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"
36 #include "OutputParams.h"
37 #include "output_latex.h"
38 #include "output_xhtml.h"
39 #include "output_docbook.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "texstream.h"
44 #include "TextClass.h"
47 #include "WordLangTuple.h"
50 #include "frontends/alert.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
55 #include "insets/InsetText.h"
57 #include "mathed/InsetMathHull.h"
59 #include "support/debug.h"
60 #include "support/docstring_list.h"
61 #include "support/ExceptionMessage.h"
62 #include "support/gettext.h"
63 #include "support/lassert.h"
64 #include "support/Length.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
87 /////////////////////////////////////////////////////////////////////
91 /////////////////////////////////////////////////////////////////////
93 class SpellResultRange {
95 SpellResultRange(FontSpan range, SpellChecker::Result result)
96 : range_(range), result_(result)
99 FontSpan const & range() const { return range_; }
101 void range(FontSpan const & r) { range_ = r; }
103 SpellChecker::Result result() const { return result_; }
105 void result(SpellChecker::Result r) { result_ = r; }
107 bool contains(pos_type pos) const { return range_.contains(pos); }
109 bool covered(FontSpan const & r) const
111 // 1. first of new range inside current range or
112 // 2. last of new range inside current range or
113 // 3. first of current range inside new range or
114 // 4. last of current range inside new range
115 //FIXME: is this the same as !range_.intersect(r).empty() ?
116 return range_.contains(r.first) || range_.contains(r.last) ||
117 r.contains(range_.first) || r.contains(range_.last);
120 void shift(pos_type pos, int offset)
122 if (range_.first > pos) {
123 range_.first += offset;
124 range_.last += offset;
125 } else if (range_.last >= pos) {
126 range_.last += offset;
131 SpellChecker::Result result_ ;
135 /////////////////////////////////////////////////////////////////////
139 /////////////////////////////////////////////////////////////////////
141 class SpellCheckerState {
145 needs_refresh_ = true;
146 current_change_number_ = 0;
149 void setRange(FontSpan const & fp, SpellChecker::Result state)
152 RangesIterator et = ranges_.end();
153 RangesIterator it = ranges_.begin();
154 for (; it != et; ++it) {
155 if (!it->covered(fp))
156 result.push_back(SpellResultRange(it->range(), it->result()));
157 else if (state == SpellChecker::WORD_OK) {
158 // trim or split the current misspelled range
159 // store misspelled ranges only
160 FontSpan range = it->range();
161 if (fp.first > range.first) {
162 // misspelled area in front of WORD_OK
163 range.last = fp.first - 1;
164 result.push_back(SpellResultRange(range, it->result()));
167 if (fp.last < range.last) {
168 // misspelled area after WORD_OK range
169 range.first = fp.last + 1;
170 result.push_back(SpellResultRange(range, it->result()));
175 if (state != SpellChecker::WORD_OK)
176 ranges_.push_back(SpellResultRange(fp, state));
179 void increasePosAfterPos(pos_type pos)
181 correctRangesAfterPos(pos, 1);
185 void decreasePosAfterPos(pos_type pos)
187 correctRangesAfterPos(pos, -1);
191 void refreshLast(pos_type pos)
193 if (pos < refresh_.last)
197 SpellChecker::Result getState(pos_type pos) const
199 SpellChecker::Result result = SpellChecker::WORD_OK;
200 RangesIterator et = ranges_.end();
201 RangesIterator it = ranges_.begin();
202 for (; it != et; ++it) {
203 if(it->contains(pos)) {
210 FontSpan const & getRange(pos_type pos) const
212 /// empty span to indicate mismatch
213 static FontSpan empty_;
214 RangesIterator et = ranges_.end();
215 RangesIterator it = ranges_.begin();
216 for (; it != et; ++it) {
217 if(it->contains(pos)) {
224 bool needsRefresh() const
226 return needs_refresh_;
229 SpellChecker::ChangeNumber currentChangeNumber() const
231 return current_change_number_;
234 void refreshRange(pos_type & first, pos_type & last) const
236 first = refresh_.first;
237 last = refresh_.last;
240 void needsRefresh(pos_type pos)
242 if (needs_refresh_ && pos != -1) {
243 if (pos < refresh_.first)
244 refresh_.first = pos;
245 if (pos > refresh_.last)
247 } else if (pos != -1) {
248 // init request check for neighbour positions too
249 refresh_.first = pos > 0 ? pos - 1 : 0;
250 // no need for special end of paragraph check
251 refresh_.last = pos + 1;
253 needs_refresh_ = pos != -1;
256 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
258 needs_refresh_ = true;
261 current_change_number_ = change_number;
264 typedef vector<SpellResultRange> Ranges;
265 typedef Ranges::const_iterator RangesIterator;
267 /// the area of the paragraph with pending spell check
269 /// spell state cache version number
270 SpellChecker::ChangeNumber current_change_number_;
274 void correctRangesAfterPos(pos_type pos, int offset)
276 RangesIterator et = ranges_.end();
277 Ranges::iterator it = ranges_.begin();
278 for (; it != et; ++it) {
279 it->shift(pos, offset);
285 /////////////////////////////////////////////////////////////////////
287 // Paragraph::Private
289 /////////////////////////////////////////////////////////////////////
291 class Paragraph::Private
293 // Enforce our own "copy" constructor
294 Private(Private const &) = delete;
295 Private & operator=(Private const &) = delete;
296 // Unique ID generator
297 static int make_id();
300 Private(Paragraph * owner, Layout const & layout);
301 /// "Copy constructor"
302 Private(Private const &, Paragraph * owner);
303 /// Copy constructor from \p beg to \p end
304 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
307 void insertChar(pos_type pos, char_type c, Change const & change);
309 /// Output the surrogate pair formed by \p c and \p next to \p os.
310 /// \return the number of characters written.
311 int latexSurrogatePair(BufferParams const &, otexstream & os,
312 char_type c, char_type next,
313 OutputParams const &);
315 /// Output a space in appropriate formatting (or a surrogate pair
316 /// if the next character is a combining character).
317 /// \return whether a surrogate pair was output.
318 bool simpleTeXBlanks(BufferParams const &,
319 OutputParams const &,
322 unsigned int & column,
324 Layout const & style);
326 /// This could go to ParagraphParameters if we want to.
327 int startTeXParParams(BufferParams const &, otexstream &,
328 OutputParams const &) const;
330 /// This could go to ParagraphParameters if we want to.
331 bool endTeXParParams(BufferParams const &, otexstream &,
332 OutputParams const &) const;
335 void latexInset(BufferParams const &,
340 Font const & outerfont,
342 Change & running_change,
343 Layout const & style,
345 unsigned int & column,
346 bool const fontswitch_inset,
347 bool const closeLanguage,
348 bool const lang_switched_at_inset);
351 void latexSpecialChar(
353 BufferParams const & bparams,
354 OutputParams const & runparams,
355 Font const & running_font,
356 string & alien_script,
357 Layout const & style,
360 unsigned int & column);
367 unsigned int & column);
373 unsigned int & column);
379 unsigned int & column);
382 void validate(LaTeXFeatures & features) const;
384 /// Checks if the paragraph contains only text and no inset or font change.
385 bool onlyText(Buffer const & buf, Font const & outerfont,
386 pos_type initial) const;
388 /// a vector of speller skip positions
389 typedef vector<FontSpan> SkipPositions;
390 typedef SkipPositions::const_iterator SkipPositionsIterator;
392 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
394 Language * getSpellLanguage(pos_type const from) const;
396 Language * locateSpellRange(pos_type & from, pos_type & to,
397 SkipPositions & skips) const;
399 bool hasSpellerChange() const
401 SpellChecker::ChangeNumber speller_change_number = 0;
402 if (theSpellChecker())
403 speller_change_number = theSpellChecker()->changeNumber();
404 return speller_change_number > speller_state_.currentChangeNumber();
407 bool ignoreWord(docstring const & word) const ;
409 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
411 pos_type textsize = owner_->size();
412 // check for sane arguments
413 if (to <= from || from >= textsize)
415 FontSpan fp = FontSpan(from, to - 1);
416 speller_state_.setRange(fp, state);
419 void requestSpellCheck(pos_type pos)
422 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
424 speller_state_.needsRefresh(pos);
427 void readySpellCheck()
429 speller_state_.needsRefresh(-1);
432 bool needsSpellCheck() const
434 return speller_state_.needsRefresh();
437 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
439 speller_state_.refreshRange(first, last);
441 last = owner_->size();
444 pos_type endpos = last;
445 owner_->locateWord(first, endpos, WHOLE_WORD, true);
448 owner_->locateWord(last, endpos, WHOLE_WORD, true);
453 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
457 while (it != et && it->first < start) {
458 long skip = it->last - it->first + 1;
466 void markMisspelledWords(pos_type const & first, pos_type const & last,
467 SpellChecker::Result result,
468 docstring const & word,
469 SkipPositions const & skips);
471 InsetCode ownerCode() const
473 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
476 /// Which Paragraph owns us?
480 Inset const * inset_owner_;
486 ParagraphParameters params_;
488 /// for recording and looking up changes
492 InsetList insetlist_;
495 pos_type begin_of_body_;
497 typedef docstring TextContainer;
501 typedef set<docstring> Words;
502 typedef map<string, Words> LangWordsMap;
506 Layout const * layout_;
508 SpellCheckerState speller_state_;
514 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
515 : owner_(owner), inset_owner_(nullptr), begin_of_body_(0), layout_(&layout), id_(-1)
522 int Paragraph::Private::make_id()
524 // The id is unique per session across buffers because it is used in
525 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
528 static int next_id(0);
533 Paragraph::Private::Private(Private const & p, Paragraph * owner)
534 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
535 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
536 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
537 layout_(p.layout_), id_(make_id())
539 requestSpellCheck(p.text_.size());
543 Paragraph::Private::Private(Private const & p, Paragraph * owner,
544 pos_type beg, pos_type end)
545 : owner_(owner), inset_owner_(p.inset_owner_),
546 params_(p.params_), changes_(p.changes_),
547 insetlist_(p.insetlist_, beg, end),
548 begin_of_body_(p.begin_of_body_), words_(p.words_),
549 layout_(p.layout_), id_(make_id())
551 if (beg >= pos_type(p.text_.size()))
553 text_ = p.text_.substr(beg, end - beg);
555 FontList::const_iterator fcit = fontlist_.begin();
556 FontList::const_iterator fend = fontlist_.end();
557 for (; fcit != fend; ++fcit) {
558 if (fcit->pos() < beg)
560 if (fcit->pos() >= end) {
561 // Add last entry in the fontlist_.
562 fontlist_.set(text_.size() - 1, fcit->font());
565 // Add a new entry in the fontlist_.
566 fontlist_.set(fcit->pos() - beg, fcit->font());
568 requestSpellCheck(p.text_.size());
572 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
573 bool output_active, TocBackend & backend) const
575 d->changes_.addToToc(cdit, buf, output_active, backend);
579 bool Paragraph::isDeleted(pos_type start, pos_type end) const
581 LASSERT(start >= 0 && start <= size(), return false);
582 LASSERT(end > start && end <= size() + 1, return false);
584 return d->changes_.isDeleted(start, end);
588 bool Paragraph::isChanged(pos_type start, pos_type end) const
590 LASSERT(start >= 0 && start <= size(), return false);
591 LASSERT(end > start && end <= size() + 1, return false);
593 return d->changes_.isChanged(start, end);
596 // FIXME: Ideally the diverse isChanged() methods should account for that!
597 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
599 LASSERT(start >= 0 && start <= size(), return false);
600 LASSERT(end > start && end <= size() + 1, return false);
602 for (auto const & icit : d->insetlist_) {
603 if (icit.pos < start)
607 if (icit.inset && icit.inset->isChanged())
613 bool Paragraph::isChanged() const
615 return d->changes_.isChanged();
619 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
621 // keep the logic here in sync with the logic of eraseChars()
625 Change const & change = d->changes_.lookup(size());
626 return change.inserted() && change.currentAuthor();
629 Change Paragraph::parEndChange() const
631 return d->changes_.lookup(size());
635 void Paragraph::setChange(Change const & change)
637 // beware of the imaginary end-of-par character!
638 d->changes_.set(change, 0, size() + 1);
641 * Propagate the change recursively - but not in case of DELETED!
643 * Imagine that your co-author makes changes in an existing inset. He
644 * sends your document to you and you come to the conclusion that the
645 * inset should go completely. If you erase it, LyX must not delete all
646 * text within the inset. Otherwise, the change tracked insertions of
647 * your co-author get lost and there is no way to restore them later.
