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 && 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;
1062 running_font.setLanguage(copy_font.language());
1063 OutputParams rp = runparams;
1064 rp.encoding = basefont.language()->encoding();
1065 // For these, we use switches, so they should be taken as
1066 // base inside the inset.
1067 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1068 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1069 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
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,
1083 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1087 runparams.local_font = &basefont;
1090 size_t const previous_row_count = os.texrow().rows();
1093 runparams.lastid = id_;
1094 runparams.lastpos = i;
1095 inset->latex(os, runparams);
1096 } catch (EncodingException & e) {
1097 // add location information and throw again.
1106 if (os.texrow().rows() > previous_row_count) {
1107 os.texrow().start(owner_->id(), i + 1);
1110 column += (unsigned int)(os.os().tellp() - len);
1113 if (owner_->isDeleted(i))
1114 --runparams.inDeletedInset;
1118 void Paragraph::Private::latexSpecialChar(otexstream & os,
1119 BufferParams const & bparams,
1120 OutputParams const & runparams,
1121 Font const & running_font,
1122 string & alien_script,
1123 Layout const & style,
1126 unsigned int & column)
1128 char_type const c = owner_->getUChar(bparams, runparams, i);
1130 if (style.pass_thru || runparams.pass_thru
1131 || contains(style.pass_thru_chars, c)
1132 || contains(runparams.pass_thru_chars, c)) {
1134 Encoding const * const enc = runparams.encoding;
1135 if (enc && !enc->encodable(c))
1136 throw EncodingException(c);
1142 // TIPA uses its own T3 encoding
1143 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1145 // If T1 font encoding is used, use the special
1146 // characters it provides.
1147 // NOTE: Some languages reset the font encoding internally to a
1148 // non-standard font encoding. If we are using such a language,
1149 // we do not output special T1 chars.
1150 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1151 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1152 && latexSpecialT1(c, os, i, column))
1154 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1155 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1156 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1159 // Otherwise, we use what LaTeX provides us.
1162 os << "\\textbackslash" << termcmd;
1166 os << "\\textless" << termcmd;
1170 os << "\\textgreater" << termcmd;
1174 os << "\\textbar" << termcmd;
1179 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1180 (end_pos == -1 || i + 1 < end_pos) &&
1181 text_[i+1] == '-') {
1182 // Prevent "--" becoming an en dash and "---" an em dash.
1183 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1189 os << "\\textquotedbl" << termcmd;
1194 case '%': case '#': case '{':
1202 os << "\\textasciitilde" << termcmd;
1207 os << "\\textasciicircum" << termcmd;
1214 // avoid being mistaken for optional arguments
1222 // Blanks are printed before font switching.
1223 // Sure? I am not! (try nice-latex)
1224 // I am sure it's correct. LyX might be smarter
1225 // in the future, but for now, nothing wrong is
1231 // XeTeX's dash behaviour is determined via a global setting
1232 if (bparams.use_dash_ligatures
1233 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1235 // TODO #10961: && not in inset Flex Code
1236 // TODO #10961: && not in layout LyXCode
1237 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1254 Encoding const & encoding = *(runparams.encoding);
1255 char_type next = '\0';
1256 if (i + 1 < int(text_.size())) {
1257 next = text_[i + 1];
1258 if (Encodings::isCombiningChar(next)) {
1259 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1264 pair<docstring, bool> latex = encoding.latexChar(c);
1265 docstring nextlatex;
1266 bool nexttipas = false;
1267 string nexttipashortcut;
1268 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1269 nextlatex = encoding.latexChar(next).first;
1270 if (runparams.inIPA) {
1271 nexttipashortcut = Encodings::TIPAShortcut(next);
1272 nexttipas = !nexttipashortcut.empty();
1276 if (runparams.inIPA) {
1277 string const tipashortcut = Encodings::TIPAShortcut(c);
1278 if (!tipashortcut.empty()) {
1279 latex.first = from_ascii(tipashortcut);
1280 latex.second = false;
1284 // eventually close "script wrapper" command (see `Paragraph::latex`)
1285 if (!alien_script.empty()
1286 && alien_script != Encodings::isKnownScriptChar(next)) {
1287 column += latex.first.length();
1288 alien_script.clear();
1289 os << latex.first << "}";
1293 && ((!prefixIs(nextlatex, '\\')
1294 && !prefixIs(nextlatex, '{')
1295 && !prefixIs(nextlatex, '}'))
1297 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1299 // Prevent eating of a following space or command corruption by
1300 // following characters
1301 if (next == ' ' || next == '\0') {
1302 column += latex.first.length() + 1;
1303 os << latex.first << "{}";
1305 column += latex.first.length();
1306 os << latex.first << " ";
1309 column += latex.first.length() - 1;
1317 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1318 pos_type i, unsigned int & column)
1324 // In T1 encoding, these characters exist
1325 // but we should avoid ligatures
1326 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1328 os << "\\textcompwordmark" << termcmd;
1335 // soul.sty breaks with \char`\"
1336 os << "\\textquotedbl" << termcmd;
1345 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1346 pos_type i, unsigned int & column)
1348 // TU encoding is currently on par with T1.
1349 return latexSpecialT1(c, os, i, column);
1353 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1354 pos_type /*i*/, unsigned int & column)
1364 os << "\\textvertline" << termcmd;
1373 void Paragraph::Private::validate(LaTeXFeatures & features) const
1375 if (layout_->inpreamble && inset_owner_) {
1376 // FIXME: Using a string stream here circumvents the encoding
1377 // switching machinery of odocstream. Therefore the
1378 // output is wrong if this paragraph contains content
1379 // that needs to switch encoding.
1380 Buffer const & buf = inset_owner_->buffer();
1381 otexstringstream os;
1382 os << layout_->preamble();
1383 size_t const length = os.length();
1384 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1385 buf.getParFromID(owner_->id()).pit(), os,
1386 features.runparams(), string(), 0, -1, true);
1387 if (os.length() > length)
1388 features.addPreambleSnippet(os.release(), true);
1391 if (features.runparams().flavor == OutputParams::HTML
1392 && layout_->htmltitle()) {
1393 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1396 // check the params.
1397 if (!params_.spacing().isDefault())
1398 features.require("setspace");
1401 features.useLayout(layout_->name());
1404 fontlist_.validate(features);
1406 // then the indentation
1407 if (!params_.leftIndent().zero())
1408 features.require("ParagraphLeftIndent");
1411 InsetList::const_iterator icit = insetlist_.begin();
1412 InsetList::const_iterator iend = insetlist_.end();
1413 for (; icit != iend; ++icit) {
1415 features.inDeletedInset(owner_->isDeleted(icit->pos));
1416 if (icit->inset->lyxCode() == FOOT_CODE) {
1417 // FIXME: an item inset would make things much easier.
1418 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1419 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1420 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1421 && (icit->pos < begin_of_body_
1422 || (icit->pos == begin_of_body_
1423 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1424 features.saveNoteEnv("description");
1426 icit->inset->validate(features);
1427 features.inDeletedInset(false);
1428 if (layout_->needprotect &&
1429 icit->inset->lyxCode() == FOOT_CODE)
1430 features.require("NeedLyXFootnoteCode");
1434 // then the contents
1435 BufferParams const bp = features.runparams().is_child
1436 ? features.buffer().masterParams() : features.buffer().params();
1437 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1438 char_type c = text_[i];
1439 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1441 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1442 features.require("textquotedblp");
1443 else if (bp.main_font_encoding() != "T1"
1444 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1445 features.require("textquotedbl");
1446 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1447 // features that depend on the font or input encoding
1448 string feats = ci.textpreamble();
1449 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1450 if (fontenc.empty())
1451 fontenc = features.runparams().main_fontenc;
1452 while (!feats.empty()) {
1454 feats = split(feats, feat, ',');
1455 if (contains(feat, "!=")) {
1456 // a feature that is required except for the spcified
1457 // font or input encodings
1459 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1460 // multiple encodings are separated by semicolon
1461 vector<string> context = getVectorFromString(contexts, ";");
1462 // require feature if the context matches neither current font
1463 // nor input encoding
1464 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1465 && std::find(context.begin(), context.end(),
1466 features.runparams().encoding->name()) == context.end())
1467 features.require(realfeature);
1468 } else if (contains(feat, '=')) {
1469 // a feature that is required only for the spcified
1470 // font or input encodings
1472 string const contexts = split(feat, realfeature, '=');
1473 // multiple encodings are separated by semicolon
1474 vector<string> context = getVectorFromString(contexts, ";");
1475 // require feature if the context matches either current font
1476 // or input encoding
1477 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1478 || std::find(context.begin(), context.end(),
1479 features.runparams().encoding->name()) != context.end())
1480 features.require(realfeature);
1483 } else if (!bp.use_dash_ligatures
1484 && (c == 0x2013 || c == 0x2014)
1485 && bp.useNonTeXFonts
1486 && features.runparams().flavor == OutputParams::XETEX)
1487 // XeTeX's dash behaviour is determined via a global setting
1488 features.require("xetexdashbreakstate");
1489 BufferEncodings::validate(c, features);
1493 /////////////////////////////////////////////////////////////////////
1497 /////////////////////////////////////////////////////////////////////
1500 Layout const emptyParagraphLayout;
1503 Paragraph::Paragraph()
1504 : d(new Paragraph::Private(this, emptyParagraphLayout))
1511 Paragraph::Paragraph(Paragraph const & par)
1512 : itemdepth(par.itemdepth),
1513 d(new Paragraph::Private(*par.d, this))
1519 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1520 : itemdepth(par.itemdepth),
1521 d(new Paragraph::Private(*par.d, this, beg, end))
1527 Paragraph & Paragraph::operator=(Paragraph const & par)
1529 // needed as we will destroy the private part before copying it
1531 itemdepth = par.itemdepth;
1535 d = new Private(*par.d, this);
1542 Paragraph::~Paragraph()
1551 // this shall be called just before every "os << ..." action.
1552 void flushString(ostream & os, docstring & s)
1561 void Paragraph::write(ostream & os, BufferParams const & bparams,
1562 depth_type & dth) const
1564 // The beginning or end of a deeper (i.e. nested) area?
