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);
348 void latexSpecialChar(
350 BufferParams const & bparams,
351 OutputParams const & runparams,
352 Font const & running_font,
353 string & alien_script,
354 Layout const & style,
357 unsigned int & column);
364 unsigned int & column);
370 unsigned int & column);
376 unsigned int & column);
379 void validate(LaTeXFeatures & features) const;
381 /// Checks if the paragraph contains only text and no inset or font change.
382 bool onlyText(Buffer const & buf, Font const & outerfont,
383 pos_type initial) const;
385 /// a vector of speller skip positions
386 typedef vector<FontSpan> SkipPositions;
387 typedef SkipPositions::const_iterator SkipPositionsIterator;
389 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
391 Language * getSpellLanguage(pos_type const from) const;
393 Language * locateSpellRange(pos_type & from, pos_type & to,
394 SkipPositions & skips) const;
396 bool hasSpellerChange() const
398 SpellChecker::ChangeNumber speller_change_number = 0;
399 if (theSpellChecker())
400 speller_change_number = theSpellChecker()->changeNumber();
401 return speller_change_number > speller_state_.currentChangeNumber();
404 bool ignoreWord(docstring const & word) const ;
406 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
408 pos_type textsize = owner_->size();
409 // check for sane arguments
410 if (to <= from || from >= textsize)
412 FontSpan fp = FontSpan(from, to - 1);
413 speller_state_.setRange(fp, state);
416 void requestSpellCheck(pos_type pos)
419 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
421 speller_state_.needsRefresh(pos);
424 void readySpellCheck()
426 speller_state_.needsRefresh(-1);
429 bool needsSpellCheck() const
431 return speller_state_.needsRefresh();
434 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
436 speller_state_.refreshRange(first, last);
438 last = owner_->size();
441 pos_type endpos = last;
442 owner_->locateWord(first, endpos, WHOLE_WORD, true);
445 owner_->locateWord(last, endpos, WHOLE_WORD, true);
450 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
454 while (it != et && it->first < start) {
455 long skip = it->last - it->first + 1;
463 void markMisspelledWords(pos_type const & first, pos_type const & last,
464 SpellChecker::Result result,
465 docstring const & word,
466 SkipPositions const & skips);
468 InsetCode ownerCode() const
470 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
473 /// Which Paragraph owns us?
477 Inset const * inset_owner_;
483 ParagraphParameters params_;
485 /// for recording and looking up changes
489 InsetList insetlist_;
492 pos_type begin_of_body_;
494 typedef docstring TextContainer;
498 typedef set<docstring> Words;
499 typedef map<string, Words> LangWordsMap;
503 Layout const * layout_;
505 SpellCheckerState speller_state_;
511 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
512 : owner_(owner), inset_owner_(nullptr), begin_of_body_(0), layout_(&layout), id_(-1)
519 int Paragraph::Private::make_id()
521 // The id is unique per session across buffers because it is used in
522 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
525 static int next_id(0);
530 Paragraph::Private::Private(Private const & p, Paragraph * owner)
531 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
532 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
533 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
534 layout_(p.layout_), id_(make_id())
536 requestSpellCheck(p.text_.size());
540 Paragraph::Private::Private(Private const & p, Paragraph * owner,
541 pos_type beg, pos_type end)
542 : owner_(owner), inset_owner_(p.inset_owner_),
543 params_(p.params_), changes_(p.changes_),
544 insetlist_(p.insetlist_, beg, end),
545 begin_of_body_(p.begin_of_body_), words_(p.words_),
546 layout_(p.layout_), id_(make_id())
548 if (beg >= pos_type(p.text_.size()))
550 text_ = p.text_.substr(beg, end - beg);
552 FontList::const_iterator fcit = fontlist_.begin();
553 FontList::const_iterator fend = fontlist_.end();
554 for (; fcit != fend; ++fcit) {
555 if (fcit->pos() < beg)
557 if (fcit->pos() >= end) {
558 // Add last entry in the fontlist_.
559 fontlist_.set(text_.size() - 1, fcit->font());
562 // Add a new entry in the fontlist_.
563 fontlist_.set(fcit->pos() - beg, fcit->font());
565 requestSpellCheck(p.text_.size());
569 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
570 bool output_active, TocBackend & backend) const
572 d->changes_.addToToc(cdit, buf, output_active, backend);
576 bool Paragraph::isDeleted(pos_type start, pos_type end) const
578 LASSERT(start >= 0 && start <= size(), return false);
579 LASSERT(end > start && end <= size() + 1, return false);
581 return d->changes_.isDeleted(start, end);
585 bool Paragraph::isChanged(pos_type start, pos_type end) const
587 LASSERT(start >= 0 && start <= size(), return false);
588 LASSERT(end > start && end <= size() + 1, return false);
590 return d->changes_.isChanged(start, end);
593 // FIXME: Ideally the diverse isChanged() methods should account for that!
594 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
596 LASSERT(start >= 0 && start <= size(), return false);
597 LASSERT(end > start && end <= size() + 1, return false);
599 for (auto const & icit : d->insetlist_) {
600 if (icit.pos < start)
604 if (icit.inset && icit.inset->isChanged())
610 bool Paragraph::isChanged() const
612 return d->changes_.isChanged();
616 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
618 // keep the logic here in sync with the logic of eraseChars()
622 Change const & change = d->changes_.lookup(size());
623 return change.inserted() && change.currentAuthor();
626 Change Paragraph::parEndChange() const
628 return d->changes_.lookup(size());
632 void Paragraph::setChange(Change const & change)
634 // beware of the imaginary end-of-par character!
635 d->changes_.set(change, 0, size() + 1);
638 * Propagate the change recursively - but not in case of DELETED!
640 * Imagine that your co-author makes changes in an existing inset. He
641 * sends your document to you and you come to the conclusion that the
642 * inset should go completely. If you erase it, LyX must not delete all
643 * text within the inset. Otherwise, the change tracked insertions of
644 * your co-author get lost and there is no way to restore them later.
646 * Conclusion: An inset's content should remain untouched if you delete it
649 if (!change.deleted()) {
650 for (pos_type pos = 0; pos < size(); ++pos) {
651 if (Inset * inset = getInset(pos))
652 inset->setChange(change);
658 void Paragraph::setChange(pos_type pos, Change const & change)
660 LASSERT(pos >= 0 && pos <= size(), return);
661 d->changes_.set(change, pos);
663 // see comment in setChange(Change const &) above
664 if (!change.deleted() && pos < size())
665 if (Inset * inset = getInset(pos))
666 inset->setChange(change);
670 Change const & Paragraph::lookupChange(pos_type pos) const
672 LBUFERR(pos >= 0 && pos <= size());
673 return d->changes_.lookup(pos);
677 void Paragraph::acceptChanges(pos_type start, pos_type end)
679 LASSERT(start >= 0 && start <= size(), return);
680 LASSERT(end > start && end <= size() + 1, return);
682 for (pos_type pos = start; pos < end; ++pos) {
683 switch (lookupChange(pos).type) {
684 case Change::UNCHANGED:
685 // accept changes in nested inset
686 if (Inset * inset = getInset(pos))
687 inset->acceptChanges();
690 case Change::INSERTED:
691 d->changes_.set(Change(Change::UNCHANGED), pos);
692 // also accept changes in nested inset
693 if (Inset * inset = getInset(pos))
694 inset->acceptChanges();
697 case Change::DELETED:
698 // Suppress access to non-existent
699 // "end-of-paragraph char"
701 eraseChar(pos, false);
712 void Paragraph::rejectChanges(pos_type start, pos_type end)
714 LASSERT(start >= 0 && start <= size(), return);
715 LASSERT(end > start && end <= size() + 1, return);
717 for (pos_type pos = start; pos < end; ++pos) {
718 switch (lookupChange(pos).type) {
719 case Change::UNCHANGED:
720 // reject changes in nested inset
721 if (Inset * inset = getInset(pos))
722 inset->rejectChanges();
725 case Change::INSERTED:
726 // Suppress access to non-existent
727 // "end-of-paragraph char"
729 eraseChar(pos, false);
735 case Change::DELETED:
736 d->changes_.set(Change(Change::UNCHANGED), pos);
738 // Do NOT reject changes within a deleted inset!
739 // There may be insertions of a co-author inside of it!
747 void Paragraph::Private::insertChar(pos_type pos, char_type c,
748 Change const & change)
750 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
753 changes_.insert(change, pos);
755 // This is actually very common when parsing buffers (and
756 // maybe inserting ascii text)
757 if (pos == pos_type(text_.size())) {
758 // when appending characters, no need to update tables
760 // but we want spell checking
761 requestSpellCheck(pos);
765 text_.insert(text_.begin() + pos, c);
767 // Update the font table.
768 fontlist_.increasePosAfterPos(pos);
771 insetlist_.increasePosAfterPos(pos);
773 // Update list of misspelled positions
774 speller_state_.increasePosAfterPos(pos);
778 bool Paragraph::insertInset(pos_type pos, Inset * inset,
779 Font const & font, Change const & change)
781 LASSERT(inset, return false);
782 LASSERT(pos >= 0 && pos <= size(), return false);
784 // Paragraph::insertInset() can be used in cut/copy/paste operation where
785 // d->inset_owner_ is not set yet.
786 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
789 d->insertChar(pos, META_INSET, change);
790 LASSERT(d->text_[pos] == META_INSET, return false);
792 // Add a new entry in the insetlist_.
793 d->insetlist_.insert(inset, pos);
795 // Some insets require run of spell checker
796 requestSpellCheck(pos);
802 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
804 LASSERT(pos >= 0 && pos <= size(), return false);
806 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
809 Change change = d->changes_.lookup(pos);
811 // set the character to DELETED if
812 // a) it was previously unchanged or
813 // b) it was inserted by a co-author
815 if (!change.changed() ||
816 (change.inserted() && !change.currentAuthor())) {
817 setChange(pos, Change(Change::DELETED));
818 // request run of spell checker
819 requestSpellCheck(pos);
823 if (change.deleted())
827 // Don't physically access the imaginary end-of-paragraph character.
828 // eraseChar() can only mark it as DELETED. A physical deletion of
829 // end-of-par must be handled externally.
835 d->changes_.erase(pos);
837 // if it is an inset, delete the inset entry
838 if (d->text_[pos] == META_INSET)
839 d->insetlist_.erase(pos);
841 d->text_.erase(d->text_.begin() + pos);
843 // Update the fontlist_
844 d->fontlist_.erase(pos);
846 // Update the insetlist_
847 d->insetlist_.decreasePosAfterPos(pos);
849 // Update list of misspelled positions
850 d->speller_state_.decreasePosAfterPos(pos);
851 d->speller_state_.refreshLast(size());
857 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
859 LASSERT(start >= 0 && start <= size(), return 0);
860 LASSERT(end >= start && end <= size() + 1, return 0);
863 for (pos_type count = end - start; count; --count) {
864 if (!eraseChar(i, trackChanges))
870 // Handle combining characters
871 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
872 otexstream & os, char_type c, char_type next,
873 OutputParams const & runparams)
875 // Writing next here may circumvent a possible font change between
876 // c and next. Since next is only output if it forms a surrogate pair
877 // with c we can ignore this:
878 // A font change inside a surrogate pair does not make sense and is
879 // hopefully impossible to input.
880 // FIXME: change tracking
881 // Is this correct WRT change tracking?
882 Encoding const & encoding = *(runparams.encoding);
883 docstring latex1 = encoding.latexChar(next).first;
884 if (runparams.inIPA) {
885 string const tipashortcut = Encodings::TIPAShortcut(next);
886 if (!tipashortcut.empty()) {
887 latex1 = from_ascii(tipashortcut);
890 docstring latex2 = encoding.latexChar(c).first;
892 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
893 // Encoding supports the combination:
894 // output as is (combining char after base char).
895 os << latex2 << latex1;
896 return latex1.length() + latex2.length();
899 os << latex1 << "{" << latex2 << "}";
900 return latex1.length() + latex2.length() + 2;
904 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
905 OutputParams const & runparams,
908 unsigned int & column,
910 Layout const & style)
912 if (style.pass_thru || runparams.pass_thru)
915 if (i + 1 < int(text_.size())) {
916 char_type next = text_[i + 1];
917 if (Encodings::isCombiningChar(next)) {
918 // This space has an accent, so we must always output it.