649 * Conclusion: An inset's content should remain untouched if you delete it
652 if (!change.deleted()) {
653 for (pos_type pos = 0; pos < size(); ++pos) {
654 if (Inset * inset = getInset(pos))
655 inset->setChange(change);
661 void Paragraph::setChange(pos_type pos, Change const & change)
663 LASSERT(pos >= 0 && pos <= size(), return);
664 d->changes_.set(change, pos);
666 // see comment in setChange(Change const &) above
667 if (!change.deleted() && pos < size())
668 if (Inset * inset = getInset(pos))
669 inset->setChange(change);
673 Change const & Paragraph::lookupChange(pos_type pos) const
675 LBUFERR(pos >= 0 && pos <= size());
676 return d->changes_.lookup(pos);
680 void Paragraph::acceptChanges(pos_type start, pos_type end)
682 LASSERT(start >= 0 && start <= size(), return);
683 LASSERT(end > start && end <= size() + 1, return);
685 for (pos_type pos = start; pos < end; ++pos) {
686 switch (lookupChange(pos).type) {
687 case Change::UNCHANGED:
688 // accept changes in nested inset
689 if (Inset * inset = getInset(pos))
690 inset->acceptChanges();
693 case Change::INSERTED:
694 d->changes_.set(Change(Change::UNCHANGED), pos);
695 // also accept changes in nested inset
696 if (Inset * inset = getInset(pos))
697 inset->acceptChanges();
700 case Change::DELETED:
701 // Suppress access to non-existent
702 // "end-of-paragraph char"
704 eraseChar(pos, false);
715 void Paragraph::rejectChanges(pos_type start, pos_type end)
717 LASSERT(start >= 0 && start <= size(), return);
718 LASSERT(end > start && end <= size() + 1, return);
720 for (pos_type pos = start; pos < end; ++pos) {
721 switch (lookupChange(pos).type) {
722 case Change::UNCHANGED:
723 // reject changes in nested inset
724 if (Inset * inset = getInset(pos))
725 inset->rejectChanges();
728 case Change::INSERTED:
729 // Suppress access to non-existent
730 // "end-of-paragraph char"
732 eraseChar(pos, false);
738 case Change::DELETED:
739 d->changes_.set(Change(Change::UNCHANGED), pos);
741 // Do NOT reject changes within a deleted inset!
742 // There may be insertions of a co-author inside of it!
750 void Paragraph::Private::insertChar(pos_type pos, char_type c,
751 Change const & change)
753 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
756 changes_.insert(change, pos);
758 // This is actually very common when parsing buffers (and
759 // maybe inserting ascii text)
760 if (pos == pos_type(text_.size())) {
761 // when appending characters, no need to update tables
763 // but we want spell checking
764 requestSpellCheck(pos);
768 text_.insert(text_.begin() + pos, c);
770 // Update the font table.
771 fontlist_.increasePosAfterPos(pos);
774 insetlist_.increasePosAfterPos(pos);
776 // Update list of misspelled positions
777 speller_state_.increasePosAfterPos(pos);
781 bool Paragraph::insertInset(pos_type pos, Inset * inset,
782 Font const & font, Change const & change)
784 LASSERT(inset, return false);
785 LASSERT(pos >= 0 && pos <= size(), return false);
787 // Paragraph::insertInset() can be used in cut/copy/paste operation where
788 // d->inset_owner_ is not set yet.
789 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
792 d->insertChar(pos, META_INSET, change);
793 LASSERT(d->text_[pos] == META_INSET, return false);
795 // Add a new entry in the insetlist_.
796 d->insetlist_.insert(inset, pos);
798 // Some insets require run of spell checker
799 requestSpellCheck(pos);
805 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
807 LASSERT(pos >= 0 && pos <= size(), return false);
809 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
812 Change change = d->changes_.lookup(pos);
814 // set the character to DELETED if
815 // a) it was previously unchanged or
816 // b) it was inserted by a co-author
818 if (!change.changed() ||
819 (change.inserted() && !change.currentAuthor())) {
820 setChange(pos, Change(Change::DELETED));
821 // request run of spell checker
822 requestSpellCheck(pos);
826 if (change.deleted())
830 // Don't physically access the imaginary end-of-paragraph character.
831 // eraseChar() can only mark it as DELETED. A physical deletion of
832 // end-of-par must be handled externally.
838 d->changes_.erase(pos);
840 // if it is an inset, delete the inset entry
841 if (d->text_[pos] == META_INSET)
842 d->insetlist_.erase(pos);
844 d->text_.erase(d->text_.begin() + pos);
846 // Update the fontlist_
847 d->fontlist_.erase(pos);
849 // Update the insetlist_
850 d->insetlist_.decreasePosAfterPos(pos);
852 // Update list of misspelled positions
853 d->speller_state_.decreasePosAfterPos(pos);
854 d->speller_state_.refreshLast(size());
860 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
862 LASSERT(start >= 0 && start <= size(), return 0);
863 LASSERT(end >= start && end <= size() + 1, return 0);
866 for (pos_type count = end - start; count; --count) {
867 if (!eraseChar(i, trackChanges))
873 // Handle combining characters
874 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
875 otexstream & os, char_type c, char_type next,
876 OutputParams const & runparams)
878 // Writing next here may circumvent a possible font change between
879 // c and next. Since next is only output if it forms a surrogate pair
880 // with c we can ignore this:
881 // A font change inside a surrogate pair does not make sense and is
882 // hopefully impossible to input.
883 // FIXME: change tracking
884 // Is this correct WRT change tracking?
885 Encoding const & encoding = *(runparams.encoding);
886 docstring latex1 = encoding.latexChar(next).first;
887 if (runparams.inIPA) {
888 string const tipashortcut = Encodings::TIPAShortcut(next);
889 if (!tipashortcut.empty()) {
890 latex1 = from_ascii(tipashortcut);
893 docstring latex2 = encoding.latexChar(c).first;
895 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
896 // Encoding supports the combination:
897 // output as is (combining char after base char).
898 os << latex2 << latex1;
899 return latex1.length() + latex2.length();
902 os << latex1 << "{" << latex2 << "}";
903 return latex1.length() + latex2.length() + 2;
907 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
908 OutputParams const & runparams,
911 unsigned int & column,
913 Layout const & style)
915 if (style.pass_thru || runparams.pass_thru)
918 if (i + 1 < int(text_.size())) {
919 char_type next = text_[i + 1];
920 if (Encodings::isCombiningChar(next)) {
921 // This space has an accent, so we must always output it.
922 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
927 if (runparams.linelen > 0
928 && column > runparams.linelen
930 && text_[i - 1] != ' '
931 && (i + 1 < int(text_.size()))
932 // same in FreeSpacing mode
933 && !owner_->isFreeSpacing()
934 // In typewriter mode, we want to avoid
935 // ! . ? : at the end of a line
936 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
937 && (text_[i - 1] == '.'
938 || text_[i - 1] == '?'
939 || text_[i - 1] == ':'
940 || text_[i - 1] == '!'))) {
942 os.texrow().start(owner_->id(), i + 1);
944 } else if (style.free_spacing) {
953 void Paragraph::Private::latexInset(BufferParams const & bparams,
955 OutputParams & runparams,
958 Font const & outerfont,
960 Change & running_change,
961 Layout const & style,
963 unsigned int & column,
964 bool const fontswitch_inset,
965 bool const closeLanguage,
966 bool const lang_switched_at_inset)
968 Inset * inset = owner_->getInset(i);
971 if (style.pass_thru) {
972 odocstringstream ods;
973 inset->plaintext(ods, runparams);
978 // FIXME: move this to InsetNewline::latex
979 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
980 // newlines are handled differently here than
981 // the default in simpleTeXSpecialChars().
982 if (!style.newline_allowed) {
986 bool needPar = false;
987 column += running_font.latexWriteEndChanges(
988 os, bparams, runparams,
989 basefont, basefont, needPar);
993 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
996 basefont = owner_->getLayoutFont(bparams, outerfont);
997 running_font = basefont;
999 if (runparams.moving_arg)
1003 os.texrow().start(owner_->id(), i + 1);
1007 if (owner_->isDeleted(i)) {
1008 if( ++runparams.inDeletedInset == 1)
1009 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1012 if (inset->canTrackChanges()) {
1013 column += Changes::latexMarkChange(os, bparams, running_change,
1014 Change(Change::UNCHANGED), runparams);
1015 running_change = Change(Change::UNCHANGED);
1019 odocstream::pos_type const len = os.os().tellp();
1021 if (inset->forceLTR(runparams)
1022 && running_font.isRightToLeft()
1023 // ERT is an exception, it should be output with no
1024 // decorations at all
1025 && inset->lyxCode() != ERT_CODE) {
1026 if (runparams.use_polyglossia) {
1028 } else if (running_font.language()->lang() == "farsi"
1029 || running_font.language()->lang() == "arabic_arabi")
1030 os << "\\textLR{" << termcmd;
1036 if (open_font && (!inset->inheritFont() || fontswitch_inset)) {
1037 bool lang_closed = false;
1038 // Close language if needed
1039 if (closeLanguage) {
1040 // We need prev_font here as language changes directly at inset
1041 // will only be started inside the inset.
1042 Font const prev_font = (i > 0) ?
1043 owner_->getFont(bparams, i - 1, outerfont)
1045 Font tmpfont(basefont);
1046 tmpfont.setLanguage(prev_font.language());
1047 bool needPar = false;
1048 unsigned int count = tmpfont.latexWriteEndChanges(os, bparams, runparams,
1050 needPar, closeLanguage);
1052 lang_closed = count > 0;
1054 // Update the running_font, making sure, however,
1055 // to leave the language as it was.
1056 // FIXME: probably a better way to keep track of the old
1057 // language, than copying the entire font?
1058 Font const copy_font(running_font);
1059 basefont = owner_->getLayoutFont(bparams, outerfont);
1060 running_font = basefont;
1061 if (!closeLanguage && !lang_switched_at_inset)
1062 running_font.setLanguage(copy_font.language());
1063 // For these, we use switches, so they should be taken as
1064 // base inside the inset.
1065 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1066 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1067 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1068 if (!closeLanguage && lang_switched_at_inset)
1069 basefont.setLanguage(copy_font.language());
1070 // Now re-do font changes in a way needed here
1071 // (using switches with multi-par insets)
1072 InsetText const * textinset = inset->asInsetText();
1073 bool const cprotect = textinset
1074 ? textinset->hasCProtectContent(runparams.moving_arg)
1075 && !textinset->text().isMainText()
1077 unsigned int count2 = basefont.latexWriteStartChanges(os, bparams,
1078 runparams, running_font,
1082 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1086 runparams.local_font = &basefont;
1089 size_t const previous_row_count = os.texrow().rows();
1092 runparams.lastid = id_;
1093 runparams.lastpos = i;
1094 inset->latex(os, runparams);
1095 } catch (EncodingException & e) {
1096 // add location information and throw again.
1105 if (os.texrow().rows() > previous_row_count) {
1106 os.texrow().start(owner_->id(), i + 1);
1109 column += (unsigned int)(os.os().tellp() - len);
1112 if (owner_->isDeleted(i))
1113 --runparams.inDeletedInset;
1117 void Paragraph::Private::latexSpecialChar(otexstream & os,
1118 BufferParams const & bparams,
1119 OutputParams const & runparams,
1120 Font const & running_font,
1121 string & alien_script,
1122 Layout const & style,
1125 unsigned int & column)
1127 char_type const c = owner_->getUChar(bparams, runparams, i);
1129 if (style.pass_thru || runparams.pass_thru
1130 || contains(style.pass_thru_chars, c)
1131 || contains(runparams.pass_thru_chars, c)) {
1133 Encoding const * const enc = runparams.encoding;
1134 if (enc && !enc->encodable(c))
1135 throw EncodingException(c);
1141 // TIPA uses its own T3 encoding
1142 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1144 // If T1 font encoding is used, use the special
1145 // characters it provides.
1146 // NOTE: Some languages reset the font encoding internally to a
1147 // non-standard font encoding. If we are using such a language,
1148 // we do not output special T1 chars.
1149 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1150 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1151 && latexSpecialT1(c, os, i, column))
1153 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1154 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1155 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1158 // Otherwise, we use what LaTeX provides us.
1161 os << "\\textbackslash" << termcmd;
1165 os << "\\textless" << termcmd;
1169 os << "\\textgreater" << termcmd;
1173 os << "\\textbar" << termcmd;
1178 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1179 (end_pos == -1 || i + 1 < end_pos) &&
1180 text_[i+1] == '-') {
1181 // Prevent "--" becoming an en dash and "---" an em dash.
1182 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1188 os << "\\textquotedbl" << termcmd;
1193 case '%': case '#': case '{':
1201 os << "\\textasciitilde" << termcmd;
1206 os << "\\textasciicircum" << termcmd;
1213 // avoid being mistaken for optional arguments
1221 // Blanks are printed before font switching.