1565 if (dth != d->params_.depth()) {
1566 if (d->params_.depth() > dth) {
1567 while (d->params_.depth() > dth) {
1568 os << "\n\\begin_deeper";
1572 while (d->params_.depth() < dth) {
1573 os << "\n\\end_deeper";
1579 // First write the layout
1580 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1582 d->params_.write(os);
1584 Font font1(inherit_font, bparams.language);
1586 Change running_change = Change(Change::UNCHANGED);
1588 // this string is used as a buffer to avoid repetitive calls
1589 // to to_utf8(), which turn out to be expensive (JMarc)
1590 docstring write_buffer;
1593 for (pos_type i = 0; i <= size(); ++i) {
1595 Change const & change = lookupChange(i);
1596 if (change != running_change)
1597 flushString(os, write_buffer);
1598 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1599 running_change = change;
1604 // Write font changes
1605 Font font2 = getFontSettings(bparams, i);
1606 if (font2 != font1) {
1607 flushString(os, write_buffer);
1608 font2.lyxWriteChanges(font1, os);
1613 char_type const c = d->text_[i];
1616 if (Inset const * inset = getInset(i)) {
1617 flushString(os, write_buffer);
1618 if (inset->directWrite()) {
1619 // international char, let it write
1620 // code directly so it's shorter in
1626 os << "\\begin_inset ";
1628 os << "\n\\end_inset\n\n";
1631 // FIXME This can be removed again once the mystery
1632 // crash has been resolved.
1637 flushString(os, write_buffer);
1638 os << "\n\\backslash\n";
1642 flushString(os, write_buffer);
1643 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1650 if ((column > 70 && c == ' ')
1652 flushString(os, write_buffer);
1656 // this check is to amend a bug. LyX sometimes
1657 // inserts '\0' this could cause problems.
1659 write_buffer.push_back(c);
1661 LYXERR0("NUL char in structure.");
1667 flushString(os, write_buffer);
1668 os << "\n\\end_layout\n";
1669 // FIXME This can be removed again once the mystery
1670 // crash has been resolved.
1675 void Paragraph::validate(LaTeXFeatures & features) const
1677 d->validate(features);
1678 bool fragile = features.runparams().moving_arg;
1679 fragile |= layout().needprotect;
1680 if (inInset().getLayout().isNeedProtect())
1682 if (needsCProtection(fragile))
1683 features.require("cprotect");
1687 void Paragraph::insert(pos_type start, docstring const & str,
1688 Font const & font, Change const & change)
1690 for (size_t i = 0, n = str.size(); i != n ; ++i)
1691 insertChar(start + i, str[i], font, change);
1695 void Paragraph::appendChar(char_type c, Font const & font,
1696 Change const & change)
1699 d->changes_.insert(change, d->text_.size());
1700 // when appending characters, no need to update tables
1701 d->text_.push_back(c);
1702 setFont(d->text_.size() - 1, font);
1703 d->requestSpellCheck(d->text_.size() - 1);
1707 void Paragraph::appendString(docstring const & s, Font const & font,
1708 Change const & change)
1710 pos_type end = s.size();
1711 size_t oldsize = d->text_.size();
1712 size_t newsize = oldsize + end;
1713 size_t capacity = d->text_.capacity();
1714 if (newsize >= capacity)
1715 d->text_.reserve(max(capacity + 100, newsize));
1717 // when appending characters, no need to update tables
1720 // FIXME: Optimize this!
1721 for (size_t i = oldsize; i != newsize; ++i) {
1723 d->changes_.insert(change, i);
1724 d->requestSpellCheck(i);
1726 d->fontlist_.set(oldsize, font);
1727 d->fontlist_.set(newsize - 1, font);
1731 void Paragraph::insertChar(pos_type pos, char_type c,
1734 d->insertChar(pos, c, Change(trackChanges ?
1735 Change::INSERTED : Change::UNCHANGED));
1739 void Paragraph::insertChar(pos_type pos, char_type c,
1740 Font const & font, bool trackChanges)
1742 d->insertChar(pos, c, Change(trackChanges ?
1743 Change::INSERTED : Change::UNCHANGED));
1748 void Paragraph::insertChar(pos_type pos, char_type c,
1749 Font const & font, Change const & change)
1751 d->insertChar(pos, c, change);
1756 void Paragraph::resetFonts(Font const & font)
1758 d->fontlist_.clear();
1759 d->fontlist_.set(0, font);
1760 d->fontlist_.set(d->text_.size() - 1, font);
1763 // Gets uninstantiated font setting at position.
1764 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1768 LYXERR0("pos: " << pos << " size: " << size());
1772 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1773 if (cit != d->fontlist_.end())
1776 if (pos == size() && !empty())
1777 return getFontSettings(bparams, pos - 1);
1779 // Optimisation: avoid a full font instantiation if there is no
1780 // language change from previous call.
1781 static Font previous_font;
1782 static Language const * previous_lang = nullptr;
1783 Language const * lang = getParLanguage(bparams);
1784 if (lang != previous_lang) {
1785 previous_lang = lang;
1786 previous_font = Font(inherit_font, lang);
1788 return previous_font;
1792 FontSpan Paragraph::fontSpan(pos_type pos) const
1794 LBUFERR(pos <= size());
1797 return FontSpan(pos, pos);
1800 FontList::const_iterator cit = d->fontlist_.begin();
1801 FontList::const_iterator end = d->fontlist_.end();
1802 for (; cit != end; ++cit) {
1803 if (cit->pos() >= pos) {
1804 if (pos >= beginOfBody())
1805 return FontSpan(max(start, beginOfBody()),
1808 return FontSpan(start,
1809 min(beginOfBody() - 1,
1812 start = cit->pos() + 1;
1815 // This should not happen, but if so, we take no chances.
1816 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1817 LASSERT(false, return FontSpan(pos, pos));
1821 // Gets uninstantiated font setting at position 0
1822 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1824 if (!empty() && !d->fontlist_.empty())
1825 return d->fontlist_.begin()->font();
1827 // Optimisation: avoid a full font instantiation if there is no
1828 // language change from previous call.
1829 static Font previous_font;
1830 static Language const * previous_lang = nullptr;
1831 if (bparams.language != previous_lang) {
1832 previous_lang = bparams.language;
1833 previous_font = Font(inherit_font, bparams.language);
1836 return previous_font;
1840 // Gets the fully instantiated font at a given position in a paragraph
1841 // This is basically the same function as Text::GetFont() in text2.cpp.
1842 // The difference is that this one is used for generating the LaTeX file,
1843 // and thus cosmetic "improvements" are disallowed: This has to deliver
1844 // the true picture of the buffer. (Asger)
1845 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1846 Font const & outerfont) const
1850 Font font = getFontSettings(bparams, pos);
1852 pos_type const body_pos = beginOfBody();
1853 FontInfo & fi = font.fontInfo();
1855 fi.realize(d->layout_->labelfont);
1857 fi.realize(d->layout_->font);
1859 fi.realize(outerfont.fontInfo());
1860 fi.realize(bparams.getFont().fontInfo());
1866 Font const Paragraph::getLabelFont
1867 (BufferParams const & bparams, Font const & outerfont) const
1869 FontInfo tmpfont = d->layout_->labelfont;
1870 tmpfont.realize(outerfont.fontInfo());
1871 tmpfont.realize(bparams.getFont().fontInfo());
1872 return Font(tmpfont, getParLanguage(bparams));
1876 Font const Paragraph::getLayoutFont
1877 (BufferParams const & bparams, Font const & outerfont) const
1879 FontInfo tmpfont = d->layout_->font;
1880 tmpfont.realize(outerfont.fontInfo());
1881 tmpfont.realize(bparams.getFont().fontInfo());
1882 return Font(tmpfont, getParLanguage(bparams));
1886 char_type Paragraph::getUChar(BufferParams const & bparams,
1887 OutputParams const & rp,
1890 char_type c = d->text_[pos];
1892 // Return unchanged character in LTR languages
1893 // or if we use poylglossia/bidi (XeTeX).
1894 if (rp.useBidiPackage()
1895 || !getFontSettings(bparams, pos).isRightToLeft())
1898 // Without polyglossia/bidi, we need to account for some special cases.
1899 // FIXME This needs to be audited!
1901 // * The input is as expected for all delimiters
1902 // => checked for Hebrew!
1903 // * The output matches the display in the LyX workarea
1904 // => checked for Hebrew!
1905 // * The special cases below are really necessary
1906 // => checked for Hebrew!
1907 // * In arabic_arabi, brackets are transformed to Arabic
1908 // Ornate Parentheses. Is this is really wanted?
1910 string const & lang = getFontSettings(bparams, pos).language()->lang();
1913 // 1. In the following languages, parentheses need to be reversed.
1914 // Also with polyglodia/luabidi
1915 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1917 // 2. In the following languages, brackets don't need to be reversed.
1918 bool const reversebrackets = lang != "arabic_arabtex"
1919 && lang != "arabic_arabi"
1922 // Now swap delimiters if needed.