919 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
924 if (runparams.linelen > 0
925 && column > runparams.linelen
927 && text_[i - 1] != ' '
928 && (i + 1 < int(text_.size()))
929 // same in FreeSpacing mode
930 && !owner_->isFreeSpacing()
931 // In typewriter mode, we want to avoid
932 // ! . ? : at the end of a line
933 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
934 && (text_[i - 1] == '.'
935 || text_[i - 1] == '?'
936 || text_[i - 1] == ':'
937 || text_[i - 1] == '!'))) {
939 os.texrow().start(owner_->id(), i + 1);
941 } else if (style.free_spacing) {
950 void Paragraph::Private::latexInset(BufferParams const & bparams,
952 OutputParams & runparams,
955 Font const & outerfont,
957 Change & running_change,
958 Layout const & style,
960 unsigned int & column)
962 Inset * inset = owner_->getInset(i);
965 if (style.pass_thru) {
966 odocstringstream ods;
967 inset->plaintext(ods, runparams);
972 // FIXME: move this to InsetNewline::latex
973 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
974 // newlines are handled differently here than
975 // the default in simpleTeXSpecialChars().
976 if (!style.newline_allowed) {
980 bool needPar = false;
981 column += running_font.latexWriteEndChanges(
982 os, bparams, runparams,
983 basefont, basefont, needPar);
987 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
990 basefont = owner_->getLayoutFont(bparams, outerfont);
991 running_font = basefont;
993 if (runparams.moving_arg)
997 os.texrow().start(owner_->id(), i + 1);
1001 if (owner_->isDeleted(i)) {
1002 if( ++runparams.inDeletedInset == 1)
1003 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1006 if (inset->canTrackChanges()) {
1007 column += Changes::latexMarkChange(os, bparams, running_change,
1008 Change(Change::UNCHANGED), runparams);
1009 running_change = Change(Change::UNCHANGED);
1013 odocstream::pos_type const len = os.os().tellp();
1015 if (inset->forceLTR(runparams)
1016 && running_font.isRightToLeft()
1017 // ERT is an exception, it should be output with no
1018 // decorations at all
1019 && inset->lyxCode() != ERT_CODE) {
1020 if (runparams.use_polyglossia) {
1022 } else if (running_font.language()->lang() == "farsi"
1023 || running_font.language()->lang() == "arabic_arabi")
1024 os << "\\textLR{" << termcmd;
1030 if (open_font && (!inset->inheritFont() || inset->allowMultiPar())) {
1031 // Some insets cannot be inside a font change command.
1032 // However, even such insets *can* be placed in \L or \R
1033 // or their equivalents (for RTL language switches),
1034 // so we don't close the language in those cases
1035 // (= differing isRightToLeft()).
1036 // ArabTeX, though, doesn't seem to handle this special behavior.
1037 bool const inRLSwitch =
1038 basefont.isRightToLeft() != running_font.isRightToLeft()
1039 && basefont.language()->lang() != "arabic_arabtex"
1040 && running_font.language()->lang() != "arabic_arabtex";
1041 // Having said that, PassThru insets must be inside a font change command,
1042 // as we do not re-open the font inside. So:
1043 bool const closeLanguage = !inset->isPassThru() && !inRLSwitch;
1044 bool lang_closed = false;
1045 bool lang_switched_at_inset = false;
1046 // Close language if needed
1047 if (closeLanguage) {
1048 // We need prev_font here as language changes directly at inset
1049 // will only be started inside the inset.
1050 Font const prev_font = (i > 0) ?
1051 owner_->getFont(bparams, i - 1, outerfont)
1053 Font tmpfont(basefont);
1054 tmpfont.setLanguage(prev_font.language());
1055 bool needPar = false;
1056 unsigned int count = tmpfont.latexWriteEndChanges(os, bparams, runparams,
1058 needPar, closeLanguage);
1060 lang_closed = count > 0;
1061 lang_switched_at_inset = prev_font.language() != running_font.language();
1063 // Now re-do font changes in a way needed here
1064 // (using switches with multi-par insets)
1065 InsetText const * textinset = inset->asInsetText();
1066 bool const cprotect = textinset
1067 ? textinset->hasCProtectContent(runparams.moving_arg)
1068 && !textinset->text().isMainText()
1070 bool const multipar_inset = inset->allowMultiPar();
1071 unsigned int count2 = running_font.latexWriteStartChanges(os, bparams,
1072 runparams, basefont,
1073 running_font, multipar_inset,
1076 // Update the running_font, making sure, however,
1077 // to leave the language as it was.
1078 // FIXME: probably a better way to keep track of the old
1079 // language, than copying the entire font?
1080 Font const copy_font(running_font);
1081 basefont = owner_->getLayoutFont(bparams, outerfont);
1082 running_font = basefont;
1084 running_font.setLanguage(copy_font.language());
1085 // For these, we use switches, so they should be taken as
1086 // base inside the inset.
1087 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1088 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1089 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1090 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1094 runparams.local_font = &basefont;
1097 size_t const previous_row_count = os.texrow().rows();
1100 runparams.lastid = id_;
1101 runparams.lastpos = i;
1102 inset->latex(os, runparams);
1103 } catch (EncodingException & e) {
1104 // add location information and throw again.
1113 if (os.texrow().rows() > previous_row_count) {
1114 os.texrow().start(owner_->id(), i + 1);
1117 column += (unsigned int)(os.os().tellp() - len);
1120 if (owner_->isDeleted(i))
1121 --runparams.inDeletedInset;
1125 void Paragraph::Private::latexSpecialChar(otexstream & os,
1126 BufferParams const & bparams,
1127 OutputParams const & runparams,
1128 Font const & running_font,
1129 string & alien_script,
1130 Layout const & style,
1133 unsigned int & column)
1135 char_type const c = owner_->getUChar(bparams, runparams, i);
1137 if (style.pass_thru || runparams.pass_thru
1138 || contains(style.pass_thru_chars, c)
1139 || contains(runparams.pass_thru_chars, c)) {
1141 Encoding const * const enc = runparams.encoding;
1142 if (enc && !enc->encodable(c))
1143 throw EncodingException(c);
1149 // TIPA uses its own T3 encoding
1150 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1152 // If T1 font encoding is used, use the special
1153 // characters it provides.
1154 // NOTE: Some languages reset the font encoding internally to a
1155 // non-standard font encoding. If we are using such a language,
1156 // we do not output special T1 chars.
1157 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1158 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1159 && latexSpecialT1(c, os, i, column))
1161 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1162 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1163 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1166 // Otherwise, we use what LaTeX provides us.
1169 os << "\\textbackslash" << termcmd;
1173 os << "\\textless" << termcmd;
1177 os << "\\textgreater" << termcmd;
1181 os << "\\textbar" << termcmd;
1186 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1187 (end_pos == -1 || i + 1 < end_pos) &&
1188 text_[i+1] == '-') {
1189 // Prevent "--" becoming an en dash and "---" an em dash.
1190 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1196 os << "\\textquotedbl" << termcmd;
1201 case '%': case '#': case '{':
1209 os << "\\textasciitilde" << termcmd;
1214 os << "\\textasciicircum" << termcmd;
1221 // avoid being mistaken for optional arguments
1229 // Blanks are printed before font switching.
1230 // Sure? I am not! (try nice-latex)
1231 // I am sure it's correct. LyX might be smarter
1232 // in the future, but for now, nothing wrong is
1238 // XeTeX's dash behaviour is determined via a global setting
1239 if (bparams.use_dash_ligatures
1240 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1242 // TODO #10961: && not in inset Flex Code
1243 // TODO #10961: && not in layout LyXCode
1244 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1261 Encoding const & encoding = *(runparams.encoding);
1262 char_type next = '\0';
1263 if (i + 1 < int(text_.size())) {
1264 next = text_[i + 1];
1265 if (Encodings::isCombiningChar(next)) {
1266 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1271 pair<docstring, bool> latex = encoding.latexChar(c);
1272 docstring nextlatex;
1273 bool nexttipas = false;
1274 string nexttipashortcut;
1275 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1276 nextlatex = encoding.latexChar(next).first;
1277 if (runparams.inIPA) {
1278 nexttipashortcut = Encodings::TIPAShortcut(next);
1279 nexttipas = !nexttipashortcut.empty();
1283 if (runparams.inIPA) {
1284 string const tipashortcut = Encodings::TIPAShortcut(c);
1285 if (!tipashortcut.empty()) {
1286 latex.first = from_ascii(tipashortcut);
1287 latex.second = false;
1291 // eventually close "script wrapper" command (see `Paragraph::latex`)
1292 if (!alien_script.empty()
1293 && alien_script != Encodings::isKnownScriptChar(next)) {
1294 column += latex.first.length();
1295 alien_script.clear();
1296 os << latex.first << "}";
1300 && ((!prefixIs(nextlatex, '\\')
1301 && !prefixIs(nextlatex, '{')
1302 && !prefixIs(nextlatex, '}'))
1304 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1306 // Prevent eating of a following space or command corruption by
1307 // following characters
1308 if (next == ' ' || next == '\0') {
1309 column += latex.first.length() + 1;
1310 os << latex.first << "{}";
1312 column += latex.first.length();
1313 os << latex.first << " ";
1316 column += latex.first.length() - 1;
1324 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1325 pos_type i, unsigned int & column)
1331 // In T1 encoding, these characters exist
1332 // but we should avoid ligatures
1333 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1335 os << "\\textcompwordmark" << termcmd;
1342 // soul.sty breaks with \char`\"
1343 os << "\\textquotedbl" << termcmd;
1352 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1353 pos_type i, unsigned int & column)
1355 // TU encoding is currently on par with T1.
1356 return latexSpecialT1(c, os, i, column);
1360 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1361 pos_type /*i*/, unsigned int & column)
1371 os << "\\textvertline" << termcmd;
1380 void Paragraph::Private::validate(LaTeXFeatures & features) const
1382 if (layout_->inpreamble && inset_owner_) {
1383 // FIXME: Using a string stream here circumvents the encoding
1384 // switching machinery of odocstream. Therefore the
1385 // output is wrong if this paragraph contains content
1386 // that needs to switch encoding.
1387 Buffer const & buf = inset_owner_->buffer();
1388 otexstringstream os;
1389 os << layout_->preamble();
1390 size_t const length = os.length();
1391 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1392 buf.getParFromID(owner_->id()).pit(), os,
1393 features.runparams(), string(), 0, -1, true);
1394 if (os.length() > length)
1395 features.addPreambleSnippet(os.release(), true);
1398 if (features.runparams().flavor == OutputParams::HTML
1399 && layout_->htmltitle()) {
1400 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1403 // check the params.
1404 if (!params_.spacing().isDefault())
1405 features.require("setspace");
1408 features.useLayout(layout_->name());
1411 fontlist_.validate(features);
1413 // then the indentation
1414 if (!params_.leftIndent().zero())
1415 features.require("ParagraphLeftIndent");
1418 InsetList::const_iterator icit = insetlist_.begin();
1419 InsetList::const_iterator iend = insetlist_.end();
1420 for (; icit != iend; ++icit) {
1422 features.inDeletedInset(owner_->isDeleted(icit->pos));
1423 if (icit->inset->lyxCode() == FOOT_CODE) {
1424 // FIXME: an item inset would make things much easier.