1222 // Sure? I am not! (try nice-latex)
1223 // I am sure it's correct. LyX might be smarter
1224 // in the future, but for now, nothing wrong is
1230 // XeTeX's dash behaviour is determined via a global setting
1231 if (bparams.use_dash_ligatures
1232 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1234 // TODO #10961: && not in inset Flex Code
1235 // TODO #10961: && not in layout LyXCode
1236 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1253 Encoding const & encoding = *(runparams.encoding);
1254 char_type next = '\0';
1255 if (i + 1 < int(text_.size())) {
1256 next = text_[i + 1];
1257 if (Encodings::isCombiningChar(next)) {
1258 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1263 pair<docstring, bool> latex = encoding.latexChar(c);
1264 docstring nextlatex;
1265 bool nexttipas = false;
1266 string nexttipashortcut;
1267 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1268 nextlatex = encoding.latexChar(next).first;
1269 if (runparams.inIPA) {
1270 nexttipashortcut = Encodings::TIPAShortcut(next);
1271 nexttipas = !nexttipashortcut.empty();
1275 if (runparams.inIPA) {
1276 string const tipashortcut = Encodings::TIPAShortcut(c);
1277 if (!tipashortcut.empty()) {
1278 latex.first = from_ascii(tipashortcut);
1279 latex.second = false;
1283 // eventually close "script wrapper" command (see `Paragraph::latex`)
1284 if (!alien_script.empty()
1285 && alien_script != Encodings::isKnownScriptChar(next)) {
1286 column += latex.first.length();
1287 alien_script.clear();
1288 os << latex.first << "}";
1292 && ((!prefixIs(nextlatex, '\\')
1293 && !prefixIs(nextlatex, '{')
1294 && !prefixIs(nextlatex, '}'))
1296 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1298 // Prevent eating of a following space or command corruption by
1299 // following characters
1300 if (next == ' ' || next == '\0') {
1301 column += latex.first.length() + 1;
1302 os << latex.first << "{}";
1304 column += latex.first.length();
1305 os << latex.first << " ";
1308 column += latex.first.length() - 1;
1316 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1317 pos_type i, unsigned int & column)
1323 // In T1 encoding, these characters exist
1324 // but we should avoid ligatures
1325 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1327 os << "\\textcompwordmark" << termcmd;
1334 // soul.sty breaks with \char`\"
1335 os << "\\textquotedbl" << termcmd;
1344 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1345 pos_type i, unsigned int & column)
1347 // TU encoding is currently on par with T1.
1348 return latexSpecialT1(c, os, i, column);
1352 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1353 pos_type /*i*/, unsigned int & column)
1363 os << "\\textvertline" << termcmd;
1372 void Paragraph::Private::validate(LaTeXFeatures & features) const
1374 if (layout_->inpreamble && inset_owner_) {
1375 // FIXME: Using a string stream here circumvents the encoding
1376 // switching machinery of odocstream. Therefore the
1377 // output is wrong if this paragraph contains content
1378 // that needs to switch encoding.
1379 Buffer const & buf = inset_owner_->buffer();
1380 otexstringstream os;
1381 os << layout_->preamble();
1382 size_t const length = os.length();
1383 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1384 buf.getParFromID(owner_->id()).pit(), os,
1385 features.runparams(), string(), 0, -1, true);
1386 if (os.length() > length)
1387 features.addPreambleSnippet(os.release(), true);
1390 if (features.runparams().flavor == OutputParams::HTML
1391 && layout_->htmltitle()) {
1392 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1395 // check the params.
1396 if (!params_.spacing().isDefault())
1397 features.require("setspace");
1400 features.useLayout(layout_->name());
1403 fontlist_.validate(features);
1405 // then the indentation
1406 if (!params_.leftIndent().zero())
1407 features.require("ParagraphLeftIndent");
1410 InsetList::const_iterator icit = insetlist_.begin();
1411 InsetList::const_iterator iend = insetlist_.end();
1412 for (; icit != iend; ++icit) {
1414 features.inDeletedInset(owner_->isDeleted(icit->pos));
1415 if (icit->inset->lyxCode() == FOOT_CODE) {
1416 // FIXME: an item inset would make things much easier.
1417 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1418 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1419 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1420 && (icit->pos < begin_of_body_
1421 || (icit->pos == begin_of_body_
1422 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1423 features.saveNoteEnv("description");
1425 icit->inset->validate(features);
1426 features.inDeletedInset(false);
1427 if (layout_->needprotect &&
1428 icit->inset->lyxCode() == FOOT_CODE)
1429 features.require("NeedLyXFootnoteCode");
1433 // then the contents
1434 BufferParams const bp = features.runparams().is_child
1435 ? features.buffer().masterParams() : features.buffer().params();
1436 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1437 char_type c = text_[i];
1438 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1440 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1441 features.require("textquotedblp");
1442 else if (bp.main_font_encoding() != "T1"
1443 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1444 features.require("textquotedbl");
1445 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1446 // features that depend on the font or input encoding
1447 string feats = ci.textpreamble();
1448 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1449 if (fontenc.empty())
1450 fontenc = features.runparams().main_fontenc;
1451 while (!feats.empty()) {
1453 feats = split(feats, feat, ',');
1454 if (contains(feat, "!=")) {
1455 // a feature that is required except for the spcified
1456 // font or input encodings
1458 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1459 // multiple encodings are separated by semicolon
1460 vector<string> context = getVectorFromString(contexts, ";");
1461 // require feature if the context matches neither current font
1462 // nor input encoding
1463 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1464 && std::find(context.begin(), context.end(),
1465 features.runparams().encoding->name()) == context.end())
1466 features.require(realfeature);
1467 } else if (contains(feat, '=')) {
1468 // a feature that is required only for the spcified
1469 // font or input encodings
1471 string const contexts = split(feat, realfeature, '=');
1472 // multiple encodings are separated by semicolon
1473 vector<string> context = getVectorFromString(contexts, ";");
1474 // require feature if the context matches either current font
1475 // or input encoding
1476 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1477 || std::find(context.begin(), context.end(),
1478 features.runparams().encoding->name()) != context.end())
1479 features.require(realfeature);
1482 } else if (!bp.use_dash_ligatures
1483 && (c == 0x2013 || c == 0x2014)
1484 && bp.useNonTeXFonts
1485 && features.runparams().flavor == OutputParams::XETEX)
1486 // XeTeX's dash behaviour is determined via a global setting
1487 features.require("xetexdashbreakstate");
1488 BufferEncodings::validate(c, features);
1492 /////////////////////////////////////////////////////////////////////
1496 /////////////////////////////////////////////////////////////////////
1499 Layout const emptyParagraphLayout;
1502 Paragraph::Paragraph()
1503 : d(new Paragraph::Private(this, emptyParagraphLayout))
1510 Paragraph::Paragraph(Paragraph const & par)
1511 : itemdepth(par.itemdepth),
1512 d(new Paragraph::Private(*par.d, this))
1518 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1519 : itemdepth(par.itemdepth),
1520 d(new Paragraph::Private(*par.d, this, beg, end))
1526 Paragraph & Paragraph::operator=(Paragraph const & par)
1528 // needed as we will destroy the private part before copying it
1530 itemdepth = par.itemdepth;
1534 d = new Private(*par.d, this);
1541 Paragraph::~Paragraph()
1550 // this shall be called just before every "os << ..." action.
1551 void flushString(ostream & os, docstring & s)
1560 void Paragraph::write(ostream & os, BufferParams const & bparams,
1561 depth_type & dth) const
1563 // The beginning or end of a deeper (i.e. nested) area?
1564 if (dth != d->params_.depth()) {
1565 if (d->params_.depth() > dth) {
1566 while (d->params_.depth() > dth) {
1567 os << "\n\\begin_deeper";
1571 while (d->params_.depth() < dth) {
1572 os << "\n\\end_deeper";
1578 // First write the layout
1579 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1581 d->params_.write(os);
1583 Font font1(inherit_font, bparams.language);
1585 Change running_change = Change(Change::UNCHANGED);
1587 // this string is used as a buffer to avoid repetitive calls
1588 // to to_utf8(), which turn out to be expensive (JMarc)
1589 docstring write_buffer;
1592 for (pos_type i = 0; i <= size(); ++i) {
1594 Change const & change = lookupChange(i);
1595 if (change != running_change)
1596 flushString(os, write_buffer);
1597 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1598 running_change = change;
1603 // Write font changes
1604 Font font2 = getFontSettings(bparams, i);
1605 if (font2 != font1) {
1606 flushString(os, write_buffer);
1607 font2.lyxWriteChanges(font1, os);
1612 char_type const c = d->text_[i];
1615 if (Inset const * inset = getInset(i)) {
1616 flushString(os, write_buffer);
1617 if (inset->directWrite()) {
1618 // international char, let it write
1619 // code directly so it's shorter in
1625 os << "\\begin_inset ";
1627 os << "\n\\end_inset\n\n";
1630 // FIXME This can be removed again once the mystery
1631 // crash has been resolved.
1636 flushString(os, write_buffer);
1637 os << "\n\\backslash\n";
1641 flushString(os, write_buffer);
1642 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1649 if ((column > 70 && c == ' ')
1651 flushString(os, write_buffer);
1655 // this check is to amend a bug. LyX sometimes
1656 // inserts '\0' this could cause problems.
1658 write_buffer.push_back(c);
1660 LYXERR0("NUL char in structure.");
1666 flushString(os, write_buffer);
1667 os << "\n\\end_layout\n";
1668 // FIXME This can be removed again once the mystery
1669 // crash has been resolved.
1674 void Paragraph::validate(LaTeXFeatures & features) const
1676 d->validate(features);
1677 bool fragile = features.runparams().moving_arg;
1678 fragile |= layout().needprotect;
1679 if (inInset().getLayout().isNeedProtect())
1681 if (needsCProtection(fragile))
1682 features.require("cprotect");
1686 void Paragraph::insert(pos_type start, docstring const & str,
1687 Font const & font, Change const & change)
1689 for (size_t i = 0, n = str.size(); i != n ; ++i)
1690 insertChar(start + i, str[i], font, change);
1694 void Paragraph::appendChar(char_type c, Font const & font,
1695 Change const & change)
1698 d->changes_.insert(change, d->text_.size());
1699 // when appending characters, no need to update tables
1700 d->text_.push_back(c);
1701 setFont(d->text_.size() - 1, font);
1702 d->requestSpellCheck(d->text_.size() - 1);
1706 void Paragraph::appendString(docstring const & s, Font const & font,
1707 Change const & change)
1709 pos_type end = s.size();
1710 size_t oldsize = d->text_.size();
1711 size_t newsize = oldsize + end;
1712 size_t capacity = d->text_.capacity();
1713 if (newsize >= capacity)
1714 d->text_.reserve(max(capacity + 100, newsize));
1716 // when appending characters, no need to update tables
1719 // FIXME: Optimize this!
1720 for (size_t i = oldsize; i != newsize; ++i) {
1722 d->changes_.insert(change, i);
1723 d->requestSpellCheck(i);
1725 d->fontlist_.set(oldsize, font);
1726 d->fontlist_.set(newsize - 1, font);
1730 void Paragraph::insertChar(pos_type pos, char_type c,
1733 d->insertChar(pos, c, Change(trackChanges ?
1734 Change::INSERTED : Change::UNCHANGED));
1738 void Paragraph::insertChar(pos_type pos, char_type c,
1739 Font const & font, bool trackChanges)
1741 d->insertChar(pos, c, Change(trackChanges ?
1742 Change::INSERTED : Change::UNCHANGED));
1747 void Paragraph::insertChar(pos_type pos, char_type c,
1748 Font const & font, Change const & change)
1750 d->insertChar(pos, c, change);
1755 void Paragraph::resetFonts(Font const & font)
1757 d->fontlist_.clear();
1758 d->fontlist_.set(0, font);
1759 d->fontlist_.set(d->text_.size() - 1, font);
1762 // Gets uninstantiated font setting at position.
1763 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1767 LYXERR0("pos: " << pos << " size: " << size());
1771 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1772 if (cit != d->fontlist_.end())
1775 if (pos == size() && !empty())
1776 return getFontSettings(bparams, pos - 1);
1778 // Optimisation: avoid a full font instantiation if there is no
1779 // language change from previous call.
1780 static Font previous_font;
1781 static Language const * previous_lang = nullptr;
1782 Language const * lang = getParLanguage(bparams);
1783 if (lang != previous_lang) {
1784 previous_lang = lang;
1785 previous_font = Font(inherit_font, lang);
1787 return previous_font;
1791 FontSpan Paragraph::fontSpan(pos_type pos) const
1793 LBUFERR(pos <= size());
1796 return FontSpan(pos, pos);
1799 FontList::const_iterator cit = d->fontlist_.begin();
1800 FontList::const_iterator end = d->fontlist_.end();
1801 for (; cit != end; ++cit) {
1802 if (cit->pos() >= pos) {
1803 if (pos >= beginOfBody())
1804 return FontSpan(max(start, beginOfBody()),
1807 return FontSpan(start,
1808 min(beginOfBody() - 1,
1811 start = cit->pos() + 1;
1814 // This should not happen, but if so, we take no chances.
1815 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1816 LASSERT(false, return FontSpan(pos, pos));
1820 // Gets uninstantiated font setting at position 0
1821 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1823 if (!empty() && !d->fontlist_.empty())
1824 return d->fontlist_.begin()->font();
1826 // Optimisation: avoid a full font instantiation if there is no
1827 // language change from previous call.
1828 static Font previous_font;
1829 static Language const * previous_lang = nullptr;
1830 if (bparams.language != previous_lang) {
1831 previous_lang = bparams.language;
1832 previous_font = Font(inherit_font, bparams.language);
1835 return previous_font;
1839 // Gets the fully instantiated font at a given position in a paragraph
1840 // This is basically the same function as Text::GetFont() in text2.cpp.