1933 if (reversebrackets)
1937 if (reversebrackets)
1958 void Paragraph::setFont(pos_type pos, Font const & font)
1960 LASSERT(pos <= size(), return);
1962 // First, reduce font against layout/label font
1963 // Update: The setCharFont() routine in text2.cpp already
1964 // reduces font, so we don't need to do that here. (Asger)
1966 d->fontlist_.set(pos, font);
1970 void Paragraph::makeSameLayout(Paragraph const & par)
1972 d->layout_ = par.d->layout_;
1973 d->params_ = par.d->params_;
1977 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1979 if (isFreeSpacing())
1985 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1986 if (eraseChar(pos, trackChanges))
1992 return count > 0 || pos > 0;
1996 bool Paragraph::hasSameLayout(Paragraph const & par) const
1998 return par.d->layout_ == d->layout_
1999 && d->params_.sameLayout(par.d->params_);
2003 depth_type Paragraph::getDepth() const
2005 return d->params_.depth();
2009 depth_type Paragraph::getMaxDepthAfter() const
2011 if (d->layout_->isEnvironment())
2012 return d->params_.depth() + 1;
2014 return d->params_.depth();
2018 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2020 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2021 return getDefaultAlign(bparams);
2023 return d->params_.align();
2027 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2029 LyXAlignment res = layout().align;
2030 if (isRTL(bparams)) {
2032 if (res == LYX_ALIGN_LEFT)
2033 res = LYX_ALIGN_RIGHT;
2034 else if (res == LYX_ALIGN_RIGHT)
2035 res = LYX_ALIGN_LEFT;
2041 docstring const & Paragraph::labelString() const
2043 return d->params_.labelString();
2047 // the next two functions are for the manual labels
2048 docstring const Paragraph::getLabelWidthString() const
2050 if (d->layout_->margintype == MARGIN_MANUAL
2051 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2052 return d->params_.labelWidthString();
2054 return _("Senseless with this layout!");
2058 void Paragraph::setLabelWidthString(docstring const & s)
2060 d->params_.labelWidthString(s);
2064 docstring Paragraph::expandLabel(Layout const & layout,
2065 BufferParams const & bparams) const
2067 return expandParagraphLabel(layout, bparams, true);
2071 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2072 BufferParams const & bparams, bool process_appendix) const
2074 DocumentClass const & tclass = bparams.documentClass();
2075 string const & lang = getParLanguage(bparams)->code();
2076 bool const in_appendix = process_appendix && d->params_.appendix();
2077 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2079 if (fmt.empty() && !layout.counter.empty())
2080 return tclass.counters().theCounter(layout.counter, lang);
2082 // handle 'inherited level parts' in 'fmt',
2083 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2084 size_t const i = fmt.find('@', 0);
2085 if (i != docstring::npos) {
2086 size_t const j = fmt.find('@', i + 1);
2087 if (j != docstring::npos) {
2088 docstring parent(fmt, i + 1, j - i - 1);
2089 docstring label = from_ascii("??");
2090 if (tclass.hasLayout(parent))
2091 label = expandParagraphLabel(tclass[parent], bparams,
2093 fmt = docstring(fmt, 0, i) + label
2094 + docstring(fmt, j + 1, docstring::npos);
2098 return tclass.counters().counterLabel(fmt, lang);
2102 void Paragraph::applyLayout(Layout const & new_layout)
2104 d->layout_ = &new_layout;
2105 LyXAlignment const oldAlign = d->params_.align();
2107 if (!(oldAlign & d->layout_->alignpossible)) {
2108 frontend::Alert::warning(_("Alignment not permitted"),
2109 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2110 d->params_.align(LYX_ALIGN_LAYOUT);
2115 pos_type Paragraph::beginOfBody() const
2117 return d->begin_of_body_;
2121 void Paragraph::setBeginOfBody()
2123 if (d->layout_->labeltype != LABEL_MANUAL) {
2124 d->begin_of_body_ = 0;
2128 // Unroll the first two cycles of the loop
2129 // and remember the previous character to
2130 // remove unnecessary getChar() calls
2132 pos_type end = size();
2133 bool prev_char_deleted = false;
2134 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2137 char_type previous_char = d->text_[i];
2138 if (!(isNewline(i) || isEnvSeparator(i))) {
2140 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2141 char_type temp = d->text_[i];
2142 prev_char_deleted = isDeleted(i);
2143 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2146 previous_char = temp;
2152 d->begin_of_body_ = i;
2156 bool Paragraph::allowParagraphCustomization() const
2158 return inInset().allowParagraphCustomization();
2162 bool Paragraph::usePlainLayout() const
2164 return inInset().usePlainLayout();
2168 bool Paragraph::isPassThru() const
2170 return inInset().isPassThru() || d->layout_->pass_thru;
2175 // paragraphs inside floats need different alignment tags to avoid
2178 bool noTrivlistCentering(InsetCode code)
2180 return code == FLOAT_CODE
2181 || code == WRAP_CODE
2182 || code == CELL_CODE;
2186 string correction(string const & orig)
2188 if (orig == "flushleft")
2189 return "raggedright";
2190 if (orig == "flushright")
2191 return "raggedleft";
2192 if (orig == "center")
2198 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2199 InsetCode code, bool const lastpar, int & col)
2201 string macro = suffix + "{";
2202 if (noTrivlistCentering(code)) {
2204 // the last paragraph in non-trivlist-aligned
2205 // context is special (to avoid unwanted whitespace)
2206 if (suffix == "\\begin") {
2207 macro = "\\" + correction(env) + "{}";
2208 os << from_ascii(macro);
2209 col += macro.size();
2214 macro += correction(env);
2218 if (suffix == "\\par\\end") {
2222 os << from_ascii(macro);
2223 col += macro.size();
2224 if (suffix == "\\begin") {
2234 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2235 otexstream & os, OutputParams const & runparams) const
2240 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2241 (layout_->toggle_indent != ITOGGLE_NEVER) :
2242 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2244 if (canindent && params_.noindent() && !layout_->pass_thru) {
2245 os << "\\noindent ";
2249 LyXAlignment const curAlign = params_.align();
2251 if (curAlign == layout_->align)
2255 case LYX_ALIGN_NONE:
2256 case LYX_ALIGN_BLOCK:
2257 case LYX_ALIGN_LAYOUT:
2258 case LYX_ALIGN_SPECIAL:
2259 case LYX_ALIGN_DECIMAL:
2261 case LYX_ALIGN_LEFT:
2262 case LYX_ALIGN_RIGHT:
2263 case LYX_ALIGN_CENTER:
2264 if (runparams.moving_arg) {
2271 string const begin_tag = "\\begin";
2272 InsetCode code = ownerCode();
2273 bool const lastpar = runparams.isLastPar;
2274 // RTL in classic (PDF)LaTeX (without the Bidi package)
2275 // Luabibdi (used by LuaTeX) behaves like classic
2276 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2277 && !runparams.useBidiPackage();
2280 case LYX_ALIGN_NONE:
2281 case LYX_ALIGN_BLOCK:
2282 case LYX_ALIGN_LAYOUT:
2283 case LYX_ALIGN_SPECIAL:
2284 case LYX_ALIGN_DECIMAL:
2286 case LYX_ALIGN_LEFT: {
2288 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2289 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2291 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2293 } case LYX_ALIGN_RIGHT: {
2295 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2296 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2298 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2300 } case LYX_ALIGN_CENTER: {
2301 corrected_env(os, begin_tag, "center", code, lastpar, column);
2310 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2311 otexstream & os, OutputParams const & runparams) const
2313 LyXAlignment const curAlign = params_.align();
2315 if (curAlign == layout_->align)
2319 case LYX_ALIGN_NONE:
2320 case LYX_ALIGN_BLOCK:
2321 case LYX_ALIGN_LAYOUT:
2322 case LYX_ALIGN_SPECIAL:
2323 case LYX_ALIGN_DECIMAL:
2325 case LYX_ALIGN_LEFT:
2326 case LYX_ALIGN_RIGHT:
2327 case LYX_ALIGN_CENTER:
2328 if (runparams.moving_arg)
2333 bool output = false;
2335 string const end_tag = "\\par\\end";
2336 InsetCode code = ownerCode();
2337 bool const lastpar = runparams.isLastPar;
2338 // RTL in classic (PDF)LaTeX (without the Bidi package)
2339 // Luabibdi (used by LuaTeX) behaves like classic
2340 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2341 && !runparams.useBidiPackage();
2344 case LYX_ALIGN_NONE:
2345 case LYX_ALIGN_BLOCK:
2346 case LYX_ALIGN_LAYOUT:
2347 case LYX_ALIGN_SPECIAL:
2348 case LYX_ALIGN_DECIMAL:
2350 case LYX_ALIGN_LEFT: {
2352 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2353 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2355 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2357 } case LYX_ALIGN_RIGHT: {
2359 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2360 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2362 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2364 } case LYX_ALIGN_CENTER: {
2365 corrected_env(os, end_tag, "center", code, lastpar, col);
2370 return output || lastpar;
2374 // This one spits out the text of the paragraph
2375 void Paragraph::latex(BufferParams const & bparams,
2376 Font const & outerfont,
2378 OutputParams const & runparams,
2379 int start_pos, int end_pos, bool force) const
2381 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2383 // FIXME This check should not be needed. Perhaps issue an
2384 // error if it triggers.
2385 Layout const & style = inInset().forcePlainLayout() ?
2386 bparams.documentClass().plainLayout() : *d->layout_;
2388 if (!force && style.inpreamble)
2391 bool const allowcust = allowParagraphCustomization();
2393 // Current base font for all inherited font changes, without any
2394 // change caused by an individual character, except for the language:
2395 // It is set to the language of the first character.
2396 // As long as we are in the label, this font is the base font of the
2397 // label. Before the first body character it is set to the base font
2401 // If there is an open font-encoding changing command (script wrapper),
2402 // alien_script is set to its name
2403 string alien_script;
2406 // Maybe we have to create a optional argument.
2407 pos_type body_pos = beginOfBody();
2408 unsigned int column = 0;
2410 // If we are inside an non inheritFont() inset, the real outerfont is local_font
2411 Font const real_outerfont = (!inInset().inheritFont()
2412 && runparams.local_font != nullptr)
2413 ? Font(runparams.local_font->fontInfo()) : outerfont;
2416 // the optional argument is kept in curly brackets in
2417 // case it contains a ']'
2418 // This is not strictly needed, but if this is changed it
2419 // would be a file format change, and tex2lyx would need
2420 // to be adjusted, since it unconditionally removes the
2421 // braces when it parses \item.
2424 basefont = getLabelFont(bparams, real_outerfont);
2426 basefont = getLayoutFont(bparams, real_outerfont);
2429 // Which font is currently active?
2430 Font running_font(basefont);
2431 // Do we have an open font change?
2432 bool open_font = false;
2434 Change runningChange = Change(Change::UNCHANGED);
2436 Encoding const * const prev_encoding = runparams.encoding;
2438 os.texrow().start(id(), 0);
2440 // if the paragraph is empty, the loop will not be entered at all
2442 // For InTitle commands, we have already opened a group
2443 // in output_latex::TeXOnePar.
2444 if (style.isCommand() && !style.intitle) {
2448 if (!style.leftdelim().empty()) {
2449 os << style.leftdelim();
2450 column += style.leftdelim().size();
2453 column += d->startTeXParParams(bparams, os, runparams);
2456 // Whether a \par can be issued for insets typeset inline with text.
2457 // Yes if greater than 0. This has to be static.
2458 THREAD_LOCAL_STATIC int parInline = 0;
2460 for (pos_type i = 0; i < size(); ++i) {
2461 // First char in paragraph or after label?
2462 if (i == body_pos) {
2465 bool needPar = false;
2466 column += running_font.latexWriteEndChanges(
2467 os, bparams, runparams,
2468 basefont, basefont, needPar);
2471 basefont = getLayoutFont(bparams, real_outerfont);
2472 running_font = basefont;
2474 column += Changes::latexMarkChange(os, bparams,
2475 runningChange, Change(Change::UNCHANGED),
2477 runningChange = Change(Change::UNCHANGED);
2482 // For InTitle commands, we have already opened a group
2483 // in output_latex::TeXOnePar.