1425 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1426 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1427 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1428 && (icit->pos < begin_of_body_
1429 || (icit->pos == begin_of_body_
1430 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1431 features.saveNoteEnv("description");
1433 icit->inset->validate(features);
1434 features.inDeletedInset(false);
1435 if (layout_->needprotect &&
1436 icit->inset->lyxCode() == FOOT_CODE)
1437 features.require("NeedLyXFootnoteCode");
1441 // then the contents
1442 BufferParams const bp = features.runparams().is_child
1443 ? features.buffer().masterParams() : features.buffer().params();
1444 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1445 char_type c = text_[i];
1446 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1448 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1449 features.require("textquotedblp");
1450 else if (bp.main_font_encoding() != "T1"
1451 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1452 features.require("textquotedbl");
1453 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1454 // features that depend on the font or input encoding
1455 string feats = ci.textpreamble();
1456 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1457 if (fontenc.empty())
1458 fontenc = features.runparams().main_fontenc;
1459 while (!feats.empty()) {
1461 feats = split(feats, feat, ',');
1462 if (contains(feat, "!=")) {
1463 // a feature that is required except for the spcified
1464 // font or input encodings
1466 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1467 // multiple encodings are separated by semicolon
1468 vector<string> context = getVectorFromString(contexts, ";");
1469 // require feature if the context matches neither current font
1470 // nor input encoding
1471 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1472 && std::find(context.begin(), context.end(),
1473 features.runparams().encoding->name()) == context.end())
1474 features.require(realfeature);
1475 } else if (contains(feat, '=')) {
1476 // a feature that is required only for the spcified
1477 // font or input encodings
1479 string const contexts = split(feat, realfeature, '=');
1480 // multiple encodings are separated by semicolon
1481 vector<string> context = getVectorFromString(contexts, ";");
1482 // require feature if the context matches either current font
1483 // or input encoding
1484 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1485 || std::find(context.begin(), context.end(),
1486 features.runparams().encoding->name()) != context.end())
1487 features.require(realfeature);
1490 } else if (!bp.use_dash_ligatures
1491 && (c == 0x2013 || c == 0x2014)
1492 && bp.useNonTeXFonts
1493 && features.runparams().flavor == OutputParams::XETEX)
1494 // XeTeX's dash behaviour is determined via a global setting
1495 features.require("xetexdashbreakstate");
1496 BufferEncodings::validate(c, features);
1500 /////////////////////////////////////////////////////////////////////
1504 /////////////////////////////////////////////////////////////////////
1507 Layout const emptyParagraphLayout;
1510 Paragraph::Paragraph()
1511 : d(new Paragraph::Private(this, emptyParagraphLayout))
1518 Paragraph::Paragraph(Paragraph const & par)
1519 : itemdepth(par.itemdepth),
1520 d(new Paragraph::Private(*par.d, this))
1526 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1527 : itemdepth(par.itemdepth),
1528 d(new Paragraph::Private(*par.d, this, beg, end))
1534 Paragraph & Paragraph::operator=(Paragraph const & par)
1536 // needed as we will destroy the private part before copying it
1538 itemdepth = par.itemdepth;
1542 d = new Private(*par.d, this);
1549 Paragraph::~Paragraph()
1558 // this shall be called just before every "os << ..." action.
1559 void flushString(ostream & os, docstring & s)
1568 void Paragraph::write(ostream & os, BufferParams const & bparams,
1569 depth_type & dth) const
1571 // The beginning or end of a deeper (i.e. nested) area?
1572 if (dth != d->params_.depth()) {
1573 if (d->params_.depth() > dth) {
1574 while (d->params_.depth() > dth) {
1575 os << "\n\\begin_deeper";
1579 while (d->params_.depth() < dth) {
1580 os << "\n\\end_deeper";
1586 // First write the layout
1587 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1589 d->params_.write(os);
1591 Font font1(inherit_font, bparams.language);
1593 Change running_change = Change(Change::UNCHANGED);
1595 // this string is used as a buffer to avoid repetitive calls
1596 // to to_utf8(), which turn out to be expensive (JMarc)
1597 docstring write_buffer;
1600 for (pos_type i = 0; i <= size(); ++i) {
1602 Change const & change = lookupChange(i);
1603 if (change != running_change)
1604 flushString(os, write_buffer);
1605 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1606 running_change = change;
1611 // Write font changes
1612 Font font2 = getFontSettings(bparams, i);
1613 if (font2 != font1) {
1614 flushString(os, write_buffer);
1615 font2.lyxWriteChanges(font1, os);
1620 char_type const c = d->text_[i];
1623 if (Inset const * inset = getInset(i)) {
1624 flushString(os, write_buffer);
1625 if (inset->directWrite()) {
1626 // international char, let it write
1627 // code directly so it's shorter in
1633 os << "\\begin_inset ";
1635 os << "\n\\end_inset\n\n";
1638 // FIXME This can be removed again once the mystery
1639 // crash has been resolved.
1644 flushString(os, write_buffer);
1645 os << "\n\\backslash\n";
1649 flushString(os, write_buffer);
1650 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1657 if ((column > 70 && c == ' ')
1659 flushString(os, write_buffer);
1663 // this check is to amend a bug. LyX sometimes
1664 // inserts '\0' this could cause problems.
1666 write_buffer.push_back(c);
1668 LYXERR0("NUL char in structure.");
1674 flushString(os, write_buffer);
1675 os << "\n\\end_layout\n";
1676 // FIXME This can be removed again once the mystery
1677 // crash has been resolved.
1682 void Paragraph::validate(LaTeXFeatures & features) const
1684 d->validate(features);
1685 bool fragile = features.runparams().moving_arg;
1686 fragile |= layout().needprotect;
1687 if (inInset().getLayout().isNeedProtect())
1689 if (needsCProtection(fragile))
1690 features.require("cprotect");
1694 void Paragraph::insert(pos_type start, docstring const & str,
1695 Font const & font, Change const & change)
1697 for (size_t i = 0, n = str.size(); i != n ; ++i)
1698 insertChar(start + i, str[i], font, change);
1702 void Paragraph::appendChar(char_type c, Font const & font,
1703 Change const & change)
1706 d->changes_.insert(change, d->text_.size());
1707 // when appending characters, no need to update tables
1708 d->text_.push_back(c);
1709 setFont(d->text_.size() - 1, font);
1710 d->requestSpellCheck(d->text_.size() - 1);
1714 void Paragraph::appendString(docstring const & s, Font const & font,
1715 Change const & change)
1717 pos_type end = s.size();
1718 size_t oldsize = d->text_.size();
1719 size_t newsize = oldsize + end;
1720 size_t capacity = d->text_.capacity();
1721 if (newsize >= capacity)
1722 d->text_.reserve(max(capacity + 100, newsize));
1724 // when appending characters, no need to update tables
1727 // FIXME: Optimize this!
1728 for (size_t i = oldsize; i != newsize; ++i) {
1730 d->changes_.insert(change, i);
1731 d->requestSpellCheck(i);
1733 d->fontlist_.set(oldsize, font);
1734 d->fontlist_.set(newsize - 1, font);
1738 void Paragraph::insertChar(pos_type pos, char_type c,
1741 d->insertChar(pos, c, Change(trackChanges ?
1742 Change::INSERTED : Change::UNCHANGED));
1746 void Paragraph::insertChar(pos_type pos, char_type c,
1747 Font const & font, bool trackChanges)
1749 d->insertChar(pos, c, Change(trackChanges ?
1750 Change::INSERTED : Change::UNCHANGED));
1755 void Paragraph::insertChar(pos_type pos, char_type c,
1756 Font const & font, Change const & change)
1758 d->insertChar(pos, c, change);
1763 void Paragraph::resetFonts(Font const & font)
1765 d->fontlist_.clear();
1766 d->fontlist_.set(0, font);
1767 d->fontlist_.set(d->text_.size() - 1, font);
1770 // Gets uninstantiated font setting at position.
1771 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1775 LYXERR0("pos: " << pos << " size: " << size());
1779 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1780 if (cit != d->fontlist_.end())
1783 if (pos == size() && !empty())
1784 return getFontSettings(bparams, pos - 1);
1786 // Optimisation: avoid a full font instantiation if there is no
1787 // language change from previous call.
1788 static Font previous_font;
1789 static Language const * previous_lang = nullptr;
1790 Language const * lang = getParLanguage(bparams);
1791 if (lang != previous_lang) {
1792 previous_lang = lang;
1793 previous_font = Font(inherit_font, lang);
1795 return previous_font;
1799 FontSpan Paragraph::fontSpan(pos_type pos) const
1801 LBUFERR(pos <= size());
1804 return FontSpan(pos, pos);
1807 FontList::const_iterator cit = d->fontlist_.begin();
1808 FontList::const_iterator end = d->fontlist_.end();
1809 for (; cit != end; ++cit) {
1810 if (cit->pos() >= pos) {
1811 if (pos >= beginOfBody())
1812 return FontSpan(max(start, beginOfBody()),
1815 return FontSpan(start,
1816 min(beginOfBody() - 1,
1819 start = cit->pos() + 1;
1822 // This should not happen, but if so, we take no chances.
1823 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1824 LASSERT(false, return FontSpan(pos, pos));
1828 // Gets uninstantiated font setting at position 0
1829 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1831 if (!empty() && !d->fontlist_.empty())
1832 return d->fontlist_.begin()->font();
1834 // Optimisation: avoid a full font instantiation if there is no
1835 // language change from previous call.
1836 static Font previous_font;
1837 static Language const * previous_lang = nullptr;
1838 if (bparams.language != previous_lang) {
1839 previous_lang = bparams.language;
1840 previous_font = Font(inherit_font, bparams.language);
1843 return previous_font;
1847 // Gets the fully instantiated font at a given position in a paragraph
1848 // This is basically the same function as Text::GetFont() in text2.cpp.
1849 // The difference is that this one is used for generating the LaTeX file,
1850 // and thus cosmetic "improvements" are disallowed: This has to deliver
1851 // the true picture of the buffer. (Asger)
1852 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1853 Font const & outerfont) const
1857 Font font = getFontSettings(bparams, pos);
1859 pos_type const body_pos = beginOfBody();
1860 FontInfo & fi = font.fontInfo();
1862 fi.realize(d->layout_->labelfont);
1864 fi.realize(d->layout_->font);
1866 fi.realize(outerfont.fontInfo());
1867 fi.realize(bparams.getFont().fontInfo());
1873 Font const Paragraph::getLabelFont
1874 (BufferParams const & bparams, Font const & outerfont) const
1876 FontInfo tmpfont = d->layout_->labelfont;
1877 tmpfont.realize(outerfont.fontInfo());
1878 tmpfont.realize(bparams.getFont().fontInfo());
1879 return Font(tmpfont, getParLanguage(bparams));
1883 Font const Paragraph::getLayoutFont
1884 (BufferParams const & bparams, Font const & outerfont) const
1886 FontInfo tmpfont = d->layout_->font;
1887 tmpfont.realize(outerfont.fontInfo());
1888 tmpfont.realize(bparams.getFont().fontInfo());
1889 return Font(tmpfont, getParLanguage(bparams));
1893 char_type Paragraph::getUChar(BufferParams const & bparams,
1894 OutputParams const & rp,
1897 char_type c = d->text_[pos];
1899 // Return unchanged character in LTR languages
1900 // or if we use poylglossia/bidi (XeTeX).
1901 if (rp.useBidiPackage()
1902 || !getFontSettings(bparams, pos).isRightToLeft())
1905 // Without polyglossia/bidi, we need to account for some special cases.
1906 // FIXME This needs to be audited!
1908 // * The input is as expected for all delimiters
1909 // => checked for Hebrew!
1910 // * The output matches the display in the LyX workarea
1911 // => checked for Hebrew!
1912 // * The special cases below are really necessary
1913 // => checked for Hebrew!
1914 // * In arabic_arabi, brackets are transformed to Arabic
1915 // Ornate Parentheses. Is this is really wanted?
1917 string const & lang = getFontSettings(bparams, pos).language()->lang();
1920 // 1. In the following languages, parentheses need to be reversed.
1921 // Also with polyglodia/luabidi
1922 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1924 // 2. In the following languages, brackets don't need to be reversed.
1925 bool const reversebrackets = lang != "arabic_arabtex"
1926 && lang != "arabic_arabi"
1929 // Now swap delimiters if needed.
