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 Kimberly 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"
24 #include "BufferParams.h"
25 #include "BufferEncodings.h"
28 #include "InsetList.h"
30 #include "LaTeXFeatures.h"
35 #include "OutputParams.h"
36 #include "output_latex.h"
37 #include "output_xhtml.h"
38 #include "output_docbook.h"
39 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
42 #include "texstream.h"
45 #include "TextClass.h"
46 #include "WordLangTuple.h"
49 #include "frontends/alert.h"
51 #include "insets/InsetArgument.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/lstrings.h"
65 #include "support/textutils.h"
72 using namespace lyx::support;
74 // OSX clang and msvc < 2015 do not support C++11 thread_local
75 #if defined(__APPLE__)
76 #define THREAD_LOCAL_STATIC static __thread
77 #elif defined(_MSC_VER) && (_MSC_VER < 1900)
78 #define THREAD_LOCAL_STATIC static __declspec(thread)
80 #define THREAD_LOCAL_STATIC thread_local static
85 /////////////////////////////////////////////////////////////////////
89 /////////////////////////////////////////////////////////////////////
91 class SpellResultRange {
93 SpellResultRange(FontSpan range, SpellChecker::Result result)
94 : range_(range), result_(result)
97 FontSpan const & range() const { return range_; }
99 void range(FontSpan const & r) { range_ = r; }
101 SpellChecker::Result result() const { return result_; }
103 void result(SpellChecker::Result r) { result_ = r; }
105 bool contains(pos_type pos) const { return range_.contains(pos); }
107 bool covered(FontSpan const & r) const
109 // 1. first of new range inside current range or
110 // 2. last of new range inside current range or
111 // 3. first of current range inside new range or
112 // 4. last of current range inside new range
113 //FIXME: is this the same as !range_.intersect(r).empty() ?
114 return range_.contains(r.first) || range_.contains(r.last) ||
115 r.contains(range_.first) || r.contains(range_.last);
118 void shift(pos_type pos, int offset)
120 if (range_.first > pos) {
121 range_.first += offset;
122 range_.last += offset;
123 } else if (range_.last >= pos) {
124 range_.last += offset;
129 SpellChecker::Result result_ ;
133 /////////////////////////////////////////////////////////////////////
137 /////////////////////////////////////////////////////////////////////
139 class SpellCheckerState {
143 needs_refresh_ = true;
144 current_change_number_ = 0;
147 void setRange(FontSpan const & fp, SpellChecker::Result state)
150 RangesIterator et = ranges_.end();
151 RangesIterator it = ranges_.begin();
152 for (; it != et; ++it) {
153 if (!it->covered(fp))
154 result.push_back(SpellResultRange(it->range(), it->result()));
155 else if (state == SpellChecker::WORD_OK) {
156 // trim or split the current misspelled range
157 // store misspelled ranges only
158 FontSpan range = it->range();
159 if (fp.first > range.first) {
160 // misspelled area in front of WORD_OK
161 range.last = fp.first - 1;
162 result.push_back(SpellResultRange(range, it->result()));
165 if (fp.last < range.last) {
166 // misspelled area after WORD_OK range
167 range.first = fp.last + 1;
168 result.push_back(SpellResultRange(range, it->result()));
173 if (state != SpellChecker::WORD_OK)
174 ranges_.push_back(SpellResultRange(fp, state));
177 void increasePosAfterPos(pos_type pos)
179 correctRangesAfterPos(pos, 1);
183 void decreasePosAfterPos(pos_type pos)
185 correctRangesAfterPos(pos, -1);
189 void refreshLast(pos_type pos)
191 if (pos < refresh_.last)
195 SpellChecker::Result getState(pos_type pos) const
197 SpellChecker::Result result = SpellChecker::WORD_OK;
198 RangesIterator et = ranges_.end();
199 RangesIterator it = ranges_.begin();
200 for (; it != et; ++it) {
201 if(it->contains(pos)) {
208 FontSpan const & getRange(pos_type pos) const
210 /// empty span to indicate mismatch
211 static FontSpan empty_;
212 RangesIterator et = ranges_.end();
213 RangesIterator it = ranges_.begin();
214 for (; it != et; ++it) {
215 if(it->contains(pos)) {
222 bool needsRefresh() const
224 return needs_refresh_;
227 SpellChecker::ChangeNumber currentChangeNumber() const
229 return current_change_number_;
232 void refreshRange(pos_type & first, pos_type & last) const
234 first = refresh_.first;
235 last = refresh_.last;
238 void needsRefresh(pos_type pos)
240 if (needs_refresh_ && pos != -1) {
241 if (pos < refresh_.first)
242 refresh_.first = pos;
243 if (pos > refresh_.last)
245 } else if (pos != -1) {
246 // init request check for neighbour positions too
247 refresh_.first = pos > 0 ? pos - 1 : 0;
248 // no need for special end of paragraph check
249 refresh_.last = pos + 1;
251 needs_refresh_ = pos != -1;
254 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
256 needs_refresh_ = true;
259 current_change_number_ = change_number;
262 typedef vector<SpellResultRange> Ranges;
263 typedef Ranges::const_iterator RangesIterator;
265 /// the area of the paragraph with pending spell check
267 /// spell state cache version number
268 SpellChecker::ChangeNumber current_change_number_;
272 void correctRangesAfterPos(pos_type pos, int offset)
274 RangesIterator et = ranges_.end();
275 Ranges::iterator it = ranges_.begin();
276 for (; it != et; ++it) {
277 it->shift(pos, offset);
283 /////////////////////////////////////////////////////////////////////
285 // Paragraph::Private
287 /////////////////////////////////////////////////////////////////////
289 class Paragraph::Private
291 // Enforce our own "copy" constructor
292 Private(Private const &) = delete;
293 Private & operator=(Private const &) = delete;
294 // Unique ID generator
295 static int make_id();
298 Private(Paragraph * owner, Layout const & layout);
299 /// "Copy constructor"
300 Private(Private const &, Paragraph * owner);
301 /// Copy constructor from \p beg to \p end
302 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
305 void insertChar(pos_type pos, char_type c, Change const & change);
307 /// Output the surrogate pair formed by \p c and \p next to \p os.
308 /// \return the number of characters written.
309 int latexSurrogatePair(BufferParams const &, otexstream & os,
310 char_type c, char_type next,
311 OutputParams const &) const;
313 /// Output a space in appropriate formatting (or a surrogate pair
314 /// if the next character is a combining character).
315 /// \return whether a surrogate pair was output.
316 bool simpleTeXBlanks(BufferParams const &,
317 OutputParams const &,
320 unsigned int & column,
322 Layout const & style) const;
324 /// This could go to ParagraphParameters if we want to.
325 int startTeXParParams(BufferParams const &, otexstream &,
326 OutputParams const &) const;
328 /// This could go to ParagraphParameters if we want to.
329 bool endTeXParParams(BufferParams const &, otexstream &,
330 OutputParams const &) const;
333 void latexInset(BufferParams const &,
338 Font const & outerfont,
340 Change & running_change,
341 Layout const & style,
343 unsigned int & column,
344 bool const fontswitch_inset,
345 bool const closeLanguage,
346 bool const lang_switched_at_inset) const;
349 void latexSpecialChar(
351 BufferParams const & bparams,
352 OutputParams const & runparams,
353 Font const & running_font,
354 string & alien_script,
355 Layout const & style,
358 unsigned int & column) const;
365 unsigned int & column) const;
371 unsigned int & column) const;
377 unsigned int & column) const;
380 void validate(LaTeXFeatures & features) const;
382 /// Checks if the paragraph contains only text and no inset or font change.
383 bool onlyText(Buffer const & buf, Font const & outerfont,
384 pos_type initial) const;
386 /// a vector of speller skip positions
387 typedef vector<FontSpan> SkipPositions;
388 typedef SkipPositions::const_iterator SkipPositionsIterator;
390 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
392 Language * getSpellLanguage(pos_type const from) const;
394 Language * locateSpellRange(pos_type & from, pos_type & to,
395 SkipPositions & skips) const;
397 bool hasSpellerChange() const
399 SpellChecker::ChangeNumber speller_change_number = 0;
400 if (theSpellChecker())
401 speller_change_number = theSpellChecker()->changeNumber();
402 return speller_change_number > speller_state_.currentChangeNumber();
405 bool ignoreWord(docstring const & word) const;
407 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
409 pos_type textsize = owner_->size();
410 // check for sane arguments
411 if (to <= from || from >= textsize)
413 FontSpan fp = FontSpan(from, to - 1);
414 speller_state_.setRange(fp, state);
417 void requestSpellCheck(pos_type pos)
420 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
422 speller_state_.needsRefresh(pos);
425 void readySpellCheck()
427 speller_state_.needsRefresh(-1);
430 bool needsSpellCheck() const
432 return speller_state_.needsRefresh();
435 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
437 speller_state_.refreshRange(first, last);
439 last = owner_->size();
442 pos_type endpos = last;
443 owner_->locateWord(first, endpos, WHOLE_WORD, true);
446 owner_->locateWord(last, endpos, WHOLE_WORD, true);
451 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
455 while (it != et && it->first < start) {
456 long skip = it->last - it->first + 1;
464 void markMisspelledWords(Language const * lang,
465 pos_type const & first, pos_type const & last,
466 SpellChecker::Result result,
467 docstring const & word,
468 SkipPositions const & skips);
470 InsetCode ownerCode() const
472 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
475 /// Which Paragraph owns us?
479 Inset const * inset_owner_;
485 ParagraphParameters params_;
487 /// for recording and looking up changes
491 InsetList insetlist_;
494 pos_type begin_of_body_;
496 typedef docstring TextContainer;
500 typedef set<docstring> Words;
501 typedef map<string, Words> LangWordsMap;
505 Layout const * layout_;
507 SpellCheckerState speller_state_;
513 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
514 : owner_(owner), inset_owner_(nullptr), begin_of_body_(0), layout_(&layout), id_(-1)
521 int Paragraph::Private::make_id()
523 // The id is unique per session across buffers because it is used in
524 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
527 static int next_id(0);
532 Paragraph::Private::Private(Private const & p, Paragraph * owner)
533 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
534 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
535 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
536 layout_(p.layout_), id_(make_id())
538 requestSpellCheck(p.text_.size());
542 Paragraph::Private::Private(Private const & p, Paragraph * owner,
543 pos_type beg, pos_type end)
544 : owner_(owner), inset_owner_(p.inset_owner_),
545 params_(p.params_), changes_(p.changes_),
546 insetlist_(p.insetlist_, beg, end),
547 begin_of_body_(p.begin_of_body_), words_(p.words_),
548 layout_(p.layout_), id_(make_id())
550 if (beg >= pos_type(p.text_.size()))
552 text_ = p.text_.substr(beg, end - beg);
554 FontList::const_iterator fcit = fontlist_.begin();
555 FontList::const_iterator fend = fontlist_.end();
556 for (; fcit != fend; ++fcit) {
557 if (fcit->pos() < beg)
559 if (fcit->pos() >= end) {
560 // Add last entry in the fontlist_.
561 fontlist_.set(text_.size() - 1, fcit->font());
564 // Add a new entry in the fontlist_.
565 fontlist_.set(fcit->pos() - beg, fcit->font());
567 requestSpellCheck(p.text_.size());
571 /////////////////////////////////////////////////////////////////////
575 /////////////////////////////////////////////////////////////////////
579 /** This helper class should be instantiated at the start of methods
580 * that can create or merge changes. If as a result the value of
581 * Paragraph::isChanged is modified, it makes sure that updateBuffer()
584 struct ChangesMonitor {
586 ChangesMonitor(Paragraph & par)
587 : par_(par), was_changed_(par.isChanged()) {}
591 /* We may need to run updateBuffer to check whether the buffer
592 * contains changes (and toggle the changes toolbar). We do it
594 * 1. the `changedness' of the paragraph has changed,
595 * 2. and we are not in the situation where the buffer has changes
596 * and new changes are added to the paragraph.
599 if (par_.isChanged() != was_changed_
600 && par_.inInset().isBufferValid()
601 && !(par_.inInset().buffer().areChangesPresent() && par_.isChanged()))
602 par_.inInset().buffer().forceUpdate();
603 } catch(support::ExceptionMessage const &) {}
608 Paragraph const & par_;
615 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
616 bool output_active, TocBackend & backend) const
618 d->changes_.addToToc(cdit, buf, output_active, backend);
622 bool Paragraph::isDeleted(pos_type start, pos_type end) const
624 LASSERT(start >= 0 && start <= size(), return false);
625 LASSERT(end > start && end <= size() + 1, return false);
627 return d->changes_.isDeleted(start, end);
631 bool Paragraph::isChanged(pos_type start, pos_type end) const
633 LASSERT(start >= 0 && start <= size(), return false);
634 LASSERT(end > start && end <= size() + 1, return false);
636 return d->changes_.isChanged(start, end);
639 // FIXME: Ideally the diverse isChanged() methods should account for that!