2484 if (style.isCommand() && !style.intitle) {
2489 if (!style.leftdelim().empty()) {
2490 os << style.leftdelim();
2491 column += style.leftdelim().size();
2495 column += d->startTeXParParams(bparams, os,
2499 runparams.wasDisplayMath = runparams.inDisplayMath;
2500 runparams.inDisplayMath = false;
2501 bool deleted_display_math = false;
2502 Change const & change = runparams.inDeletedInset
2503 ? runparams.changeOfDeletedInset : lookupChange(i);
2505 char_type const c = d->text_[i];
2507 // Check whether a display math inset follows
2509 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2511 runparams.ctObject = getInset(i)->CtObject(runparams);
2513 InsetMath const * im = getInset(i)->asInsetMath();
2514 if (im && im->asHullInset()
2515 && im->asHullInset()->outerDisplay()) {
2516 runparams.inDisplayMath = true;
2517 // runparams.inDeletedInset will be set by
2518 // latexInset later, but we need this info
2519 // before it is called. On the other hand, we
2520 // cannot set it here because it is a counter.
2521 deleted_display_math = isDeleted(i);
2523 if (bparams.output_changes && deleted_display_math
2524 && runningChange == change
2525 && change.type == Change::DELETED
2526 && !os.afterParbreak()) {
2527 // A display math in the same paragraph follows.
2528 // We have to close and then reopen \lyxdeleted,
2529 // otherwise the math will be shifted up.
2530 OutputParams rp = runparams;
2532 bool needPar = false;
2533 column += running_font.latexWriteEndChanges(
2534 os, bparams, rp, basefont,
2538 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2539 : getLayoutFont(bparams, real_outerfont);
2540 running_font = basefont;
2541 column += Changes::latexMarkChange(os, bparams,
2542 Change(Change::INSERTED), change, rp);
2546 if (bparams.output_changes && runningChange != change) {
2547 if (!alien_script.empty()) {
2550 alien_script.clear();
2553 bool needPar = false;
2554 column += running_font.latexWriteEndChanges(
2555 os, bparams, runparams,
2556 basefont, basefont, needPar);
2559 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2560 : getLayoutFont(bparams, real_outerfont);
2561 running_font = basefont;
2562 column += Changes::latexMarkChange(os, bparams, runningChange,
2564 runningChange = change;
2567 // do not output text which is marked deleted
2568 // if change tracking output is disabled
2569 if (!bparams.output_changes && change.deleted()) {
2575 // Fully instantiated font
2576 Font current_font = getFont(bparams, i, outerfont);
2578 Font const prev_font = (i > 0) ?
2579 getFont(bparams, i - 1, outerfont)
2582 Font const last_font = running_font;
2583 bool const in_ct_deletion = (bparams.output_changes
2584 && runningChange == change
2585 && change.type == Change::DELETED
2586 && !os.afterParbreak());
2587 // Insets where font switches are used (rather than font commands)
2588 bool const fontswitch_inset =
2591 && getInset(i)->allowMultiPar()
2592 && getInset(i)->lyxCode() != ERT_CODE;
2594 bool closeLanguage = false;
2595 bool lang_switched_at_inset = false;
2596 if (fontswitch_inset) {
2597 // Some insets cannot be inside a font change command.
2598 // However, even such insets *can* be placed in \L or \R
2599 // or their equivalents (for RTL language switches),
2600 // so we don't close the language in those cases
2601 // (= differing isRightToLeft()).
2602 // ArabTeX, though, doesn't seem to handle this special behavior.
2603 closeLanguage = basefont.isRightToLeft() == current_font.isRightToLeft()
2604 || basefont.language()->lang() == "arabic_arabtex"
2605 || current_font.language()->lang() == "arabic_arabtex";
2606 // We need to check prev_font as language changes directly at inset
2607 // will only be started inside the inset.
2608 lang_switched_at_inset = prev_font.language() != current_font.language();
2611 // Do we need to close the previous font?
2613 ((current_font != running_font
2614 || current_font.language() != running_font.language())
2615 || (fontswitch_inset
2616 && (current_font == prev_font))))
2618 // ensure there is no open script-wrapper
2619 if (!alien_script.empty()) {
2622 alien_script.clear();
2624 bool needPar = false;
2625 if (in_ct_deletion) {
2626 // We have to close and then reopen \lyxdeleted,
2627 // as strikeout needs to be on lowest level.
2632 // Force language closing
2633 current_font.setLanguage(basefont.language());
2634 column += running_font.latexWriteEndChanges(
2635 os, bparams, runparams, basefont,
2636 (i == body_pos-1) ? basefont : current_font,
2638 if (in_ct_deletion) {
2639 // We have to close and then reopen \lyxdeleted,
2640 // as strikeout needs to be on lowest level.
2641 OutputParams rp = runparams;
2642 column += Changes::latexMarkChange(os, bparams,
2643 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2645 running_font = basefont;
2649 // if necessary, close language environment before opening CJK
2650 string const running_lang = running_font.language()->babel();
2651 string const lang_end_command = lyxrc.language_command_end;
2652 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2653 && !running_lang.empty()
2654 && running_lang == openLanguageName()
2655 && current_font.language()->encoding()->package() == Encoding::CJK) {
2656 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2657 os << from_ascii(end_tag);
2658 column += end_tag.length();
2662 // Switch file encoding if necessary (and allowed)
2663 if (!runparams.pass_thru && !style.pass_thru &&
2664 runparams.encoding->package() != Encoding::none &&
2665 current_font.language()->encoding()->package() != Encoding::none) {
2666 pair<bool, int> const enc_switch =
2667 switchEncoding(os.os(), bparams, runparams,
2668 *(current_font.language()->encoding()));
2669 if (enc_switch.first) {
2670 column += enc_switch.second;
2671 runparams.encoding = current_font.language()->encoding();
2675 // A display math inset inside an ulem command will be output
2676 // as a box of width \linewidth, so we have to either disable
2677 // indentation if the inset starts a paragraph, or start a new
2678 // line to accommodate such box. This has to be done before
2679 // writing any font changing commands.
2680 if (runparams.inDisplayMath && !deleted_display_math
2681 && runparams.inulemcmd) {
2682 if (os.afterParbreak())
2688 // Do we need to change font?
2689 if (!fontswitch_inset &&
2690 (current_font != running_font ||
2691 current_font.language() != running_font.language())
2692 && i != body_pos - 1)
2694 if (in_ct_deletion) {
2695 // We have to close and then reopen \lyxdeleted,
2696 // as strikeout needs to be on lowest level.
2697 bool needPar = false;
2698 OutputParams rp = runparams;
2699 column += running_font.latexWriteEndChanges(
2700 os, bparams, rp, basefont,
2705 otexstringstream ots;
2706 InsetText const * textinset = inInset().asInsetText();
2707 bool const cprotect = textinset
2708 ? textinset->hasCProtectContent(runparams.moving_arg)
2709 && !textinset->text().isMainText()
2711 column += current_font.latexWriteStartChanges(ots, bparams,
2712 runparams, basefont, last_font, false,
2714 // Check again for display math in ulem commands as a
2715 // font change may also occur just before a math inset.
2716 if (runparams.inDisplayMath && !deleted_display_math
2717 && runparams.inulemcmd) {
2718 if (os.afterParbreak())
2723 running_font = current_font;
2725 docstring fontchange = ots.str();
2727 // check whether the fontchange ends with a \\textcolor
2728 // modifier and the text starts with a space. If so we
2729 // need to add } in order to prevent \\textcolor from gobbling
2730 // the space (bug 4473).
2731 docstring const last_modifier = rsplit(fontchange, '\\');
2732 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2733 os << from_ascii("{}");
2734 else if (ots.terminateCommand())
2736 if (in_ct_deletion) {
2737 // We have to close and then reopen \lyxdeleted,
2738 // as strikeout needs to be on lowest level.
2739 OutputParams rp = runparams;
2740 column += Changes::latexMarkChange(os, bparams,
2741 Change(Change::UNCHANGED), change, rp);
2745 // FIXME: think about end_pos implementation...
2746 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2747 // FIXME: integrate this case in latexSpecialChar
2748 // Do not print the separation of the optional argument
2749 // if style.pass_thru is false. This works because
2750 // latexSpecialChar ignores spaces if
2751 // style.pass_thru is false.
2752 if (i != body_pos - 1) {
2753 if (d->simpleTeXBlanks(bparams, runparams, os,
2754 i, column, current_font, style)) {
2755 // A surrogate pair was output. We
2756 // must not call latexSpecialChar
2757 // in this iteration, since it would output
2758 // the combining character again.
2765 OutputParams rp = runparams;
2766 rp.free_spacing = style.free_spacing;
2767 rp.local_font = ¤t_font;
2768 rp.intitle = style.intitle;
2770 // Two major modes: LaTeX or plain
2771 // Handle here those cases common to both modes
2772 // and then split to handle the two modes separately.
2773 if (c == META_INSET) {
2774 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2775 // Greyedout notes and, in general, all insets
2776 // with InsetLayout::isDisplay() == false,
2777 // are typeset inline with the text. So, we
2778 // can add a \par to the last paragraph of
2779 // such insets only if nothing else follows.