1940 if (reversebrackets)
1944 if (reversebrackets)
1965 void Paragraph::setFont(pos_type pos, Font const & font)
1967 LASSERT(pos <= size(), return);
1969 // First, reduce font against layout/label font
1970 // Update: The setCharFont() routine in text2.cpp already
1971 // reduces font, so we don't need to do that here. (Asger)
1973 d->fontlist_.set(pos, font);
1977 void Paragraph::makeSameLayout(Paragraph const & par)
1979 d->layout_ = par.d->layout_;
1980 d->params_ = par.d->params_;
1984 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1986 if (isFreeSpacing())
1992 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1993 if (eraseChar(pos, trackChanges))
1999 return count > 0 || pos > 0;
2003 bool Paragraph::hasSameLayout(Paragraph const & par) const
2005 return par.d->layout_ == d->layout_
2006 && d->params_.sameLayout(par.d->params_);
2010 depth_type Paragraph::getDepth() const
2012 return d->params_.depth();
2016 depth_type Paragraph::getMaxDepthAfter() const
2018 if (d->layout_->isEnvironment())
2019 return d->params_.depth() + 1;
2021 return d->params_.depth();
2025 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2027 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2028 return getDefaultAlign(bparams);
2030 return d->params_.align();
2034 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2036 LyXAlignment res = layout().align;
2037 if (isRTL(bparams)) {
2039 if (res == LYX_ALIGN_LEFT)
2040 res = LYX_ALIGN_RIGHT;
2041 else if (res == LYX_ALIGN_RIGHT)
2042 res = LYX_ALIGN_LEFT;
2048 docstring const & Paragraph::labelString() const
2050 return d->params_.labelString();
2054 // the next two functions are for the manual labels
2055 docstring const Paragraph::getLabelWidthString() const
2057 if (d->layout_->margintype == MARGIN_MANUAL
2058 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2059 return d->params_.labelWidthString();
2061 return _("Senseless with this layout!");
2065 void Paragraph::setLabelWidthString(docstring const & s)
2067 d->params_.labelWidthString(s);
2071 docstring Paragraph::expandLabel(Layout const & layout,
2072 BufferParams const & bparams) const
2074 return expandParagraphLabel(layout, bparams, true);
2078 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2079 BufferParams const & bparams, bool process_appendix) const
2081 DocumentClass const & tclass = bparams.documentClass();
2082 string const & lang = getParLanguage(bparams)->code();
2083 bool const in_appendix = process_appendix && d->params_.appendix();
2084 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2086 if (fmt.empty() && !layout.counter.empty())
2087 return tclass.counters().theCounter(layout.counter, lang);
2089 // handle 'inherited level parts' in 'fmt',
2090 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2091 size_t const i = fmt.find('@', 0);
2092 if (i != docstring::npos) {
2093 size_t const j = fmt.find('@', i + 1);
2094 if (j != docstring::npos) {
2095 docstring parent(fmt, i + 1, j - i - 1);
2096 docstring label = from_ascii("??");
2097 if (tclass.hasLayout(parent))
2098 label = expandParagraphLabel(tclass[parent], bparams,
2100 fmt = docstring(fmt, 0, i) + label
2101 + docstring(fmt, j + 1, docstring::npos);
2105 return tclass.counters().counterLabel(fmt, lang);
2109 void Paragraph::applyLayout(Layout const & new_layout)
2111 d->layout_ = &new_layout;
2112 LyXAlignment const oldAlign = d->params_.align();
2114 if (!(oldAlign & d->layout_->alignpossible)) {
2115 frontend::Alert::warning(_("Alignment not permitted"),
2116 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2117 d->params_.align(LYX_ALIGN_LAYOUT);
2122 pos_type Paragraph::beginOfBody() const
2124 return d->begin_of_body_;
2128 void Paragraph::setBeginOfBody()
2130 if (d->layout_->labeltype != LABEL_MANUAL) {
2131 d->begin_of_body_ = 0;
2135 // Unroll the first two cycles of the loop
2136 // and remember the previous character to
2137 // remove unnecessary getChar() calls
2139 pos_type end = size();
2140 bool prev_char_deleted = false;
2141 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2144 char_type previous_char = d->text_[i];
2145 if (!(isNewline(i) || isEnvSeparator(i))) {
2147 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2148 char_type temp = d->text_[i];
2149 prev_char_deleted = isDeleted(i);
2150 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2153 previous_char = temp;
2159 d->begin_of_body_ = i;
2163 bool Paragraph::allowParagraphCustomization() const
2165 return inInset().allowParagraphCustomization();
2169 bool Paragraph::usePlainLayout() const
2171 return inInset().usePlainLayout();
2175 bool Paragraph::isPassThru() const
2177 return inInset().isPassThru() || d->layout_->pass_thru;
2182 // paragraphs inside floats need different alignment tags to avoid
2185 bool noTrivlistCentering(InsetCode code)
2187 return code == FLOAT_CODE
2188 || code == WRAP_CODE
2189 || code == CELL_CODE;
2193 string correction(string const & orig)
2195 if (orig == "flushleft")
2196 return "raggedright";
2197 if (orig == "flushright")
2198 return "raggedleft";
2199 if (orig == "center")
2205 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2206 InsetCode code, bool const lastpar, int & col)
2208 string macro = suffix + "{";
2209 if (noTrivlistCentering(code)) {
2211 // the last paragraph in non-trivlist-aligned
2212 // context is special (to avoid unwanted whitespace)
2213 if (suffix == "\\begin") {
2214 macro = "\\" + correction(env) + "{}";
2215 os << from_ascii(macro);
2216 col += macro.size();
2221 macro += correction(env);
2225 if (suffix == "\\par\\end") {
2229 os << from_ascii(macro);
2230 col += macro.size();
2231 if (suffix == "\\begin") {
2241 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2242 otexstream & os, OutputParams const & runparams) const
2247 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2248 (layout_->toggle_indent != ITOGGLE_NEVER) :
2249 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2251 if (canindent && params_.noindent() && !layout_->pass_thru) {
2252 os << "\\noindent ";
2256 LyXAlignment const curAlign = params_.align();
2258 if (curAlign == layout_->align)
2262 case LYX_ALIGN_NONE:
2263 case LYX_ALIGN_BLOCK:
2264 case LYX_ALIGN_LAYOUT:
2265 case LYX_ALIGN_SPECIAL:
2266 case LYX_ALIGN_DECIMAL:
2268 case LYX_ALIGN_LEFT:
2269 case LYX_ALIGN_RIGHT:
2270 case LYX_ALIGN_CENTER:
2271 if (runparams.moving_arg) {
2278 string const begin_tag = "\\begin";
2279 InsetCode code = ownerCode();
2280 bool const lastpar = runparams.isLastPar;
2281 // RTL in classic (PDF)LaTeX (without the Bidi package)
2282 // Luabibdi (used by LuaTeX) behaves like classic
2283 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2284 && !runparams.useBidiPackage();
2287 case LYX_ALIGN_NONE:
2288 case LYX_ALIGN_BLOCK:
2289 case LYX_ALIGN_LAYOUT:
2290 case LYX_ALIGN_SPECIAL:
2291 case LYX_ALIGN_DECIMAL:
2293 case LYX_ALIGN_LEFT: {
2295 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2296 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2298 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2300 } case LYX_ALIGN_RIGHT: {
2302 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2303 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2305 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2307 } case LYX_ALIGN_CENTER: {
2308 corrected_env(os, begin_tag, "center", code, lastpar, column);
2317 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2318 otexstream & os, OutputParams const & runparams) const
2320 LyXAlignment const curAlign = params_.align();
2322 if (curAlign == layout_->align)
2326 case LYX_ALIGN_NONE:
2327 case LYX_ALIGN_BLOCK:
2328 case LYX_ALIGN_LAYOUT:
2329 case LYX_ALIGN_SPECIAL:
2330 case LYX_ALIGN_DECIMAL:
2332 case LYX_ALIGN_LEFT:
2333 case LYX_ALIGN_RIGHT:
2334 case LYX_ALIGN_CENTER:
2335 if (runparams.moving_arg)
2340 bool output = false;
2342 string const end_tag = "\\par\\end";
2343 InsetCode code = ownerCode();
2344 bool const lastpar = runparams.isLastPar;
2345 // RTL in classic (PDF)LaTeX (without the Bidi package)
2346 // Luabibdi (used by LuaTeX) behaves like classic
2347 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2348 && !runparams.useBidiPackage();
2351 case LYX_ALIGN_NONE:
2352 case LYX_ALIGN_BLOCK:
2353 case LYX_ALIGN_LAYOUT:
2354 case LYX_ALIGN_SPECIAL:
2355 case LYX_ALIGN_DECIMAL:
2357 case LYX_ALIGN_LEFT: {
2359 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2360 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2362 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2364 } case LYX_ALIGN_RIGHT: {
2366 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2367 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2369 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2371 } case LYX_ALIGN_CENTER: {
2372 corrected_env(os, end_tag, "center", code, lastpar, col);
2377 return output || lastpar;
2381 // This one spits out the text of the paragraph
2382 void Paragraph::latex(BufferParams const & bparams,
2383 Font const & outerfont,
2385 OutputParams const & runparams,
2386 int start_pos, int end_pos, bool force) const
2388 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2390 // FIXME This check should not be needed. Perhaps issue an
2391 // error if it triggers.
2392 Layout const & style = inInset().forcePlainLayout() ?
2393 bparams.documentClass().plainLayout() : *d->layout_;
2395 if (!force && style.inpreamble)
2398 bool const allowcust = allowParagraphCustomization();
2400 // Current base font for all inherited font changes, without any
2401 // change caused by an individual character, except for the language:
2402 // It is set to the language of the first character.
2403 // As long as we are in the label, this font is the base font of the
2404 // label. Before the first body character it is set to the base font
2408 // If there is an open font-encoding changing command (script wrapper),
2409 // alien_script is set to its name
2410 string alien_script;
2413 // Maybe we have to create a optional argument.
2414 pos_type body_pos = beginOfBody();
2415 unsigned int column = 0;
2417 Font real_outerfont = outerfont;
2418 // If we are inside an non inheritFont() inset, the real main
2419 // properties of the outerfont are those of the local_font
2420 if (!inInset().inheritFont() && runparams.local_font != nullptr)
2421 real_outerfont.setProperties(runparams.local_font->fontInfo());
2424 // the optional argument is kept in curly brackets in
2425 // case it contains a ']'
2426 // This is not strictly needed, but if this is changed it
2427 // would be a file format change, and tex2lyx would need
2428 // to be adjusted, since it unconditionally removes the
2429 // braces when it parses \item.
2432 basefont = getLabelFont(bparams, real_outerfont);
2434 basefont = getLayoutFont(bparams, real_outerfont);
2437 // Which font is currently active?
2438 Font running_font(basefont);
2439 // Do we have an open font change?
2440 bool open_font = false;
2442 Change runningChange = Change(Change::UNCHANGED);
2444 Encoding const * const prev_encoding = runparams.encoding;
2446 os.texrow().start(id(), 0);
2448 // if the paragraph is empty, the loop will not be entered at all
2450 // For InTitle commands, we have already opened a group
2451 // in output_latex::TeXOnePar.
2452 if (style.isCommand() && !style.intitle) {
2456 if (!style.leftdelim().empty()) {
2457 os << style.leftdelim();
2458 column += style.leftdelim().size();
2461 column += d->startTeXParParams(bparams, os, runparams);
2464 // Whether a \par can be issued for insets typeset inline with text.
2465 // Yes if greater than 0. This has to be static.
2466 THREAD_LOCAL_STATIC int parInline = 0;
2468 bool multipar_inset = false;
2470 for (pos_type i = 0; i < size(); ++i) {
2471 // First char in paragraph or after label?
2472 if (i == body_pos) {
2475 bool needPar = false;
2476 column += running_font.latexWriteEndChanges(
2477 os, bparams, runparams,
2478 basefont, basefont, needPar,
2482 basefont = getLayoutFont(bparams, real_outerfont);
2483 running_font = basefont;
2485 column += Changes::latexMarkChange(os, bparams,
2486 runningChange, Change(Change::UNCHANGED),
2488 runningChange = Change(Change::UNCHANGED);
2493 // For InTitle commands, we have already opened a group
2494 // in output_latex::TeXOnePar.
2495 if (style.isCommand() && !style.intitle) {
2500 if (!style.leftdelim().empty()) {
2501 os << style.leftdelim();
2502 column += style.leftdelim().size();
2506 column += d->startTeXParParams(bparams, os,
2510 runparams.wasDisplayMath = runparams.inDisplayMath;
2511 runparams.inDisplayMath = false;
2512 bool deleted_display_math = false;
2513 Change const & change = runparams.inDeletedInset
2514 ? runparams.changeOfDeletedInset : lookupChange(i);
2516 char_type const c = d->text_[i];
2518 // Check whether a display math inset follows
2520 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2522 runparams.ctObject = getInset(i)->CtObject(runparams);
2524 InsetMath const * im = getInset(i)->asInsetMath();
2525 if (im && im->asHullInset()
2526 && im->asHullInset()->outerDisplay()) {
2527 runparams.inDisplayMath = true;
2528 // runparams.inDeletedInset will be set by
2529 // latexInset later, but we need this info
2530 // before it is called. On the other hand, we
2531 // cannot set it here because it is a counter.