1841 // The difference is that this one is used for generating the LaTeX file,
1842 // and thus cosmetic "improvements" are disallowed: This has to deliver
1843 // the true picture of the buffer. (Asger)
1844 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1845 Font const & outerfont) const
1849 Font font = getFontSettings(bparams, pos);
1851 pos_type const body_pos = beginOfBody();
1852 FontInfo & fi = font.fontInfo();
1854 fi.realize(d->layout_->labelfont);
1856 fi.realize(d->layout_->font);
1858 fi.realize(outerfont.fontInfo());
1859 fi.realize(bparams.getFont().fontInfo());
1865 Font const Paragraph::getLabelFont
1866 (BufferParams const & bparams, Font const & outerfont) const
1868 FontInfo tmpfont = d->layout_->labelfont;
1869 tmpfont.realize(outerfont.fontInfo());
1870 tmpfont.realize(bparams.getFont().fontInfo());
1871 return Font(tmpfont, getParLanguage(bparams));
1875 Font const Paragraph::getLayoutFont
1876 (BufferParams const & bparams, Font const & outerfont) const
1878 FontInfo tmpfont = d->layout_->font;
1879 tmpfont.realize(outerfont.fontInfo());
1880 tmpfont.realize(bparams.getFont().fontInfo());
1881 return Font(tmpfont, getParLanguage(bparams));
1885 char_type Paragraph::getUChar(BufferParams const & bparams,
1886 OutputParams const & rp,
1889 char_type c = d->text_[pos];
1891 // Return unchanged character in LTR languages
1892 // or if we use poylglossia/bidi (XeTeX).
1893 if (rp.useBidiPackage()
1894 || !getFontSettings(bparams, pos).isRightToLeft())
1897 // Without polyglossia/bidi, we need to account for some special cases.
1898 // FIXME This needs to be audited!
1900 // * The input is as expected for all delimiters
1901 // => checked for Hebrew!
1902 // * The output matches the display in the LyX workarea
1903 // => checked for Hebrew!
1904 // * The special cases below are really necessary
1905 // => checked for Hebrew!
1906 // * In arabic_arabi, brackets are transformed to Arabic
1907 // Ornate Parentheses. Is this is really wanted?
1909 string const & lang = getFontSettings(bparams, pos).language()->lang();
1912 // 1. In the following languages, parentheses need to be reversed.
1913 // Also with polyglodia/luabidi
1914 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1916 // 2. In the following languages, brackets don't need to be reversed.
1917 bool const reversebrackets = lang != "arabic_arabtex"
1918 && lang != "arabic_arabi"
1921 // Now swap delimiters if needed.
1932 if (reversebrackets)
1936 if (reversebrackets)
1957 void Paragraph::setFont(pos_type pos, Font const & font)
1959 LASSERT(pos <= size(), return);
1961 // First, reduce font against layout/label font
1962 // Update: The setCharFont() routine in text2.cpp already
1963 // reduces font, so we don't need to do that here. (Asger)
1965 d->fontlist_.set(pos, font);
1969 void Paragraph::makeSameLayout(Paragraph const & par)
1971 d->layout_ = par.d->layout_;
1972 d->params_ = par.d->params_;
1976 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1978 if (isFreeSpacing())
1984 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1985 if (eraseChar(pos, trackChanges))
1991 return count > 0 || pos > 0;
1995 bool Paragraph::hasSameLayout(Paragraph const & par) const
1997 return par.d->layout_ == d->layout_
1998 && d->params_.sameLayout(par.d->params_);
2002 depth_type Paragraph::getDepth() const
2004 return d->params_.depth();
2008 depth_type Paragraph::getMaxDepthAfter() const
2010 if (d->layout_->isEnvironment())
2011 return d->params_.depth() + 1;
2013 return d->params_.depth();
2017 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2019 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2020 return getDefaultAlign(bparams);
2022 return d->params_.align();
2026 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2028 LyXAlignment res = layout().align;
2029 if (isRTL(bparams)) {
2031 if (res == LYX_ALIGN_LEFT)
2032 res = LYX_ALIGN_RIGHT;
2033 else if (res == LYX_ALIGN_RIGHT)
2034 res = LYX_ALIGN_LEFT;
2040 docstring const & Paragraph::labelString() const
2042 return d->params_.labelString();
2046 // the next two functions are for the manual labels
2047 docstring const Paragraph::getLabelWidthString() const
2049 if (d->layout_->margintype == MARGIN_MANUAL
2050 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2051 return d->params_.labelWidthString();
2053 return _("Senseless with this layout!");
2057 void Paragraph::setLabelWidthString(docstring const & s)
2059 d->params_.labelWidthString(s);
2063 docstring Paragraph::expandLabel(Layout const & layout,
2064 BufferParams const & bparams) const
2066 return expandParagraphLabel(layout, bparams, true);
2070 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2071 BufferParams const & bparams, bool process_appendix) const
2073 DocumentClass const & tclass = bparams.documentClass();
2074 string const & lang = getParLanguage(bparams)->code();
2075 bool const in_appendix = process_appendix && d->params_.appendix();
2076 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2078 if (fmt.empty() && !layout.counter.empty())
2079 return tclass.counters().theCounter(layout.counter, lang);
2081 // handle 'inherited level parts' in 'fmt',
2082 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2083 size_t const i = fmt.find('@', 0);
2084 if (i != docstring::npos) {
2085 size_t const j = fmt.find('@', i + 1);
2086 if (j != docstring::npos) {
2087 docstring parent(fmt, i + 1, j - i - 1);
2088 docstring label = from_ascii("??");
2089 if (tclass.hasLayout(parent))
2090 label = expandParagraphLabel(tclass[parent], bparams,
2092 fmt = docstring(fmt, 0, i) + label
2093 + docstring(fmt, j + 1, docstring::npos);
2097 return tclass.counters().counterLabel(fmt, lang);
2101 void Paragraph::applyLayout(Layout const & new_layout)
2103 d->layout_ = &new_layout;
2104 LyXAlignment const oldAlign = d->params_.align();
2106 if (!(oldAlign & d->layout_->alignpossible)) {
2107 frontend::Alert::warning(_("Alignment not permitted"),
2108 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2109 d->params_.align(LYX_ALIGN_LAYOUT);
2114 pos_type Paragraph::beginOfBody() const
2116 return d->begin_of_body_;
2120 void Paragraph::setBeginOfBody()
2122 if (d->layout_->labeltype != LABEL_MANUAL) {
2123 d->begin_of_body_ = 0;
2127 // Unroll the first two cycles of the loop
2128 // and remember the previous character to
2129 // remove unnecessary getChar() calls
2131 pos_type end = size();
2132 bool prev_char_deleted = false;
2133 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2136 char_type previous_char = d->text_[i];
2137 if (!(isNewline(i) || isEnvSeparator(i))) {
2139 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2140 char_type temp = d->text_[i];
2141 prev_char_deleted = isDeleted(i);
2142 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2145 previous_char = temp;
2151 d->begin_of_body_ = i;
2155 bool Paragraph::allowParagraphCustomization() const
2157 return inInset().allowParagraphCustomization();
2161 bool Paragraph::usePlainLayout() const
2163 return inInset().usePlainLayout();
2167 bool Paragraph::isPassThru() const
2169 return inInset().isPassThru() || d->layout_->pass_thru;
2174 // paragraphs inside floats need different alignment tags to avoid
2177 bool noTrivlistCentering(InsetCode code)
2179 return code == FLOAT_CODE
2180 || code == WRAP_CODE
2181 || code == CELL_CODE;
2185 string correction(string const & orig)
2187 if (orig == "flushleft")
2188 return "raggedright";
2189 if (orig == "flushright")
2190 return "raggedleft";
2191 if (orig == "center")
2197 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2198 InsetCode code, bool const lastpar, int & col)
2200 string macro = suffix + "{";
2201 if (noTrivlistCentering(code)) {
2203 // the last paragraph in non-trivlist-aligned
2204 // context is special (to avoid unwanted whitespace)
2205 if (suffix == "\\begin") {
2206 macro = "\\" + correction(env) + "{}";
2207 os << from_ascii(macro);
2208 col += macro.size();
2213 macro += correction(env);
2217 if (suffix == "\\par\\end") {
2221 os << from_ascii(macro);
2222 col += macro.size();
2223 if (suffix == "\\begin") {
2233 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2234 otexstream & os, OutputParams const & runparams) const
2239 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2240 (layout_->toggle_indent != ITOGGLE_NEVER) :
2241 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2243 if (canindent && params_.noindent() && !layout_->pass_thru) {
2244 os << "\\noindent ";
2248 LyXAlignment const curAlign = params_.align();
2250 if (curAlign == layout_->align)
2254 case LYX_ALIGN_NONE:
2255 case LYX_ALIGN_BLOCK:
2256 case LYX_ALIGN_LAYOUT:
2257 case LYX_ALIGN_SPECIAL:
2258 case LYX_ALIGN_DECIMAL:
2260 case LYX_ALIGN_LEFT:
2261 case LYX_ALIGN_RIGHT:
2262 case LYX_ALIGN_CENTER:
2263 if (runparams.moving_arg) {
2270 string const begin_tag = "\\begin";
2271 InsetCode code = ownerCode();
2272 bool const lastpar = runparams.isLastPar;
2273 // RTL in classic (PDF)LaTeX (without the Bidi package)
2274 // Luabibdi (used by LuaTeX) behaves like classic
2275 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2276 && !runparams.useBidiPackage();
2279 case LYX_ALIGN_NONE:
2280 case LYX_ALIGN_BLOCK:
2281 case LYX_ALIGN_LAYOUT:
2282 case LYX_ALIGN_SPECIAL:
2283 case LYX_ALIGN_DECIMAL:
2285 case LYX_ALIGN_LEFT: {
2287 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2288 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2290 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2292 } case LYX_ALIGN_RIGHT: {
2294 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2295 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2297 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2299 } case LYX_ALIGN_CENTER: {
2300 corrected_env(os, begin_tag, "center", code, lastpar, column);
2309 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2310 otexstream & os, OutputParams const & runparams) const
2312 LyXAlignment const curAlign = params_.align();
2314 if (curAlign == layout_->align)
2318 case LYX_ALIGN_NONE:
2319 case LYX_ALIGN_BLOCK:
2320 case LYX_ALIGN_LAYOUT:
2321 case LYX_ALIGN_SPECIAL:
2322 case LYX_ALIGN_DECIMAL:
2324 case LYX_ALIGN_LEFT:
2325 case LYX_ALIGN_RIGHT:
2326 case LYX_ALIGN_CENTER:
2327 if (runparams.moving_arg)
2332 bool output = false;
2334 string const end_tag = "\\par\\end";
2335 InsetCode code = ownerCode();
2336 bool const lastpar = runparams.isLastPar;
2337 // RTL in classic (PDF)LaTeX (without the Bidi package)
2338 // Luabibdi (used by LuaTeX) behaves like classic
2339 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2340 && !runparams.useBidiPackage();
2343 case LYX_ALIGN_NONE:
2344 case LYX_ALIGN_BLOCK:
2345 case LYX_ALIGN_LAYOUT:
2346 case LYX_ALIGN_SPECIAL:
2347 case LYX_ALIGN_DECIMAL:
2349 case LYX_ALIGN_LEFT: {
2351 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2352 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2354 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2356 } case LYX_ALIGN_RIGHT: {
2358 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2359 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2361 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2363 } case LYX_ALIGN_CENTER: {
2364 corrected_env(os, end_tag, "center", code, lastpar, col);
2369 return output || lastpar;
2373 // This one spits out the text of the paragraph
2374 void Paragraph::latex(BufferParams const & bparams,
2375 Font const & outerfont,
2377 OutputParams const & runparams,
2378 int start_pos, int end_pos, bool force) const
2380 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2382 // FIXME This check should not be needed. Perhaps issue an
2383 // error if it triggers.
2384 Layout const & style = inInset().forcePlainLayout() ?
2385 bparams.documentClass().plainLayout() : *d->layout_;
2387 if (!force && style.inpreamble)
2390 bool const allowcust = allowParagraphCustomization();
2392 // Current base font for all inherited font changes, without any
2393 // change caused by an individual character, except for the language:
2394 // It is set to the language of the first character.
2395 // As long as we are in the label, this font is the base font of the
2396 // label. Before the first body character it is set to the base font
2400 // If there is an open font-encoding changing command (script wrapper),
2401 // alien_script is set to its name
2402 string alien_script;
2405 // Maybe we have to create a optional argument.
2406 pos_type body_pos = beginOfBody();
2407 unsigned int column = 0;
2409 // If we are inside an non inheritFont() inset, the real outerfont is local_font
2410 Font const real_outerfont = (!inInset().inheritFont()
2411 && runparams.local_font != nullptr)
2412 ? Font(runparams.local_font->fontInfo()) : outerfont;
2415 // the optional argument is kept in curly brackets in
2416 // case it contains a ']'
2417 // This is not strictly needed, but if this is changed it
2418 // would be a file format change, and tex2lyx would need
2419 // to be adjusted, since it unconditionally removes the
2420 // braces when it parses \item.
2423 basefont = getLabelFont(bparams, real_outerfont);
2425 basefont = getLayoutFont(bparams, real_outerfont);
2428 // Which font is currently active?
2429 Font running_font(basefont);
2430 // Do we have an open font change?
2431 bool open_font = false;
2433 Change runningChange = Change(Change::UNCHANGED);
2435 Encoding const * const prev_encoding = runparams.encoding;
2437 os.texrow().start(id(), 0);
2439 // if the paragraph is empty, the loop will not be entered at all
2441 // For InTitle commands, we have already opened a group
2442 // in output_latex::TeXOnePar.