2780 bool incremented = false;
2781 Inset const * inset = getInset(i);
2782 InsetText const * textinset = inset
2783 ? inset->asInsetText()
2785 if (i + 1 == size() && textinset
2786 && !inset->getLayout().isDisplay()) {
2787 ParagraphList const & pars =
2788 textinset->text().paragraphs();
2789 pit_type const pit = pars.size() - 1;
2790 Font const lastfont =
2791 pit < 0 || pars[pit].empty()
2792 ? pars[pit].getLayoutFont(
2795 : pars[pit].getFont(bparams,
2796 pars[pit].size() - 1,
2798 if (lastfont.fontInfo().size() !=
2799 basefont.fontInfo().size()) {
2804 // We need to restore parts of this after insets with
2805 // allowMultiPar() true
2806 Font const save_basefont = basefont;
2807 d->latexInset(bparams, os, rp, running_font,
2808 basefont, real_outerfont, open_font,
2809 runningChange, style, i, column, fontswitch_inset,
2810 closeLanguage, lang_switched_at_inset);
2811 if (fontswitch_inset) {
2813 bool needPar = false;
2814 column += running_font.latexWriteEndChanges(
2815 os, bparams, runparams,
2816 basefont, basefont, needPar);
2819 basefont.fontInfo().setSize(save_basefont.fontInfo().size());
2820 basefont.fontInfo().setFamily(save_basefont.fontInfo().family());
2821 basefont.fontInfo().setSeries(save_basefont.fontInfo().series());
2826 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2827 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2828 // Close \lyx*deleted and force its
2829 // reopening (if needed)
2832 runningChange = Change(Change::UNCHANGED);
2833 runparams.ctObject = OutputParams::CT_NORMAL;
2836 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2837 if (!bparams.useNonTeXFonts)
2838 script = Encodings::isKnownScriptChar(c);
2839 if (script != alien_script) {
2840 if (!alien_script.empty()) {
2842 alien_script.clear();
2844 string fontenc = running_font.language()->fontenc(bparams);
2846 && !Encodings::fontencSupportsScript(fontenc, script)) {
2847 column += script.length() + 2;
2848 os << "\\" << script << "{";
2849 alien_script = script;
2853 d->latexSpecialChar(os, bparams, rp, running_font,
2854 alien_script, style, i, end_pos, column);
2855 } catch (EncodingException & e) {
2856 if (runparams.dryrun) {
2857 os << "<" << _("LyX Warning: ")
2858 << _("uncodable character") << " '";
2862 // add location information and throw again.
2870 // Set the encoding to that returned from latexSpecialChar (see
2871 // comment for encoding member in OutputParams.h)
2872 runparams.encoding = rp.encoding;
2874 // Also carry on the info on a closed ulem command for insets
2875 // such as Note that do not produce any output, so that no
2876 // command is ever executed but its opening was recorded.
2877 runparams.inulemcmd = rp.inulemcmd;
2879 // These need to be passed upstream as well
2880 runparams.need_maketitle = rp.need_maketitle;
2881 runparams.have_maketitle = rp.have_maketitle;
2883 // And finally, pass the post_macros upstream
2884 runparams.post_macro = rp.post_macro;
2887 // Close wrapper for alien script
2888 if (!alien_script.empty()) {
2890 alien_script.clear();
2893 // If we have an open font definition, we have to close it
2895 // Make sure that \\par is done with the font of the last
2896 // character if this has another size as the default.
2897 // This is necessary because LaTeX (and LyX on the screen)
2898 // calculates the space between the baselines according
2899 // to this font. (Matthias)
2901 // We must not change the font for the last paragraph
2902 // of non-multipar insets, tabular cells or commands,
2903 // since this produces unwanted whitespace.
2905 Font const font = empty()
2906 ? getLayoutFont(bparams, real_outerfont)
2907 : getFont(bparams, size() - 1, real_outerfont);
2909 InsetText const * textinset = inInset().asInsetText();
2911 bool const maintext = textinset
2912 ? textinset->text().isMainText()
2915 size_t const numpars = textinset
2916 ? textinset->text().paragraphs().size()
2919 bool needPar = false;
2921 if (style.resfont.size() != font.fontInfo().size()
2922 && (!runparams.isLastPar || maintext
2923 || (numpars > 1 && d->ownerCode() != CELL_CODE
2924 && (inInset().getLayout().isDisplay()
2926 && !style.isCommand()) {
2929 #ifdef FIXED_LANGUAGE_END_DETECTION
2931 running_font.latexWriteEndChanges(os, bparams,
2932 runparams, basefont,
2933 next_->getFont(bparams, 0, outerfont),
2936 running_font.latexWriteEndChanges(os, bparams,
2937 runparams, basefont, basefont, needPar);
2940 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2941 //FIXME: there as we start another \selectlanguage with the next paragraph if
2942 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2943 running_font.latexWriteEndChanges(os, bparams, runparams,
2944 basefont, basefont, needPar);
2947 // The \par could not be inserted at the same nesting
2948 // level of the font size change, so do it now.
2949 os << "{\\" << font.latexSize() << "\\par}";
2953 column += Changes::latexMarkChange(os, bparams, runningChange,
2954 Change(Change::UNCHANGED), runparams);
2956 // Needed if there is an optional argument but no contents.
2957 if (body_pos > 0 && body_pos == size()) {
2961 if (!style.rightdelim().empty()) {
2962 os << style.rightdelim();
2963 column += style.rightdelim().size();
2966 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2967 && runparams.encoding != prev_encoding) {
2968 runparams.encoding = prev_encoding;
2969 os << setEncoding(prev_encoding->iconvName());
2972 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2976 bool Paragraph::emptyTag() const
2978 for (pos_type i = 0; i < size(); ++i) {
2979 if (Inset const * inset = getInset(i)) {
2980 InsetCode lyx_code = inset->lyxCode();
2981 // FIXME testing like that is wrong. What is
2983 if (lyx_code != TOC_CODE &&
2984 lyx_code != INCLUDE_CODE &&
2985 lyx_code != GRAPHICS_CODE &&
2986 lyx_code != ERT_CODE &&
2987 lyx_code != LISTINGS_CODE &&
2988 lyx_code != FLOAT_CODE &&
2989 lyx_code != TABULAR_CODE) {
2993 char_type c = d->text_[i];
2994 if (c != ' ' && c != '\t')
3002 string Paragraph::getID(Buffer const &, OutputParams const &)
3005 for (pos_type i = 0; i < size(); ++i) {
3006 if (Inset const * inset = getInset(i)) {
3007 InsetCode lyx_code = inset->lyxCode();
3008 if (lyx_code == LABEL_CODE) {
3009 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
3010 docstring const & id = il->getParam("name");
3011 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3019 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3022 for (i = 0; i < size(); ++i) {
3023 if (Inset const * inset = getInset(i)) {
3024 inset->docbook(xs, runparams);
3026 char_type c = d->text_[i];
3036 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3040 for (i = 0; i < size(); ++i) {
3041 if (Inset const * inset = getInset(i)) {
3042 inset->xhtml(xs, runparams);
3044 char_type c = d->text_[i];
3054 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3057 pos_type size = text_.size();
3058 for (pos_type i = initial; i < size; ++i) {
3059 Font font = owner_->getFont(buf.params(), i, outerfont);
3060 if (text_[i] == META_INSET)
3062 if (i != initial && font != font_old)
3073 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3074 vector<xml::EndFontTag> & tagsToClose,
3075 bool & flag, FontState curstate, xml::FontTypes type)
3077 if (curstate == FONT_ON) {
3078 tagsToOpen.push_back(docbookStartFontTag(type));
3081 tagsToClose.push_back(docbookEndFontTag(type));
3086 class OptionalFontType {
3091 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3092 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3095 OptionalFontType fontShapeToXml(FontShape fs)
3099 return {xml::FT_ITALIC};
3101 return {xml::FT_SLANTED};
3102 case SMALLCAPS_SHAPE:
3103 return {xml::FT_SMALLCAPS};
3108 // the other tags are for internal use
3114 OptionalFontType fontFamilyToXml(FontFamily fm)
3118 return {xml::FT_ROMAN};
3120 return {xml::FT_SANS};
3121 case TYPEWRITER_FAMILY:
3122 return {xml::FT_TYPE};
3123 case INHERIT_FAMILY:
3126 // the other tags are for internal use
3132 OptionalFontType fontSizeToXml(FontSize fs)
3136 return {xml::FT_SIZE_TINY};
3138 return {xml::FT_SIZE_SCRIPT};
3140 return {xml::FT_SIZE_FOOTNOTE};
3142 return {xml::FT_SIZE_SMALL};
3144 return {xml::FT_SIZE_LARGE};
3146 return {xml::FT_SIZE_LARGER};
3148 return {xml::FT_SIZE_LARGEST};
3150 return {xml::FT_SIZE_HUGE};
3152 return {xml::FT_SIZE_HUGER};
3154 return {xml::FT_SIZE_INCREASE};
3156 return {xml::FT_SIZE_DECREASE};
3161 // the other tags are for internal use
3167 struct DocBookFontState
3169 FontShape curr_fs = INHERIT_SHAPE;
3170 FontFamily curr_fam = INHERIT_FAMILY;
3171 FontSize curr_size = INHERIT_SIZE;
3173 // track whether we have opened these tags
3174 bool emph_flag = false;
3175 bool bold_flag = false;
3176 bool noun_flag = false;
3177 bool ubar_flag = false;
3178 bool dbar_flag = false;
3179 bool sout_flag = false;
3180 bool xout_flag = false;
3181 bool wave_flag = false;
3183 bool shap_flag = false;
3185 bool faml_flag = false;
3187 bool size_flag = false;
3190 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3192 std::string const & default_family,
3193 DocBookFontState & fs)
3195 vector<xml::FontTag> tagsToOpen;
3196 vector<xml::EndFontTag> tagsToClose;
3199 FontState curstate = font.fontInfo().emph();
3200 if (font_old.emph() != curstate)
3201 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3204 curstate = font.fontInfo().noun();
3205 if (font_old.noun() != curstate)
3206 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3209 curstate = font.fontInfo().underbar();
3210 if (font_old.underbar() != curstate)
3211 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3214 curstate = font.fontInfo().strikeout();
3215 if (font_old.strikeout() != curstate)
3216 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3219 curstate = font.fontInfo().xout();
3220 if (font_old.xout() != curstate)
3221 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3224 curstate = font.fontInfo().uuline();
3225 if (font_old.uuline() != curstate)
3226 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3229 curstate = font.fontInfo().uwave();
3230 if (font_old.uwave() != curstate)
3231 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3234 // a little hackish, but allows us to reuse what we have.
3235 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3236 if (font_old.series() != font.fontInfo().series())
3237 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3240 fs.curr_fs = font.fontInfo().shape();
3241 FontShape old_fs = font_old.shape();
3242 if (old_fs != fs.curr_fs) {
3244 OptionalFontType tag = fontShapeToXml(old_fs);
3246 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3247 fs.shap_flag = false;
3250 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3252 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3256 fs.curr_fam = font.fontInfo().family();
3257 FontFamily old_fam = font_old.family();
3258 if (old_fam != fs.curr_fam) {
3260 OptionalFontType tag = fontFamilyToXml(old_fam);
3262 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3263 fs.faml_flag = false;
3265 switch (fs.curr_fam) {
3267 // we will treat a "default" font family as roman, since we have
3268 // no other idea what to do.