2532 deleted_display_math = isDeleted(i);
2534 if (bparams.output_changes && deleted_display_math
2535 && runningChange == change
2536 && change.type == Change::DELETED
2537 && !os.afterParbreak()) {
2538 // A display math in the same paragraph follows.
2539 // We have to close and then reopen \lyxdeleted,
2540 // otherwise the math will be shifted up.
2541 OutputParams rp = runparams;
2543 bool needPar = false;
2544 column += running_font.latexWriteEndChanges(
2545 os, bparams, rp, basefont,
2546 basefont, needPar, multipar_inset);
2549 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2550 : getLayoutFont(bparams, real_outerfont);
2551 running_font = basefont;
2552 column += Changes::latexMarkChange(os, bparams,
2553 Change(Change::INSERTED), change, rp);
2557 if (bparams.output_changes && runningChange != change) {
2558 if (!alien_script.empty()) {
2561 alien_script.clear();
2564 bool needPar = false;
2565 column += running_font.latexWriteEndChanges(
2566 os, bparams, runparams,
2567 basefont, basefont, needPar,
2571 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2572 : getLayoutFont(bparams, real_outerfont);
2573 running_font = basefont;
2574 column += Changes::latexMarkChange(os, bparams, runningChange,
2576 runningChange = change;
2579 // do not output text which is marked deleted
2580 // if change tracking output is disabled
2581 if (!bparams.output_changes && change.deleted()) {
2587 // Fully instantiated font
2588 Font const current_font = getFont(bparams, i, outerfont);
2590 Font const prev_font = (i > 0) ?
2591 getFont(bparams, i - 1, outerfont)
2594 Font const last_font = running_font;
2595 bool const in_ct_deletion = (bparams.output_changes
2596 && runningChange == change
2597 && change.type == Change::DELETED
2598 && !os.afterParbreak());
2600 (c == META_INSET && getInset(i) && getInset(i)->allowMultiPar());
2602 // Do we need to close the previous font?
2604 ((current_font != running_font
2605 || current_font.language() != running_font.language())
2607 && (current_font == prev_font
2608 || current_font.language() == prev_font.language()))))
2610 // ensure there is no open script-wrapper
2611 if (!alien_script.empty()) {
2614 alien_script.clear();
2616 bool needPar = false;
2617 if (in_ct_deletion) {
2618 // We have to close and then reopen \lyxdeleted,
2619 // as strikeout needs to be on lowest level.
2623 column += running_font.latexWriteEndChanges(
2624 os, bparams, runparams, basefont,
2625 (i == body_pos-1) ? basefont : current_font,
2626 needPar, multipar_inset);
2627 if (in_ct_deletion) {
2628 // We have to close and then reopen \lyxdeleted,
2629 // as strikeout needs to be on lowest level.
2630 OutputParams rp = runparams;
2631 column += Changes::latexMarkChange(os, bparams,
2632 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2634 running_font = basefont;
2638 // if necessary, close language environment before opening CJK
2639 string const running_lang = running_font.language()->babel();
2640 string const lang_end_command = lyxrc.language_command_end;
2641 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2642 && !running_lang.empty()
2643 && running_lang == openLanguageName()
2644 && current_font.language()->encoding()->package() == Encoding::CJK) {
2645 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2646 os << from_ascii(end_tag);
2647 column += end_tag.length();
2651 // Switch file encoding if necessary (and allowed)
2652 if (!runparams.pass_thru && !style.pass_thru &&
2653 runparams.encoding->package() != Encoding::none &&
2654 current_font.language()->encoding()->package() != Encoding::none) {
2655 pair<bool, int> const enc_switch =
2656 switchEncoding(os.os(), bparams, runparams,
2657 *(current_font.language()->encoding()));
2658 if (enc_switch.first) {
2659 column += enc_switch.second;
2660 runparams.encoding = current_font.language()->encoding();
2664 // A display math inset inside an ulem command will be output
2665 // as a box of width \linewidth, so we have to either disable
2666 // indentation if the inset starts a paragraph, or start a new
2667 // line to accommodate such box. This has to be done before
2668 // writing any font changing commands.
2669 if (runparams.inDisplayMath && !deleted_display_math
2670 && runparams.inulemcmd) {
2671 if (os.afterParbreak())
2677 // Do we need to change font?
2678 if ((current_font != running_font ||
2679 current_font.language() != running_font.language())
2680 && i != body_pos - 1)
2682 if (in_ct_deletion) {
2683 // We have to close and then reopen \lyxdeleted,
2684 // as strikeout needs to be on lowest level.
2685 bool needPar = false;
2686 OutputParams rp = runparams;
2687 column += running_font.latexWriteEndChanges(
2688 os, bparams, rp, basefont,
2689 basefont, needPar, multipar_inset);
2693 otexstringstream ots;
2694 if (!multipar_inset) {
2695 InsetText const * textinset = inInset().asInsetText();
2696 bool const cprotect = textinset
2697 ? textinset->hasCProtectContent(runparams.moving_arg)
2698 && !textinset->text().isMainText()
2700 column += current_font.latexWriteStartChanges(ots, bparams,
2701 runparams, basefont, last_font, false,
2704 // Check again for display math in ulem commands as a
2705 // font change may also occur just before a math inset.
2706 if (runparams.inDisplayMath && !deleted_display_math
2707 && runparams.inulemcmd) {
2708 if (os.afterParbreak())
2713 running_font = current_font;
2715 docstring fontchange = ots.str();
2717 // check whether the fontchange ends with a \\textcolor
2718 // modifier and the text starts with a space. If so we
2719 // need to add } in order to prevent \\textcolor from gobbling
2720 // the space (bug 4473).
2721 docstring const last_modifier = rsplit(fontchange, '\\');
2722 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2723 os << from_ascii("{}");
2724 else if (ots.terminateCommand())
2726 if (in_ct_deletion) {
2727 // We have to close and then reopen \lyxdeleted,
2728 // as strikeout needs to be on lowest level.
2729 OutputParams rp = runparams;
2730 column += Changes::latexMarkChange(os, bparams,
2731 Change(Change::UNCHANGED), change, rp);
2735 // FIXME: think about end_pos implementation...
2736 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2737 // FIXME: integrate this case in latexSpecialChar
2738 // Do not print the separation of the optional argument
2739 // if style.pass_thru is false. This works because
2740 // latexSpecialChar ignores spaces if
2741 // style.pass_thru is false.
2742 if (i != body_pos - 1) {
2743 if (d->simpleTeXBlanks(bparams, runparams, os,
2744 i, column, current_font, style)) {
2745 // A surrogate pair was output. We
2746 // must not call latexSpecialChar
2747 // in this iteration, since it would output
2748 // the combining character again.
2755 OutputParams rp = runparams;
2756 rp.free_spacing = style.free_spacing;
2757 rp.local_font = ¤t_font;
2758 rp.intitle = style.intitle;
2760 // Two major modes: LaTeX or plain
2761 // Handle here those cases common to both modes
2762 // and then split to handle the two modes separately.
2763 if (c == META_INSET) {
2764 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2765 // Greyedout notes and, in general, all insets
2766 // with InsetLayout::isDisplay() == false,
2767 // are typeset inline with the text. So, we
2768 // can add a \par to the last paragraph of
2769 // such insets only if nothing else follows.
2770 bool incremented = false;
2771 Inset const * inset = getInset(i);
2772 InsetText const * textinset = inset
2773 ? inset->asInsetText()
2775 if (i + 1 == size() && textinset
2776 && !inset->getLayout().isDisplay()) {
2777 ParagraphList const & pars =
2778 textinset->text().paragraphs();
2779 pit_type const pit = pars.size() - 1;
2780 Font const lastfont =
2781 pit < 0 || pars[pit].empty()
2782 ? pars[pit].getLayoutFont(
2785 : pars[pit].getFont(bparams,
2786 pars[pit].size() - 1,
2788 if (lastfont.fontInfo().size() !=
2789 basefont.fontInfo().size()) {
2794 // We need to restore the main properties of
2795 // these fonts after allowMultiPar() insets
2796 FontInfo const running_font_info = running_font.fontInfo();
2797 FontInfo const basefont_info = basefont.fontInfo();
2798 d->latexInset(bparams, os, rp, running_font,
2799 basefont, real_outerfont, open_font,
2800 runningChange, style, i, column);
2801 if (multipar_inset) {
2802 running_font.setProperties(running_font_info);
2803 basefont.setProperties(basefont_info);
2808 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2809 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2810 // Close \lyx*deleted and force its
2811 // reopening (if needed)
2814 runningChange = Change(Change::UNCHANGED);
2815 runparams.ctObject = OutputParams::CT_NORMAL;
2818 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2819 if (!bparams.useNonTeXFonts)
2820 script = Encodings::isKnownScriptChar(c);
2821 if (script != alien_script) {
2822 if (!alien_script.empty()) {
2824 alien_script.clear();
2826 string fontenc = running_font.language()->fontenc(bparams);
2828 && !Encodings::fontencSupportsScript(fontenc, script)) {
2829 column += script.length() + 2;
2830 os << "\\" << script << "{";
2831 alien_script = script;
2835 d->latexSpecialChar(os, bparams, rp, running_font,
2836 alien_script, style, i, end_pos, column);
2837 } catch (EncodingException & e) {
2838 if (runparams.dryrun) {
2839 os << "<" << _("LyX Warning: ")
2840 << _("uncodable character") << " '";
2844 // add location information and throw again.
2852 // Set the encoding to that returned from latexSpecialChar (see
2853 // comment for encoding member in OutputParams.h)
2854 runparams.encoding = rp.encoding;
2856 // Also carry on the info on a closed ulem command for insets
2857 // such as Note that do not produce any output, so that no
2858 // command is ever executed but its opening was recorded.
2859 runparams.inulemcmd = rp.inulemcmd;
2861 // These need to be passed upstream as well
2862 runparams.need_maketitle = rp.need_maketitle;
2863 runparams.have_maketitle = rp.have_maketitle;
2865 // And finally, pass the post_macros upstream
2866 runparams.post_macro = rp.post_macro;
2869 // Close wrapper for alien script
2870 if (!alien_script.empty()) {
2872 alien_script.clear();
2875 // If we have an open font definition, we have to close it
2877 // Make sure that \\par is done with the font of the last
2878 // character if this has another size as the default.
2879 // This is necessary because LaTeX (and LyX on the screen)
2880 // calculates the space between the baselines according
2881 // to this font. (Matthias)
2883 // We must not change the font for the last paragraph
2884 // of non-multipar insets, tabular cells or commands,
2885 // since this produces unwanted whitespace.
2887 Font const font = empty()
2888 ? getLayoutFont(bparams, real_outerfont)
2889 : getFont(bparams, size() - 1, real_outerfont);
2891 InsetText const * textinset = inInset().asInsetText();
2893 bool const maintext = textinset
2894 ? textinset->text().isMainText()
2897 size_t const numpars = textinset
2898 ? textinset->text().paragraphs().size()
2901 bool needPar = false;
2903 if (style.resfont.size() != font.fontInfo().size()
2904 && (!runparams.isLastPar || maintext
2905 || (numpars > 1 && d->ownerCode() != CELL_CODE
2906 && (inInset().getLayout().isDisplay()
2908 && !style.isCommand()) {
2911 #ifdef FIXED_LANGUAGE_END_DETECTION
2913 running_font.latexWriteEndChanges(os, bparams,
2914 runparams, basefont,
2915 next_->getFont(bparams, 0, outerfont),
2918 running_font.latexWriteEndChanges(os, bparams,
2919 runparams, basefont, basefont, needPar);
2922 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2923 //FIXME: there as we start another \selectlanguage with the next paragraph if
2924 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2925 running_font.latexWriteEndChanges(os, bparams, runparams,
2926 basefont, basefont, needPar, multipar_inset);
2929 // The \par could not be inserted at the same nesting
2930 // level of the font size change, so do it now.
2931 os << "{\\" << font.latexSize() << "\\par}";
2935 column += Changes::latexMarkChange(os, bparams, runningChange,
2936 Change(Change::UNCHANGED), runparams);
2938 // Needed if there is an optional argument but no contents.