640 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
642 LASSERT(start >= 0 && start <= size(), return false);
643 LASSERT(end > start && end <= size() + 1, return false);
645 for (auto const & icit : d->insetlist_) {
646 if (icit.pos < start)
650 if (icit.inset && icit.inset->isChanged())
656 bool Paragraph::isChanged() const
658 return d->changes_.isChanged();
662 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
664 // keep the logic here in sync with the logic of eraseChars()
668 Change const & change = d->changes_.lookup(size());
669 return change.inserted() && change.currentAuthor();
672 Change Paragraph::parEndChange() const
674 return d->changes_.lookup(size());
678 void Paragraph::setChange(Change const & change)
680 // Make sure that Buffer::hasChangesPresent is updated
681 ChangesMonitor cm(*this);
683 // beware of the imaginary end-of-par character!
684 d->changes_.set(change, 0, size() + 1);
687 * Propagate the change recursively - but not in case of DELETED!
689 * Imagine that your co-author makes changes in an existing inset. He
690 * sends your document to you and you come to the conclusion that the
691 * inset should go completely. If you erase it, LyX must not delete all
692 * text within the inset. Otherwise, the change tracked insertions of
693 * your co-author get lost and there is no way to restore them later.
695 * Conclusion: An inset's content should remain untouched if you delete it
698 if (!change.deleted()) {
699 for (pos_type pos = 0; pos < size(); ++pos) {
700 if (Inset * inset = getInset(pos))
701 inset->setChange(change);
707 void Paragraph::setChange(pos_type pos, Change const & change)
709 // Make sure that Buffer::hasChangesPresent is updated
710 ChangesMonitor cm(*this);
712 LASSERT(pos >= 0 && pos <= size(), return);
713 d->changes_.set(change, pos);
715 // see comment in setChange(Change const &) above
716 if (!change.deleted() && pos < size())
717 if (Inset * inset = getInset(pos))
718 inset->setChange(change);
722 Change const & Paragraph::lookupChange(pos_type pos) const
724 LBUFERR(pos >= 0 && pos <= size());
725 return d->changes_.lookup(pos);
729 void Paragraph::acceptChanges(pos_type start, pos_type end)
731 // Make sure that Buffer::hasChangesPresent is updated
732 ChangesMonitor cm(*this);
734 LASSERT(start >= 0 && start <= size(), return);
735 LASSERT(end > start && end <= size() + 1, return);
737 for (pos_type pos = start; pos < end; ++pos) {
738 switch (lookupChange(pos).type) {
739 case Change::UNCHANGED:
740 // accept changes in nested inset
741 if (Inset * inset = getInset(pos))
742 inset->acceptChanges();
745 case Change::INSERTED:
746 d->changes_.set(Change(Change::UNCHANGED), pos);
747 // also accept changes in nested inset
748 if (Inset * inset = getInset(pos))
749 inset->acceptChanges();
752 case Change::DELETED:
753 // Suppress access to non-existent
754 // "end-of-paragraph char"
756 eraseChar(pos, false);
767 void Paragraph::rejectChanges(pos_type start, pos_type end)
769 LASSERT(start >= 0 && start <= size(), return);
770 LASSERT(end > start && end <= size() + 1, return);
772 // Make sure that Buffer::hasChangesPresent is updated
773 ChangesMonitor cm(*this);
775 for (pos_type pos = start; pos < end; ++pos) {
776 switch (lookupChange(pos).type) {
777 case Change::UNCHANGED:
778 // reject changes in nested inset
779 if (Inset * inset = getInset(pos))
780 inset->rejectChanges();
783 case Change::INSERTED:
784 // Suppress access to non-existent
785 // "end-of-paragraph char"
787 eraseChar(pos, false);
793 case Change::DELETED:
794 d->changes_.set(Change(Change::UNCHANGED), pos);
796 // Do NOT reject changes within a deleted inset!
797 // There may be insertions of a co-author inside of it!
805 void Paragraph::Private::insertChar(pos_type pos, char_type c,
806 Change const & change)
808 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
810 // Make sure that Buffer::hasChangesPresent is updated
811 ChangesMonitor cm(*owner_);
814 changes_.insert(change, pos);
816 // This is actually very common when parsing buffers (and
817 // maybe inserting ascii text)
818 if (pos == pos_type(text_.size())) {
819 // when appending characters, no need to update tables
821 // but we want spell checking
822 requestSpellCheck(pos);
826 text_.insert(text_.begin() + pos, c);
828 // Update the font table.
829 fontlist_.increasePosAfterPos(pos);
832 insetlist_.increasePosAfterPos(pos);
834 // Update list of misspelled positions
835 speller_state_.increasePosAfterPos(pos);
838 if (inset_owner_ && inset_owner_->isBufferValid())
839 theSession().bookmarks().adjustPosAfterPos(inset_owner_->buffer().fileName(),
844 bool Paragraph::insertInset(pos_type pos, Inset * inset,
845 Font const & font, Change const & change)
847 LASSERT(inset, return false);
848 LASSERT(pos >= 0 && pos <= size(), return false);
850 // Make sure that Buffer::hasChangesPresent is updated
851 ChangesMonitor cm(*this);
853 // Paragraph::insertInset() can be used in cut/copy/paste operation where
854 // d->inset_owner_ is not set yet.
855 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
858 d->insertChar(pos, META_INSET, change);
859 LASSERT(d->text_[pos] == META_INSET, return false);
861 // Add a new entry in the insetlist_.
862 d->insetlist_.insert(inset, pos);
864 // Some insets require run of spell checker
865 requestSpellCheck(pos);
871 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
873 LASSERT(pos >= 0 && pos <= size(), return false);
875 // Make sure that Buffer::hasChangesPresent is updated
876 ChangesMonitor cm(*this);
878 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
881 Change change = d->changes_.lookup(pos);
883 // set the character to DELETED if
884 // a) it was previously unchanged or
885 // b) it was inserted by a co-author
887 if (!change.changed() ||
888 (change.inserted() && !change.currentAuthor())) {
889 setChange(pos, Change(Change::DELETED));
890 // request run of spell checker
891 requestSpellCheck(pos);
895 if (change.deleted())
899 // Don't physically access the imaginary end-of-paragraph character.
900 // eraseChar() can only mark it as DELETED. A physical deletion of
901 // end-of-par must be handled externally.
907 d->changes_.erase(pos);
909 // if it is an inset, delete the inset entry
910 if (d->text_[pos] == META_INSET)
911 d->insetlist_.erase(pos);
913 d->text_.erase(d->text_.begin() + pos);
915 // Update the fontlist_
916 d->fontlist_.erase(pos);
918 // Update the insetlist_
919 d->insetlist_.decreasePosAfterPos(pos);
921 // Update list of misspelled positions
922 d->speller_state_.decreasePosAfterPos(pos);
923 d->speller_state_.refreshLast(size());
926 if (d->inset_owner_ && d->inset_owner_->isBufferValid())
927 theSession().bookmarks().adjustPosAfterPos(d->inset_owner_->buffer().fileName(),
934 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
936 LASSERT(start >= 0 && start <= size(), return 0);
937 LASSERT(end >= start && end <= size() + 1, return 0);
940 for (pos_type count = end - start; count; --count) {
941 if (!eraseChar(i, trackChanges))
948 // Handle combining characters
949 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
950 otexstream & os, char_type c, char_type next,
951 OutputParams const & runparams) const
953 // Writing next here may circumvent a possible font change between
954 // c and next. Since next is only output if it forms a surrogate pair
955 // with c we can ignore this:
956 // A font change inside a surrogate pair does not make sense and is
957 // hopefully impossible to input.
958 // FIXME: change tracking
959 // Is this correct WRT change tracking?
960 Encoding const & encoding = *(runparams.encoding);
961 docstring latex1 = encoding.latexChar(next).first;
962 if (runparams.inIPA) {
963 string const tipashortcut = Encodings::TIPAShortcut(next);
964 if (!tipashortcut.empty()) {
965 latex1 = from_ascii(tipashortcut);
968 docstring latex2 = encoding.latexChar(c).first;
970 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
971 // Encoding supports the combination:
972 // output as is (combining char after base char).
973 os << latex2 << latex1;
974 return latex1.length() + latex2.length();
977 os << latex1 << "{" << latex2 << "}";
978 return latex1.length() + latex2.length() + 2;
982 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
983 OutputParams const & runparams,
986 unsigned int & column,
988 Layout const & style) const
990 if (style.pass_thru || runparams.pass_thru)
993 if (i + 1 < int(text_.size())) {
994 char_type next = text_[i + 1];
995 if (Encodings::isCombiningChar(next)) {
996 // This space has an accent, so we must always output it.
997 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
1002 if (runparams.linelen > 0
1003 && column > runparams.linelen
1005 && text_[i - 1] != ' '
1006 && (i + 1 < int(text_.size()))
1007 // same in FreeSpacing mode
1008 && !owner_->isFreeSpacing()
1009 // In typewriter mode, we want to avoid
1010 // ! . ? : at the end of a line
1011 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
1012 && (text_[i - 1] == '.'
1013 || text_[i - 1] == '?'
1014 || text_[i - 1] == ':'
1015 || text_[i - 1] == '!'))) {
1017 os.texrow().start(owner_->id(), i + 1);
1019 } else if (style.free_spacing) {
1028 void Paragraph::Private::latexInset(BufferParams const & bparams,
1030 OutputParams & runparams,
1031 Font & running_font,
1033 Font const & outerfont,
1035 Change & running_change,
1036 Layout const & style,
1038 unsigned int & column,
1039 bool const fontswitch_inset,
1040 bool const closeLanguage,
1041 bool const lang_switched_at_inset) const
1043 Inset * inset = owner_->getInset(i);
1046 if (style.pass_thru) {
1047 odocstringstream ods;
1048 inset->plaintext(ods, runparams);
1053 // FIXME: move this to InsetNewline::latex
1054 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1055 // newlines are handled differently here than
1056 // the default in simpleTeXSpecialChars().
1057 if (!style.newline_allowed) {
1061 bool needPar = false;
1062 column += running_font.latexWriteEndChanges(
1063 os, bparams, runparams,
1064 basefont, basefont, needPar);
1068 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1071 basefont = owner_->getLayoutFont(bparams, outerfont);
1072 running_font = basefont;
1074 if (runparams.moving_arg)
1078 os.texrow().start(owner_->id(), i + 1);
1082 if (owner_->isDeleted(i)) {
1083 if( ++runparams.inDeletedInset == 1)
1084 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1087 if (inset->canTrackChanges()) {
1088 column += Changes::latexMarkChange(os, bparams, running_change,
1089 Change(Change::UNCHANGED), runparams);
1090 running_change = Change(Change::UNCHANGED);
1094 odocstream::pos_type const len = os.os().tellp();
1096 if (inset->forceLTR(runparams)
1097 && running_font.isRightToLeft()
1098 // ERT is an exception, it should be output with no
1099 // decorations at all
1100 && inset->lyxCode() != ERT_CODE) {
1101 if (runparams.use_polyglossia) {
1103 } else if (running_font.language()->lang() == "farsi"
1104 || running_font.language()->lang() == "arabic_arabi")
1105 os << "\\textLR{" << termcmd;
1111 if (open_font && fontswitch_inset) {
1112 bool lang_closed = false;
1113 // Close language if needed
1114 if (closeLanguage && !lang_switched_at_inset) {
1115 // We need prev_font here as language changes directly at inset
1116 // will only be started inside the inset.
1117 Font const prev_font = (i > 0) ?
1118 owner_->getFont(bparams, i - 1, outerfont)
1120 Font tmpfont(basefont);
1121 tmpfont.setLanguage(prev_font.language());
1122 bool needPar = false;
1123 unsigned int count = tmpfont.latexWriteEndChanges(os, bparams, runparams,
1125 needPar, closeLanguage);
1127 lang_closed = count > 0;
1129 // Update the running_font, making sure, however,
1130 // to leave the language as it was.
1131 // FIXME: probably a better way to keep track of the old
1132 // language, than copying the entire font?
1133 Font const copy_font(running_font);
1134 basefont = owner_->getLayoutFont(bparams, outerfont);
1135 running_font = basefont;
1137 running_font.setLanguage(copy_font.language());
1138 OutputParams rp = runparams;
1139 rp.encoding = basefont.language()->encoding();
1140 // For these, we use switches, so they should be taken as
1141 // base inside the inset.
1142 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1143 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1144 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1145 // Now re-do font changes in a way needed here
1146 // (using switches with multi-par insets)
1147 InsetText const * textinset = inset->asInsetText();
1148 bool const cprotect = textinset
1149 ? textinset->hasCProtectContent(runparams.moving_arg)
1150 && !textinset->text().isMainText()
1151 && inset->lyxCode() != BRANCH_CODE
1153 unsigned int count2 = basefont.latexWriteStartChanges(os, bparams,
1159 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1163 runparams.local_font = &basefont;
1166 if (fontswitch_inset && !closeLanguage) {
1167 // The directionality has been switched at inset.
