prevpos = pos > 0 ? pos - 1 : 0;
3758 if ((nextpos == psize || isSpace(nextpos))
3759 && (pos == 0 || isSpace(prevpos)))
3765 bool Paragraph::needsCProtection(bool const fragile) const
3767 // first check the layout of the paragraph, but only in insets
3768 InsetText const * textinset = inInset().asInsetText();
3769 bool const maintext = textinset
3770 ? textinset->text().isMainText()
3773 if (!maintext && layout().needcprotect) {
3774 // Environments need cprotection regardless the content
3775 if (layout().latextype == LATEX_ENVIRONMENT)
3778 // Commands need cprotection if they contain specific chars
3779 int const nchars_escape = 9;
3780 static char_type const chars_escape[nchars_escape] = {
3781 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3783 docstring const pars = asString();
3784 for (int k = 0; k < nchars_escape; k++) {
3785 if (contains(pars, chars_escape[k]))
3790 // now check whether we have insets that need cprotection
3791 pos_type size = pos_type(d->text_.size());
3792 for (pos_type i = 0; i < size; ++i) {
3795 Inset const * ins = getInset(i);
3796 if (ins->needsCProtection(maintext, fragile))
3798 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3799 // Environments need cprotection regardless the content
3801 // Now check math environments
3802 InsetMath const * im = getInset(i)->asInsetMath();
3803 if (!im || im->cell(0).empty())
3805 switch(im->cell(0)[0]->lyxCode()) {
3806 case MATH_AMSARRAY_CODE:
3807 case MATH_SUBSTACK_CODE:
3809 case MATH_XYMATRIX_CODE:
3810 // these need cprotection
3821 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3823 return d->speller_state_.getRange(pos);
3827 bool Paragraph::isChar(pos_type pos) const
3829 if (Inset const * inset = getInset(pos))
3830 return inset->isChar();
3831 char_type const c = d->text_[pos];
3832 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3836 bool Paragraph::isSpace(pos_type pos) const
3838 if (Inset const * inset = getInset(pos))
3839 return inset->isSpace();
3840 char_type const c = d->text_[pos];
3841 return lyx::isSpace(c);
3846 Paragraph::getParLanguage(BufferParams const & bparams) const
3849 return getFirstFontSettings(bparams).language();
3850 // FIXME: we should check the prev par as well (Lgb)
3851 return bparams.language;
3855 bool Paragraph::isRTL(BufferParams const & bparams) const
3857 return getParLanguage(bparams)->rightToLeft()
3858 && !inInset().getLayout().forceLTR();
3862 void Paragraph::changeLanguage(BufferParams const & bparams,
3863 Language const * from, Language const * to)
3865 // change language including dummy font change at the end
3866 for (pos_type i = 0; i <= size(); ++i) {
3867 Font font = getFontSettings(bparams, i);
3868 if (font.language() == from) {
3869 font.setLanguage(to);
3871 d->requestSpellCheck(i);
3877 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3879 Language const * doc_language = bparams.language;
3880 for (auto const & f : d->fontlist_)
3881 if (f.font().language() != ignore_language &&
3882 f.font().language() != latex_language &&
3883 f.font().language() != doc_language)
3889 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3891 for (auto const & f : d->fontlist_) {
3892 Language const * lang = f.font().language();
3893 if (lang != ignore_language &&
3894 lang != latex_language)
3900 docstring Paragraph::asString(int options) const
3902 return asString(0, size(), options);
3906 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3908 odocstringstream os;
3911 && options & AS_STR_LABEL
3912 && !d->params_.labelString().empty())
3913 os << d->params_.labelString() << ' ';
3915 for (pos_type i = beg; i < end; ++i) {
3916 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3918 char_type const c = d->text_[i];
3919 if (isPrintable(c) || c == '\t'
3920 || (c == '\n' && (options & AS_STR_NEWLINES)))
3922 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3923 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3924 LASSERT(runparams != nullptr, return docstring());
3925 getInset(i)->plaintext(os, *runparams);
3927 getInset(i)->toString(os);
3936 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3937 bool const shorten, bool const label) const
3939 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3940 if (label && !labelString().empty())
3941 os += labelString() + ' ';
3942 if (!layout().isTocCaption())
3944 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3947 char_type const c = d->text_[i];