2939 if (body_pos > 0 && body_pos == size()) {
2943 if (!style.rightdelim().empty()) {
2944 os << style.rightdelim();
2945 column += style.rightdelim().size();
2948 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2949 && runparams.encoding != prev_encoding) {
2950 runparams.encoding = prev_encoding;
2951 os << setEncoding(prev_encoding->iconvName());
2954 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2958 bool Paragraph::emptyTag() const
2960 for (pos_type i = 0; i < size(); ++i) {
2961 if (Inset const * inset = getInset(i)) {
2962 InsetCode lyx_code = inset->lyxCode();
2963 // FIXME testing like that is wrong. What is
2965 if (lyx_code != TOC_CODE &&
2966 lyx_code != INCLUDE_CODE &&
2967 lyx_code != GRAPHICS_CODE &&
2968 lyx_code != ERT_CODE &&
2969 lyx_code != LISTINGS_CODE &&
2970 lyx_code != FLOAT_CODE &&
2971 lyx_code != TABULAR_CODE) {
2975 char_type c = d->text_[i];
2976 if (c != ' ' && c != '\t')
2984 string Paragraph::getID(Buffer const &, OutputParams const &)
2987 for (pos_type i = 0; i < size(); ++i) {
2988 if (Inset const * inset = getInset(i)) {
2989 InsetCode lyx_code = inset->lyxCode();
2990 if (lyx_code == LABEL_CODE) {
2991 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2992 docstring const & id = il->getParam("name");
2993 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3001 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3004 for (i = 0; i < size(); ++i) {
3005 if (Inset const * inset = getInset(i)) {
3006 inset->docbook(xs, runparams);
3008 char_type c = d->text_[i];
3018 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3022 for (i = 0; i < size(); ++i) {
3023 if (Inset const * inset = getInset(i)) {
3024 inset->xhtml(xs, runparams);
3026 char_type c = d->text_[i];
3036 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3039 pos_type size = text_.size();
3040 for (pos_type i = initial; i < size; ++i) {
3041 Font font = owner_->getFont(buf.params(), i, outerfont);
3042 if (text_[i] == META_INSET)
3044 if (i != initial && font != font_old)
3055 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3056 vector<xml::EndFontTag> & tagsToClose,
3057 bool & flag, FontState curstate, xml::FontTypes type)
3059 if (curstate == FONT_ON) {
3060 tagsToOpen.push_back(docbookStartFontTag(type));
3063 tagsToClose.push_back(docbookEndFontTag(type));
3068 class OptionalFontType {
3073 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3074 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3077 OptionalFontType fontShapeToXml(FontShape fs)
3081 return {xml::FT_ITALIC};
3083 return {xml::FT_SLANTED};
3084 case SMALLCAPS_SHAPE:
3085 return {xml::FT_SMALLCAPS};
3090 // the other tags are for internal use
3096 OptionalFontType fontFamilyToXml(FontFamily fm)
3100 return {xml::FT_ROMAN};
3102 return {xml::FT_SANS};
3103 case TYPEWRITER_FAMILY:
3104 return {xml::FT_TYPE};
3105 case INHERIT_FAMILY:
3108 // the other tags are for internal use
3114 OptionalFontType fontSizeToXml(FontSize fs)
3118 return {xml::FT_SIZE_TINY};
3120 return {xml::FT_SIZE_SCRIPT};
3122 return {xml::FT_SIZE_FOOTNOTE};
3124 return {xml::FT_SIZE_SMALL};
3126 return {xml::FT_SIZE_LARGE};
3128 return {xml::FT_SIZE_LARGER};
3130 return {xml::FT_SIZE_LARGEST};
3132 return {xml::FT_SIZE_HUGE};
3134 return {xml::FT_SIZE_HUGER};
3136 return {xml::FT_SIZE_INCREASE};
3138 return {xml::FT_SIZE_DECREASE};
3143 // the other tags are for internal use
3149 struct DocBookFontState
3151 FontShape curr_fs = INHERIT_SHAPE;
3152 FontFamily curr_fam = INHERIT_FAMILY;
3153 FontSize curr_size = INHERIT_SIZE;
3155 // track whether we have opened these tags
3156 bool emph_flag = false;
3157 bool bold_flag = false;
3158 bool noun_flag = false;
3159 bool ubar_flag = false;
3160 bool dbar_flag = false;
3161 bool sout_flag = false;
3162 bool xout_flag = false;
3163 bool wave_flag = false;
3165 bool shap_flag = false;
3167 bool faml_flag = false;
3169 bool size_flag = false;
3172 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3174 std::string const & default_family,
3175 DocBookFontState & fs)
3177 vector<xml::FontTag> tagsToOpen;
3178 vector<xml::EndFontTag> tagsToClose;
3181 FontState curstate = font.fontInfo().emph();
3182 if (font_old.emph() != curstate)
3183 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3186 curstate = font.fontInfo().noun();
3187 if (font_old.noun() != curstate)
3188 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3191 curstate = font.fontInfo().underbar();
3192 if (font_old.underbar() != curstate)
3193 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3196 curstate = font.fontInfo().strikeout();
3197 if (font_old.strikeout() != curstate)
3198 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3201 curstate = font.fontInfo().xout();
3202 if (font_old.xout() != curstate)
3203 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3206 curstate = font.fontInfo().uuline();
3207 if (font_old.uuline() != curstate)
3208 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3211 curstate = font.fontInfo().uwave();
3212 if (font_old.uwave() != curstate)
3213 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3216 // a little hackish, but allows us to reuse what we have.
3217 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3218 if (font_old.series() != font.fontInfo().series())
3219 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3222 fs.curr_fs = font.fontInfo().shape();
3223 FontShape old_fs = font_old.shape();
3224 if (old_fs != fs.curr_fs) {
3226 OptionalFontType tag = fontShapeToXml(old_fs);
3228 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3229 fs.shap_flag = false;
3232 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3234 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3238 fs.curr_fam = font.fontInfo().family();
3239 FontFamily old_fam = font_old.family();
3240 if (old_fam != fs.curr_fam) {
3242 OptionalFontType tag = fontFamilyToXml(old_fam);
3244 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3245 fs.faml_flag = false;
3247 switch (fs.curr_fam) {
3249 // we will treat a "default" font family as roman, since we have
3250 // no other idea what to do.
3251 if (default_family != "rmdefault" && default_family != "default") {
3252 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3253 fs.faml_flag = true;
3257 if (default_family != "sfdefault") {
3258 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3259 fs.faml_flag = true;
3262 case TYPEWRITER_FAMILY:
3263 if (default_family != "ttdefault") {
3264 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3265 fs.faml_flag = true;
3268 case INHERIT_FAMILY:
3271 // the other tags are for internal use
3278 fs.curr_size = font.fontInfo().size();
3279 FontSize old_size = font_old.size();
3280 if (old_size != fs.curr_size) {
3282 OptionalFontType tag = fontSizeToXml(old_size);
3284 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3285 fs.size_flag = false;
3288 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3289 if (tag.has_value) {
3290 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3291 fs.size_flag = true;
3295 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3298 } // anonymous namespace
3301 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3303 OutputParams const & runparams,
3304 Font const & outerfont,
3305 bool start_paragraph, bool close_paragraph,
3306 pos_type initial) const
3308 // track whether we have opened these tags
3309 DocBookFontState fs;
3311 if (start_paragraph)
3312 xs.startDivision(allowEmpty());
3314 Layout const & style = *d->layout_;
3316 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3318 string const default_family =
3319 buf.masterBuffer()->params().fonts_default_family;
3321 vector<xml::FontTag> tagsToOpen;
3322 vector<xml::EndFontTag> tagsToClose;
3324 // parsing main loop
3325 for (pos_type i = initial; i < size(); ++i) {
3326 // let's not show deleted material in the output
3330 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3332 // Determine which tags should be opened or closed.
3333 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs);
3336 // Other such tags? What about the other text ranges?
3338 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3339 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3340 for (; cit != cen; ++cit)
3343 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3344 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3345 for (; sit != sen; ++sit)
3348 tagsToClose.clear();
3351 if (Inset const * inset = getInset(i)) {
3352 if (!runparams.for_toc || inset->isInToc()) {
3353 OutputParams np = runparams;
3354 np.local_font = &font;
3355 // If the paragraph has size 1, then we are in the "special
3356 // case" where we do not output the containing paragraph info.
3357 // This "special case" is defined in more details in output_docbook.cpp, makeParagraphs. The results
3358 // of that brittle logic is passed to this function through open_par.
3359 if (!inset->getLayout().htmlisblock() && size() != 1) // TODO: htmlisblock here too!
3360 np.docbook_in_par = true;
3361 inset->docbook(xs, np);
3364 char_type c = getUChar(buf.masterBuffer()->params(), runparams, i);
3367 font_old = font.fontInfo();
3370 // FIXME, this code is just imported from XHTML
3371 // I'm worried about what happens if a branch, say, is itself
3372 // wrapped in some font stuff. I think that will not work.
3374 if (runparams.docbook_in_listing)
3376 if (close_paragraph)
3383 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3384 vector<xml::EndFontTag> & tagsToClose,
3385 bool & flag, FontState curstate, xml::FontTypes type)
3387 if (curstate == FONT_ON) {
3388 tagsToOpen.push_back(xhtmlStartFontTag(type));
3391 tagsToClose.push_back(xhtmlEndFontTag(type));
3396 } // anonymous namespace
3399 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3401 OutputParams const & runparams,
3402 Font const & outerfont,
3403 bool start_paragraph, bool close_paragraph,
3404 pos_type initial) const
3408 // track whether we have opened these tags
3409 bool emph_flag = false;
3410 bool bold_flag = false;
3411 bool noun_flag = false;
3412 bool ubar_flag = false;
3413 bool dbar_flag = false;
3414 bool sout_flag = false;
3415 bool xout_flag = false;
3416 bool wave_flag = false;
3418 bool shap_flag = false;
3420 bool faml_flag = false;
3422 bool size_flag = false;
3424 Layout const & style = *d->layout_;
3426 if (start_paragraph)
3427 xs.startDivision(allowEmpty());
3430 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3432 FontShape curr_fs = INHERIT_SHAPE;
3433 FontFamily curr_fam = INHERIT_FAMILY;
3434 FontSize curr_size = INHERIT_SIZE;
3436 string const default_family =
3437 buf.masterBuffer()->params().fonts_default_family;
3439 vector<xml::FontTag> tagsToOpen;
3440 vector<xml::EndFontTag> tagsToClose;
3442 // parsing main loop
3443 for (pos_type i = initial; i < size(); ++i) {
3444 // let's not show deleted material in the output
3448 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3451 FontState curstate = font.fontInfo().emph();
3452 if (font_old.emph() != curstate)
3453 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3456 curstate = font.fontInfo().noun();
3457 if (font_old.noun() != curstate)
3458 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3461 curstate = font.fontInfo().underbar();
3462 if (font_old.underbar() != curstate)
3463 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3466 curstate = font.fontInfo().strikeout();
3467 if (font_old.strikeout() != curstate)
3468 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3471 curstate = font.fontInfo().xout();
3472 if (font_old.xout() != curstate)
3473 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3476 curstate = font.fontInfo().uuline();
3477 if (font_old.uuline() != curstate)
3478 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3481 curstate = font.fontInfo().uwave();
3482 if (font_old.uwave() != curstate)
3483 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3486 // a little hackish, but allows us to reuse what we have.
3487 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3488 if (font_old.series() != font.fontInfo().series())
3489 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3492 curr_fs = font.fontInfo().shape();
3493 FontShape old_fs = font_old.shape();
3494 if (old_fs != curr_fs) {
3498 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3501 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3503 case SMALLCAPS_SHAPE:
3504 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3510 // the other tags are for internal use
3518 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3522 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3525 case SMALLCAPS_SHAPE:
3526 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3533 // the other tags are for internal use
3540 curr_fam = font.fontInfo().family();
3541 FontFamily old_fam = font_old.family();
3542 if (old_fam != curr_fam) {
3546 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3549 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3551 case TYPEWRITER_FAMILY:
3552 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3554 case INHERIT_FAMILY:
3557 // the other tags are for internal use
3565 // we will treat a "default" font family as roman, since we have
3566 // no other idea what to do.