1168 // Force markup inside.
1169 runparams.local_font = &basefont;
1172 size_t const previous_row_count = os.texrow().rows();
1175 runparams.lastid = id_;
1176 runparams.lastpos = i;
1177 inset->latex(os, runparams);
1178 } catch (EncodingException & e) {
1179 // add location information and throw again.
1188 if (os.texrow().rows() > previous_row_count) {
1189 os.texrow().start(owner_->id(), i + 1);
1192 column += (unsigned int)(os.os().tellp() - len);
1195 if (owner_->isDeleted(i))
1196 --runparams.inDeletedInset;
1200 void Paragraph::Private::latexSpecialChar(otexstream & os,
1201 BufferParams const & bparams,
1202 OutputParams const & runparams,
1203 Font const & running_font,
1204 string & alien_script,
1205 Layout const & style,
1208 unsigned int & column) const
1210 char_type const c = owner_->getUChar(bparams, runparams, i);
1212 if (style.pass_thru || runparams.pass_thru || (runparams.for_searchAdv != OutputParams::NoSearch)
1213 || contains(style.pass_thru_chars, c)
1214 || contains(runparams.pass_thru_chars, c)) {
1215 if (runparams.for_searchAdv != OutputParams::NoSearch) {
1219 os << "\\braceleft ";
1221 os << "\\braceright ";
1225 else if (c != '\0') {
1226 Encoding const * const enc = runparams.encoding;
1227 if (enc && !enc->encodable(c))
1228 throw EncodingException(c);
1234 // TIPA uses its own T3 encoding
1235 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1237 // If T1 font encoding is used, use the special
1238 // characters it provides.
1239 // NOTE: Some languages reset the font encoding internally to a
1240 // non-standard font encoding. If we are using such a language,
1241 // we do not output special T1 chars.
1242 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1243 && !runparams.isFullUnicode() && runparams.main_fontenc == "T1"
1244 && latexSpecialT1(c, os, i, column))
1246 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1247 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1248 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1251 // Otherwise, we use what LaTeX provides us.
1254 os << "\\textbackslash" << termcmd;
1258 os << "\\textless" << termcmd;
1262 os << "\\textgreater" << termcmd;
1266 os << "\\textbar" << termcmd;
1271 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1272 (end_pos == -1 || i + 1 < end_pos) &&
1273 text_[i+1] == '-') {
1274 // Prevent "--" becoming an en dash and "---" an em dash.
1275 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1281 os << "\\textquotedbl" << termcmd;
1286 case '%': case '#': case '{':
1294 os << "\\textasciitilde" << termcmd;
1299 os << "\\textasciicircum" << termcmd;
1306 // avoid being mistaken for optional arguments
1314 // Blanks are printed before font switching.
1315 // Sure? I am not! (try nice-latex)
1316 // I am sure it's correct. LyX might be smarter
1317 // in the future, but for now, nothing wrong is
1323 // XeTeX's dash behaviour is determined via a global setting
1324 if (bparams.use_dash_ligatures
1325 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1327 // TODO #10961: && not in inset Flex Code
1328 // TODO #10961: && not in layout LyXCode
1329 && (!bparams.useNonTeXFonts || runparams.flavor != Flavor::XeTeX)) {
1346 Encoding const & encoding = *(runparams.encoding);
1347 char_type next = '\0';
1348 if (i + 1 < int(text_.size())) {
1349 next = text_[i + 1];
1350 if (Encodings::isCombiningChar(next)) {
1351 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1356 pair<docstring, bool> latex = encoding.latexChar(c);
1357 docstring nextlatex;
1358 bool nexttipas = false;
1359 string nexttipashortcut;
1360 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1361 nextlatex = encoding.latexChar(next).first;
1362 if (runparams.inIPA) {
1363 nexttipashortcut = Encodings::TIPAShortcut(next);
1364 nexttipas = !nexttipashortcut.empty();
1368 if (runparams.inIPA) {
1369 string const tipashortcut = Encodings::TIPAShortcut(c);
1370 if (!tipashortcut.empty()) {
1371 latex.first = from_ascii(tipashortcut);
1372 latex.second = false;
1376 // eventually close "script wrapper" command (see `Paragraph::latex`)
1377 if (!alien_script.empty()
1378 && alien_script != Encodings::isKnownScriptChar(next)) {
1379 column += latex.first.length();
1380 alien_script.clear();
1381 os << latex.first << "}";
1385 && ((!prefixIs(nextlatex, '\\')
1386 && !prefixIs(nextlatex, '{')
1387 && !prefixIs(nextlatex, '}'))
1389 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1391 // Prevent eating of a following space or command corruption by
1392 // following characters
1393 if (next == ' ' || next == '\0') {
1394 column += latex.first.length() + 1;
1395 os << latex.first << "{}";
1397 column += latex.first.length();
1398 os << latex.first << " ";
1401 column += latex.first.length() - 1;
1409 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1410 pos_type i, unsigned int & column) const
1416 // In T1 encoding, these characters exist
1417 // but we should avoid ligatures
1418 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1420 os << "\\textcompwordmark" << termcmd;
1427 // soul.sty breaks with \char`\"
1428 os << "\\textquotedbl" << termcmd;
1437 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1438 pos_type i, unsigned int & column) const
1440 // TU encoding is currently on par with T1.
1441 return latexSpecialT1(c, os, i, column);
1445 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1446 pos_type /*i*/, unsigned int & column) const
1456 os << "\\textvertline" << termcmd;
1465 void Paragraph::Private::validate(LaTeXFeatures & features) const
1467 if (layout_->inpreamble && inset_owner_) {
1468 // FIXME: Using a string stream here circumvents the encoding
1469 // switching machinery of odocstream. Therefore the
1470 // output is wrong if this paragraph contains content
1471 // that needs to switch encoding.
1472 Buffer const & buf = inset_owner_->buffer();
1473 otexstringstream os;
1474 os << layout_->preamble();
1475 size_t const length = os.length();
1476 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1477 buf.getParFromID(owner_->id()).pit(), os,
1478 features.runparams(), string(), 0, -1, true);
1479 if (os.length() > length)
1480 features.addPreambleSnippet(os.release(), true);
1483 if (features.runparams().flavor == Flavor::Html
1484 && layout_->htmltitle()) {
1485 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1488 // check the params.
1489 if (!params_.spacing().isDefault())
1490 features.require("setspace");
1493 features.useLayout(layout_->name());
1496 fontlist_.validate(features);
1498 // then the indentation
1499 if (!params_.leftIndent().zero())
1500 features.require("ParagraphLeftIndent");
1503 InsetList::const_iterator icit = insetlist_.begin();
1504 InsetList::const_iterator iend = insetlist_.end();
1505 for (; icit != iend; ++icit) {
1507 features.inDeletedInset(owner_->isDeleted(icit->pos));
1508 if (icit->inset->lyxCode() == FOOT_CODE) {
1509 // FIXME: an item inset would make things much easier.
1510 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1511 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1512 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1513 && (icit->pos < begin_of_body_
1514 || (icit->pos == begin_of_body_
1515 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1516 features.saveNoteEnv("description");
1518 icit->inset->validate(features);
1519 features.inDeletedInset(false);
1520 if (layout_->needprotect &&
1521 icit->inset->lyxCode() == FOOT_CODE)
1522 features.require("NeedLyXFootnoteCode");
1526 // then the contents
1527 BufferParams const bp = features.runparams().is_child
1528 ? features.buffer().masterParams() : features.buffer().params();
1529 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1530 char_type c = text_[i];
1531 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1533 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1534 features.require("textquotedblp");
1535 else if (features.runparams().main_fontenc != "T1"
1536 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1537 features.require("textquotedbl");
1538 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1539 // features that depend on the font or input encoding
1540 string feats = ci.textpreamble();
1541 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1542 if (fontenc.empty())
1543 fontenc = features.runparams().main_fontenc;
1544 while (!feats.empty()) {
1546 feats = split(feats, feat, ',');
1547 if (contains(feat, "!=")) {
1548 // a feature that is required except for the spcified
1549 // font or input encodings
1551 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1552 // multiple encodings are separated by semicolon
1553 vector<string> context = getVectorFromString(contexts, ";");
1554 // require feature if the context matches neither current font
1555 // nor input encoding
1556 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1557 && std::find(context.begin(), context.end(),
1558 features.runparams().encoding->name()) == context.end())
1559 features.require(realfeature);
1560 } else if (contains(feat, '=')) {
1561 // a feature that is required only for the spcified
1562 // font or input encodings
1564 string const contexts = split(feat, realfeature, '=');
1565 // multiple encodings are separated by semicolon
1566 vector<string> context = getVectorFromString(contexts, ";");
1567 // require feature if the context matches either current font
1568 // or input encoding
1569 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1570 || std::find(context.begin(), context.end(),
1571 features.runparams().encoding->name()) != context.end())
1572 features.require(realfeature);
1575 } else if (!bp.use_dash_ligatures
1576 && (c == 0x2013 || c == 0x2014)
1577 && bp.useNonTeXFonts
1578 && features.runparams().flavor == Flavor::XeTeX)
1579 // XeTeX's dash behaviour is determined via a global setting
1580 features.require("xetexdashbreakstate");
1581 BufferEncodings::validate(c, features);
1585 /////////////////////////////////////////////////////////////////////
1589 /////////////////////////////////////////////////////////////////////
1592 Layout const emptyParagraphLayout;
1595 Paragraph::Paragraph()
1596 : d(new Paragraph::Private(this, emptyParagraphLayout))
1603 Paragraph::Paragraph(Paragraph const & par)
1604 : itemdepth(par.itemdepth),
1605 d(new Paragraph::Private(*par.d, this))
1611 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1612 : itemdepth(par.itemdepth),
1613 d(new Paragraph::Private(*par.d, this, beg, end))
1619 Paragraph & Paragraph::operator=(Paragraph const & par)
1621 // needed as we will destroy the private part before copying it
1623 itemdepth = par.itemdepth;
1627 d = new Private(*par.d, this);
1634 Paragraph::~Paragraph()
1643 // this shall be called just before every "os << ..." action.
1644 void flushString(ostream & os, docstring & s)
1653 void Paragraph::write(ostream & os, BufferParams const & bparams,
1654 depth_type & depth) const
1656 // The beginning or end of a deeper (i.e. nested) area?
1657 if (depth != d->params_.depth()) {
1658 if (d->params_.depth() > depth) {
1659 while (d->params_.depth() > depth) {
1660 os << "\n\\begin_deeper";
1664 while (d->params_.depth() < depth) {
1665 os << "\n\\end_deeper";
1671 // First write the layout
1672 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1674 d->params_.write(os);
1676 Font font1(inherit_font, bparams.language);
1678 Change running_change = Change(Change::UNCHANGED);
1680 // this string is used as a buffer to avoid repetitive calls
1681 // to to_utf8(), which turn out to be expensive (JMarc)
1682 docstring write_buffer;
1685 for (pos_type i = 0; i <= size(); ++i) {
1687 Change const & change = lookupChange(i);
1688 if (change != running_change)
1689 flushString(os, write_buffer);
1690 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1691 running_change = change;
1696 // Write font changes
1697 Font font2 = getFontSettings(bparams, i);
1698 if (font2 != font1) {
1699 flushString(os, write_buffer);
1700 font2.lyxWriteChanges(font1, os);
1705 char_type const c = d->text_[i];
1708 if (Inset const * inset = getInset(i)) {
1709 flushString(os, write_buffer);
1710 if (inset->directWrite()) {
1711 // international char, let it write
1712 // code directly so it's shorter in
1718 os << "\\begin_inset ";
1720 os << "\n\\end_inset\n\n";
1723 // FIXME This can be removed again once the mystery
1724 // crash has been resolved.
1729 flushString(os, write_buffer);
1730 os << "\n\\backslash\n";
1734 flushString(os, write_buffer);
1735 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1742 if ((column > 70 && c == ' ')
1744 flushString(os, write_buffer);
1748 // this check is to amend a bug. LyX sometimes
1749 // inserts '\0' this could cause problems.
1751 write_buffer.push_back(c);
1753 LYXERR0("NUL char in structure.");
1759 flushString(os, write_buffer);
1760 os << "\n\\end_layout\n";
1761 // FIXME This can be removed again once the mystery
1762 // crash has been resolved.