3950 else if (c == META_INSET)
3951 getInset(i)->forOutliner(os, tmplen, false);
3954 Text::shortenForOutliner(os, maxlen);
3958 void Paragraph::setInsetOwner(Inset const * inset)
3960 d->inset_owner_ = inset;
3964 int Paragraph::id() const
3970 void Paragraph::setId(int id)
3976 Layout const & Paragraph::layout() const
3982 void Paragraph::setLayout(Layout const & layout)
3984 d->layout_ = &layout;
3988 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3990 setLayout(tc.defaultLayout());
3994 void Paragraph::setPlainLayout(DocumentClass const & tc)
3996 setLayout(tc.plainLayout());
4000 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
4002 if (usePlainLayout())
4003 setPlainLayout(tclass);
4005 setDefaultLayout(tclass);
4009 Inset const & Paragraph::inInset() const
4011 LBUFERR(d->inset_owner_);
4012 return *d->inset_owner_;
4016 ParagraphParameters & Paragraph::params()
4022 ParagraphParameters const & Paragraph::params() const
4028 bool Paragraph::isFreeSpacing() const
4030 if (d->layout_->free_spacing)
4032 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4036 bool Paragraph::allowEmpty() const
4038 if (d->layout_->keepempty)
4040 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4044 bool Paragraph::brokenBiblio() const
4046 // There is a problem if there is no bibitem at position 0 in
4047 // paragraphs that need one, if there is another bibitem in the
4048 // paragraph or if this paragraph is not supposed to have
4049 // a bibitem inset at all.
4050 return ((d->layout_->labeltype == LABEL_BIBLIO
4051 && (d->insetlist_.find(BIBITEM_CODE) != 0
4052 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4053 || (d->layout_->labeltype != LABEL_BIBLIO
4054 && d->insetlist_.find(BIBITEM_CODE) != -1));
4058 int Paragraph::fixBiblio(Buffer const & buffer)
4060 // FIXME: when there was already an inset at 0, the return value is 1,
4061 // which does not tell whether another inset has been remove; the
4062 // cursor cannot be correctly updated.
4064 bool const track_changes = buffer.params().track_changes;
4065 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4067 // The case where paragraph is not BIBLIO
4068 if (d->layout_->labeltype != LABEL_BIBLIO) {
4069 if (bibitem_pos == -1)
4070 // No InsetBibitem => OK
4072 // There is an InsetBibitem: remove it!
4073 d->insetlist_.release(bibitem_pos);
4074 eraseChar(bibitem_pos, track_changes);
4075 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4078 bool const hasbibitem0 = bibitem_pos == 0;
4080 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4081 // There was an InsetBibitem at pos 0,
4082 // and no other one => OK
4083 if (bibitem_pos == -1)
4085 // there is a bibitem at the 0 position, but since
4086 // there is a second one, we copy the second on the
4087 // first. We're assuming there are at most two of
4088 // these, which there should be.
4089 // FIXME: why does it make sense to do that rather
4090 // than keep the first? (JMarc)
4091 Inset * inset = releaseInset(bibitem_pos);
4092 d->insetlist_.begin()->inset = inset;
4093 return -bibitem_pos;
4096 // We need to create an inset at the beginning
4097 Inset * inset = nullptr;
4098 if (bibitem_pos > 0) {
4099 // there was one somewhere in the paragraph, let's move it
4100 inset = d->insetlist_.release(bibitem_pos);
4101 eraseChar(bibitem_pos, track_changes);
4104 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4105 InsetCommandParams(BIBITEM_CODE));
4107 Font font(inherit_font, buffer.params().language);
4108 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4109 : Change::UNCHANGED));
4111 // This is needed to get the counters right
4112 buffer.updateBuffer();
4117 void Paragraph::checkAuthors(AuthorList const & authorList)
4119 d->changes_.checkAuthors(authorList);
4123 bool Paragraph::isChanged(pos_type pos) const
4125 return lookupChange(pos).changed();
4129 bool Paragraph::isInserted(pos_type pos) const
4131 return lookupChange(pos).inserted();
4135 bool Paragraph::isDeleted(pos_type pos) const
4137 return lookupChange(pos).deleted();
4141 InsetList const & Paragraph::insetList() const
4143 return d->insetlist_;
4147 void Paragraph::setInsetBuffers(Buffer & b)
4149 d->insetlist_.setBuffer(b);
4153 void Paragraph::resetBuffer()
4155 d->insetlist_.resetBuffer();
4159 Inset * Paragraph::releaseInset(pos_type pos)
4161 Inset * inset = d->insetlist_.release(pos);
4162 /// does not honour change tracking!