2443 if (style.isCommand() && !style.intitle) {
2447 if (!style.leftdelim().empty()) {
2448 os << style.leftdelim();
2449 column += style.leftdelim().size();
2452 column += d->startTeXParParams(bparams, os, runparams);
2455 // Whether a \par can be issued for insets typeset inline with text.
2456 // Yes if greater than 0. This has to be static.
2457 THREAD_LOCAL_STATIC int parInline = 0;
2459 for (pos_type i = 0; i < size(); ++i) {
2460 // First char in paragraph or after label?
2461 if (i == body_pos) {
2464 bool needPar = false;
2465 column += running_font.latexWriteEndChanges(
2466 os, bparams, runparams,
2467 basefont, basefont, needPar);
2470 basefont = getLayoutFont(bparams, real_outerfont);
2471 running_font = basefont;
2473 column += Changes::latexMarkChange(os, bparams,
2474 runningChange, Change(Change::UNCHANGED),
2476 runningChange = Change(Change::UNCHANGED);
2481 // For InTitle commands, we have already opened a group
2482 // in output_latex::TeXOnePar.
2483 if (style.isCommand() && !style.intitle) {
2488 if (!style.leftdelim().empty()) {
2489 os << style.leftdelim();
2490 column += style.leftdelim().size();
2494 column += d->startTeXParParams(bparams, os,
2498 runparams.wasDisplayMath = runparams.inDisplayMath;
2499 runparams.inDisplayMath = false;
2500 bool deleted_display_math = false;
2501 Change const & change = runparams.inDeletedInset
2502 ? runparams.changeOfDeletedInset : lookupChange(i);
2504 char_type const c = d->text_[i];
2506 // Check whether a display math inset follows
2508 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2510 runparams.ctObject = getInset(i)->CtObject(runparams);
2512 InsetMath const * im = getInset(i)->asInsetMath();
2513 if (im && im->asHullInset()
2514 && im->asHullInset()->outerDisplay()) {
2515 runparams.inDisplayMath = true;
2516 // runparams.inDeletedInset will be set by
2517 // latexInset later, but we need this info
2518 // before it is called. On the other hand, we
2519 // cannot set it here because it is a counter.
2520 deleted_display_math = isDeleted(i);
2522 if (bparams.output_changes && deleted_display_math
2523 && runningChange == change
2524 && change.type == Change::DELETED
2525 && !os.afterParbreak()) {
2526 // A display math in the same paragraph follows.
2527 // We have to close and then reopen \lyxdeleted,
2528 // otherwise the math will be shifted up.
2529 OutputParams rp = runparams;
2531 bool needPar = false;
2532 column += running_font.latexWriteEndChanges(
2533 os, bparams, rp, basefont,
2537 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2538 : getLayoutFont(bparams, real_outerfont);
2539 running_font = basefont;
2540 column += Changes::latexMarkChange(os, bparams,
2541 Change(Change::INSERTED), change, rp);
2545 if (bparams.output_changes && runningChange != change) {
2546 if (!alien_script.empty()) {
2549 alien_script.clear();
2552 bool needPar = false;
2553 column += running_font.latexWriteEndChanges(
2554 os, bparams, runparams,
2555 basefont, basefont, needPar);
2558 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2559 : getLayoutFont(bparams, real_outerfont);
2560 running_font = basefont;
2561 column += Changes::latexMarkChange(os, bparams, runningChange,
2563 runningChange = change;
2566 // do not output text which is marked deleted
2567 // if change tracking output is disabled
2568 if (!bparams.output_changes && change.deleted()) {
2574 // Fully instantiated font
2575 Font const current_font = getFont(bparams, i, outerfont);
2577 Font const prev_font = (i > 0) ?
2578 getFont(bparams, i - 1, outerfont)
2581 Font const last_font = running_font;
2582 bool const in_ct_deletion = (bparams.output_changes
2583 && runningChange == change
2584 && change.type == Change::DELETED
2585 && !os.afterParbreak());
2586 // Insets where font switches are used (rather than font commands)
2587 bool const fontswitch_inset =
2590 && getInset(i)->allowMultiPar()
2591 && !getInset(i)->isPassThru();
2593 // Do we need to close the previous font?
2595 ((current_font != running_font
2596 || current_font.language() != running_font.language())
2597 || (fontswitch_inset
2598 && (current_font == prev_font
2599 || current_font.language() == prev_font.language()))))
2601 // ensure there is no open script-wrapper
2602 if (!alien_script.empty()) {
2605 alien_script.clear();
2607 bool needPar = false;
2608 if (in_ct_deletion) {
2609 // We have to close and then reopen \lyxdeleted,
2610 // as strikeout needs to be on lowest level.
2614 column += running_font.latexWriteEndChanges(
2615 os, bparams, runparams, basefont,
2616 (i == body_pos-1) ? basefont : current_font,
2618 if (in_ct_deletion) {
2619 // We have to close and then reopen \lyxdeleted,
2620 // as strikeout needs to be on lowest level.
2621 OutputParams rp = runparams;
2622 column += Changes::latexMarkChange(os, bparams,
2623 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2625 running_font = basefont;
2629 // if necessary, close language environment before opening CJK
2630 string const running_lang = running_font.language()->babel();
2631 string const lang_end_command = lyxrc.language_command_end;
2632 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2633 && !running_lang.empty()
2634 && running_lang == openLanguageName()
2635 && current_font.language()->encoding()->package() == Encoding::CJK) {
2636 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2637 os << from_ascii(end_tag);
2638 column += end_tag.length();
2642 // Switch file encoding if necessary (and allowed)
2643 if (!runparams.pass_thru && !style.pass_thru &&
2644 runparams.encoding->package() != Encoding::none &&
2645 current_font.language()->encoding()->package() != Encoding::none) {
2646 pair<bool, int> const enc_switch =
2647 switchEncoding(os.os(), bparams, runparams,
2648 *(current_font.language()->encoding()));
2649 if (enc_switch.first) {
2650 column += enc_switch.second;
2651 runparams.encoding = current_font.language()->encoding();
2655 // A display math inset inside an ulem command will be output
2656 // as a box of width \linewidth, so we have to either disable
2657 // indentation if the inset starts a paragraph, or start a new
2658 // line to accommodate such box. This has to be done before
2659 // writing any font changing commands.
2660 if (runparams.inDisplayMath && !deleted_display_math
2661 && runparams.inulemcmd) {
2662 if (os.afterParbreak())
2668 // Do we need to change font?
2669 if ((current_font != running_font ||
2670 current_font.language() != running_font.language())
2671 && i != body_pos - 1)
2673 if (in_ct_deletion) {
2674 // We have to close and then reopen \lyxdeleted,
2675 // as strikeout needs to be on lowest level.
2676 bool needPar = false;
2677 OutputParams rp = runparams;
2678 column += running_font.latexWriteEndChanges(
2679 os, bparams, rp, basefont,
2684 otexstringstream ots;
2685 if (!fontswitch_inset) {
2686 InsetText const * textinset = inInset().asInsetText();
2687 bool const cprotect = textinset
2688 ? textinset->hasCProtectContent(runparams.moving_arg)
2689 && !textinset->text().isMainText()
2691 column += current_font.latexWriteStartChanges(ots, bparams,
2692 runparams, basefont, last_font, false,
2695 // Check again for display math in ulem commands as a
2696 // font change may also occur just before a math inset.
2697 if (runparams.inDisplayMath && !deleted_display_math
2698 && runparams.inulemcmd) {
2699 if (os.afterParbreak())
2704 running_font = current_font;
2706 docstring fontchange = ots.str();
2708 // check whether the fontchange ends with a \\textcolor
2709 // modifier and the text starts with a space. If so we
2710 // need to add } in order to prevent \\textcolor from gobbling
2711 // the space (bug 4473).
2712 docstring const last_modifier = rsplit(fontchange, '\\');
2713 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2714 os << from_ascii("{}");
2715 else if (ots.terminateCommand())
2717 if (in_ct_deletion) {
2718 // We have to close and then reopen \lyxdeleted,
2719 // as strikeout needs to be on lowest level.
2720 OutputParams rp = runparams;
2721 column += Changes::latexMarkChange(os, bparams,
2722 Change(Change::UNCHANGED), change, rp);
2726 // FIXME: think about end_pos implementation...
2727 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2728 // FIXME: integrate this case in latexSpecialChar
2729 // Do not print the separation of the optional argument
2730 // if style.pass_thru is false. This works because
2731 // latexSpecialChar ignores spaces if
2732 // style.pass_thru is false.
2733 if (i != body_pos - 1) {
2734 if (d->simpleTeXBlanks(bparams, runparams, os,
2735 i, column, current_font, style)) {
2736 // A surrogate pair was output. We
2737 // must not call latexSpecialChar
2738 // in this iteration, since it would output
2739 // the combining character again.
2746 OutputParams rp = runparams;
2747 rp.free_spacing = style.free_spacing;
2748 rp.local_font = ¤t_font;
2749 rp.intitle = style.intitle;
2751 // Two major modes: LaTeX or plain
2752 // Handle here those cases common to both modes
2753 // and then split to handle the two modes separately.
2754 if (c == META_INSET) {
2755 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2756 // Greyedout notes and, in general, all insets
2757 // with InsetLayout::isDisplay() == false,
2758 // are typeset inline with the text. So, we
2759 // can add a \par to the last paragraph of
2760 // such insets only if nothing else follows.
2761 bool incremented = false;
2762 Inset const * inset = getInset(i);
2763 InsetText const * textinset = inset
2764 ? inset->asInsetText()
2766 if (i + 1 == size() && textinset
2767 && !inset->getLayout().isDisplay()) {
2768 ParagraphList const & pars =
2769 textinset->text().paragraphs();
2770 pit_type const pit = pars.size() - 1;
2771 Font const lastfont =
2772 pit < 0 || pars[pit].empty()
2773 ? pars[pit].getLayoutFont(
2776 : pars[pit].getFont(bparams,
2777 pars[pit].size() - 1,
2779 if (lastfont.fontInfo().size() !=
2780 basefont.fontInfo().size()) {
2785 // We need to restore parts of this after insets with
2786 // allowMultiPar() true
2787 Font const save_basefont = basefont;
2788 Font const save_runningfont = running_font;
2789 bool closeLanguage = false;
2790 bool lang_switched_at_inset = false;
2791 if (fontswitch_inset) {
2792 // Some insets cannot be inside a font change command.
2793 // However, even such insets *can* be placed in \L or \R
2794 // or their equivalents (for RTL language switches),
2795 // so we don't close the language in those cases
2796 // (= differing isRightToLeft()).
2797 // ArabTeX, though, doesn't seem to handle this special behavior.
2798 closeLanguage = basefont.isRightToLeft() != running_font.isRightToLeft()
2799 && basefont.language()->lang() != "arabic_arabtex"
2800 && running_font.language()->lang() != "arabic_arabtex";
2801 // We need to check prev_font as language changes directly at inset
2802 // will only be started inside the inset.
2803 lang_switched_at_inset = prev_font.language() != running_font.language();
2805 d->latexInset(bparams, os, rp, running_font,
2806 basefont, real_outerfont, open_font,
2807 runningChange, style, i, column,
2808 fontswitch_inset, closeLanguage, lang_switched_at_inset);
2809 if (fontswitch_inset) {
2811 bool needPar = false;
2812 column += running_font.latexWriteEndChanges(
2813 os, bparams, runparams,
2814 basefont, basefont, needPar);
2817 basefont.fontInfo().setSize(save_basefont.fontInfo().size());
2818 basefont.fontInfo().setFamily(save_basefont.fontInfo().family());
2819 basefont.fontInfo().setSeries(save_basefont.fontInfo().series());
2820 if (!closeLanguage && lang_switched_at_inset) {
2821 basefont.setLanguage(save_basefont.language());
2822 running_font.setLanguage(save_runningfont.language());
2828 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2829 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2830 // Close \lyx*deleted and force its
2831 // reopening (if needed)
2834 runningChange = Change(Change::UNCHANGED);
2835 runparams.ctObject = OutputParams::CT_NORMAL;
2838 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2839 if (!bparams.useNonTeXFonts)
2840 script = Encodings::isKnownScriptChar(c);
2841 if (script != alien_script) {
2842 if (!alien_script.empty()) {
2844 alien_script.clear();
2846 string fontenc = running_font.language()->fontenc(bparams);
2848 && !Encodings::fontencSupportsScript(fontenc, script)) {
2849 column += script.length() + 2;
2850 os << "\\" << script << "{";
2851 alien_script = script;
2855 d->latexSpecialChar(os, bparams, rp, running_font,
2856 alien_script, style, i, end_pos, column);
2857 } catch (EncodingException & e) {
2858 if (runparams.dryrun) {
2859 os << "<" << _("LyX Warning: ")
2860 << _("uncodable character") << " '";
2864 // add location information and throw again.
2872 // Set the encoding to that returned from latexSpecialChar (see
2873 // comment for encoding member in OutputParams.h)
2874 runparams.encoding = rp.encoding;
2876 // Also carry on the info on a closed ulem command for insets
2877 // such as Note that do not produce any output, so that no
2878 // command is ever executed but its opening was recorded.