3567 if (default_family != "rmdefault" && default_family != "default") {
3568 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3573 if (default_family != "sfdefault") {
3574 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3578 case TYPEWRITER_FAMILY:
3579 if (default_family != "ttdefault") {
3580 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3584 case INHERIT_FAMILY:
3587 // the other tags are for internal use
3594 curr_size = font.fontInfo().size();
3595 FontSize old_size = font_old.size();
3596 if (old_size != curr_size) {
3600 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3603 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3606 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3609 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3612 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3615 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3618 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3621 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3624 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3627 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3630 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3636 // the other tags are for internal use
3642 switch (curr_size) {
3644 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3648 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3652 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3656 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3660 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3664 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3668 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3672 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3676 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3680 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3684 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3691 // the other tags are for internal use
3698 // Other such tags? What about the other text ranges?
3700 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3701 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3702 for (; cit != cen; ++cit)
3705 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3706 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3707 for (; sit != sen; ++sit)
3710 tagsToClose.clear();
3713 Inset const * inset = getInset(i);
3715 if (!runparams.for_toc || inset->isInToc()) {
3716 OutputParams np = runparams;
3717 np.local_font = &font;
3718 // If the paragraph has size 1, then we are in the "special
3719 // case" where we do not output the containing paragraph info
3720 if (!inset->getLayout().htmlisblock() && size() != 1)
3721 np.html_in_par = true;
3722 retval += inset->xhtml(xs, np);
3725 char_type c = getUChar(buf.masterBuffer()->params(),
3727 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3728 xs << XMLStream::ESCAPE_NONE << " ";
3732 font_old = font.fontInfo();
3736 // I'm worried about what happens if a branch, say, is itself
3737 // wrapped in some font stuff. I think that will not work.
3739 if (close_paragraph)
3746 bool Paragraph::isHfill(pos_type pos) const
3748 Inset const * inset = getInset(pos);
3749 return inset && inset->isHfill();
3753 bool Paragraph::isNewline(pos_type pos) const
3755 // U+2028 LINE SEPARATOR
3756 // U+2029 PARAGRAPH SEPARATOR
3757 char_type const c = d->text_[pos];
3758 if (c == 0x2028 || c == 0x2029)
3760 Inset const * inset = getInset(pos);
3761 return inset && inset->lyxCode() == NEWLINE_CODE;
3765 bool Paragraph::isEnvSeparator(pos_type pos) const
3767 Inset const * inset = getInset(pos);
3768 return inset && inset->lyxCode() == SEPARATOR_CODE;
3772 bool Paragraph::isLineSeparator(pos_type pos) const
3774 char_type const c = d->text_[pos];
3775 if (isLineSeparatorChar(c))
3777 Inset const * inset = getInset(pos);
3778 return inset && inset->isLineSeparator();
3782 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3786 if (ignore_deleted && isDeleted(pos))
3788 if (Inset const * inset = getInset(pos))
3789 return !inset->isLetter();
3790 // if we have a hard hyphen (no en- or emdash) or apostrophe
3791 // we pass this to the spell checker
3792 // FIXME: this method is subject to change, visit
3793 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3794 // to get an impression how complex this is.
3795 if (isHardHyphenOrApostrophe(pos))
3797 char_type const c = d->text_[pos];
3798 // We want to pass the escape chars to the spellchecker
3799 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3800 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3804 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3806 pos_type const psize = size();
3809 char_type const c = d->text_[pos];
3810 if (c != '-' && c != '\'')
3812 int nextpos = pos + 1;
3813 int prevpos = pos > 0 ? pos - 1 : 0;
3814 if ((nextpos == psize || isSpace(nextpos))
3815 && (pos == 0 || isSpace(prevpos)))
3821 bool Paragraph::needsCProtection(bool const fragile) const
3823 // first check the layout of the paragraph, but only in insets
3824 InsetText const * textinset = inInset().asInsetText();
3825 bool const maintext = textinset
3826 ? textinset->text().isMainText()
3829 if (!maintext && layout().needcprotect) {
3830 // Environments need cprotection regardless the content
3831 if (layout().latextype == LATEX_ENVIRONMENT)
3834 // Commands need cprotection if they contain specific chars
3835 int const nchars_escape = 9;
3836 static char_type const chars_escape[nchars_escape] = {
3837 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3839 docstring const pars = asString();
3840 for (int k = 0; k < nchars_escape; k++) {
3841 if (contains(pars, chars_escape[k]))
3846 // now check whether we have insets that need cprotection
3847 pos_type size = pos_type(d->text_.size());
3848 for (pos_type i = 0; i < size; ++i) {
3851 Inset const * ins = getInset(i);
3852 if (ins->needsCProtection(maintext, fragile))
3854 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3855 // Environments need cprotection regardless the content
3857 // Now check math environments
3858 InsetMath const * im = getInset(i)->asInsetMath();
3859 if (!im || im->cell(0).empty())
3861 switch(im->cell(0)[0]->lyxCode()) {
3862 case MATH_AMSARRAY_CODE:
3863 case MATH_SUBSTACK_CODE:
3865 case MATH_XYMATRIX_CODE:
3866 // these need cprotection
3877 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3879 return d->speller_state_.getRange(pos);
3883 bool Paragraph::isChar(pos_type pos) const
3885 if (Inset const * inset = getInset(pos))
3886 return inset->isChar();
3887 char_type const c = d->text_[pos];
3888 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3892 bool Paragraph::isSpace(pos_type pos) const
3894 if (Inset const * inset = getInset(pos))
3895 return inset->isSpace();
3896 char_type const c = d->text_[pos];
3897 return lyx::isSpace(c);
3902 Paragraph::getParLanguage(BufferParams const & bparams) const
3905 return getFirstFontSettings(bparams).language();
3906 // FIXME: we should check the prev par as well (Lgb)
3907 return bparams.language;
3911 bool Paragraph::isRTL(BufferParams const & bparams) const
3913 return getParLanguage(bparams)->rightToLeft()
3914 && !inInset().getLayout().forceLTR();
3918 void Paragraph::changeLanguage(BufferParams const & bparams,
3919 Language const * from, Language const * to)
3921 // change language including dummy font change at the end
3922 for (pos_type i = 0; i <= size(); ++i) {
3923 Font font = getFontSettings(bparams, i);
3924 if (font.language() == from) {
3925 font.setLanguage(to);
3927 d->requestSpellCheck(i);
3933 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3935 Language const * doc_language = bparams.language;
3936 for (auto const & f : d->fontlist_)
3937 if (f.font().language() != ignore_language &&
3938 f.font().language() != latex_language &&
3939 f.font().language() != doc_language)
3945 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3947 for (auto const & f : d->fontlist_) {
3948 Language const * lang = f.font().language();
3949 if (lang != ignore_language &&
3950 lang != latex_language)
3956 docstring Paragraph::asString(int options) const
3958 return asString(0, size(), options);
3962 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3964 odocstringstream os;
3967 && options & AS_STR_LABEL
3968 && !d->params_.labelString().empty())
3969 os << d->params_.labelString() << ' ';
3971 for (pos_type i = beg; i < end; ++i) {
3972 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3974 char_type const c = d->text_[i];
3975 if (isPrintable(c) || c == '\t'
3976 || (c == '\n' && (options & AS_STR_NEWLINES)))
3978 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3979 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3980 LASSERT(runparams != nullptr, return docstring());
3981 getInset(i)->plaintext(os, *runparams);
3983 getInset(i)->toString(os);
3992 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3993 bool const shorten, bool const label) const
3995 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3996 if (label && !labelString().empty())
3997 os += labelString() + ' ';
3998 if (!layout().isTocCaption())
4000 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4003 char_type const c = d->text_[i];
4006 else if (c == META_INSET)
4007 getInset(i)->forOutliner(os, tmplen, false);
4010 Text::shortenForOutliner(os, maxlen);
4014 void Paragraph::setInsetOwner(Inset const * inset)
4016 d->inset_owner_ = inset;
4020 int Paragraph::id() const
4026 void Paragraph::setId(int id)
4032 Layout const & Paragraph::layout() const
4038 void Paragraph::setLayout(Layout const & layout)
4040 d->layout_ = &layout;
4044 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4046 setLayout(tc.defaultLayout());
4050 void Paragraph::setPlainLayout(DocumentClass const & tc)
4052 setLayout(tc.plainLayout());
4056 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
4058 if (usePlainLayout())
4059 setPlainLayout(tclass);
4061 setDefaultLayout(tclass);
4065 Inset const & Paragraph::inInset() const
4067 LBUFERR(d->inset_owner_);
4068 return *d->inset_owner_;
4072 ParagraphParameters & Paragraph::params()
4078 ParagraphParameters const & Paragraph::params() const
4084 bool Paragraph::isFreeSpacing() const
4086 if (d->layout_->free_spacing)
4088 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4092 bool Paragraph::allowEmpty() const
4094 if (d->layout_->keepempty)
4096 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4100 bool Paragraph::brokenBiblio() const
4102 // There is a problem if there is no bibitem at position 0 in
4103 // paragraphs that need one, if there is another bibitem in the
4104 // paragraph or if this paragraph is not supposed to have
4105 // a bibitem inset at all.
4106 return ((d->layout_->labeltype == LABEL_BIBLIO
4107 && (d->insetlist_.find(BIBITEM_CODE) != 0
4108 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4109 || (d->layout_->labeltype != LABEL_BIBLIO
4110 && d->insetlist_.find(BIBITEM_CODE) != -1));
4114 int Paragraph::fixBiblio(Buffer const & buffer)
4116 // FIXME: when there was already an inset at 0, the return value is 1,
4117 // which does not tell whether another inset has been removed; the
4118 // cursor cannot be correctly updated.
4120 bool const track_changes = buffer.params().track_changes;
4121 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4123 // The case where paragraph is not BIBLIO
4124 if (d->layout_->labeltype != LABEL_BIBLIO) {
4125 if (bibitem_pos == -1)
4126 // No InsetBibitem => OK
4128 // There is an InsetBibitem: remove it!
4129 d->insetlist_.release(bibitem_pos);
4130 eraseChar(bibitem_pos, track_changes);
4131 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4134 bool const hasbibitem0 = bibitem_pos == 0;
4136 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4137 // There was an InsetBibitem at pos 0,
4138 // and no other one => OK
4139 if (bibitem_pos == -1)
4141 // there is a bibitem at the 0 position, but since
4142 // there is a second one, we copy the second on the
4143 // first. We're assuming there are at most two of
4144 // these, which there should be.
4145 // FIXME: why does it make sense to do that rather
4146 // than keep the first? (JMarc)
4147 Inset * inset = releaseInset(bibitem_pos);
4148 d->insetlist_.begin()->inset = inset;
4149 // This needs to be done to update the counter (#8499)
4150 buffer.updateBuffer();
4151 return -bibitem_pos;
4154 // We need to create an inset at the beginning
4155 Inset * inset = nullptr;
4156 if (bibitem_pos > 0) {
4157 // there was one somewhere in the paragraph, let's move it
4158 inset = d->insetlist_.release(bibitem_pos);
4159 eraseChar(bibitem_pos, track_changes);
4162 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4163 InsetCommandParams(BIBITEM_CODE));
4165 Font font(inherit_font, buffer.params().language);
4166 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4167 : Change::UNCHANGED));
4169 // This is needed to get the counters right
4170 buffer.updateBuffer();
4175 void Paragraph::checkAuthors(AuthorList const & authorList)
4177 d->changes_.checkAuthors(authorList);
4181 bool Paragraph::isChanged(pos_type pos) const
4183 return lookupChange(pos).changed();
4187 bool Paragraph::isInserted(pos_type pos) const
4189 return lookupChange(pos).inserted();
4193 bool Paragraph::isDeleted(pos_type pos) const
4195 return lookupChange(pos).deleted();
4199 InsetList const & Paragraph::insetList() const
4201 return d->insetlist_;
4205 void Paragraph::setInsetBuffers(Buffer & b)
4207 d->insetlist_.setBuffer(b);
4211 void Paragraph::resetBuffer()
4213 d->insetlist_.resetBuffer();
4217 Inset * Paragraph::releaseInset(pos_type pos)
4219 Inset * inset = d->insetlist_.release(pos);
4220 /// does not honour change tracking!
4221 eraseChar(pos, false);
4226 Inset * Paragraph::getInset(pos_type pos)
4228 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4229 ? d->insetlist_.get(pos) : nullptr;
4233 Inset const * Paragraph::getInset(pos_type pos) const
4235 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4236 ? d->insetlist_.get(pos) : nullptr;
4240 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4241 pos_type & right, TextCase action)
4243 // process sequences of modified characters; in change
4244 // tracking mode, this approach results in much better
4245 // usability than changing case on a char-by-char basis
4246 // We also need to track the current font, since font
4247 // changes within sequences can occur.