4163 eraseChar(pos, false);
4168 Inset * Paragraph::getInset(pos_type pos)
4170 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4171 ? d->insetlist_.get(pos) : nullptr;
4175 Inset const * Paragraph::getInset(pos_type pos) const
4177 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4178 ? d->insetlist_.get(pos) : nullptr;
4182 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4183 pos_type & right, TextCase action)
4185 // process sequences of modified characters; in change
4186 // tracking mode, this approach results in much better
4187 // usability than changing case on a char-by-char basis
4188 // We also need to track the current font, since font
4189 // changes within sequences can occur.
4190 vector<pair<char_type, Font> > changes;
4192 bool const trackChanges = bparams.track_changes;
4194 bool capitalize = true;
4196 for (; pos < right; ++pos) {
4197 char_type oldChar = d->text_[pos];
4198 char_type newChar = oldChar;
4200 // ignore insets and don't play with deleted text!
4201 if (oldChar != META_INSET && !isDeleted(pos)) {
4203 case text_lowercase:
4204 newChar = lowercase(oldChar);
4206 case text_capitalization:
4208 newChar = uppercase(oldChar);
4212 case text_uppercase:
4213 newChar = uppercase(oldChar);
4218 if (isWordSeparator(pos) || isDeleted(pos)) {
4219 // permit capitalization again
4223 if (oldChar != newChar) {
4224 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4225 if (pos != right - 1)
4227 // step behind the changing area
4231 int erasePos = pos - changes.size();
4232 for (size_t i = 0; i < changes.size(); i++) {
4233 insertChar(pos, changes[i].first,
4236 if (!eraseChar(erasePos, trackChanges)) {
4239 ++right; // expand selection
4247 int Paragraph::find(docstring const & str, bool cs, bool mw,
4248 pos_type start_pos, bool del) const
4250 pos_type pos = start_pos;
4251 int const strsize = str.length();
4253 pos_type const parsize = d->text_.size();
4254 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4255 // Ignore "invisible" letters such as ligature breaks
4256 // and hyphenation chars while searching
4257 while (pos < parsize - 1 && isInset(pos)) {
4258 odocstringstream os;
4259 getInset(pos)->toString(os);
4260 if (!getInset(pos)->isLetter() || !os.str().empty())
4264 if (cs && str[i] != d->text_[pos])
4266 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4268 if (!del && isDeleted(pos))
4275 // if necessary, check whether string matches word
4277 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4280 && !isWordSeparator(pos))
4284 return pos - start_pos;
4288 char_type Paragraph::getChar(pos_type pos) const
4290 return d->text_[pos];
4294 pos_type Paragraph::size() const
4296 return d->text_.size();
4300 bool Paragraph::empty() const
4302 return d->text_.empty();
4306 bool Paragraph::isInset(pos_type pos) const
4308 return d->text_[pos] == META_INSET;
4312 bool Paragraph::isSeparator(pos_type pos) const
4314 //FIXME: Are we sure this can be the only separator?
4315 return d->text_[pos] == ' ';
4319 void Paragraph::deregisterWords()
4321 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4322 Private::LangWordsMap::const_iterator ite = d->words_.end();
4323 for (; itl != ite; ++itl) {
4324 WordList & wl = theWordList(itl->first);
4325 Private::Words::const_iterator it = (itl->second).begin();
4326 Private::Words::const_iterator et = (itl->second).end();
4327 for (; it != et; ++it)
4334 void Paragraph::locateWord(pos_type & from, pos_type & to,
4335 word_location const loc, bool const ignore_deleted) const
4338 case WHOLE_WORD_STRICT:
4339 if (from == 0 || from == size()
4340 || isWordSeparator(from, ignore_deleted)
4341 || isWordSeparator(from - 1, ignore_deleted)) {
4348 // If we are already at the beginning of a word, do nothing
4349 if (!from || isWordSeparator(from - 1, ignore_deleted))
4354 // always move the cursor to the beginning of previous word
4355 while (from && !isWordSeparator(from - 1, ignore_deleted))
4359 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4362 // no need to move the 'from' cursor
4366 while (to < size() && !isWordSeparator(to, ignore_deleted))
4371 void Paragraph::collectWords()
4373 for (pos_type pos = 0; pos < size(); ++pos) {
4374 if (isWordSeparator(pos))
4376 pos_type from = pos;
4377 locateWord(from, pos, WHOLE_WORD);
4378 // Work around MSVC warning: The statement
4379 // if (pos < from + lyxrc.completion_minlength)
4380 // triggers a signed vs. unsigned warning.