2879 runparams.inulemcmd = rp.inulemcmd;
2881 // These need to be passed upstream as well
2882 runparams.need_maketitle = rp.need_maketitle;
2883 runparams.have_maketitle = rp.have_maketitle;
2885 // And finally, pass the post_macros upstream
2886 runparams.post_macro = rp.post_macro;
2889 // Close wrapper for alien script
2890 if (!alien_script.empty()) {
2892 alien_script.clear();
2895 // If we have an open font definition, we have to close it
2897 // Make sure that \\par is done with the font of the last
2898 // character if this has another size as the default.
2899 // This is necessary because LaTeX (and LyX on the screen)
2900 // calculates the space between the baselines according
2901 // to this font. (Matthias)
2903 // We must not change the font for the last paragraph
2904 // of non-multipar insets, tabular cells or commands,
2905 // since this produces unwanted whitespace.
2907 Font const font = empty()
2908 ? getLayoutFont(bparams, real_outerfont)
2909 : getFont(bparams, size() - 1, real_outerfont);
2911 InsetText const * textinset = inInset().asInsetText();
2913 bool const maintext = textinset
2914 ? textinset->text().isMainText()
2917 size_t const numpars = textinset
2918 ? textinset->text().paragraphs().size()
2921 bool needPar = false;
2923 if (style.resfont.size() != font.fontInfo().size()
2924 && (!runparams.isLastPar || maintext
2925 || (numpars > 1 && d->ownerCode() != CELL_CODE
2926 && (inInset().getLayout().isDisplay()
2928 && !style.isCommand()) {
2931 #ifdef FIXED_LANGUAGE_END_DETECTION
2933 running_font.latexWriteEndChanges(os, bparams,
2934 runparams, basefont,
2935 next_->getFont(bparams, 0, outerfont),
2938 running_font.latexWriteEndChanges(os, bparams,
2939 runparams, basefont, basefont, needPar);
2942 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2943 //FIXME: there as we start another \selectlanguage with the next paragraph if
2944 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2945 running_font.latexWriteEndChanges(os, bparams, runparams,
2946 basefont, basefont, needPar);
2949 // The \par could not be inserted at the same nesting
2950 // level of the font size change, so do it now.
2951 os << "{\\" << font.latexSize() << "\\par}";
2955 column += Changes::latexMarkChange(os, bparams, runningChange,
2956 Change(Change::UNCHANGED), runparams);
2958 // Needed if there is an optional argument but no contents.
2959 if (body_pos > 0 && body_pos == size()) {
2963 if (!style.rightdelim().empty()) {
2964 os << style.rightdelim();
2965 column += style.rightdelim().size();
2968 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2969 && runparams.encoding != prev_encoding) {
2970 runparams.encoding = prev_encoding;
2971 os << setEncoding(prev_encoding->iconvName());
2974 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2978 bool Paragraph::emptyTag() const
2980 for (pos_type i = 0; i < size(); ++i) {
2981 if (Inset const * inset = getInset(i)) {
2982 InsetCode lyx_code = inset->lyxCode();
2983 // FIXME testing like that is wrong. What is
2985 if (lyx_code != TOC_CODE &&
2986 lyx_code != INCLUDE_CODE &&
2987 lyx_code != GRAPHICS_CODE &&
2988 lyx_code != ERT_CODE &&
2989 lyx_code != LISTINGS_CODE &&
2990 lyx_code != FLOAT_CODE &&
2991 lyx_code != TABULAR_CODE) {
2995 char_type c = d->text_[i];
2996 if (c != ' ' && c != '\t')
3004 string Paragraph::getID(Buffer const &, OutputParams const &)
3007 for (pos_type i = 0; i < size(); ++i) {
3008 if (Inset const * inset = getInset(i)) {
3009 InsetCode lyx_code = inset->lyxCode();
3010 if (lyx_code == LABEL_CODE) {
3011 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
3012 docstring const & id = il->getParam("name");
3013 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3021 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3024 for (i = 0; i < size(); ++i) {
3025 if (Inset const * inset = getInset(i)) {
3026 inset->docbook(xs, runparams);
3028 char_type c = d->text_[i];
3038 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3042 for (i = 0; i < size(); ++i) {
3043 if (Inset const * inset = getInset(i)) {
3044 inset->xhtml(xs, runparams);
3046 char_type c = d->text_[i];
3056 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3059 pos_type size = text_.size();
3060 for (pos_type i = initial; i < size; ++i) {
3061 Font font = owner_->getFont(buf.params(), i, outerfont);
3062 if (text_[i] == META_INSET)
3064 if (i != initial && font != font_old)
3075 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3076 vector<xml::EndFontTag> & tagsToClose,
3077 bool & flag, FontState curstate, xml::FontTypes type)
3079 if (curstate == FONT_ON) {
3080 tagsToOpen.push_back(docbookStartFontTag(type));
3083 tagsToClose.push_back(docbookEndFontTag(type));
3088 class OptionalFontType {
3093 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3094 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3097 OptionalFontType fontShapeToXml(FontShape fs)
3101 return {xml::FT_ITALIC};
3103 return {xml::FT_SLANTED};
3104 case SMALLCAPS_SHAPE:
3105 return {xml::FT_SMALLCAPS};
3110 // the other tags are for internal use
3116 OptionalFontType fontFamilyToXml(FontFamily fm)
3120 return {xml::FT_ROMAN};
3122 return {xml::FT_SANS};
3123 case TYPEWRITER_FAMILY:
3124 return {xml::FT_TYPE};
3125 case INHERIT_FAMILY:
3128 // the other tags are for internal use
3134 OptionalFontType fontSizeToXml(FontSize fs)
3138 return {xml::FT_SIZE_TINY};
3140 return {xml::FT_SIZE_SCRIPT};
3142 return {xml::FT_SIZE_FOOTNOTE};
3144 return {xml::FT_SIZE_SMALL};
3146 return {xml::FT_SIZE_LARGE};
3148 return {xml::FT_SIZE_LARGER};
3150 return {xml::FT_SIZE_LARGEST};
3152 return {xml::FT_SIZE_HUGE};
3154 return {xml::FT_SIZE_HUGER};
3156 return {xml::FT_SIZE_INCREASE};
3158 return {xml::FT_SIZE_DECREASE};
3163 // the other tags are for internal use
3169 struct DocBookFontState
3171 FontShape curr_fs = INHERIT_SHAPE;
3172 FontFamily curr_fam = INHERIT_FAMILY;
3173 FontSize curr_size = INHERIT_SIZE;
3175 // track whether we have opened these tags
3176 bool emph_flag = false;
3177 bool bold_flag = false;
3178 bool noun_flag = false;
3179 bool ubar_flag = false;
3180 bool dbar_flag = false;
3181 bool sout_flag = false;
3182 bool xout_flag = false;
3183 bool wave_flag = false;
3185 bool shap_flag = false;
3187 bool faml_flag = false;
3189 bool size_flag = false;
3192 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3194 std::string const & default_family,
3195 DocBookFontState & fs)
3197 vector<xml::FontTag> tagsToOpen;
3198 vector<xml::EndFontTag> tagsToClose;
3201 FontState curstate = font.fontInfo().emph();
3202 if (font_old.emph() != curstate)
3203 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3206 curstate = font.fontInfo().noun();
3207 if (font_old.noun() != curstate)
3208 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3211 curstate = font.fontInfo().underbar();
3212 if (font_old.underbar() != curstate)
3213 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3216 curstate = font.fontInfo().strikeout();
3217 if (font_old.strikeout() != curstate)
3218 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3221 curstate = font.fontInfo().xout();
3222 if (font_old.xout() != curstate)
3223 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3226 curstate = font.fontInfo().uuline();
3227 if (font_old.uuline() != curstate)
3228 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3231 curstate = font.fontInfo().uwave();
3232 if (font_old.uwave() != curstate)
3233 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3236 // a little hackish, but allows us to reuse what we have.
3237 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3238 if (font_old.series() != font.fontInfo().series())
3239 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3242 fs.curr_fs = font.fontInfo().shape();
3243 FontShape old_fs = font_old.shape();
3244 if (old_fs != fs.curr_fs) {
3246 OptionalFontType tag = fontShapeToXml(old_fs);
3248 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3249 fs.shap_flag = false;
3252 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3254 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3258 fs.curr_fam = font.fontInfo().family();
3259 FontFamily old_fam = font_old.family();
3260 if (old_fam != fs.curr_fam) {
3262 OptionalFontType tag = fontFamilyToXml(old_fam);
3264 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3265 fs.faml_flag = false;
3267 switch (fs.curr_fam) {
3269 // we will treat a "default" font family as roman, since we have
3270 // no other idea what to do.
3271 if (default_family != "rmdefault" && default_family != "default") {
3272 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3273 fs.faml_flag = true;
3277 if (default_family != "sfdefault") {
3278 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3279 fs.faml_flag = true;
3282 case TYPEWRITER_FAMILY:
3283 if (default_family != "ttdefault") {
3284 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3285 fs.faml_flag = true;
3288 case INHERIT_FAMILY:
3291 // the other tags are for internal use
3298 fs.curr_size = font.fontInfo().size();
3299 FontSize old_size = font_old.size();
3300 if (old_size != fs.curr_size) {
3302 OptionalFontType tag = fontSizeToXml(old_size);
3304 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3305 fs.size_flag = false;
3308 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3309 if (tag.has_value) {
3310 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3311 fs.size_flag = true;
3315 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3318 } // anonymous namespace
3321 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3323 OutputParams const & runparams,
3324 Font const & outerfont,
3325 bool start_paragraph, bool close_paragraph,
3326 pos_type initial) const
3328 // track whether we have opened these tags
3329 DocBookFontState fs;
3331 if (start_paragraph)
3332 xs.startDivision(allowEmpty());
3334 Layout const & style = *d->layout_;
3336 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3338 string const default_family =
3339 buf.masterBuffer()->params().fonts_default_family;
3341 vector<xml::FontTag> tagsToOpen;
3342 vector<xml::EndFontTag> tagsToClose;
3344 // parsing main loop
3345 for (pos_type i = initial; i < size(); ++i) {
3346 // let's not show deleted material in the output
3350 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3352 // Determine which tags should be opened or closed.
3353 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs);
3356 // Other such tags? What about the other text ranges?
3358 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3359 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3360 for (; cit != cen; ++cit)
3363 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3364 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3365 for (; sit != sen; ++sit)
3368 tagsToClose.clear();
3371 if (Inset const * inset = getInset(i)) {
3372 if (!runparams.for_toc || inset->isInToc()) {
3373 OutputParams np = runparams;
3374 np.local_font = &font;
3375 // If the paragraph has size 1, then we are in the "special
3376 // case" where we do not output the containing paragraph info.
3377 // This "special case" is defined in more details in output_docbook.cpp, makeParagraphs. The results
3378 // of that brittle logic is passed to this function through open_par.
3379 if (!inset->getLayout().htmlisblock() && size() != 1) // TODO: htmlisblock here too!
3380 np.docbook_in_par = true;
3381 inset->docbook(xs, np);
3384 char_type c = getUChar(buf.masterBuffer()->params(), runparams, i);
3387 font_old = font.fontInfo();
3390 // FIXME, this code is just imported from XHTML
3391 // I'm worried about what happens if a branch, say, is itself
3392 // wrapped in some font stuff. I think that will not work.
3394 if (runparams.docbook_in_listing)
3396 if (close_paragraph)
3403 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3404 vector<xml::EndFontTag> & tagsToClose,
3405 bool & flag, FontState curstate, xml::FontTypes type)
3407 if (curstate == FONT_ON) {
3408 tagsToOpen.push_back(xhtmlStartFontTag(type));
3411 tagsToClose.push_back(xhtmlEndFontTag(type));
3416 } // anonymous namespace
3419 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3421 OutputParams const & runparams,
3422 Font const & outerfont,
3423 bool start_paragraph, bool close_paragraph,
3424 pos_type initial) const
3428 // track whether we have opened these tags
3429 bool emph_flag = false;
3430 bool bold_flag = false;
3431 bool noun_flag = false;
3432 bool ubar_flag = false;
3433 bool dbar_flag = false;
3434 bool sout_flag = false;
3435 bool xout_flag = false;
3436 bool wave_flag = false;
3438 bool shap_flag = false;
3440 bool faml_flag = false;
3442 bool size_flag = false;
3444 Layout const & style = *d->layout_;
3446 if (start_paragraph)
3447 xs.startDivision(allowEmpty());
3450 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3452 FontShape curr_fs = INHERIT_SHAPE;
3453 FontFamily curr_fam = INHERIT_FAMILY;
3454 FontSize curr_size = INHERIT_SIZE;
3456 string const default_family =
3457 buf.masterBuffer()->params().fonts_default_family;
3459 vector<xml::FontTag> tagsToOpen;
3460 vector<xml::EndFontTag> tagsToClose;
3462 // parsing main loop
3463 for (pos_type i = initial; i < size(); ++i) {
3464 // let's not show deleted material in the output
3468 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3471 FontState curstate = font.fontInfo().emph();
3472 if (font_old.emph() != curstate)
3473 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3476 curstate = font.fontInfo().noun();
3477 if (font_old.noun() != curstate)
3478 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3481 curstate = font.fontInfo().underbar();
3482 if (font_old.underbar() != curstate)
3483 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3486 curstate = font.fontInfo().strikeout();
3487 if (font_old.strikeout() != curstate)
3488 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3491 curstate = font.fontInfo().xout();
3492 if (font_old.xout() != curstate)
3493 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3496 curstate = font.fontInfo().uuline();
3497 if (font_old.uuline() != curstate)
3498 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3501 curstate = font.fontInfo().uwave();
3502 if (font_old.uwave() != curstate)
3503 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3506 // a little hackish, but allows us to reuse what we have.