4248 vector<pair<char_type, Font> > changes;
4250 bool const trackChanges = bparams.track_changes;
4252 bool capitalize = true;
4254 for (; pos < right; ++pos) {
4255 char_type oldChar = d->text_[pos];
4256 char_type newChar = oldChar;
4258 // ignore insets and don't play with deleted text!
4259 if (oldChar != META_INSET && !isDeleted(pos)) {
4261 case text_lowercase:
4262 newChar = lowercase(oldChar);
4264 case text_capitalization:
4266 newChar = uppercase(oldChar);
4270 case text_uppercase:
4271 newChar = uppercase(oldChar);
4276 if (isWordSeparator(pos) || isDeleted(pos)) {
4277 // permit capitalization again
4281 if (oldChar != newChar) {
4282 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4283 if (pos != right - 1)
4285 // step behind the changing area
4289 int erasePos = pos - changes.size();
4290 for (size_t i = 0; i < changes.size(); i++) {
4291 insertChar(pos, changes[i].first,
4294 if (!eraseChar(erasePos, trackChanges)) {
4297 ++right; // expand selection
4305 int Paragraph::find(docstring const & str, bool cs, bool mw,
4306 pos_type start_pos, bool del) const
4308 pos_type pos = start_pos;
4309 int const strsize = str.length();
4311 pos_type const parsize = d->text_.size();
4312 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4313 // Ignore "invisible" letters such as ligature breaks
4314 // and hyphenation chars while searching
4315 while (pos < parsize - 1 && isInset(pos)) {
4316 odocstringstream os;
4317 getInset(pos)->toString(os);
4318 if (!getInset(pos)->isLetter() || !os.str().empty())
4322 if (cs && str[i] != d->text_[pos])
4324 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4326 if (!del && isDeleted(pos))
4333 // if necessary, check whether string matches word
4335 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4338 && !isWordSeparator(pos))
4342 return pos - start_pos;
4346 char_type Paragraph::getChar(pos_type pos) const
4348 return d->text_[pos];
4352 pos_type Paragraph::size() const
4354 return d->text_.size();
4358 bool Paragraph::empty() const
4360 return d->text_.empty();
4364 bool Paragraph::isInset(pos_type pos) const
4366 return d->text_[pos] == META_INSET;
4370 bool Paragraph::isSeparator(pos_type pos) const
4372 //FIXME: Are we sure this can be the only separator?
4373 return d->text_[pos] == ' ';
4377 void Paragraph::deregisterWords()
4379 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4380 Private::LangWordsMap::const_iterator ite = d->words_.end();
4381 for (; itl != ite; ++itl) {
4382 WordList & wl = theWordList(itl->first);
4383 Private::Words::const_iterator it = (itl->second).begin();
4384 Private::Words::const_iterator et = (itl->second).end();
4385 for (; it != et; ++it)
4392 void Paragraph::locateWord(pos_type & from, pos_type & to,
4393 word_location const loc, bool const ignore_deleted) const
4396 case WHOLE_WORD_STRICT:
4397 if (from == 0 || from == size()
4398 || isWordSeparator(from, ignore_deleted)
4399 || isWordSeparator(from - 1, ignore_deleted)) {
4406 // If we are already at the beginning of a word, do nothing
4407 if (!from || isWordSeparator(from - 1, ignore_deleted))
4412 // always move the cursor to the beginning of previous word
4413 while (from && !isWordSeparator(from - 1, ignore_deleted))
4417 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4420 // no need to move the 'from' cursor
4424 while (to < size() && !isWordSeparator(to, ignore_deleted))
4429 void Paragraph::collectWords()
4431 for (pos_type pos = 0; pos < size(); ++pos) {
4432 if (isWordSeparator(pos))
4434 pos_type from = pos;
4435 locateWord(from, pos, WHOLE_WORD);
4436 // Work around MSVC warning: The statement
4437 // if (pos < from + lyxrc.completion_minlength)
4438 // triggers a signed vs. unsigned warning.
4439 // I don't know why this happens, it could be a MSVC bug, or
4440 // related to LLP64 (windows) vs. LP64 (unix) programming
4441 // model, or the C++ standard might be ambigous in the section
4442 // defining the "usual arithmetic conversions". However, using
4443 // a temporary variable is safe and works on all compilers.
4444 pos_type const endpos = from + lyxrc.completion_minlength;
4447 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4448 if (cit == d->fontlist_.end())
4450 Language const * lang = cit->font().language();
4451 docstring const word = asString(from, pos, AS_STR_NONE);
4452 d->words_[lang->lang()].insert(word);
4457 void Paragraph::registerWords()
4459 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4460 Private::LangWordsMap::const_iterator ite = d->words_.end();
4461 for (; itl != ite; ++itl) {
4462 WordList & wl = theWordList(itl->first);
4463 Private::Words::const_iterator it = (itl->second).begin();
4464 Private::Words::const_iterator et = (itl->second).end();
4465 for (; it != et; ++it)
4471 void Paragraph::updateWords()
4479 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4481 SkipPositionsIterator begin = skips.begin();
4482 SkipPositions::iterator end = skips.end();
4483 if (pos > 0 && begin < end) {
4485 if (end->last == pos - 1) {
4490 skips.insert(end, FontSpan(pos, pos));
4494 Language * Paragraph::Private::locateSpellRange(
4495 pos_type & from, pos_type & to,
4496 SkipPositions & skips) const
4498 // skip leading white space
4499 while (from < to && owner_->isWordSeparator(from))
4501 // don't check empty range
4504 // get current language
4505 Language * lang = getSpellLanguage(from);
4506 pos_type last = from;
4507 bool samelang = true;
4508 bool sameinset = true;
4509 while (last < to && samelang && sameinset) {
4510 // hop to end of word
4511 while (last < to && !owner_->isWordSeparator(last)) {
4512 if (owner_->getInset(last)) {
4513 appendSkipPosition(skips, last);
4514 } else if (owner_->isDeleted(last)) {
4515 appendSkipPosition(skips, last);
4519 // hop to next word while checking for insets
4520 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4521 if (Inset const * inset = owner_->getInset(last))
4522 sameinset = inset->isChar() && inset->isLetter();
4523 if (sameinset && owner_->isDeleted(last)) {
4524 appendSkipPosition(skips, last);
4529 if (sameinset && last < to) {
4530 // now check for language change
4531 samelang = lang == getSpellLanguage(last);
4534 // if language change detected backstep is needed
4542 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4545 const_cast<Language *>(owner_->getFontSettings(
4546 inset_owner_->buffer().params(), from).language());
4547 if (lang == inset_owner_->buffer().params().language
4548 && !lyxrc.spellchecker_alt_lang.empty()) {
4550 string const lang_variety =
4551 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4552 lang->setCode(lang_code);
4553 lang->setVariety(lang_variety);
4559 void Paragraph::requestSpellCheck(pos_type pos)
4561 d->requestSpellCheck(pos);
4565 bool Paragraph::needsSpellCheck() const
4567 SpellChecker::ChangeNumber speller_change_number = 0;
4568 if (theSpellChecker())
4569 speller_change_number = theSpellChecker()->changeNumber();
4570 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4571 d->speller_state_.needsCompleteRefresh(speller_change_number);
4573 return d->needsSpellCheck();
4577 bool Paragraph::Private::ignoreWord(docstring const & word) const
4579 // Ignore words with digits
4580 // FIXME: make this customizable
4581 // (note that some checkers ignore words with digits by default)
4582 docstring::const_iterator cit = word.begin();
4583 docstring::const_iterator const end = word.end();
4584 for (; cit != end; ++cit) {
4585 if (isNumber((*cit)))
4592 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4593 WordLangTuple & wl, docstring_list & suggestions,
4594 bool do_suggestion, bool check_learned) const
4596 SpellChecker::Result result = SpellChecker::WORD_OK;
4597 SpellChecker * speller = theSpellChecker();
4601 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4604 locateWord(from, to, WHOLE_WORD, true);
4605 if (from == to || from >= size())
4608 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4609 Language * lang = d->getSpellLanguage(from);
4611 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4614 wl = WordLangTuple(word, lang);
4619 if (needsSpellCheck() || check_learned) {
4621 if (!d->ignoreWord(word)) {
4622 bool const trailing_dot = to < size() && d->text_[to] == '.';
4623 result = speller->check(wl);
4624 if (SpellChecker::misspelled(result) && trailing_dot) {
4625 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4626 result = speller->check(wl);
4627 if (!SpellChecker::misspelled(result)) {
4628 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4630 from << ".." << to << "]");
4632 // spell check with dot appended failed too
4633 // restore original word/lang value
4634 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4635 wl = WordLangTuple(word, lang);
4639 if (!SpellChecker::misspelled(result)) {
4640 // area up to the begin of the next word is not misspelled
4641 while (end < size() && isWordSeparator(end))
4644 d->setMisspelled(from, end, result);
4646 result = d->speller_state_.getState(from);
4650 suggestions.clear();
4652 if (SpellChecker::misspelled(result)) {
4653 LYXERR(Debug::GUI, "misspelled word: \"" <<
4655 from << ".." << to << "]");
4657 speller->suggest(wl, suggestions);
4663 void Paragraph::anonymize()
4665 // This is a very crude anonymization for now
4666 for (char_type & c : d->text_)
4667 if (isLetterChar(c) || isNumber(c))
4672 void Paragraph::Private::markMisspelledWords(
4673 pos_type const & first, pos_type const & last,
4674 SpellChecker::Result result,
4675 docstring const & word,
4676 SkipPositions const & skips)
4678 if (!SpellChecker::misspelled(result)) {
4679 setMisspelled(first, last, SpellChecker::WORD_OK);
4683 SpellChecker * speller = theSpellChecker();
4684 // locate and enumerate the error positions
4685 int nerrors = speller->numMisspelledWords();
4687 SkipPositionsIterator it = skips.begin();
4688 SkipPositionsIterator et = skips.end();
4689 for (int index = 0; index < nerrors; ++index) {
4692 speller->misspelledWord(index, wstart, wlen);
4693 /// should not happen if speller supports range checks
4694 if (!wlen) continue;
4695 docstring const misspelled = word.substr(wstart, wlen);
4696 wstart += first + numskipped;
4697 if (snext < wstart) {
4698 /// mark the range of correct spelling
4699 numskipped += countSkips(it, et, wstart);
4700 setMisspelled(snext,
4701 wstart - 1, SpellChecker::WORD_OK);
4703 snext = wstart + wlen;
4704 numskipped += countSkips(it, et, snext);
4705 /// mark the range of misspelling
4706 setMisspelled(wstart, snext, result);
4707 LYXERR(Debug::GUI, "misspelled word: \"" <<
4708 misspelled << "\" [" <<
4709 wstart << ".." << (snext-1) << "]");
4712 if (snext <= last) {
4713 /// mark the range of correct spelling at end
4714 setMisspelled(snext, last, SpellChecker::WORD_OK);
4719 void Paragraph::spellCheck() const
4721 SpellChecker * speller = theSpellChecker();
4722 if (!speller || empty() ||!needsSpellCheck())
4726 d->rangeOfSpellCheck(start, endpos);
4727 if (speller->canCheckParagraph()) {
4728 // loop until we leave the range
4729 for (pos_type first = start; first < endpos; ) {
4730 pos_type last = endpos;
4731 Private::SkipPositions skips;
4732 Language * lang = d->locateSpellRange(first, last, skips);
4733 if (first >= endpos)
4735 // start the spell checker on the unit of meaning
4736 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4737 WordLangTuple wl = WordLangTuple(word, lang);
4738 SpellChecker::Result result = word.size() ?
4739 speller->check(wl) : SpellChecker::WORD_OK;
4740 d->markMisspelledWords(first, last, result, word, skips);
4744 static docstring_list suggestions;
4745 pos_type to = endpos;
4746 while (start < endpos) {
4748 spellCheck(start, to, wl, suggestions, false);
4752 d->readySpellCheck();
4756 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4758 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4759 if (result || pos <= 0 || pos > size())
4761 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4762 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4767 string Paragraph::magicLabel() const
4770 ss << "magicparlabel-" << id();