4381 // I don't know why this happens, it could be a MSVC bug, or
4382 // related to LLP64 (windows) vs. LP64 (unix) programming
4383 // model, or the C++ standard might be ambigous in the section
4384 // defining the "usual arithmetic conversions". However, using
4385 // a temporary variable is safe and works on all compilers.
4386 pos_type const endpos = from + lyxrc.completion_minlength;
4389 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4390 if (cit == d->fontlist_.end())
4392 Language const * lang = cit->font().language();
4393 docstring const word = asString(from, pos, AS_STR_NONE);
4394 d->words_[lang->lang()].insert(word);
4399 void Paragraph::registerWords()
4401 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4402 Private::LangWordsMap::const_iterator ite = d->words_.end();
4403 for (; itl != ite; ++itl) {
4404 WordList & wl = theWordList(itl->first);
4405 Private::Words::const_iterator it = (itl->second).begin();
4406 Private::Words::const_iterator et = (itl->second).end();
4407 for (; it != et; ++it)
4413 void Paragraph::updateWords()
4421 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4423 SkipPositionsIterator begin = skips.begin();
4424 SkipPositions::iterator end = skips.end();
4425 if (pos > 0 && begin < end) {
4427 if (end->last == pos - 1) {
4432 skips.insert(end, FontSpan(pos, pos));
4436 Language * Paragraph::Private::locateSpellRange(
4437 pos_type & from, pos_type & to,
4438 SkipPositions & skips) const
4440 // skip leading white space
4441 while (from < to && owner_->isWordSeparator(from))
4443 // don't check empty range
4446 // get current language
4447 Language * lang = getSpellLanguage(from);
4448 pos_type last = from;
4449 bool samelang = true;
4450 bool sameinset = true;
4451 while (last < to && samelang && sameinset) {
4452 // hop to end of word
4453 while (last < to && !owner_->isWordSeparator(last)) {
4454 if (owner_->getInset(last)) {
4455 appendSkipPosition(skips, last);
4456 } else if (owner_->isDeleted(last)) {
4457 appendSkipPosition(skips, last);
4461 // hop to next word while checking for insets
4462 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4463 if (Inset const * inset = owner_->getInset(last))
4464 sameinset = inset->isChar() && inset->isLetter();
4465 if (sameinset && owner_->isDeleted(last)) {
4466 appendSkipPosition(skips, last);
4471 if (sameinset && last < to) {
4472 // now check for language change
4473 samelang = lang == getSpellLanguage(last);
4476 // if language change detected backstep is needed
4484 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4487 const_cast<Language *>(owner_->getFontSettings(
4488 inset_owner_->buffer().params(), from).language());
4489 if (lang == inset_owner_->buffer().params().language
4490 && !lyxrc.spellchecker_alt_lang.empty()) {
4492 string const lang_variety =
4493 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4494 lang->setCode(lang_code);
4495 lang->setVariety(lang_variety);
4501 void Paragraph::requestSpellCheck(pos_type pos)
4503 d->requestSpellCheck(pos);
4507 bool Paragraph::needsSpellCheck() const
4509 SpellChecker::ChangeNumber speller_change_number = 0;
4510 if (theSpellChecker())
4511 speller_change_number = theSpellChecker()->changeNumber();
4512 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4513 d->speller_state_.needsCompleteRefresh(speller_change_number);
4515 return d->needsSpellCheck();
4519 bool Paragraph::Private::ignoreWord(docstring const & word) const
4521 // Ignore words with digits
4522 // FIXME: make this customizable
4523 // (note that some checkers ignore words with digits by default)
4524 docstring::const_iterator cit = word.begin();
4525 docstring::const_iterator const end = word.end();
4526 for (; cit != end; ++cit) {
4527 if (isNumber((*cit)))
4534 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4535 WordLangTuple & wl, docstring_list & suggestions,
4536 bool do_suggestion, bool check_learned) const
4538 SpellChecker::Result result = SpellChecker::WORD_OK;
4539 SpellChecker * speller = theSpellChecker();
4543 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4546 locateWord(from, to, WHOLE_WORD, true);
4547 if (from == to || from >= size())