3269 if (default_family != "rmdefault" && default_family != "default") {
3270 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3271 fs.faml_flag = true;
3275 if (default_family != "sfdefault") {
3276 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3277 fs.faml_flag = true;
3280 case TYPEWRITER_FAMILY:
3281 if (default_family != "ttdefault") {
3282 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3283 fs.faml_flag = true;
3286 case INHERIT_FAMILY:
3289 // the other tags are for internal use
3296 fs.curr_size = font.fontInfo().size();
3297 FontSize old_size = font_old.size();
3298 if (old_size != fs.curr_size) {
3300 OptionalFontType tag = fontSizeToXml(old_size);
3302 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3303 fs.size_flag = false;
3306 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3307 if (tag.has_value) {
3308 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3309 fs.size_flag = true;
3313 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3316 } // anonymous namespace
3319 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3321 OutputParams const & runparams,
3322 Font const & outerfont,
3323 bool start_paragraph, bool close_paragraph,
3324 pos_type initial) const
3326 // track whether we have opened these tags
3327 DocBookFontState fs;
3329 if (start_paragraph)
3330 xs.startDivision(allowEmpty());
3332 Layout const & style = *d->layout_;
3334 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3336 string const default_family =
3337 buf.masterBuffer()->params().fonts_default_family;
3339 vector<xml::FontTag> tagsToOpen;
3340 vector<xml::EndFontTag> tagsToClose;
3342 // parsing main loop
3343 for (pos_type i = initial; i < size(); ++i) {
3344 // let's not show deleted material in the output
3348 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3350 // Determine which tags should be opened or closed.
3351 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs);
3354 // Other such tags? What about the other text ranges?
3356 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3357 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3358 for (; cit != cen; ++cit)
3361 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3362 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3363 for (; sit != sen; ++sit)
3366 tagsToClose.clear();
3369 if (Inset const * inset = getInset(i)) {
3370 if (!runparams.for_toc || inset->isInToc()) {
3371 OutputParams np = runparams;
3372 np.local_font = &font;
3373 // If the paragraph has size 1, then we are in the "special
3374 // case" where we do not output the containing paragraph info.
3375 // This "special case" is defined in more details in output_docbook.cpp, makeParagraphs. The results
3376 // of that brittle logic is passed to this function through open_par.
3377 if (!inset->getLayout().htmlisblock() && size() != 1) // TODO: htmlisblock here too!
3378 np.docbook_in_par = true;
3379 inset->docbook(xs, np);
3382 char_type c = getUChar(buf.masterBuffer()->params(), runparams, i);
3385 font_old = font.fontInfo();
3388 // FIXME, this code is just imported from XHTML
3389 // I'm worried about what happens if a branch, say, is itself
3390 // wrapped in some font stuff. I think that will not work.
3392 if (runparams.docbook_in_listing)
3394 if (close_paragraph)
3401 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3402 vector<xml::EndFontTag> & tagsToClose,
3403 bool & flag, FontState curstate, xml::FontTypes type)
3405 if (curstate == FONT_ON) {
3406 tagsToOpen.push_back(xhtmlStartFontTag(type));
3409 tagsToClose.push_back(xhtmlEndFontTag(type));
3414 } // anonymous namespace
3417 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3419 OutputParams const & runparams,
3420 Font const & outerfont,
3421 bool start_paragraph, bool close_paragraph,
3422 pos_type initial) const
3426 // track whether we have opened these tags
3427 bool emph_flag = false;
3428 bool bold_flag = false;
3429 bool noun_flag = false;
3430 bool ubar_flag = false;
3431 bool dbar_flag = false;
3432 bool sout_flag = false;
3433 bool xout_flag = false;
3434 bool wave_flag = false;
3436 bool shap_flag = false;
3438 bool faml_flag = false;
3440 bool size_flag = false;
3442 Layout const & style = *d->layout_;
3444 if (start_paragraph)
3445 xs.startDivision(allowEmpty());
3448 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3450 FontShape curr_fs = INHERIT_SHAPE;
3451 FontFamily curr_fam = INHERIT_FAMILY;
3452 FontSize curr_size = INHERIT_SIZE;
3454 string const default_family =
3455 buf.masterBuffer()->params().fonts_default_family;
3457 vector<xml::FontTag> tagsToOpen;
3458 vector<xml::EndFontTag> tagsToClose;
3460 // parsing main loop
3461 for (pos_type i = initial; i < size(); ++i) {
3462 // let's not show deleted material in the output
3466 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3469 FontState curstate = font.fontInfo().emph();
3470 if (font_old.emph() != curstate)
3471 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3474 curstate = font.fontInfo().noun();
3475 if (font_old.noun() != curstate)
3476 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3479 curstate = font.fontInfo().underbar();
3480 if (font_old.underbar() != curstate)
3481 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3484 curstate = font.fontInfo().strikeout();
3485 if (font_old.strikeout() != curstate)
3486 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3489 curstate = font.fontInfo().xout();
3490 if (font_old.xout() != curstate)
3491 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3494 curstate = font.fontInfo().uuline();
3495 if (font_old.uuline() != curstate)
3496 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3499 curstate = font.fontInfo().uwave();
3500 if (font_old.uwave() != curstate)
3501 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3504 // a little hackish, but allows us to reuse what we have.
3505 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3506 if (font_old.series() != font.fontInfo().series())
3507 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3510 curr_fs = font.fontInfo().shape();
3511 FontShape old_fs = font_old.shape();
3512 if (old_fs != curr_fs) {
3516 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3519 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3521 case SMALLCAPS_SHAPE:
3522 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3528 // the other tags are for internal use
3536 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3540 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3543 case SMALLCAPS_SHAPE:
3544 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3551 // the other tags are for internal use
3558 curr_fam = font.fontInfo().family();
3559 FontFamily old_fam = font_old.family();
3560 if (old_fam != curr_fam) {
3564 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3567 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3569 case TYPEWRITER_FAMILY:
3570 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3572 case INHERIT_FAMILY:
3575 // the other tags are for internal use
3583 // we will treat a "default" font family as roman, since we have
3584 // no other idea what to do.
3585 if (default_family != "rmdefault" && default_family != "default") {
3586 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3591 if (default_family != "sfdefault") {
3592 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3596 case TYPEWRITER_FAMILY:
3597 if (default_family != "ttdefault") {
3598 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3602 case INHERIT_FAMILY:
3605 // the other tags are for internal use
3612 curr_size = font.fontInfo().size();
3613 FontSize old_size = font_old.size();
3614 if (old_size != curr_size) {
3618 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3621 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3624 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3627 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3630 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3633 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3636 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3639 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3642 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3645 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3648 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3654 // the other tags are for internal use
3660 switch (curr_size) {
3662 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3666 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3670 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3674 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3678 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3682 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3686 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3690 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3694 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3698 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3702 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3709 // the other tags are for internal use
3716 // Other such tags? What about the other text ranges?
3718 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3719 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3720 for (; cit != cen; ++cit)
3723 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3724 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3725 for (; sit != sen; ++sit)
3728 tagsToClose.clear();
3731 Inset const * inset = getInset(i);
3733 if (!runparams.for_toc || inset->isInToc()) {
3734 OutputParams np = runparams;
3735 np.local_font = &font;
3736 // If the paragraph has size 1, then we are in the "special
3737 // case" where we do not output the containing paragraph info
3738 if (!inset->getLayout().htmlisblock() && size() != 1)
3739 np.html_in_par = true;
3740 retval += inset->xhtml(xs, np);
3743 char_type c = getUChar(buf.masterBuffer()->params(),
3745 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3746 xs << XMLStream::ESCAPE_NONE << " ";
3750 font_old = font.fontInfo();
3754 // I'm worried about what happens if a branch, say, is itself
3755 // wrapped in some font stuff. I think that will not work.
3757 if (close_paragraph)
3764 bool Paragraph::isHfill(pos_type pos) const
3766 Inset const * inset = getInset(pos);
3767 return inset && inset->isHfill();
3771 bool Paragraph::isNewline(pos_type pos) const
3773 // U+2028 LINE SEPARATOR
3774 // U+2029 PARAGRAPH SEPARATOR
3775 char_type const c = d->text_[pos];
3776 if (c == 0x2028 || c == 0x2029)
3778 Inset const * inset = getInset(pos);
3779 return inset && inset->lyxCode() == NEWLINE_CODE;
3783 bool Paragraph::isEnvSeparator(pos_type pos) const
3785 Inset const * inset = getInset(pos);
3786 return inset && inset->lyxCode() == SEPARATOR_CODE;
3790 bool Paragraph::isLineSeparator(pos_type pos) const
3792 char_type const c = d->text_[pos];
3793 if (isLineSeparatorChar(c))
3795 Inset const * inset = getInset(pos);
3796 return inset && inset->isLineSeparator();
3800 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3804 if (ignore_deleted && isDeleted(pos))
3806 if (Inset const * inset = getInset(pos))
3807 return !inset->isLetter();
3808 // if we have a hard hyphen (no en- or emdash) or apostrophe
3809 // we pass this to the spell checker
3810 // FIXME: this method is subject to change, visit
3811 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3812 // to get an impression how complex this is.