1767 void Paragraph::validate(LaTeXFeatures & features) const
1769 d->validate(features);
1770 bool fragile = features.runparams().moving_arg;
1771 fragile |= layout().needprotect;
1772 if (inInset().getLayout().isNeedProtect())
1774 if (needsCProtection(fragile))
1775 features.require("cprotect");
1779 void Paragraph::insert(pos_type pos, docstring const & str,
1780 Font const & font, Change const & change)
1782 for (size_t i = 0, n = str.size(); i != n ; ++i)
1783 insertChar(pos + i, str[i], font, change);
1787 void Paragraph::appendChar(char_type c, Font const & font,
1788 Change const & change)
1790 // Make sure that Buffer::hasChangesPresent is updated
1791 ChangesMonitor cm(*this);
1794 d->changes_.insert(change, d->text_.size());
1795 // when appending characters, no need to update tables
1796 d->text_.push_back(c);
1797 setFont(d->text_.size() - 1, font);
1798 d->requestSpellCheck(d->text_.size() - 1);
1802 void Paragraph::appendString(docstring const & s, Font const & font,
1803 Change const & change)
1805 // Make sure that Buffer::hasChangesPresent is updated
1806 ChangesMonitor cm(*this);
1808 pos_type end = s.size();
1809 size_t oldsize = d->text_.size();
1810 size_t newsize = oldsize + end;
1811 size_t capacity = d->text_.capacity();
1812 if (newsize >= capacity)
1813 d->text_.reserve(max(capacity + 100, newsize));
1815 // when appending characters, no need to update tables
1818 // FIXME: Optimize this!
1819 for (size_t i = oldsize; i != newsize; ++i) {
1821 d->changes_.insert(change, i);
1822 d->requestSpellCheck(i);
1824 d->fontlist_.set(oldsize, font);
1825 d->fontlist_.set(newsize - 1, font);
1829 void Paragraph::insertChar(pos_type pos, char_type c,
1832 d->insertChar(pos, c, Change(trackChanges ?
1833 Change::INSERTED : Change::UNCHANGED));
1837 void Paragraph::insertChar(pos_type pos, char_type c,
1838 Font const & font, bool trackChanges)
1840 d->insertChar(pos, c, Change(trackChanges ?
1841 Change::INSERTED : Change::UNCHANGED));
1846 void Paragraph::insertChar(pos_type pos, char_type c,
1847 Font const & font, Change const & change)
1849 d->insertChar(pos, c, change);
1854 void Paragraph::resetFonts(Font const & font)
1856 d->fontlist_.clear();
1857 d->fontlist_.set(0, font);
1858 d->fontlist_.set(d->text_.size() - 1, font);
1861 // Gets uninstantiated font setting at position.
1862 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1866 LYXERR0("pos: " << pos << " size: " << size());
1870 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1871 if (cit != d->fontlist_.end())
1874 if (pos == size() && !empty())
1875 return getFontSettings(bparams, pos - 1);
1877 // Optimisation: avoid a full font instantiation if there is no
1878 // language change from previous call.
1879 static Font previous_font;
1880 static Language const * previous_lang = nullptr;
1881 Language const * lang = getParLanguage(bparams);
1882 if (lang != previous_lang) {
1883 previous_lang = lang;
1884 previous_font = Font(inherit_font, lang);
1886 return previous_font;
1890 FontSpan Paragraph::fontSpan(pos_type pos) const
1892 LBUFERR(pos <= size());
1895 return FontSpan(pos, pos);
1898 FontList::const_iterator cit = d->fontlist_.begin();
1899 FontList::const_iterator end = d->fontlist_.end();
1900 for (; cit != end; ++cit) {
1901 if (cit->pos() >= pos) {
1902 if (pos >= beginOfBody())
1903 return FontSpan(max(start, beginOfBody()),
1906 return FontSpan(start,
1907 min(beginOfBody() - 1,
1910 start = cit->pos() + 1;
1913 // This should not happen, but if so, we take no chances.
1914 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1915 LASSERT(false, return FontSpan(pos, pos));
1919 // Gets uninstantiated font setting at position 0
1920 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1922 if (!empty() && !d->fontlist_.empty())
1923 return d->fontlist_.begin()->font();
1925 // Optimisation: avoid a full font instantiation if there is no
1926 // language change from previous call.
1927 static Font previous_font;
1928 static Language const * previous_lang = nullptr;
1929 if (bparams.language != previous_lang) {
1930 previous_lang = bparams.language;
1931 previous_font = Font(inherit_font, bparams.language);
1934 return previous_font;
1938 // Gets the fully instantiated font at a given position in a paragraph
1939 // This is basically the same function as Text::GetFont() in text2.cpp.
1940 // The difference is that this one is used for generating the LaTeX file,
1941 // and thus cosmetic "improvements" are disallowed: This has to deliver
1942 // the true picture of the buffer. (Asger)
1943 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1944 Font const & outerfont) const
1948 Font font = getFontSettings(bparams, pos);
1950 pos_type const body_pos = beginOfBody();
1951 FontInfo & fi = font.fontInfo();
1953 fi.realize(d->layout_->labelfont);
1955 fi.realize(d->layout_->font);
1957 fi.realize(outerfont.fontInfo());
1958 fi.realize(bparams.getFont().fontInfo());
1964 Font const Paragraph::getLabelFont
1965 (BufferParams const & bparams, Font const & outerfont) const
1967 FontInfo tmpfont = d->layout_->labelfont;
1968 tmpfont.realize(outerfont.fontInfo());
1969 tmpfont.realize(bparams.getFont().fontInfo());
1970 return Font(tmpfont, getParLanguage(bparams));
1974 Font const Paragraph::getLayoutFont
1975 (BufferParams const & bparams, Font const & outerfont) const
1977 FontInfo tmpfont = d->layout_->font;
1978 tmpfont.realize(outerfont.fontInfo());
1979 tmpfont.realize(bparams.getFont().fontInfo());
1980 return Font(tmpfont, getParLanguage(bparams));
1984 char_type Paragraph::getUChar(BufferParams const & bparams,
1985 OutputParams const & rp,
1988 char_type c = d->text_[pos];
1990 // Return unchanged character in LTR languages
1991 // or if we use poylglossia/bidi (XeTeX).
1992 if (rp.useBidiPackage()
1993 || !getFontSettings(bparams, pos).isRightToLeft())
1996 // Without polyglossia/bidi, we need to account for some special cases.
1997 // FIXME This needs to be audited!
1999 // * The input is as expected for all delimiters
2000 // => checked for Hebrew!
2001 // * The output matches the display in the LyX workarea
2002 // => checked for Hebrew!
2003 // * The special cases below are really necessary
2004 // => checked for Hebrew!
2005 // * In arabic_arabi, brackets are transformed to Arabic
2006 // Ornate Parentheses. Is this is really wanted?
2008 string const & lang = getFontSettings(bparams, pos).language()->lang();
2011 // 1. In the following languages, parentheses need to be reversed.
2012 // Also with polyglodia/luabidi
2013 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
2015 // 2. In the following languages, brackets don't need to be reversed.
2016 bool const reversebrackets = lang != "arabic_arabtex"
2017 && lang != "arabic_arabi"
2020 // Now swap delimiters if needed.
2031 if (reversebrackets)
2035 if (reversebrackets)
2056 void Paragraph::setFont(pos_type pos, Font const & font)
2058 LASSERT(pos <= size(), return);
2060 // First, reduce font against layout/label font
2061 // Update: The setCharFont() routine in text2.cpp already
2062 // reduces font, so we don't need to do that here. (Asger)
2064 d->fontlist_.set(pos, font);
2068 void Paragraph::makeSameLayout(Paragraph const & par)
2070 d->layout_ = par.d->layout_;
2071 d->params_ = par.d->params_;
2075 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2077 if (isFreeSpacing())
2083 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2084 if (eraseChar(pos, trackChanges))
2090 return count > 0 || pos > 0;
2094 bool Paragraph::hasSameLayout(Paragraph const & par) const
2096 return par.d->layout_ == d->layout_
2097 && d->params_.sameLayout(par.d->params_);
2101 depth_type Paragraph::getDepth() const
2103 return d->params_.depth();
2107 depth_type Paragraph::getMaxDepthAfter() const
2109 if (d->layout_->isEnvironment())
2110 return d->params_.depth() + 1;
2112 return d->params_.depth();
2116 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2118 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2119 return getDefaultAlign(bparams);
2121 return d->params_.align();
2125 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2127 LyXAlignment res = layout().align;
2128 if (isRTL(bparams)) {
2130 if (res == LYX_ALIGN_LEFT)
2131 res = LYX_ALIGN_RIGHT;
2132 else if (res == LYX_ALIGN_RIGHT)
2133 res = LYX_ALIGN_LEFT;
2139 docstring const & Paragraph::labelString() const
2141 return d->params_.labelString();
2145 // the next two functions are for the manual labels
2146 docstring const Paragraph::getLabelWidthString() const
2148 if (d->layout_->margintype == MARGIN_MANUAL
2149 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2150 return d->params_.labelWidthString();
2152 return _("Senseless with this layout!");
2156 void Paragraph::setLabelWidthString(docstring const & s)
2158 d->params_.labelWidthString(s);
2162 docstring Paragraph::expandLabel(Layout const & layout,
2163 BufferParams const & bparams) const
2165 return expandParagraphLabel(layout, bparams, true);
2169 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2170 BufferParams const & bparams, bool process_appendix) const
2172 DocumentClass const & tclass = bparams.documentClass();
2173 string const & lang = getParLanguage(bparams)->code();
2174 bool const in_appendix = process_appendix && d->params_.appendix();
2175 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2177 if (fmt.empty() && !layout.counter.empty())
2178 return tclass.counters().theCounter(layout.counter, lang);
2180 // handle 'inherited level parts' in 'fmt',
2181 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2182 size_t const i = fmt.find('@', 0);
2183 if (i != docstring::npos) {
2184 size_t const j = fmt.find('@', i + 1);
2185 if (j != docstring::npos) {
2186 docstring parent(fmt, i + 1, j - i - 1);
2187 docstring label = from_ascii("??");
2188 if (tclass.hasLayout(parent))
2189 label = expandParagraphLabel(tclass[parent], bparams,
2191 fmt = docstring(fmt, 0, i) + label
2192 + docstring(fmt, j + 1, docstring::npos);
2196 return tclass.counters().counterLabel(fmt, lang);
2200 void Paragraph::applyLayout(Layout const & new_layout)
2202 d->layout_ = &new_layout;
2203 LyXAlignment const oldAlign = d->params_.align();
2205 if (!(oldAlign & d->layout_->alignpossible)) {
2206 frontend::Alert::warning(_("Alignment not permitted"),
2207 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2208 d->params_.align(LYX_ALIGN_LAYOUT);
2213 pos_type Paragraph::beginOfBody() const
2215 return d->begin_of_body_;
2219 void Paragraph::setBeginOfBody()
2221 if (d->layout_->labeltype != LABEL_MANUAL) {
2222 d->begin_of_body_ = 0;
2226 // Unroll the first two cycles of the loop
2227 // and remember the previous character to
2228 // remove unnecessary getChar() calls
2230 pos_type end = size();
2231 bool prev_char_deleted = false;
2232 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2235 char_type previous_char = d->text_[i];
2236 if (!(isNewline(i) || isEnvSeparator(i))) {
2238 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2239 char_type temp = d->text_[i];
2240 prev_char_deleted = isDeleted(i);
2241 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2244 previous_char = temp;
2250 d->begin_of_body_ = i;
2254 bool Paragraph::allowParagraphCustomization() const
2256 return inInset().allowParagraphCustomization();
2260 bool Paragraph::usePlainLayout() const
2262 return inInset().usePlainLayout();
2266 bool Paragraph::isPassThru() const
2268 return inInset().isPassThru() || d->layout_->pass_thru;
2272 bool Paragraph::parbreakIsNewline() const
2274 return inInset().getLayout().parbreakIsNewline() || d->layout_->parbreak_is_newline;
2278 bool Paragraph::isPartOfTextSequence() const
2280 for (pos_type i = 0; i < size(); ++i) {
2281 if (!isInset(i) || getInset(i)->isPartOfTextSequence())
2289 // paragraphs inside floats need different alignment tags to avoid
2292 bool noTrivlistCentering(InsetCode code)
2294 return code == FLOAT_CODE
2295 || code == WRAP_CODE
2296 || code == CELL_CODE;
2300 string correction(string const & orig)
2302 if (orig == "flushleft")
2303 return "raggedright";
2304 if (orig == "flushright")
2305 return "raggedleft";
2306 if (orig == "center")
2312 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2313 InsetCode code, bool const lastpar, int & col)
2315 string macro = suffix + "{";
2316 if (noTrivlistCentering(code)) {
2318 // the last paragraph in non-trivlist-aligned
2319 // context is special (to avoid unwanted whitespace)
2320 if (suffix == "\\begin") {
2321 macro = "\\" + correction(env) + "{}";
2322 os << from_ascii(macro);
2323 col += macro.size();
2328 macro += correction(env);
2332 if (suffix == "\\par\\end") {
2336 os << from_ascii(macro);
2337 col += macro.size();
2338 if (suffix == "\\begin") {
2348 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2349 otexstream & os, OutputParams const & runparams) const
2354 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2355 (layout_->toggle_indent != ITOGGLE_NEVER) :
2356 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2358 LyXAlignment const curAlign = params_.align();
2360 // Do not output \\noindent for paragraphs
2361 // 1. that cannot have indentation or are indented always,
2362 // 2. that are not part of the immediate text sequence (e.g., contain only floats),
2363 // 3. that are PassThru,
2364 // 4. or that are centered.