3507 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3508 if (font_old.series() != font.fontInfo().series())
3509 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3512 curr_fs = font.fontInfo().shape();
3513 FontShape old_fs = font_old.shape();
3514 if (old_fs != curr_fs) {
3518 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3521 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3523 case SMALLCAPS_SHAPE:
3524 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3530 // the other tags are for internal use
3538 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3542 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3545 case SMALLCAPS_SHAPE:
3546 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3553 // the other tags are for internal use
3560 curr_fam = font.fontInfo().family();
3561 FontFamily old_fam = font_old.family();
3562 if (old_fam != curr_fam) {
3566 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3569 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3571 case TYPEWRITER_FAMILY:
3572 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3574 case INHERIT_FAMILY:
3577 // the other tags are for internal use
3585 // we will treat a "default" font family as roman, since we have
3586 // no other idea what to do.
3587 if (default_family != "rmdefault" && default_family != "default") {
3588 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3593 if (default_family != "sfdefault") {
3594 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3598 case TYPEWRITER_FAMILY:
3599 if (default_family != "ttdefault") {
3600 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3604 case INHERIT_FAMILY:
3607 // the other tags are for internal use
3614 curr_size = font.fontInfo().size();
3615 FontSize old_size = font_old.size();
3616 if (old_size != curr_size) {
3620 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3623 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3626 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3629 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3632 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3635 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3638 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3641 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3644 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3647 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3650 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3656 // the other tags are for internal use
3662 switch (curr_size) {
3664 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3668 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3672 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3676 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3680 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3684 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3688 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3692 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3696 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3700 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3704 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3711 // the other tags are for internal use
3718 // Other such tags? What about the other text ranges?
3720 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3721 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3722 for (; cit != cen; ++cit)
3725 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3726 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3727 for (; sit != sen; ++sit)
3730 tagsToClose.clear();
3733 Inset const * inset = getInset(i);
3735 if (!runparams.for_toc || inset->isInToc()) {
3736 OutputParams np = runparams;
3737 np.local_font = &font;
3738 // If the paragraph has size 1, then we are in the "special
3739 // case" where we do not output the containing paragraph info
3740 if (!inset->getLayout().htmlisblock() && size() != 1)
3741 np.html_in_par = true;
3742 retval += inset->xhtml(xs, np);
3745 char_type c = getUChar(buf.masterBuffer()->params(),
3747 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3748 xs << XMLStream::ESCAPE_NONE << " ";
3752 font_old = font.fontInfo();
3756 // I'm worried about what happens if a branch, say, is itself
3757 // wrapped in some font stuff. I think that will not work.
3759 if (close_paragraph)
3766 bool Paragraph::isHfill(pos_type pos) const
3768 Inset const * inset = getInset(pos);
3769 return inset && inset->isHfill();
3773 bool Paragraph::isNewline(pos_type pos) const
3775 // U+2028 LINE SEPARATOR
3776 // U+2029 PARAGRAPH SEPARATOR
3777 char_type const c = d->text_[pos];
3778 if (c == 0x2028 || c == 0x2029)
3780 Inset const * inset = getInset(pos);
3781 return inset && inset->lyxCode() == NEWLINE_CODE;
3785 bool Paragraph::isEnvSeparator(pos_type pos) const
3787 Inset const * inset = getInset(pos);
3788 return inset && inset->lyxCode() == SEPARATOR_CODE;
3792 bool Paragraph::isLineSeparator(pos_type pos) const
3794 char_type const c = d->text_[pos];
3795 if (isLineSeparatorChar(c))
3797 Inset const * inset = getInset(pos);
3798 return inset && inset->isLineSeparator();
3802 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3806 if (ignore_deleted && isDeleted(pos))
3808 if (Inset const * inset = getInset(pos))
3809 return !inset->isLetter();
3810 // if we have a hard hyphen (no en- or emdash) or apostrophe
3811 // we pass this to the spell checker
3812 // FIXME: this method is subject to change, visit
3813 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3814 // to get an impression how complex this is.
3815 if (isHardHyphenOrApostrophe(pos))
3817 char_type const c = d->text_[pos];
3818 // We want to pass the escape chars to the spellchecker
3819 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3820 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3824 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3826 pos_type const psize = size();
3829 char_type const c = d->text_[pos];
3830 if (c != '-' && c != '\'')
3832 int nextpos = pos + 1;
3833 int prevpos = pos > 0 ? pos - 1 : 0;
3834 if ((nextpos == psize || isSpace(nextpos))
3835 && (pos == 0 || isSpace(prevpos)))
3841 bool Paragraph::needsCProtection(bool const fragile) const
3843 // first check the layout of the paragraph, but only in insets
3844 InsetText const * textinset = inInset().asInsetText();
3845 bool const maintext = textinset
3846 ? textinset->text().isMainText()
3849 if (!maintext && layout().needcprotect) {
3850 // Environments need cprotection regardless the content
3851 if (layout().latextype == LATEX_ENVIRONMENT)
3854 // Commands need cprotection if they contain specific chars
3855 int const nchars_escape = 9;
3856 static char_type const chars_escape[nchars_escape] = {
3857 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3859 docstring const pars = asString();
3860 for (int k = 0; k < nchars_escape; k++) {
3861 if (contains(pars, chars_escape[k]))
3866 // now check whether we have insets that need cprotection
3867 pos_type size = pos_type(d->text_.size());
3868 for (pos_type i = 0; i < size; ++i) {
3871 Inset const * ins = getInset(i);
3872 if (ins->needsCProtection(maintext, fragile))
3874 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3875 // Environments need cprotection regardless the content
3877 // Now check math environments
3878 InsetMath const * im = getInset(i)->asInsetMath();
3879 if (!im || im->cell(0).empty())
3881 switch(im->cell(0)[0]->lyxCode()) {
3882 case MATH_AMSARRAY_CODE:
3883 case MATH_SUBSTACK_CODE:
3885 case MATH_XYMATRIX_CODE:
3886 // these need cprotection
3897 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3899 return d->speller_state_.getRange(pos);
3903 bool Paragraph::isChar(pos_type pos) const
3905 if (Inset const * inset = getInset(pos))
3906 return inset->isChar();
3907 char_type const c = d->text_[pos];
3908 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3912 bool Paragraph::isSpace(pos_type pos) const
3914 if (Inset const * inset = getInset(pos))
3915 return inset->isSpace();
3916 char_type const c = d->text_[pos];
3917 return lyx::isSpace(c);
3922 Paragraph::getParLanguage(BufferParams const & bparams) const
3925 return getFirstFontSettings(bparams).language();
3926 // FIXME: we should check the prev par as well (Lgb)
3927 return bparams.language;
3931 bool Paragraph::isRTL(BufferParams const & bparams) const
3933 return getParLanguage(bparams)->rightToLeft()
3934 && !inInset().getLayout().forceLTR();
3938 void Paragraph::changeLanguage(BufferParams const & bparams,
3939 Language const * from, Language const * to)
3941 // change language including dummy font change at the end
3942 for (pos_type i = 0; i <= size(); ++i) {
3943 Font font = getFontSettings(bparams, i);
3944 if (font.language() == from) {
3945 font.setLanguage(to);
3947 d->requestSpellCheck(i);
3953 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3955 Language const * doc_language = bparams.language;
3956 for (auto const & f : d->fontlist_)
3957 if (f.font().language() != ignore_language &&
3958 f.font().language() != latex_language &&
3959 f.font().language() != doc_language)
3965 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3967 for (auto const & f : d->fontlist_) {
3968 Language const * lang = f.font().language();
3969 if (lang != ignore_language &&
3970 lang != latex_language)
3976 docstring Paragraph::asString(int options) const
3978 return asString(0, size(), options);
3982 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3984 odocstringstream os;
3987 && options & AS_STR_LABEL
3988 && !d->params_.labelString().empty())
3989 os << d->params_.labelString() << ' ';
3991 for (pos_type i = beg; i < end; ++i) {
3992 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3994 char_type const c = d->text_[i];
3995 if (isPrintable(c) || c == '\t'
3996 || (c == '\n' && (options & AS_STR_NEWLINES)))
3998 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3999 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
4000 LASSERT(runparams != nullptr, return docstring());
4001 getInset(i)->plaintext(os, *runparams);
4003 getInset(i)->toString(os);
4012 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
4013 bool const shorten, bool const label) const
4015 size_t tmplen = shorten ? maxlen + 1 : maxlen;
4016 if (label && !labelString().empty())
4017 os += labelString() + ' ';
4018 if (!layout().isTocCaption())
4020 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4023 char_type const c = d->text_[i];
4026 else if (c == META_INSET)
4027 getInset(i)->forOutliner(os, tmplen, false);
4030 Text::shortenForOutliner(os, maxlen);
4034 void Paragraph::setInsetOwner(Inset const * inset)
4036 d->inset_owner_ = inset;
4040 int Paragraph::id() const
4046 void Paragraph::setId(int id)
4052 Layout const & Paragraph::layout() const
4058 void Paragraph::setLayout(Layout const & layout)
4060 d->layout_ = &layout;
4064 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4066 setLayout(tc.defaultLayout());
4070 void Paragraph::setPlainLayout(DocumentClass const & tc)
4072 setLayout(tc.plainLayout());
4076 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
4078 if (usePlainLayout())
4079 setPlainLayout(tclass);
4081 setDefaultLayout(tclass);
4085 Inset const & Paragraph::inInset() const
4087 LBUFERR(d->inset_owner_);
4088 return *d->inset_owner_;
4092 ParagraphParameters & Paragraph::params()
4098 ParagraphParameters const & Paragraph::params() const
4104 bool Paragraph::isFreeSpacing() const
4106 if (d->layout_->free_spacing)
4108 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4112 bool Paragraph::allowEmpty() const
4114 if (d->layout_->keepempty)
4116 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4120 bool Paragraph::brokenBiblio() const
4122 // There is a problem if there is no bibitem at position 0 in
4123 // paragraphs that need one, if there is another bibitem in the
4124 // paragraph or if this paragraph is not supposed to have
4125 // a bibitem inset at all.
4126 return ((d->layout_->labeltype == LABEL_BIBLIO
4127 && (d->insetlist_.find(BIBITEM_CODE) != 0
4128 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4129 || (d->layout_->labeltype != LABEL_BIBLIO
4130 && d->insetlist_.find(BIBITEM_CODE) != -1));
4134 int Paragraph::fixBiblio(Buffer const & buffer)
4136 // FIXME: when there was already an inset at 0, the return value is 1,
4137 // which does not tell whether another inset has been removed; the
4138 // cursor cannot be correctly updated.
4140 bool const track_changes = buffer.params().track_changes;
4141 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4143 // The case where paragraph is not BIBLIO
4144 if (d->layout_->labeltype != LABEL_BIBLIO) {
4145 if (bibitem_pos == -1)
4146 // No InsetBibitem => OK
4148 // There is an InsetBibitem: remove it!
4149 d->insetlist_.release(bibitem_pos);
4150 eraseChar(bibitem_pos, track_changes);
4151 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4154 bool const hasbibitem0 = bibitem_pos == 0;
4156 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4157 // There was an InsetBibitem at pos 0,
4158 // and no other one => OK
4159 if (bibitem_pos == -1)
4161 // there is a bibitem at the 0 position, but since
4162 // there is a second one, we copy the second on the
4163 // first. We're assuming there are at most two of
4164 // these, which there should be.
4165 // FIXME: why does it make sense to do that rather
4166 // than keep the first? (JMarc)
4167 Inset * inset = releaseInset(bibitem_pos);
4168 d->insetlist_.begin()->inset = inset;
4169 // This needs to be done to update the counter (#8499)
4170 buffer.updateBuffer();
4171 return -bibitem_pos;
4174 // We need to create an inset at the beginning
4175 Inset * inset = nullptr;
4176 if (bibitem_pos > 0) {
4177 // there was one somewhere in the paragraph, let's move it
4178 inset = d->insetlist_.release(bibitem_pos);
4179 eraseChar(bibitem_pos, track_changes);
4182 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4183 InsetCommandParams(BIBITEM_CODE));
4185 Font font(inherit_font, buffer.params().language);
4186 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4187 : Change::UNCHANGED));
4189 // This is needed to get the counters right
4190 buffer.updateBuffer();
4195 void Paragraph::checkAuthors(AuthorList const & authorList)
4197 d->changes_.checkAuthors(authorList);
4201 bool Paragraph::isChanged(pos_type pos) const
4203 return lookupChange(pos).changed();
4207 bool Paragraph::isInserted(pos_type pos) const
4209 return lookupChange(pos).inserted();
4213 bool Paragraph::isDeleted(pos_type pos) const
4215 return lookupChange(pos).deleted();
4219 InsetList const & Paragraph::insetList() const
4221 return d->insetlist_;
4225 void Paragraph::setInsetBuffers(Buffer & b)
4227 d->insetlist_.setBuffer(b);
4231 void Paragraph::resetBuffer()
4233 d->insetlist_.resetBuffer();
4237 Inset * Paragraph::releaseInset(pos_type pos)
4239 Inset * inset = d->insetlist_.release(pos);
4240 /// does not honour change tracking!