3813 if (isHardHyphenOrApostrophe(pos))
3815 char_type const c = d->text_[pos];
3816 // We want to pass the escape chars to the spellchecker
3817 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3818 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3822 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3824 pos_type const psize = size();
3827 char_type const c = d->text_[pos];
3828 if (c != '-' && c != '\'')
3830 int nextpos = pos + 1;
3831 int prevpos = pos > 0 ? pos - 1 : 0;
3832 if ((nextpos == psize || isSpace(nextpos))
3833 && (pos == 0 || isSpace(prevpos)))
3839 bool Paragraph::needsCProtection(bool const fragile) const
3841 // first check the layout of the paragraph, but only in insets
3842 InsetText const * textinset = inInset().asInsetText();
3843 bool const maintext = textinset
3844 ? textinset->text().isMainText()
3847 if (!maintext && layout().needcprotect) {
3848 // Environments need cprotection regardless the content
3849 if (layout().latextype == LATEX_ENVIRONMENT)
3852 // Commands need cprotection if they contain specific chars
3853 int const nchars_escape = 9;
3854 static char_type const chars_escape[nchars_escape] = {
3855 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3857 docstring const pars = asString();
3858 for (int k = 0; k < nchars_escape; k++) {
3859 if (contains(pars, chars_escape[k]))
3864 // now check whether we have insets that need cprotection
3865 pos_type size = pos_type(d->text_.size());
3866 for (pos_type i = 0; i < size; ++i) {
3869 Inset const * ins = getInset(i);
3870 if (ins->needsCProtection(maintext, fragile))
3872 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3873 // Environments need cprotection regardless the content
3875 // Now check math environments
3876 InsetMath const * im = getInset(i)->asInsetMath();
3877 if (!im || im->cell(0).empty())
3879 switch(im->cell(0)[0]->lyxCode()) {
3880 case MATH_AMSARRAY_CODE:
3881 case MATH_SUBSTACK_CODE:
3883 case MATH_XYMATRIX_CODE:
3884 // these need cprotection
3895 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3897 return d->speller_state_.getRange(pos);
3901 bool Paragraph::isChar(pos_type pos) const
3903 if (Inset const * inset = getInset(pos))
3904 return inset->isChar();
3905 char_type const c = d->text_[pos];
3906 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3910 bool Paragraph::isSpace(pos_type pos) const
3912 if (Inset const * inset = getInset(pos))
3913 return inset->isSpace();
3914 char_type const c = d->text_[pos];
3915 return lyx::isSpace(c);
3920 Paragraph::getParLanguage(BufferParams const & bparams) const
3923 return getFirstFontSettings(bparams).language();
3924 // FIXME: we should check the prev par as well (Lgb)
3925 return bparams.language;
3929 bool Paragraph::isRTL(BufferParams const & bparams) const
3931 return getParLanguage(bparams)->rightToLeft()
3932 && !inInset().getLayout().forceLTR();
3936 void Paragraph::changeLanguage(BufferParams const & bparams,
3937 Language const * from, Language const * to)
3939 // change language including dummy font change at the end
3940 for (pos_type i = 0; i <= size(); ++i) {
3941 Font font = getFontSettings(bparams, i);
3942 if (font.language() == from) {
3943 font.setLanguage(to);
3945 d->requestSpellCheck(i);
3951 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3953 Language const * doc_language = bparams.language;
3954 for (auto const & f : d->fontlist_)
3955 if (f.font().language() != ignore_language &&
3956 f.font().language() != latex_language &&
3957 f.font().language() != doc_language)
3963 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3965 for (auto const & f : d->fontlist_) {
3966 Language const * lang = f.font().language();
3967 if (lang != ignore_language &&
3968 lang != latex_language)
3974 docstring Paragraph::asString(int options) const
3976 return asString(0, size(), options);
3980 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3982 odocstringstream os;
3985 && options & AS_STR_LABEL
3986 && !d->params_.labelString().empty())
3987 os << d->params_.labelString() << ' ';
3989 for (pos_type i = beg; i < end; ++i) {
3990 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3992 char_type const c = d->text_[i];
3993 if (isPrintable(c) || c == '\t'
3994 || (c == '\n' && (options & AS_STR_NEWLINES)))
3996 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3997 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3998 LASSERT(runparams != nullptr, return docstring());
3999 getInset(i)->plaintext(os, *runparams);
4001 getInset(i)->toString(os);
4010 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
4011 bool const shorten, bool const label) const
4013 size_t tmplen = shorten ? maxlen + 1 : maxlen;
4014 if (label && !labelString().empty())
4015 os += labelString() + ' ';
4016 if (!layout().isTocCaption())
4018 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4021 char_type const c = d->text_[i];
4024 else if (c == META_INSET)
4025 getInset(i)->forOutliner(os, tmplen, false);
4028 Text::shortenForOutliner(os, maxlen);
4032 void Paragraph::setInsetOwner(Inset const * inset)
4034 d->inset_owner_ = inset;
4038 int Paragraph::id() const
4044 void Paragraph::setId(int id)
4050 Layout const & Paragraph::layout() const
4056 void Paragraph::setLayout(Layout const & layout)
4058 d->layout_ = &layout;
4062 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4064 setLayout(tc.defaultLayout());
4068 void Paragraph::setPlainLayout(DocumentClass const & tc)
4070 setLayout(tc.plainLayout());
4074 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
4076 if (usePlainLayout())
4077 setPlainLayout(tclass);
4079 setDefaultLayout(tclass);
4083 Inset const & Paragraph::inInset() const
4085 LBUFERR(d->inset_owner_);
4086 return *d->inset_owner_;
4090 ParagraphParameters & Paragraph::params()
4096 ParagraphParameters const & Paragraph::params() const
4102 bool Paragraph::isFreeSpacing() const
4104 if (d->layout_->free_spacing)
4106 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4110 bool Paragraph::allowEmpty() const
4112 if (d->layout_->keepempty)
4114 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4118 bool Paragraph::brokenBiblio() const
4120 // There is a problem if there is no bibitem at position 0 in
4121 // paragraphs that need one, if there is another bibitem in the
4122 // paragraph or if this paragraph is not supposed to have
4123 // a bibitem inset at all.
4124 return ((d->layout_->labeltype == LABEL_BIBLIO
4125 && (d->insetlist_.find(BIBITEM_CODE) != 0
4126 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4127 || (d->layout_->labeltype != LABEL_BIBLIO
4128 && d->insetlist_.find(BIBITEM_CODE) != -1));
4132 int Paragraph::fixBiblio(Buffer const & buffer)
4134 // FIXME: when there was already an inset at 0, the return value is 1,
4135 // which does not tell whether another inset has been removed; the
4136 // cursor cannot be correctly updated.
4138 bool const track_changes = buffer.params().track_changes;
4139 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4141 // The case where paragraph is not BIBLIO
4142 if (d->layout_->labeltype != LABEL_BIBLIO) {
4143 if (bibitem_pos == -1)
4144 // No InsetBibitem => OK
4146 // There is an InsetBibitem: remove it!
4147 d->insetlist_.release(bibitem_pos);
4148 eraseChar(bibitem_pos, track_changes);
4149 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4152 bool const hasbibitem0 = bibitem_pos == 0;
4154 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4155 // There was an InsetBibitem at pos 0,
4156 // and no other one => OK
4157 if (bibitem_pos == -1)
4159 // there is a bibitem at the 0 position, but since
4160 // there is a second one, we copy the second on the
4161 // first. We're assuming there are at most two of
4162 // these, which there should be.
4163 // FIXME: why does it make sense to do that rather
4164 // than keep the first? (JMarc)
4165 Inset * inset = releaseInset(bibitem_pos);
4166 d->insetlist_.begin()->inset = inset;
4167 // This needs to be done to update the counter (#8499)
4168 buffer.updateBuffer();
4169 return -bibitem_pos;
4172 // We need to create an inset at the beginning
4173 Inset * inset = nullptr;
4174 if (bibitem_pos > 0) {
4175 // there was one somewhere in the paragraph, let's move it
4176 inset = d->insetlist_.release(bibitem_pos);
4177 eraseChar(bibitem_pos, track_changes);
4180 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4181 InsetCommandParams(BIBITEM_CODE));
4183 Font font(inherit_font, buffer.params().language);
4184 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4185 : Change::UNCHANGED));
4187 // This is needed to get the counters right
4188 buffer.updateBuffer();
4193 void Paragraph::checkAuthors(AuthorList const & authorList)
4195 d->changes_.checkAuthors(authorList);
4199 bool Paragraph::isChanged(pos_type pos) const
4201 return lookupChange(pos).changed();
4205 bool Paragraph::isInserted(pos_type pos) const
4207 return lookupChange(pos).inserted();
4211 bool Paragraph::isDeleted(pos_type pos) const
4213 return lookupChange(pos).deleted();
4217 InsetList const & Paragraph::insetList() const
4219 return d->insetlist_;
4223 void Paragraph::setInsetBuffers(Buffer & b)
4225 d->insetlist_.setBuffer(b);
4229 void Paragraph::resetBuffer()
4231 d->insetlist_.resetBuffer();
4235 Inset * Paragraph::releaseInset(pos_type pos)
4237 Inset * inset = d->insetlist_.release(pos);
4238 /// does not honour change tracking!
4239 eraseChar(pos, false);
4244 Inset * Paragraph::getInset(pos_type pos)
4246 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4247 ? d->insetlist_.get(pos) : nullptr;
4251 Inset const * Paragraph::getInset(pos_type pos) const
4253 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4254 ? d->insetlist_.get(pos) : nullptr;
4258 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4259 pos_type & right, TextCase action)
4261 // process sequences of modified characters; in change
4262 // tracking mode, this approach results in much better
4263 // usability than changing case on a char-by-char basis
4264 // We also need to track the current font, since font
4265 // changes within sequences can occur.
4266 vector<pair<char_type, Font> > changes;
4268 bool const trackChanges = bparams.track_changes;
4270 bool capitalize = true;
4272 for (; pos < right; ++pos) {
4273 char_type oldChar = d->text_[pos];
4274 char_type newChar = oldChar;
4276 // ignore insets and don't play with deleted text!
4277 if (oldChar != META_INSET && !isDeleted(pos)) {
4279 case text_lowercase:
4280 newChar = lowercase(oldChar);
4282 case text_capitalization:
4284 newChar = uppercase(oldChar);
4288 case text_uppercase:
4289 newChar = uppercase(oldChar);
4294 if (isWordSeparator(pos) || isDeleted(pos)) {
4295 // permit capitalization again
4299 if (oldChar != newChar) {
4300 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4301 if (pos != right - 1)
4303 // step behind the changing area
4307 int erasePos = pos - changes.size();
4308 for (size_t i = 0; i < changes.size(); i++) {
4309 insertChar(pos, changes[i].first,
4312 if (!eraseChar(erasePos, trackChanges)) {
4315 ++right; // expand selection
4323 int Paragraph::find(docstring const & str, bool cs, bool mw,
4324 pos_type start_pos, bool del) const
4326 pos_type pos = start_pos;
4327 int const strsize = str.length();
4329 pos_type const parsize = d->text_.size();
4330 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4331 // Ignore "invisible" letters such as ligature breaks
4332 // and hyphenation chars while searching
4333 while (pos < parsize - 1 && isInset(pos)) {
4334 Inset const * inset = getInset(pos);
4335 if (!inset->isLetter())
4337 odocstringstream os;
4338 inset->toString(os);
4339 if (!os.str().empty())
4343 if (cs && str[i] != d->text_[pos])
4345 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4347 if (!del && isDeleted(pos))
4354 // if necessary, check whether string matches word
4356 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4359 && !isWordSeparator(pos))
4363 return pos - start_pos;
4367 char_type Paragraph::getChar(pos_type pos) const
4369 return d->text_[pos];
4373 pos_type Paragraph::size() const
4375 return d->text_.size();
4379 bool Paragraph::empty() const
4381 return d->text_.empty();
4385 bool Paragraph::isInset(pos_type pos) const
4387 return d->text_[pos] == META_INSET;
4391 bool Paragraph::isSeparator(pos_type pos) const
4393 //FIXME: Are we sure this can be the only separator?