2365 if (canindent && params_.noindent()
2366 && owner_->isPartOfTextSequence()
2367 && !layout_->pass_thru
2368 && curAlign != LYX_ALIGN_CENTER) {
2369 if (!owner_->empty()
2370 && (owner_->isInset(0)
2371 && owner_->getInset(0)->lyxCode() == VSPACE_CODE))
2372 // If the paragraph starts with a vspace, the \\noindent
2373 // needs to come after that (as it leaves vmode).
2374 // If the paragraph consists only of the vspace,
2375 // \\noindent is not needed at all.
2376 runparams.need_noindent = owner_->size() > 1;
2378 os << "\\noindent" << termcmd;
2383 if (curAlign == layout_->align)
2387 case LYX_ALIGN_NONE:
2388 case LYX_ALIGN_BLOCK:
2389 case LYX_ALIGN_LAYOUT:
2390 case LYX_ALIGN_SPECIAL:
2391 case LYX_ALIGN_DECIMAL:
2393 case LYX_ALIGN_LEFT:
2394 case LYX_ALIGN_RIGHT:
2395 case LYX_ALIGN_CENTER:
2396 if (runparams.moving_arg) {
2403 string const begin_tag = "\\begin";
2404 InsetCode code = ownerCode();
2405 bool const lastpar = runparams.isLastPar;
2406 // RTL in classic (PDF)LaTeX (without the Bidi package)
2407 // Luabibdi (used by LuaTeX) behaves like classic
2408 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2409 && !runparams.useBidiPackage();
2412 case LYX_ALIGN_NONE:
2413 case LYX_ALIGN_BLOCK:
2414 case LYX_ALIGN_LAYOUT:
2415 case LYX_ALIGN_SPECIAL:
2416 case LYX_ALIGN_DECIMAL:
2418 case LYX_ALIGN_LEFT: {
2420 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2421 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2423 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2425 } case LYX_ALIGN_RIGHT: {
2427 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2428 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2430 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2432 } case LYX_ALIGN_CENTER: {
2433 corrected_env(os, begin_tag, "center", code, lastpar, column);
2442 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2443 otexstream & os, OutputParams const & runparams) const
2445 LyXAlignment const curAlign = params_.align();
2447 if (curAlign == layout_->align)
2451 case LYX_ALIGN_NONE:
2452 case LYX_ALIGN_BLOCK:
2453 case LYX_ALIGN_LAYOUT:
2454 case LYX_ALIGN_SPECIAL:
2455 case LYX_ALIGN_DECIMAL:
2457 case LYX_ALIGN_LEFT:
2458 case LYX_ALIGN_RIGHT:
2459 case LYX_ALIGN_CENTER:
2460 if (runparams.moving_arg)
2465 bool output = false;
2467 string const end_tag = "\\par\\end";
2468 InsetCode code = ownerCode();
2469 bool const lastpar = runparams.isLastPar;
2470 // RTL in classic (PDF)LaTeX (without the Bidi package)
2471 // Luabibdi (used by LuaTeX) behaves like classic
2472 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2473 && !runparams.useBidiPackage();
2476 case LYX_ALIGN_NONE:
2477 case LYX_ALIGN_BLOCK:
2478 case LYX_ALIGN_LAYOUT:
2479 case LYX_ALIGN_SPECIAL:
2480 case LYX_ALIGN_DECIMAL:
2482 case LYX_ALIGN_LEFT: {
2484 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2485 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2487 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2489 } case LYX_ALIGN_RIGHT: {
2491 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2492 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2494 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2496 } case LYX_ALIGN_CENTER: {
2497 corrected_env(os, end_tag, "center", code, lastpar, col);
2502 return output || lastpar;
2506 // This one spits out the text of the paragraph
2507 void Paragraph::latex(BufferParams const & bparams,
2508 Font const & outerfont,
2510 OutputParams const & runparams,
2511 int start_pos, int end_pos, bool force) const
2513 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2515 // FIXME This check should not be needed. Perhaps issue an
2516 // error if it triggers.
2517 Layout const & style = inInset().forcePlainLayout() ?
2518 bparams.documentClass().plainLayout() : *d->layout_;
2520 if (!force && style.inpreamble)
2523 bool const allowcust = allowParagraphCustomization();
2525 // Current base font for all inherited font changes, without any
2526 // change caused by an individual character, except for the language:
2527 // It is set to the language of the first character.
2528 // As long as we are in the label, this font is the base font of the
2529 // label. Before the first body character it is set to the base font
2533 // If there is an open font-encoding changing command (script wrapper),
2534 // alien_script is set to its name
2535 string alien_script;
2538 // Maybe we have to create a optional argument.
2539 pos_type body_pos = beginOfBody();
2540 unsigned int column = 0;
2542 // If we are inside an non inheritFont() inset, the real outerfont is local_font
2543 Font const real_outerfont = (!inInset().inheritFont()
2544 && runparams.local_font != nullptr)
2545 ? Font(runparams.local_font->fontInfo()) : outerfont;
2548 // the optional argument is kept in curly brackets in
2549 // case it contains a ']'
2550 // This is not strictly needed, but if this is changed it
2551 // would be a file format change, and tex2lyx would need
2552 // to be adjusted, since it unconditionally removes the
2553 // braces when it parses \item.
2556 basefont = getLabelFont(bparams, real_outerfont);
2558 basefont = getLayoutFont(bparams, real_outerfont);
2561 // Which font is currently active?
2562 Font running_font(basefont);
2563 // Do we have an open font change?
2564 bool open_font = false;
2566 Change runningChange =
2567 runparams.inDeletedInset && !inInset().canTrackChanges()
2568 ? runparams.changeOfDeletedInset : Change(Change::UNCHANGED);
2570 Encoding const * const prev_encoding = runparams.encoding;
2572 os.texrow().start(id(), 0);
2574 // if the paragraph is empty, the loop will not be entered at all
2576 // For InTitle commands, we have already opened a group
2577 // in output_latex::TeXOnePar.
2578 if (style.isCommand() && !style.intitle) {
2582 if (!style.leftdelim().empty()) {
2583 os << style.leftdelim();
2584 column += style.leftdelim().size();
2587 column += d->startTeXParParams(bparams, os, runparams);
2590 // Whether a \par can be issued for insets typeset inline with text.
2591 // Yes if greater than 0. This has to be static.
2592 THREAD_LOCAL_STATIC int parInline = 0;
2594 for (pos_type i = 0; i < size(); ++i) {
2595 // First char in paragraph or after label?
2596 if (i == body_pos) {
2599 bool needPar = false;
2600 column += running_font.latexWriteEndChanges(
2601 os, bparams, runparams,
2602 basefont, basefont, needPar);
2605 basefont = getLayoutFont(bparams, real_outerfont);
2606 running_font = basefont;
2608 column += Changes::latexMarkChange(os, bparams,
2609 runningChange, Change(Change::UNCHANGED),
2611 runningChange = Change(Change::UNCHANGED);
2613 os << (isEnvSeparator(i) ? "}]~" : "}] ");
2616 // For InTitle commands, we have already opened a group
2617 // in output_latex::TeXOnePar.
2618 if (style.isCommand() && !style.intitle) {
2623 if (!style.leftdelim().empty()) {
2624 os << style.leftdelim();
2625 column += style.leftdelim().size();
2629 column += d->startTeXParParams(bparams, os,
2633 runparams.wasDisplayMath = runparams.inDisplayMath;
2634 runparams.inDisplayMath = false;
2635 bool deleted_display_math = false;
2636 Change const & change = runparams.inDeletedInset
2637 ? runparams.changeOfDeletedInset : lookupChange(i);
2639 char_type const c = d->text_[i];
2641 // Check whether a display math inset follows
2642 bool output_changes;
2643 if (runparams.for_searchAdv == OutputParams::NoSearch)
2644 output_changes = bparams.output_changes;
2646 output_changes = (runparams.for_searchAdv == OutputParams::SearchWithDeleted);
2648 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2650 runparams.ctObject = getInset(i)->getCtObject(runparams);
2652 InsetMath const * im = getInset(i)->asInsetMath();
2653 if (im && im->asHullInset()
2654 && im->asHullInset()->outerDisplay()) {
2655 runparams.inDisplayMath = true;
2656 // runparams.inDeletedInset will be set by
2657 // latexInset later, but we need this info
2658 // before it is called. On the other hand, we
2659 // cannot set it here because it is a counter.
2660 deleted_display_math = isDeleted(i);
2662 if (output_changes && deleted_display_math
2663 && runningChange == change
2664 && change.type == Change::DELETED
2665 && !os.afterParbreak()) {
2666 // A display math in the same paragraph follows.
2667 // We have to close and then reopen \lyxdeleted,
2668 // otherwise the math will be shifted up.
2669 OutputParams rp = runparams;
2671 bool needPar = false;
2672 column += running_font.latexWriteEndChanges(
2673 os, bparams, rp, basefont,
2677 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2678 : getLayoutFont(bparams, real_outerfont);
2679 running_font = basefont;
2680 column += Changes::latexMarkChange(os, bparams,
2681 Change(Change::INSERTED), change, rp);
2685 if (output_changes && runningChange != change) {
2686 if (!alien_script.empty()) {
2689 alien_script.clear();
2692 bool needPar = false;
2693 column += running_font.latexWriteEndChanges(
2694 os, bparams, runparams,
2695 basefont, basefont, needPar);
2698 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2699 : getLayoutFont(bparams, real_outerfont);
2700 running_font = basefont;
2701 column += Changes::latexMarkChange(os, bparams, runningChange,
2703 runningChange = change;
2706 // do not output text which is marked deleted
2707 // if change tracking output is disabled
2708 if (!output_changes && change.deleted()) {
2714 // Fully instantiated font
2715 Font current_font = getFont(bparams, i, outerfont);
2717 Font const prev_font = (i > 0) ?
2718 getFont(bparams, i - 1, outerfont)
2721 Font const last_font = running_font;
2722 bool const in_ct_deletion = (output_changes
2723 && runningChange == change
2724 && change.type == Change::DELETED
2725 && !os.afterParbreak());
2726 // Insets where font switches are used (rather than font commands)
2727 bool const fontswitch_inset =
2730 && getInset(i)->allowMultiPar()
2731 && getInset(i)->lyxCode() != ERT_CODE
2732 && (getInset(i)->producesOutput()
2733 // FIXME Something more general?
2734 // Comments do not "produce output" but are still
2735 // part of the TeX source and require font switches
2736 // to be closed (otherwise LaTeX fails).
2737 || getInset(i)->layoutName() == "Note:Comment");
2739 bool closeLanguage = false;
2740 bool lang_switched_at_inset = false;
2741 if (fontswitch_inset) {
2742 // Some insets cannot be inside a font change command.
2743 // However, even such insets *can* be placed in \L or \R
2744 // or their equivalents (for RTL language switches),
2745 // so we don't close the language in those cases
2746 // (= differing isRightToLeft()).
2747 // ArabTeX, though, doesn't seem to handle this special behavior.
2748 closeLanguage = basefont.isRightToLeft() == current_font.isRightToLeft()
2749 || basefont.language()->lang() == "arabic_arabtex"
2750 || current_font.language()->lang() == "arabic_arabtex";
2751 // We need to check prev_font as language changes directly at inset
2752 // will only be started inside the inset.
2753 lang_switched_at_inset = prev_font.language() != current_font.language();
2756 // Do we need to close the previous font?
2757 bool langClosed = false;
2759 ((current_font != running_font
2760 || current_font.language() != running_font.language())
2761 || (fontswitch_inset
2762 && (current_font == prev_font))))
2764 // ensure there is no open script-wrapper
2765 if (!alien_script.empty()) {
2768 alien_script.clear();
2770 if (in_ct_deletion) {
2771 // We have to close and then reopen \lyxdeleted,
2772 // as strikeout needs to be on lowest level.
2777 // Force language closing
2778 current_font.setLanguage(basefont.language());
2779 Font const nextfont = (i == body_pos-1) ? basefont : current_font;
2780 bool needPar = false;
2781 column += running_font.latexWriteEndChanges(
2782 os, bparams, runparams, basefont,
2784 if (in_ct_deletion) {
2785 // We have to close and then reopen \lyxdeleted,
2786 // as strikeout needs to be on lowest level.
2787 OutputParams rp = runparams;
2788 column += Changes::latexMarkChange(os, bparams,
2789 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2792 // Has the language been closed in the latexWriteEndChanges() call above?