4241 eraseChar(pos, false);
4246 Inset * Paragraph::getInset(pos_type pos)
4248 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4249 ? d->insetlist_.get(pos) : nullptr;
4253 Inset const * Paragraph::getInset(pos_type pos) const
4255 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4256 ? d->insetlist_.get(pos) : nullptr;
4260 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4261 pos_type & right, TextCase action)
4263 // process sequences of modified characters; in change
4264 // tracking mode, this approach results in much better
4265 // usability than changing case on a char-by-char basis
4266 // We also need to track the current font, since font
4267 // changes within sequences can occur.
4268 vector<pair<char_type, Font> > changes;
4270 bool const trackChanges = bparams.track_changes;
4272 bool capitalize = true;
4274 for (; pos < right; ++pos) {
4275 char_type oldChar = d->text_[pos];
4276 char_type newChar = oldChar;
4278 // ignore insets and don't play with deleted text!
4279 if (oldChar != META_INSET && !isDeleted(pos)) {
4281 case text_lowercase:
4282 newChar = lowercase(oldChar);
4284 case text_capitalization:
4286 newChar = uppercase(oldChar);
4290 case text_uppercase:
4291 newChar = uppercase(oldChar);
4296 if (isWordSeparator(pos) || isDeleted(pos)) {
4297 // permit capitalization again
4301 if (oldChar != newChar) {
4302 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4303 if (pos != right - 1)
4305 // step behind the changing area
4309 int erasePos = pos - changes.size();
4310 for (size_t i = 0; i < changes.size(); i++) {
4311 insertChar(pos, changes[i].first,
4314 if (!eraseChar(erasePos, trackChanges)) {
4317 ++right; // expand selection
4325 int Paragraph::find(docstring const & str, bool cs, bool mw,
4326 pos_type start_pos, bool del) const
4328 pos_type pos = start_pos;
4329 int const strsize = str.length();
4331 pos_type const parsize = d->text_.size();
4332 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4333 // Ignore "invisible" letters such as ligature breaks
4334 // and hyphenation chars while searching
4335 while (pos < parsize - 1 && isInset(pos)) {
4336 Inset const * inset = getInset(pos);
4337 if (!inset->isLetter())
4339 odocstringstream os;
4340 inset->toString(os);
4341 if (!os.str().empty())
4345 if (cs && str[i] != d->text_[pos])
4347 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4349 if (!del && isDeleted(pos))
4356 // if necessary, check whether string matches word
4358 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4361 && !isWordSeparator(pos))
4365 return pos - start_pos;
4369 char_type Paragraph::getChar(pos_type pos) const
4371 return d->text_[pos];
4375 pos_type Paragraph::size() const
4377 return d->text_.size();
4381 bool Paragraph::empty() const
4383 return d->text_.empty();
4387 bool Paragraph::isInset(pos_type pos) const
4389 return d->text_[pos] == META_INSET;
4393 bool Paragraph::isSeparator(pos_type pos) const
4395 //FIXME: Are we sure this can be the only separator?
4396 return d->text_[pos] == ' ';
4400 void Paragraph::deregisterWords()
4402 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4403 Private::LangWordsMap::const_iterator ite = d->words_.end();
4404 for (; itl != ite; ++itl) {
4405 WordList & wl = theWordList(itl->first);
4406 Private::Words::const_iterator it = (itl->second).begin();
4407 Private::Words::const_iterator et = (itl->second).end();
4408 for (; it != et; ++it)
4415 void Paragraph::locateWord(pos_type & from, pos_type & to,
4416 word_location const loc, bool const ignore_deleted) const
4419 case WHOLE_WORD_STRICT:
4420 if (from == 0 || from == size()
4421 || isWordSeparator(from, ignore_deleted)
4422 || isWordSeparator(from - 1, ignore_deleted)) {
4429 // If we are already at the beginning of a word, do nothing
4430 if (!from || isWordSeparator(from - 1, ignore_deleted))
4435 // always move the cursor to the beginning of previous word
4436 while (from && !isWordSeparator(from - 1, ignore_deleted))
4440 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4443 // no need to move the 'from' cursor
4447 while (to < size() && !isWordSeparator(to, ignore_deleted))
4452 void Paragraph::collectWords()
4454 for (pos_type pos = 0; pos < size(); ++pos) {
4455 if (isWordSeparator(pos))
4457 pos_type from = pos;
4458 locateWord(from, pos, WHOLE_WORD);
4459 // Work around MSVC warning: The statement
4460 // if (pos < from + lyxrc.completion_minlength)
4461 // triggers a signed vs. unsigned warning.
4462 // I don't know why this happens, it could be a MSVC bug, or
4463 // related to LLP64 (windows) vs. LP64 (unix) programming
4464 // model, or the C++ standard might be ambigous in the section
4465 // defining the "usual arithmetic conversions". However, using
4466 // a temporary variable is safe and works on all compilers.
4467 pos_type const endpos = from + lyxrc.completion_minlength;
4470 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4471 if (cit == d->fontlist_.end())
4473 Language const * lang = cit->font().language();
4474 docstring const word = asString(from, pos, AS_STR_NONE);
4475 d->words_[lang->lang()].insert(word);
4480 void Paragraph::registerWords()
4482 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4483 Private::LangWordsMap::const_iterator ite = d->words_.end();
4484 for (; itl != ite; ++itl) {
4485 WordList & wl = theWordList(itl->first);
4486 Private::Words::const_iterator it = (itl->second).begin();
4487 Private::Words::const_iterator et = (itl->second).end();
4488 for (; it != et; ++it)
4494 void Paragraph::updateWords()
4502 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4504 SkipPositionsIterator begin = skips.begin();
4505 SkipPositions::iterator end = skips.end();
4506 if (pos > 0 && begin < end) {
4508 if (end->last == pos - 1) {
4513 skips.insert(end, FontSpan(pos, pos));
4517 Language * Paragraph::Private::locateSpellRange(
4518 pos_type & from, pos_type & to,
4519 SkipPositions & skips) const
4521 // skip leading white space
4522 while (from < to && owner_->isWordSeparator(from))
4524 // don't check empty range
4527 // get current language
4528 Language * lang = getSpellLanguage(from);
4529 pos_type last = from;
4530 bool samelang = true;
4531 bool sameinset = true;
4532 while (last < to && samelang && sameinset) {
4533 // hop to end of word
4534 while (last < to && !owner_->isWordSeparator(last)) {
4535 Inset const * inset = owner_->getInset(last);
4536 if (inset && inset->lyxCode() == SPECIALCHAR_CODE) {
4537 // check for "invisible" letters such as ligature breaks
4538 odocstringstream os;
4539 inset->toString(os);
4540 if (os.str().length() != 0) {
4541 // avoid spell check of visible special char insets
4542 // stop the loop in front of the special char inset
4547 appendSkipPosition(skips, last);
4548 } else if (owner_->isDeleted(last)) {
4549 appendSkipPosition(skips, last);
4553 // hop to next word while checking for insets
4554 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4555 if (Inset const * inset = owner_->getInset(last))
4556 sameinset = inset->isChar() && inset->isLetter();
4557 if (sameinset && owner_->isDeleted(last)) {
4558 appendSkipPosition(skips, last);
4563 if (sameinset && last < to) {
4564 // now check for language change
4565 samelang = lang == getSpellLanguage(last);
4568 // if language change detected backstep is needed
4576 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4579 const_cast<Language *>(owner_->getFontSettings(
4580 inset_owner_->buffer().params(), from).language());
4581 if (lang == inset_owner_->buffer().params().language
4582 && !lyxrc.spellchecker_alt_lang.empty()) {
4584 string const lang_variety =
4585 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4586 lang->setCode(lang_code);
4587 lang->setVariety(lang_variety);
4593 void Paragraph::requestSpellCheck(pos_type pos)
4595 d->requestSpellCheck(pos);
4599 bool Paragraph::needsSpellCheck() const
4601 SpellChecker::ChangeNumber speller_change_number = 0;
4602 if (theSpellChecker())
4603 speller_change_number = theSpellChecker()->changeNumber();
4604 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4605 d->speller_state_.needsCompleteRefresh(speller_change_number);
4607 return d->needsSpellCheck();
4611 bool Paragraph::Private::ignoreWord(docstring const & word) const
4613 // Ignore words with digits
4614 // FIXME: make this customizable
4615 // (note that some checkers ignore words with digits by default)
4616 docstring::const_iterator cit = word.begin();
4617 docstring::const_iterator const end = word.end();
4618 for (; cit != end; ++cit) {
4619 if (isNumber((*cit)))
4626 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4627 WordLangTuple & wl, docstring_list & suggestions,
4628 bool do_suggestion, bool check_learned) const
4630 SpellChecker::Result result = SpellChecker::WORD_OK;
4631 SpellChecker * speller = theSpellChecker();
4635 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4638 locateWord(from, to, WHOLE_WORD, true);
4639 if (from == to || from >= size())
4642 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4643 Language * lang = d->getSpellLanguage(from);
4645 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4648 wl = WordLangTuple(word, lang);
4653 if (needsSpellCheck() || check_learned) {
4655 if (!d->ignoreWord(word)) {
4656 bool const trailing_dot = to < size() && d->text_[to] == '.';
4657 result = speller->check(wl);
4658 if (SpellChecker::misspelled(result) && trailing_dot) {
4659 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4660 result = speller->check(wl);
4661 if (!SpellChecker::misspelled(result)) {
4662 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4664 from << ".." << to << "]");
4666 // spell check with dot appended failed too
4667 // restore original word/lang value
4668 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4669 wl = WordLangTuple(word, lang);
4673 if (!SpellChecker::misspelled(result)) {
4674 // area up to the begin of the next word is not misspelled
4675 while (end < size() && isWordSeparator(end))
4678 d->setMisspelled(from, end, result);
4680 result = d->speller_state_.getState(from);
4684 suggestions.clear();
4686 if (SpellChecker::misspelled(result)) {
4687 LYXERR(Debug::GUI, "misspelled word: \"" <<
4689 from << ".." << to << "]");
4691 speller->suggest(wl, suggestions);
4697 void Paragraph::anonymize()
4699 // This is a very crude anonymization for now
4700 for (char_type & c : d->text_)
4701 if (isLetterChar(c) || isNumber(c))
4706 void Paragraph::Private::markMisspelledWords(
4707 pos_type const & first, pos_type const & last,
4708 SpellChecker::Result result,
4709 docstring const & word,
4710 SkipPositions const & skips)
4712 if (!SpellChecker::misspelled(result)) {
4713 setMisspelled(first, last, SpellChecker::WORD_OK);
4717 SpellChecker * speller = theSpellChecker();
4718 // locate and enumerate the error positions
4719 int nerrors = speller->numMisspelledWords();
4721 SkipPositionsIterator it = skips.begin();
4722 SkipPositionsIterator et = skips.end();
4723 for (int index = 0; index < nerrors; ++index) {
4726 speller->misspelledWord(index, wstart, wlen);
4727 /// should not happen if speller supports range checks
4728 if (!wlen) continue;
4729 docstring const misspelled = word.substr(wstart, wlen);
4730 wstart += first + numskipped;
4731 if (snext < wstart) {
4732 /// mark the range of correct spelling
4733 numskipped += countSkips(it, et, wstart);
4734 setMisspelled(snext,
4735 wstart - 1, SpellChecker::WORD_OK);
4737 snext = wstart + wlen;
4738 numskipped += countSkips(it, et, snext);
4739 /// mark the range of misspelling
4740 setMisspelled(wstart, snext, result);
4741 LYXERR(Debug::GUI, "misspelled word: \"" <<
4742 misspelled << "\" [" <<
4743 wstart << ".." << (snext-1) << "]");
4746 if (snext <= last) {
4747 /// mark the range of correct spelling at end
4748 setMisspelled(snext, last, SpellChecker::WORD_OK);
4753 void Paragraph::spellCheck() const
4755 SpellChecker * speller = theSpellChecker();
4756 if (!speller || empty() ||!needsSpellCheck())
4760 d->rangeOfSpellCheck(start, endpos);
4761 if (speller->canCheckParagraph()) {
4762 // loop until we leave the range
4763 for (pos_type first = start; first < endpos; ) {
4764 pos_type last = endpos;
4765 Private::SkipPositions skips;
4766 Language * lang = d->locateSpellRange(first, last, skips);
4767 if (first >= endpos)
4769 // start the spell checker on the unit of meaning
4770 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4771 WordLangTuple wl = WordLangTuple(word, lang);
4772 SpellChecker::Result result = word.size() ?
4773 speller->check(wl) : SpellChecker::WORD_OK;
4774 d->markMisspelledWords(first, last, result, word, skips);
4778 static docstring_list suggestions;
4779 pos_type to = endpos;
4780 while (start < endpos) {
4782 spellCheck(start, to, wl, suggestions, false);
4786 d->readySpellCheck();
4790 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4792 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4793 if (result || pos <= 0 || pos > size())
4795 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4796 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4801 string Paragraph::magicLabel() const
4804 ss << "magicparlabel-" << id();