4550 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4551 Language * lang = d->getSpellLanguage(from);
4553 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4556 wl = WordLangTuple(word, lang);
4561 if (needsSpellCheck() || check_learned) {
4563 if (!d->ignoreWord(word)) {
4564 bool const trailing_dot = to < size() && d->text_[to] == '.';
4565 result = speller->check(wl);
4566 if (SpellChecker::misspelled(result) && trailing_dot) {
4567 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4568 result = speller->check(wl);
4569 if (!SpellChecker::misspelled(result)) {
4570 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4572 from << ".." << to << "]");
4574 // spell check with dot appended failed too
4575 // restore original word/lang value
4576 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4577 wl = WordLangTuple(word, lang);
4581 if (!SpellChecker::misspelled(result)) {
4582 // area up to the begin of the next word is not misspelled
4583 while (end < size() && isWordSeparator(end))
4586 d->setMisspelled(from, end, result);
4588 result = d->speller_state_.getState(from);
4592 suggestions.clear();
4594 if (SpellChecker::misspelled(result)) {
4595 LYXERR(Debug::GUI, "misspelled word: \"" <<
4597 from << ".." << to << "]");
4599 speller->suggest(wl, suggestions);
4605 void Paragraph::anonymize()
4607 // This is a very crude anonymization for now
4608 for (char_type & c : d->text_)
4609 if (isLetterChar(c) || isNumber(c))
4614 void Paragraph::Private::markMisspelledWords(
4615 pos_type const & first, pos_type const & last,
4616 SpellChecker::Result result,
4617 docstring const & word,
4618 SkipPositions const & skips)
4620 if (!SpellChecker::misspelled(result)) {
4621 setMisspelled(first, last, SpellChecker::WORD_OK);
4625 SpellChecker * speller = theSpellChecker();
4626 // locate and enumerate the error positions
4627 int nerrors = speller->numMisspelledWords();
4629 SkipPositionsIterator it = skips.begin();
4630 SkipPositionsIterator et = skips.end();
4631 for (int index = 0; index < nerrors; ++index) {
4634 speller->misspelledWord(index, wstart, wlen);
4635 /// should not happen if speller supports range checks
4636 if (!wlen) continue;
4637 docstring const misspelled = word.substr(wstart, wlen);
4638 wstart += first + numskipped;
4639 if (snext < wstart) {
4640 /// mark the range of correct spelling
4641 numskipped += countSkips(it, et, wstart);
4642 setMisspelled(snext,
4643 wstart - 1, SpellChecker::WORD_OK);
4645 snext = wstart + wlen;
4646 numskipped += countSkips(it, et, snext);
4647 /// mark the range of misspelling
4648 setMisspelled(wstart, snext, result);
4649 LYXERR(Debug::GUI, "misspelled word: \"" <<
4650 misspelled << "\" [" <<
4651 wstart << ".." << (snext-1) << "]");
4654 if (snext <= last) {
4655 /// mark the range of correct spelling at end
4656 setMisspelled(snext, last, SpellChecker::WORD_OK);
4661 void Paragraph::spellCheck() const
4663 SpellChecker * speller = theSpellChecker();
4664 if (!speller || empty() ||!needsSpellCheck())
4668 d->rangeOfSpellCheck(start, endpos);
4669 if (speller->canCheckParagraph()) {
4670 // loop until we leave the range
4671 for (pos_type first = start; first < endpos; ) {
4672 pos_type last = endpos;
4673 Private::SkipPositions skips;
4674 Language * lang = d->locateSpellRange(first, last, skips);
4675 if (first >= endpos)
4677 // start the spell checker on the unit of meaning
4678 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4679 WordLangTuple wl = WordLangTuple(word, lang);
4680 SpellChecker::Result result = word.size() ?
4681 speller->check(wl) : SpellChecker::WORD_OK;
4682 d->markMisspelledWords(first, last, result, word, skips);
4686 static docstring_list suggestions;
4687 pos_type to = endpos;
4688 while (start < endpos) {
4690 spellCheck(start, to, wl, suggestions, false);
4694 d->readySpellCheck();
4698 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4700 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4701 if (result || pos <= 0 || pos > size())
4703 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4704 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4709 string Paragraph::magicLabel() const
4712 ss << "magicparlabel-" << id();