2793 langClosed = running_font.language() != basefont.language()
2794 && running_font.language() != nextfont.language()
2795 && (running_font.language()->encoding()->package() != Encoding::CJK);
2796 running_font = basefont;
2799 // if necessary, close language environment before opening CJK
2800 string const running_lang = running_font.language()->babel();
2801 string const lang_end_command = lyxrc.language_command_end;
2802 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2803 && !running_lang.empty()
2804 && running_lang == openLanguageName()
2805 && current_font.language()->encoding()->package() == Encoding::CJK) {
2806 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2807 os << from_ascii(end_tag);
2808 column += end_tag.length();
2812 // Switch file encoding if necessary (and allowed)
2813 if ((!fontswitch_inset || closeLanguage)
2814 && !runparams.pass_thru && !style.pass_thru &&
2815 runparams.encoding->package() != Encoding::none &&
2816 current_font.language()->encoding()->package() != Encoding::none) {
2817 pair<bool, int> const enc_switch =
2818 switchEncoding(os.os(), bparams, runparams,
2819 *(current_font.language()->encoding()));
2820 if (enc_switch.first) {
2821 column += enc_switch.second;
2822 runparams.encoding = current_font.language()->encoding();
2826 // A display math inset inside an ulem command will be output
2827 // as a box of width \linewidth, so we have to either disable
2828 // indentation if the inset starts a paragraph, or start a new
2829 // line to accommodate such box. This has to be done before
2830 // writing any font changing commands.
2831 if (runparams.inDisplayMath && !deleted_display_math
2832 && runparams.inulemcmd) {
2833 if (os.afterParbreak())
2839 // Do we need to change font?
2840 if ((current_font != running_font ||
2841 current_font.language() != running_font.language())
2842 && i != body_pos - 1)
2844 if (!fontswitch_inset) {
2845 if (in_ct_deletion) {
2846 // We have to close and then reopen \lyxdeleted,
2847 // as strikeout needs to be on lowest level.
2848 OutputParams rp = runparams;
2849 bool needPar = false;
2850 column += running_font.latexWriteEndChanges(
2851 os, bparams, rp, basefont,
2856 otexstringstream ots;
2857 InsetText const * textinset = inInset().asInsetText();
2858 bool const cprotect = textinset
2859 ? textinset->hasCProtectContent(runparams.moving_arg)
2860 && !textinset->text().isMainText()
2861 && inInset().lyxCode() != BRANCH_CODE
2863 column += current_font.latexWriteStartChanges(ots, bparams,
2864 runparams, basefont, last_font, false,
2866 // Check again for display math in ulem commands as a
2867 // font change may also occur just before a math inset.
2868 if (runparams.inDisplayMath && !deleted_display_math
2869 && runparams.inulemcmd) {
2870 if (os.afterParbreak())
2875 running_font = current_font;
2877 docstring fontchange = ots.str();
2879 // check whether the fontchange ends with a \\textcolor
2880 // modifier and the text starts with a space. If so we
2881 // need to add } in order to prevent \\textcolor from gobbling
2882 // the space (bug 4473).
2883 docstring const last_modifier = rsplit(fontchange, '\\');
2884 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2885 os << from_ascii("{}");
2886 else if (ots.terminateCommand())
2888 if (in_ct_deletion) {
2889 // We have to close and then reopen \lyxdeleted,
2890 // as strikeout needs to be on lowest level.
2891 OutputParams rp = runparams;
2892 column += Changes::latexMarkChange(os, bparams,
2893 Change(Change::UNCHANGED), change, rp);
2896 running_font = current_font;
2897 open_font = !langClosed;
2901 // FIXME: think about end_pos implementation...
2902 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2903 // FIXME: integrate this case in latexSpecialChar
2904 // Do not print the separation of the optional argument
2905 // if style.pass_thru is false. This works because
2906 // latexSpecialChar ignores spaces if
2907 // style.pass_thru is false.
2908 if (i != body_pos - 1) {
2909 if (d->simpleTeXBlanks(bparams, runparams, os,
2910 i, column, current_font, style)) {
2911 // A surrogate pair was output. We
2912 // must not call latexSpecialChar
2913 // in this iteration, since it would output
2914 // the combining character again.
2921 OutputParams rp = runparams;
2922 rp.free_spacing = style.free_spacing;
2923 rp.local_font = ¤t_font;
2924 rp.intitle = style.intitle;
2926 // Two major modes: LaTeX or plain
2927 // Handle here those cases common to both modes
2928 // and then split to handle the two modes separately.
2929 if (c == META_INSET) {
2930 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2931 // Greyedout notes and, in general, all insets
2932 // with InsetLayout::isDisplay() == false,
2933 // are typeset inline with the text. So, we
2934 // can add a \par to the last paragraph of
2935 // such insets only if nothing else follows.
2936 bool incremented = false;
2937 Inset const * inset = getInset(i);
2938 InsetText const * textinset = inset
2939 ? inset->asInsetText()
2941 if (i + 1 == size() && textinset
2942 && !inset->getLayout().isDisplay()) {
2943 ParagraphList const & pars =
2944 textinset->text().paragraphs();
2945 pit_type const pit = pars.size() - 1;
2946 Font const lastfont =
2947 pit < 0 || pars[pit].empty()
2948 ? pars[pit].getLayoutFont(
2951 : pars[pit].getFont(bparams,
2952 pars[pit].size() - 1,
2954 if (lastfont.fontInfo().size() !=
2955 basefont.fontInfo().size()) {
2960 // We need to restore parts of this after insets with
2961 // allowMultiPar() true
2962 Font const save_basefont = basefont;
2963 d->latexInset(bparams, os, rp, running_font,
2964 basefont, real_outerfont, open_font,
2965 runningChange, style, i, column, fontswitch_inset,
2966 closeLanguage, lang_switched_at_inset);
2967 if (fontswitch_inset) {
2969 bool needPar = false;
2970 column += running_font.latexWriteEndChanges(
2971 os, bparams, runparams,
2972 basefont, basefont, needPar);
2975 basefont.fontInfo().setSize(save_basefont.fontInfo().size());
2976 basefont.fontInfo().setFamily(save_basefont.fontInfo().family());
2977 basefont.fontInfo().setSeries(save_basefont.fontInfo().series());
2982 if (runparams.ctObject == CtObject::DisplayObject
2983 || runparams.ctObject == CtObject::UDisplayObject) {
2984 // Close \lyx*deleted and force its
2985 // reopening (if needed)
2988 runningChange = Change(Change::UNCHANGED);
2989 runparams.ctObject = CtObject::Normal;
2992 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2993 if (!bparams.useNonTeXFonts)
2994 script = Encodings::isKnownScriptChar(c);
2995 if (script != alien_script) {
2996 if (!alien_script.empty()) {
2998 alien_script.clear();
3000 string fontenc = running_font.language()->fontenc(bparams);
3002 && !Encodings::fontencSupportsScript(fontenc, script)) {
3003 column += script.length() + 2;
3004 os << "\\" << script << "{";
3005 alien_script = script;
3009 d->latexSpecialChar(os, bparams, rp, running_font,
3010 alien_script, style, i, end_pos, column);
3011 } catch (EncodingException & e) {
3012 if (runparams.dryrun) {
3013 os << "<" << _("LyX Warning: ")
3014 << _("uncodable character") << " '";
3018 // add location information and throw again.
3026 // Set the encoding to that returned from latexSpecialChar (see
3027 // comment for encoding member in OutputParams.h)
3028 runparams.encoding = rp.encoding;
3030 // Also carry on the info on a closed ulem command for insets
3031 // such as Note that do not produce any output, so that no
3032 // command is ever executed but its opening was recorded.
3033 runparams.inulemcmd = rp.inulemcmd;
3035 // These need to be passed upstream as well
3036 runparams.need_maketitle = rp.need_maketitle;
3037 runparams.have_maketitle = rp.have_maketitle;
3039 // And finally, pass the post_macros upstream
3040 runparams.post_macro = rp.post_macro;
3043 // Close wrapper for alien script
3044 if (!alien_script.empty()) {
3046 alien_script.clear();
3049 Font const font = empty()
3050 ? getLayoutFont(bparams, real_outerfont)
3051 : getFont(bparams, size() - 1, real_outerfont);
3053 InsetText const * textinset = inInset().asInsetText();
3055 bool const maintext = textinset
3056 ? textinset->text().isMainText()
3059 size_t const numpars = textinset
3060 ? textinset->text().paragraphs().size()
3063 bool needPar = false;
3065 if (style.resfont.size() != font.fontInfo().size()
3066 && (!runparams.isLastPar || maintext
3067 || (numpars > 1 && d->ownerCode() != CELL_CODE
3068 && (inInset().getLayout().isDisplay()
3070 && !style.isCommand()) {
3074 // If we have an open font definition, we have to close it
3076 // Make sure that \\par is done with the font of the last
3077 // character if this has another size as the default.
3078 // This is necessary because LaTeX (and LyX on the screen)
3079 // calculates the space between the baselines according
3080 // to this font. (Matthias)
3082 // We must not change the font for the last paragraph
3083 // of non-multipar insets, tabular cells or commands,
3084 // since this produces unwanted whitespace.
3085 #ifdef FIXED_LANGUAGE_END_DETECTION
3087 running_font.latexWriteEndChanges(os, bparams,
3088 runparams, basefont,
3089 next_->getFont(bparams, 0, outerfont),
3092 running_font.latexWriteEndChanges(os, bparams,
3093 runparams, basefont, basefont, needPar);
3096 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
3097 //FIXME: there as we start another \selectlanguage with the next paragraph if
3098 //FIXME: we are in need of this. This should be fixed sometime (Jug)
3099 running_font.latexWriteEndChanges(os, bparams, runparams,
3100 basefont, basefont, needPar);
3104 // The \par could not be inserted at the same nesting
3105 // level of the font size change, so do it now.
3106 os << "{\\" << font.latexSize() << "\\par}";
3109 if (!runparams.inDeletedInset || inInset().canTrackChanges())
3110 column += Changes::latexMarkChange(os, bparams, runningChange,
3111 Change(Change::UNCHANGED), runparams);
3113 // Needed if there is an optional argument but no contents.
3114 if (body_pos > 0 && body_pos == size()) {
3118 if (!style.rightdelim().empty()) {
3119 os << style.rightdelim();
3120 column += style.rightdelim().size();
3123 if (allowcust && d->endTeXParParams(bparams, os, runparams)
3124 && runparams.encoding != prev_encoding) {
3125 runparams.encoding = prev_encoding;
3126 os << setEncoding(prev_encoding->iconvName());
3129 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
3133 bool Paragraph::emptyTag() const
3135 for (pos_type i = 0; i < size(); ++i) {
3136 if (Inset const * inset = getInset(i)) {
3137 InsetCode lyx_code = inset->lyxCode();
3138 // FIXME testing like that is wrong. What is
3140 if (lyx_code != TOC_CODE &&
3141 lyx_code != INCLUDE_CODE &&
3142 lyx_code != GRAPHICS_CODE &&
3143 lyx_code != ERT_CODE &&
3144 lyx_code != LISTINGS_CODE &&
3145 lyx_code != FLOAT_CODE &&
3146 lyx_code != TABULAR_CODE) {
3150 char_type c = d->text_[i];
3151 if (c != ' ' && c != '\t')
3159 string Paragraph::getID(Buffer const &, OutputParams const &)
3162 for (pos_type i = 0; i < size(); ++i) {
3163 if (Inset const * inset = getInset(i)) {
3164 InsetCode lyx_code = inset->lyxCode();
3165 if (lyx_code == LABEL_CODE) {
3166 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
3167 docstring const & id = il->getParam("name");
3168 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3176 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3179 for (i = 0; i < size(); ++i) {
3180 if (Inset const * inset = getInset(i)) {
3181 inset->docbook(xs, runparams);
3183 char_type c = d->text_[i];
3193 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3197 for (i = 0; i < size(); ++i) {
3198 if (Inset const * inset = getInset(i)) {
3199 inset->xhtml(xs, runparams);
3201 char_type c = d->text_[i];
3211 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3214 pos_type size = text_.size();
3215 for (pos_type i = initial; i < size; ++i) {
3216 Font font = owner_->getFont(buf.params(), i, outerfont);
3217 if (text_[i] == META_INSET)
3219 if (i != initial && font != font_old)
3230 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3231 vector<xml::EndFontTag> & tagsToClose,
3232 bool & flag, FontState curstate, xml::FontTypes type)
3234 if (curstate == FONT_ON) {
3235 tagsToOpen.push_back(docbookStartFontTag(type));
3238 tagsToClose.push_back(docbookEndFontTag(type));
3243 class OptionalFontType {
3248 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3249 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3252 OptionalFontType fontShapeToXml(FontShape fs)
3256 return {xml::FT_ITALIC};
3258 return {xml::FT_SLANTED};
3259 case SMALLCAPS_SHAPE:
3260 return {xml::FT_SMALLCAPS};
3265 // the other tags are for internal use
3271 OptionalFontType fontFamilyToXml(FontFamily fm)
3275 return {xml::FT_ROMAN};
3277 return {xml::FT_SANS};
3278 case TYPEWRITER_FAMILY:
3279 return {xml::FT_TYPE};
3280 case INHERIT_FAMILY:
3283 // the other tags are for internal use
3289 OptionalFontType fontSizeToXml(FontSize fs)
3293 return {xml::FT_SIZE_TINY};
3295 return {xml::FT_SIZE_SCRIPT};
3297 return {xml::FT_SIZE_FOOTNOTE};
3299 return {xml::FT_SIZE_SMALL};
3301 return {xml::FT_SIZE_LARGE};
3303 return {xml::FT_SIZE_LARGER};
3305 return {xml::FT_SIZE_LARGEST};
3307 return {xml::FT_SIZE_HUGE};
3309 return {xml::FT_SIZE_HUGER};
3311 return {xml::FT_SIZE_INCREASE};
3313 return {xml::FT_SIZE_DECREASE};
3318 // the other tags are for internal use
3324 struct DocBookFontState
3326 FontShape curr_fs = INHERIT_SHAPE;
3327 FontFamily curr_fam = INHERIT_FAMILY;
3328 FontSize curr_size = INHERIT_SIZE;
3330 // track whether we have opened these tags
3331 bool emph_flag = false;
3332 bool bold_flag = false;
3333 bool noun_flag = false;
3334 bool ubar_flag = false;
3335 bool dbar_flag = false;
3336 bool sout_flag = false;
3337 bool xout_flag = false;
3338 bool wave_flag = false;
3340 bool shap_flag = false;
3342 bool faml_flag = false;
3344 bool size_flag = false;
3347 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3349 std::string const & default_family,
3350 DocBookFontState & fs)
3352 vector<xml::FontTag> tagsToOpen;
3353 vector<xml::EndFontTag> tagsToClose;
3356 FontState curstate = font.fontInfo().emph();
3357 if (font_old.emph() != curstate)
3358 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3361 curstate = font.fontInfo().noun();
3362 if (font_old.noun() != curstate)
3363 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3366 curstate = font.fontInfo().underbar();
3367 if (font_old.underbar() != curstate)
3368 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3371 curstate = font.fontInfo().strikeout();
3372 if (font_old.strikeout() != curstate)
3373 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3376 curstate = font.fontInfo().xout();
3377 if (font_old.xout() != curstate)
3378 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3381 curstate = font.fontInfo().uuline();
3382 if (font_old.uuline() != curstate)
3383 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3386 curstate = font.fontInfo().uwave();
3387 if (font_old.uwave() != curstate)
3388 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3391 // a little hackish, but allows us to reuse what we have.