4394 return d->text_[pos] == ' ';
4398 void Paragraph::deregisterWords()
4400 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4401 Private::LangWordsMap::const_iterator ite = d->words_.end();
4402 for (; itl != ite; ++itl) {
4403 WordList & wl = theWordList(itl->first);
4404 Private::Words::const_iterator it = (itl->second).begin();
4405 Private::Words::const_iterator et = (itl->second).end();
4406 for (; it != et; ++it)
4413 void Paragraph::locateWord(pos_type & from, pos_type & to,
4414 word_location const loc, bool const ignore_deleted) const
4417 case WHOLE_WORD_STRICT:
4418 if (from == 0 || from == size()
4419 || isWordSeparator(from, ignore_deleted)
4420 || isWordSeparator(from - 1, ignore_deleted)) {
4427 // If we are already at the beginning of a word, do nothing
4428 if (!from || isWordSeparator(from - 1, ignore_deleted))
4433 // always move the cursor to the beginning of previous word
4434 while (from && !isWordSeparator(from - 1, ignore_deleted))
4438 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4441 // no need to move the 'from' cursor
4445 while (to < size() && !isWordSeparator(to, ignore_deleted))
4450 void Paragraph::collectWords()
4452 for (pos_type pos = 0; pos < size(); ++pos) {
4453 if (isWordSeparator(pos))
4455 pos_type from = pos;
4456 locateWord(from, pos, WHOLE_WORD);
4457 // Work around MSVC warning: The statement
4458 // if (pos < from + lyxrc.completion_minlength)
4459 // triggers a signed vs. unsigned warning.
4460 // I don't know why this happens, it could be a MSVC bug, or
4461 // related to LLP64 (windows) vs. LP64 (unix) programming
4462 // model, or the C++ standard might be ambigous in the section
4463 // defining the "usual arithmetic conversions". However, using
4464 // a temporary variable is safe and works on all compilers.
4465 pos_type const endpos = from + lyxrc.completion_minlength;
4468 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4469 if (cit == d->fontlist_.end())
4471 Language const * lang = cit->font().language();
4472 docstring const word = asString(from, pos, AS_STR_NONE);
4473 d->words_[lang->lang()].insert(word);
4478 void Paragraph::registerWords()
4480 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4481 Private::LangWordsMap::const_iterator ite = d->words_.end();
4482 for (; itl != ite; ++itl) {
4483 WordList & wl = theWordList(itl->first);
4484 Private::Words::const_iterator it = (itl->second).begin();
4485 Private::Words::const_iterator et = (itl->second).end();
4486 for (; it != et; ++it)
4492 void Paragraph::updateWords()
4500 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4502 SkipPositionsIterator begin = skips.begin();
4503 SkipPositions::iterator end = skips.end();
4504 if (pos > 0 && begin < end) {
4506 if (end->last == pos - 1) {
4511 skips.insert(end, FontSpan(pos, pos));
4515 Language * Paragraph::Private::locateSpellRange(
4516 pos_type & from, pos_type & to,
4517 SkipPositions & skips) const
4519 // skip leading white space
4520 while (from < to && owner_->isWordSeparator(from))
4522 // don't check empty range
4525 // get current language
4526 Language * lang = getSpellLanguage(from);
4527 pos_type last = from;
4528 bool samelang = true;
4529 bool sameinset = true;
4530 while (last < to && samelang && sameinset) {
4531 // hop to end of word
4532 while (last < to && !owner_->isWordSeparator(last)) {
4533 Inset const * inset = owner_->getInset(last);
4534 if (inset && inset->lyxCode() == SPECIALCHAR_CODE) {
4535 // check for "invisible" letters such as ligature breaks
4536 odocstringstream os;
4537 inset->toString(os);
4538 if (os.str().length() != 0) {
4539 // avoid spell check of visible special char insets
4540 // stop the loop in front of the special char inset
4545 appendSkipPosition(skips, last);
4546 } else if (owner_->isDeleted(last)) {
4547 appendSkipPosition(skips, last);
4551 // hop to next word while checking for insets
4552 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4553 if (Inset const * inset = owner_->getInset(last))
4554 sameinset = inset->isChar() && inset->isLetter();
4555 if (sameinset && owner_->isDeleted(last)) {
4556 appendSkipPosition(skips, last);
4561 if (sameinset && last < to) {
4562 // now check for language change
4563 samelang = lang == getSpellLanguage(last);
4566 // if language change detected backstep is needed
4574 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4577 const_cast<Language *>(owner_->getFontSettings(
4578 inset_owner_->buffer().params(), from).language());
4579 if (lang == inset_owner_->buffer().params().language
4580 && !lyxrc.spellchecker_alt_lang.empty()) {
4582 string const lang_variety =
4583 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4584 lang->setCode(lang_code);
4585 lang->setVariety(lang_variety);
4591 void Paragraph::requestSpellCheck(pos_type pos)
4593 d->requestSpellCheck(pos);
4597 bool Paragraph::needsSpellCheck() const
4599 SpellChecker::ChangeNumber speller_change_number = 0;
4600 if (theSpellChecker())
4601 speller_change_number = theSpellChecker()->changeNumber();
4602 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4603 d->speller_state_.needsCompleteRefresh(speller_change_number);
4605 return d->needsSpellCheck();
4609 bool Paragraph::Private::ignoreWord(docstring const & word) const
4611 // Ignore words with digits
4612 // FIXME: make this customizable
4613 // (note that some checkers ignore words with digits by default)
4614 docstring::const_iterator cit = word.begin();
4615 docstring::const_iterator const end = word.end();
4616 for (; cit != end; ++cit) {
4617 if (isNumber((*cit)))
4624 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4625 WordLangTuple & wl, docstring_list & suggestions,
4626 bool do_suggestion, bool check_learned) const
4628 SpellChecker::Result result = SpellChecker::WORD_OK;
4629 SpellChecker * speller = theSpellChecker();
4633 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4636 locateWord(from, to, WHOLE_WORD, true);
4637 if (from == to || from >= size())
4640 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4641 Language * lang = d->getSpellLanguage(from);
4643 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4646 wl = WordLangTuple(word, lang);
4651 if (needsSpellCheck() || check_learned) {
4653 if (!d->ignoreWord(word)) {
4654 bool const trailing_dot = to < size() && d->text_[to] == '.';
4655 result = speller->check(wl);
4656 if (SpellChecker::misspelled(result) && trailing_dot) {
4657 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4658 result = speller->check(wl);
4659 if (!SpellChecker::misspelled(result)) {
4660 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4662 from << ".." << to << "]");
4664 // spell check with dot appended failed too
4665 // restore original word/lang value
4666 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4667 wl = WordLangTuple(word, lang);
4671 if (!SpellChecker::misspelled(result)) {
4672 // area up to the begin of the next word is not misspelled
4673 while (end < size() && isWordSeparator(end))
4676 d->setMisspelled(from, end, result);
4678 result = d->speller_state_.getState(from);
4682 suggestions.clear();
4684 if (SpellChecker::misspelled(result)) {
4685 LYXERR(Debug::GUI, "misspelled word: \"" <<
4687 from << ".." << to << "]");
4689 speller->suggest(wl, suggestions);
4695 void Paragraph::anonymize()
4697 // This is a very crude anonymization for now
4698 for (char_type & c : d->text_)
4699 if (isLetterChar(c) || isNumber(c))
4704 void Paragraph::Private::markMisspelledWords(
4705 pos_type const & first, pos_type const & last,
4706 SpellChecker::Result result,
4707 docstring const & word,
4708 SkipPositions const & skips)
4710 if (!SpellChecker::misspelled(result)) {
4711 setMisspelled(first, last, SpellChecker::WORD_OK);
4715 SpellChecker * speller = theSpellChecker();
4716 // locate and enumerate the error positions
4717 int nerrors = speller->numMisspelledWords();
4719 SkipPositionsIterator it = skips.begin();
4720 SkipPositionsIterator et = skips.end();
4721 for (int index = 0; index < nerrors; ++index) {
4724 speller->misspelledWord(index, wstart, wlen);
4725 /// should not happen if speller supports range checks
4726 if (!wlen) continue;
4727 docstring const misspelled = word.substr(wstart, wlen);
4728 wstart += first + numskipped;
4729 if (snext < wstart) {
4730 /// mark the range of correct spelling
4731 numskipped += countSkips(it, et, wstart);
4732 setMisspelled(snext,
4733 wstart - 1, SpellChecker::WORD_OK);
4735 snext = wstart + wlen;
4736 numskipped += countSkips(it, et, snext);
4737 /// mark the range of misspelling
4738 setMisspelled(wstart, snext, result);
4739 LYXERR(Debug::GUI, "misspelled word: \"" <<
4740 misspelled << "\" [" <<
4741 wstart << ".." << (snext-1) << "]");
4744 if (snext <= last) {
4745 /// mark the range of correct spelling at end
4746 setMisspelled(snext, last, SpellChecker::WORD_OK);
4751 void Paragraph::spellCheck() const
4753 SpellChecker * speller = theSpellChecker();
4754 if (!speller || empty() ||!needsSpellCheck())
4758 d->rangeOfSpellCheck(start, endpos);
4759 if (speller->canCheckParagraph()) {
4760 // loop until we leave the range
4761 for (pos_type first = start; first < endpos; ) {
4762 pos_type last = endpos;
4763 Private::SkipPositions skips;
4764 Language * lang = d->locateSpellRange(first, last, skips);
4765 if (first >= endpos)
4767 // start the spell checker on the unit of meaning
4768 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4769 WordLangTuple wl = WordLangTuple(word, lang);
4770 SpellChecker::Result result = word.size() ?
4771 speller->check(wl) : SpellChecker::WORD_OK;
4772 d->markMisspelledWords(first, last, result, word, skips);
4776 static docstring_list suggestions;
4777 pos_type to = endpos;
4778 while (start < endpos) {
4780 spellCheck(start, to, wl, suggestions, false);
4784 d->readySpellCheck();
4788 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4790 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4791 if (result || pos <= 0 || pos > size())
4793 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4794 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4799 string Paragraph::magicLabel() const
4802 ss << "magicparlabel-" << id();