3392 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3393 if (font_old.series() != font.fontInfo().series())
3394 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3397 fs.curr_fs = font.fontInfo().shape();
3398 FontShape old_fs = font_old.shape();
3399 if (old_fs != fs.curr_fs) {
3401 OptionalFontType tag = fontShapeToXml(old_fs);
3403 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3404 fs.shap_flag = false;
3407 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3409 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3413 fs.curr_fam = font.fontInfo().family();
3414 FontFamily old_fam = font_old.family();
3415 if (old_fam != fs.curr_fam) {
3417 OptionalFontType tag = fontFamilyToXml(old_fam);
3419 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3420 fs.faml_flag = false;
3422 switch (fs.curr_fam) {
3424 // we will treat a "default" font family as roman, since we have
3425 // no other idea what to do.
3426 if (default_family != "rmdefault" && default_family != "default") {
3427 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3428 fs.faml_flag = true;
3432 if (default_family != "sfdefault") {
3433 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3434 fs.faml_flag = true;
3437 case TYPEWRITER_FAMILY:
3438 if (default_family != "ttdefault") {
3439 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3440 fs.faml_flag = true;
3443 case INHERIT_FAMILY:
3446 // the other tags are for internal use
3453 fs.curr_size = font.fontInfo().size();
3454 FontSize old_size = font_old.size();
3455 if (old_size != fs.curr_size) {
3457 OptionalFontType tag = fontSizeToXml(old_size);
3459 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3460 fs.size_flag = false;
3463 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3464 if (tag.has_value) {
3465 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3466 fs.size_flag = true;
3470 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3473 } // anonymous namespace
3476 std::tuple<std::vector<docstring>, std::vector<docstring>, std::vector<docstring>>
3477 Paragraph::simpleDocBookOnePar(Buffer const & buf,
3478 OutputParams const & runparams,
3479 Font const & outerfont,
3482 bool ignore_fonts) const
3484 std::vector<docstring> prependedParagraphs;
3485 std::vector<docstring> generatedParagraphs;
3486 std::vector<docstring> appendedParagraphs;
3487 odocstringstream os;
3489 // If there is an argument that must be output before the main tag, do it before handling the rest of the paragraph.
3490 // Also tag all arguments that shouldn't go in the main content right now, so that they are never generated at the
3492 OutputParams rp = runparams;
3493 for (pos_type i = initial; i < size(); ++i) {
3494 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3495 const InsetArgument * arg = getInset(i)->asInsetArgument();
3496 if (arg->docbookargumentbeforemaintag()) {
3497 auto xs_local = XMLStream(os);
3498 arg->docbook(xs_local, rp);
3500 prependedParagraphs.push_back(os.str());
3501 os.str(from_ascii(""));
3503 rp.docbook_prepended_arguments.insert(arg);
3504 } else if (arg->docbookargumentaftermaintag()) {
3505 rp.docbook_appended_arguments.insert(arg);
3510 // State variables for the main loop.
3511 auto xs = new XMLStream(os); // XMLStream has no copy constructor: to create a new object, the only solution
3512 // is to hold a pointer to the XMLStream (xs = XMLStream(os) is not allowed once the first object is built).
3513 std::vector<char_type> delayedChars; // When a font tag ends with a space, output it after the closing font tag.
3514 // This requires to store delayed characters at some point.
3516 DocBookFontState fs; // Track whether we have opened font tags
3517 DocBookFontState old_fs = fs;
3519 Layout const & style = *d->layout_;
3520 FontInfo font_old = style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3521 string const default_family = buf.masterBuffer()->params().fonts_default_family;
3523 vector<xml::FontTag> tagsToOpen;
3524 vector<xml::EndFontTag> tagsToClose;
3526 // Parsing main loop.
3527 for (pos_type i = initial; i < size(); ++i) {
3528 bool ignore_fonts_i = ignore_fonts
3529 || style.docbooknofontinside()
3530 || (getInset(i) && getInset(i)->getLayout().docbooknofontinside());
3532 // Don't show deleted material in the output.
3536 // If this is an InsetNewline, generate a new paragraph. Also reset the fonts, so that tags are closed in
3538 if (getInset(i) && getInset(i)->lyxCode() == NEWLINE_CODE) {
3539 if (!ignore_fonts_i)
3540 xs->closeFontTags();
3542 // Output one paragraph (i.e. one string entry in generatedParagraphs).
3543 generatedParagraphs.push_back(os.str());
3545 // Create a new XMLStream for the new paragraph, completely independent from the previous one. This implies
3546 // that the string stream must be reset.
3547 os.str(from_ascii(""));
3549 xs = new XMLStream(os);
3551 // Restore the fonts for the new paragraph, so that the right tags are opened for the new entry.
3552 if (!ignore_fonts_i) {
3553 font_old = outerfont.fontInfo();
3558 // Determine which tags should be opened or closed regarding fonts.
3559 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3560 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs);
3562 if (!ignore_fonts_i) {
3563 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3564 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3565 for (; cit != cen; ++cit)
3569 // Deal with the delayed characters *after* closing font tags.
3570 if (!delayedChars.empty()) {
3571 for (char_type c: delayedChars)
3573 delayedChars.clear();
3576 if (!ignore_fonts_i) {
3577 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3578 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3579 for (; sit != sen; ++sit)
3582 tagsToClose.clear();
3586 // Finally, write the next character or inset.
3587 if (Inset const * inset = getInset(i)) {
3588 bool inset_is_argument_elsewhere = getInset(i)->asInsetArgument() &&
3589 rp.docbook_appended_arguments.find(inset->asInsetArgument()) != rp.docbook_appended_arguments.end() &&
3590 rp.docbook_prepended_arguments.find(inset->asInsetArgument()) != rp.docbook_prepended_arguments.end();
3592 if ((!rp.for_toc || inset->isInToc()) && !inset_is_argument_elsewhere) {
3593 // Arguments may need to be output
3594 OutputParams np = rp;
3595 np.local_font = &font;
3597 // TODO: special case will bite here.
3598 np.docbook_in_par = true;
3599 inset->docbook(*xs, np);
3602 char_type c = getUChar(buf.masterBuffer()->params(), rp, i);
3603 if (lyx::isSpace(c) && !ignore_fonts)
3604 delayedChars.push_back(c);
3608 font_old = font.fontInfo();
3611 // FIXME, this code is just imported from XHTML
3612 // I'm worried about what happens if a branch, say, is itself
3613 // wrapped in some font stuff. I think that will not work.
3615 xs->closeFontTags();
3617 // Deal with the delayed characters *after* closing font tags.
3618 if (!delayedChars.empty())
3619 for (char_type c: delayedChars)
3622 // In listings, new lines (i.e. \n characters in the output) are very important. Avoid generating one for the
3623 // last line to get a clean output.
3624 if (rp.docbook_in_listing && !is_last_par)
3627 // Finalise the last (and most likely only) paragraph.
3628 generatedParagraphs.push_back(os.str());
3629 os.str(from_ascii(""));
3632 // If there is an argument that must be output after the main tag, do it after handling the rest of the paragraph.
3633 for (pos_type i = initial; i < size(); ++i) {
3634 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3635 const InsetArgument * arg = getInset(i)->asInsetArgument();
3636 if (arg->docbookargumentaftermaintag()) {
3637 // Don't use rp, as this argument would not generate anything.
3638 auto xs_local = XMLStream(os);
3639 arg->docbook(xs_local, runparams);
3641 appendedParagraphs.push_back(os.str());
3642 os.str(from_ascii(""));
3647 return std::make_tuple(prependedParagraphs, generatedParagraphs, appendedParagraphs);
3653 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3654 vector<xml::EndFontTag> & tagsToClose,
3655 bool & flag, FontState curstate, xml::FontTypes type)
3657 if (curstate == FONT_ON) {
3658 tagsToOpen.push_back(xhtmlStartFontTag(type));
3661 tagsToClose.push_back(xhtmlEndFontTag(type));
3666 } // anonymous namespace
3669 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3671 OutputParams const & runparams,
3672 Font const & outerfont,
3673 bool start_paragraph, bool close_paragraph,
3674 pos_type initial) const
3678 // track whether we have opened these tags
3679 bool emph_flag = false;
3680 bool bold_flag = false;
3681 bool noun_flag = false;
3682 bool ubar_flag = false;
3683 bool dbar_flag = false;
3684 bool sout_flag = false;
3685 bool xout_flag = false;
3686 bool wave_flag = false;
3688 bool shap_flag = false;
3690 bool faml_flag = false;
3692 bool size_flag = false;
3694 Layout const & style = *d->layout_;
3696 if (start_paragraph)
3697 xs.startDivision(allowEmpty());
3700 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3702 FontShape curr_fs = INHERIT_SHAPE;
3703 FontFamily curr_fam = INHERIT_FAMILY;
3704 FontSize curr_size = INHERIT_SIZE;
3706 string const default_family =
3707 buf.masterBuffer()->params().fonts_default_family;
3709 vector<xml::FontTag> tagsToOpen;
3710 vector<xml::EndFontTag> tagsToClose;
3712 // parsing main loop
3713 for (pos_type i = initial; i < size(); ++i) {
3714 // let's not show deleted material in the output
3718 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3721 FontState curstate = font.fontInfo().emph();
3722 if (font_old.emph() != curstate)
3723 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3726 curstate = font.fontInfo().noun();
3727 if (font_old.noun() != curstate)
3728 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3731 curstate = font.fontInfo().underbar();
3732 if (font_old.underbar() != curstate)
3733 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3736 curstate = font.fontInfo().strikeout();
3737 if (font_old.strikeout() != curstate)
3738 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3741 curstate = font.fontInfo().xout();
3742 if (font_old.xout() != curstate)
3743 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3746 curstate = font.fontInfo().uuline();
3747 if (font_old.uuline() != curstate)
3748 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3751 curstate = font.fontInfo().uwave();
3752 if (font_old.uwave() != curstate)
3753 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3756 // a little hackish, but allows us to reuse what we have.
3757 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3758 if (font_old.series() != font.fontInfo().series())
3759 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3762 curr_fs = font.fontInfo().shape();
3763 FontShape old_fs = font_old.shape();
3764 if (old_fs != curr_fs) {
3768 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3771 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3773 case SMALLCAPS_SHAPE:
3774 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3780 // the other tags are for internal use
3788 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3792 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3795 case SMALLCAPS_SHAPE:
3796 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3803 // the other tags are for internal use
3810 curr_fam = font.fontInfo().family();
3811 FontFamily old_fam = font_old.family();
3812 if (old_fam != curr_fam) {
3816 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3819 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3821 case TYPEWRITER_FAMILY:
3822 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3824 case INHERIT_FAMILY:
3827 // the other tags are for internal use
3835 // we will treat a "default" font family as roman, since we have
3836 // no other idea what to do.
3837 if (default_family != "rmdefault" && default_family != "default") {
3838 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3843 if (default_family != "sfdefault") {
3844 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3848 case TYPEWRITER_FAMILY:
3849 if (default_family != "ttdefault") {
3850 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3854 case INHERIT_FAMILY:
3857 // the other tags are for internal use
3864 curr_size = font.fontInfo().size();
3865 FontSize old_size = font_old.size();
3866 if (old_size != curr_size) {
3870 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3873 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3876 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3879 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3882 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3885 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3888 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3891 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3894 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3897 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3900 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3906 // the other tags are for internal use
3912 switch (curr_size) {
3914 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3918 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3922 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3926 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3930 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3934 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3938 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3942 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3946 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3950 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3954 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3961 // the other tags are for internal use
3968 // Other such tags? What about the other text ranges?
3970 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3971 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3972 for (; cit != cen; ++cit)
3975 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3976 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3977 for (; sit != sen; ++sit)
3980 tagsToClose.clear();
3983 Inset const * inset = getInset(i);
3985 if (!runparams.for_toc || inset->isInToc()) {
3986 OutputParams np = runparams;
3987 np.local_font = &font;
3988 // If the paragraph has size 1, then we are in the "special
3989 // case" where we do not output the containing paragraph info
3990 if (!inset->getLayout().htmlisblock() && size() != 1)
3991 np.html_in_par = true;
3992 retval += inset->xhtml(xs, np);
3995 char_type c = getUChar(buf.masterBuffer()->params(),
3997 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3998 xs << XMLStream::ESCAPE_NONE << " ";
4002 font_old = font.fontInfo();
4006 // I'm worried about what happens if a branch, say, is itself
4007 // wrapped in some font stuff. I think that will not work.
4009 if (close_paragraph)
4016 bool Paragraph::isHfill(pos_type pos) const
4018 Inset const * inset = getInset(pos);
4019 return inset && inset->isHfill();
4023 bool Paragraph::isNewline(pos_type pos) const
4025 // U+2028 LINE SEPARATOR
4026 // U+2029 PARAGRAPH SEPARATOR
4027 char_type const c = d->text_[pos];
4028 if (c == 0x2028 || c == 0x2029)
4030 Inset const * inset = getInset(pos);
4031 return inset && inset->lyxCode() == NEWLINE_CODE;
4035 bool Paragraph::isEnvSeparator(pos_type pos) const
4037 Inset const * inset = getInset(pos);
4038 return inset && inset->lyxCode() == SEPARATOR_CODE;
4042 bool Paragraph::isLineSeparator(pos_type pos) const
4044 char_type const c = d->text_[pos];
4045 if (isLineSeparatorChar(c))
4047 Inset const * inset = getInset(pos);
4048 return inset && inset->isLineSeparator();
4052 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
4056 if (ignore_deleted && isDeleted(pos))
4058 if (Inset const * inset = getInset(pos))
4059 return !inset->isLetter();
4060 // if we have a hard hyphen (no en- or emdash) or apostrophe
4061 // we pass this to the spell checker
4062 // FIXME: this method is subject to change, visit
4063 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
4064 // to get an impression how complex this is.
4065 if (isHardHyphenOrApostrophe(pos))
4067 char_type const c = d->text_[pos];
4068 // We want to pass the escape chars to the spellchecker
4069 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
4070 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
4074 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
4076 pos_type const psize = size();
4079 char_type const c = d->text_[pos];
4080 if (c != '-' && c != '\'')
4082 pos_type nextpos = pos + 1;
4083 pos_type prevpos = pos > 0 ? pos - 1 : 0;
4084 if ((nextpos == psize || isSpace(nextpos))
4085 && (pos == 0 || isSpace(prevpos)))
4091 bool Paragraph::needsCProtection(bool const fragile) const
4093 // first check the layout of the paragraph, but only in insets
4094 InsetText const * textinset = inInset().asInsetText();
4095 bool const maintext = textinset
4096 ? textinset->text().isMainText()
4099 if (!maintext && layout().needcprotect) {
4100 // Environments need cprotection regardless the content
4101 if (layout().latextype == LATEX_ENVIRONMENT)
4104 // Commands need cprotection if they contain specific chars
4105 int const nchars_escape = 9;
4106 static char_type const chars_escape[nchars_escape] = {
4107 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
4109 docstring const pars = asString();
4110 for (int k = 0; k < nchars_escape; k++) {
4111 if (contains(pars, chars_escape[k]))
4116 // now check whether we have insets that need cprotection
4117 for (auto const & icit : d->insetlist_) {
4118 Inset const * ins = icit.inset;
4121 if (ins->needsCProtection(maintext, fragile))
4123 // Now check math environments
4124 InsetMath const * im = ins->asInsetMath();
4125 if (!im || im->cell(0).empty())
4127 switch(im->cell(0)[0]->lyxCode()) {
4129 // these need cprotection
4140 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
4142 return d->speller_state_.getRange(pos);
4146 bool Paragraph::isChar(pos_type pos) const
4148 if (Inset const * inset = getInset(pos))
4149 return inset->isChar();
4150 char_type const c = d->text_[pos];
4151 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
4155 bool Paragraph::isSpace(pos_type pos) const
4157 if (Inset const * inset = getInset(pos))
4158 return inset->isSpace();
4159 char_type const c = d->text_[pos];
4160 return lyx::isSpace(c);
4165 Paragraph::getParLanguage(BufferParams const & bparams) const
4168 return getFirstFontSettings(bparams).language();
4169 // FIXME: we should check the prev par as well (Lgb)
4170 return bparams.language;
4174 bool Paragraph::isRTL(BufferParams const & bparams) const
4176 return getParLanguage(bparams)->rightToLeft()
4177 && !inInset().getLayout().forceLTR();
4181 void Paragraph::changeLanguage(BufferParams const & bparams,
4182 Language const * from, Language const * to)
4184 // change language including dummy font change at the end
4185 for (pos_type i = 0; i <= size(); ++i) {
4186 Font font = getFontSettings(bparams, i);
4187 if (font.language() == from) {
4188 font.setLanguage(to);
4190 d->requestSpellCheck(i);
4196 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
4198 Language const * doc_language = bparams.language;
4199 for (auto const & f : d->fontlist_)
4200 if (f.font().language() != ignore_language &&
4201 f.font().language() != latex_language &&
4202 f.font().language() != doc_language)
4208 void Paragraph::getLanguages(std::set<Language const *> & langs) const
4210 for (auto const & f : d->fontlist_) {
4211 Language const * lang = f.font().language();
4212 if (lang != ignore_language &&
4213 lang != latex_language)
4219 docstring Paragraph::asString(int options) const
4221 return asString(0, size(), options);
4225 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
4227 odocstringstream os;
4230 && options & AS_STR_LABEL
4231 && !d->params_.labelString().empty())
4232 os << d->params_.labelString() << ' ';
4234 for (pos_type i = beg; i < end; ++i) {
4235 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
4237 char_type const c = d->text_[i];
4238 if (isPrintable(c) || c == '\t'
4239 || (c == '\n' && (options & AS_STR_NEWLINES)))
4241 else if (c == META_INSET && (options & AS_STR_INSETS)) {
4242 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
4243 LASSERT(runparams != nullptr, return docstring());
4244 getInset(i)->plaintext(os, *runparams);
4245 } else if (c == META_INSET && (options & AS_STR_MATHED)
4246 && getInset(i)->lyxCode() == REF_CODE) {
4247 Buffer const & buf = getInset(i)->buffer();
4248 OutputParams rp(&buf.params().encoding());
4249 Font const font(inherit_font, buf.params().language);
4250 rp.local_font = &font;
4252 getInset(i)->latex(ots, rp);
4254 getInset(i)->toString(os);
4263 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
4264 bool const shorten, bool const label) const
4266 size_t tmplen = shorten ? maxlen + 1 : maxlen;
4267 if (label && !labelString().empty())
4268 os += labelString() + ' ';
4269 if (!layout().isTocCaption())
4271 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4274 char_type const c = d->text_[i];
4277 else if (c == META_INSET)
4278 getInset(i)->forOutliner(os, tmplen, false);
4281 Text::shortenForOutliner(os, maxlen);
4285 void Paragraph::setInsetOwner(Inset const * inset)
4287 d->inset_owner_ = inset;
4291 int Paragraph::id() const
4297 void Paragraph::setId(int id)
4303 Layout const & Paragraph::layout() const
4309 void Paragraph::setLayout(Layout const & layout)
4311 d->layout_ = &layout;
4315 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4317 setLayout(tc.defaultLayout());
4321 void Paragraph::setPlainLayout(DocumentClass const & tc)
4323 setLayout(tc.plainLayout());
4327 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tc)
4329 if (usePlainLayout())
4332 setDefaultLayout(tc);
4336 Inset const & Paragraph::inInset() const
4338 LBUFERR(d->inset_owner_);
4339 return *d->inset_owner_;
4343 ParagraphParameters & Paragraph::params()
4349 ParagraphParameters const & Paragraph::params() const
4355 bool Paragraph::isFreeSpacing() const
4357 if (d->layout_->free_spacing)
4359 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4363 bool Paragraph::allowEmpty() const
4365 if (d->layout_->keepempty)
4367 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4371 bool Paragraph::brokenBiblio() const
4373 // There is a problem if there is no bibitem at position 0 in
4374 // paragraphs that need one, if there is another bibitem in the
4375 // paragraph or if this paragraph is not supposed to have
4376 // a bibitem inset at all.
4377 return ((d->layout_->labeltype == LABEL_BIBLIO
4378 && (d->insetlist_.find(BIBITEM_CODE) != 0
4379 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4380 || (d->layout_->labeltype != LABEL_BIBLIO
4381 && d->insetlist_.find(BIBITEM_CODE) != -1));
4385 int Paragraph::fixBiblio(Buffer const & buffer)
4387 // FIXME: when there was already an inset at 0, the return value is 1,
4388 // which does not tell whether another inset has been removed; the
4389 // cursor cannot be correctly updated.
4391 bool const track_changes = buffer.params().track_changes;
4392 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4394 // The case where paragraph is not BIBLIO
4395 if (d->layout_->labeltype != LABEL_BIBLIO) {
4396 if (bibitem_pos == -1)
4397 // No InsetBibitem => OK
4399 // There is an InsetBibitem: remove it!
4400 d->insetlist_.release(bibitem_pos);
4401 eraseChar(bibitem_pos, track_changes);
4402 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4405 bool const hasbibitem0 = bibitem_pos == 0;
4407 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4408 // There was an InsetBibitem at pos 0,
4409 // and no other one => OK
4410 if (bibitem_pos == -1)
4412 // there is a bibitem at the 0 position, but since
4413 // there is a second one, we copy the second on the
4414 // first. We're assuming there are at most two of
4415 // these, which there should be.
4416 // FIXME: why does it make sense to do that rather
4417 // than keep the first? (JMarc)
4418 Inset * inset = releaseInset(bibitem_pos);
4419 d->insetlist_.begin()->inset = inset;
4420 // This needs to be done to update the counter (#8499)
4421 buffer.updateBuffer();
4422 return -bibitem_pos;
4425 // We need to create an inset at the beginning
4426 Inset * inset = nullptr;
4427 if (bibitem_pos > 0) {
4428 // there was one somewhere in the paragraph, let's move it
4429 inset = d->insetlist_.release(bibitem_pos);
4430 eraseChar(bibitem_pos, track_changes);
4433 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4434 InsetCommandParams(BIBITEM_CODE));
4436 Font font(inherit_font, buffer.params().language);
4437 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4438 : Change::UNCHANGED));
4440 // This is needed to get the counters right
4441 buffer.updateBuffer();
4446 void Paragraph::checkAuthors(AuthorList const & authorList)
4448 d->changes_.checkAuthors(authorList);
4452 bool Paragraph::isChanged(pos_type pos) const
4454 return lookupChange(pos).changed();
4458 bool Paragraph::isInserted(pos_type pos) const
4460 return lookupChange(pos).inserted();
4464 bool Paragraph::isDeleted(pos_type pos) const
4466 return lookupChange(pos).deleted();
4470 InsetList const & Paragraph::insetList() const
4472 return d->insetlist_;
4476 void Paragraph::setInsetBuffers(Buffer & b)
4478 d->insetlist_.setBuffer(b);
4482 void Paragraph::resetBuffer()
4484 d->insetlist_.resetBuffer();
4488 Inset * Paragraph::releaseInset(pos_type pos)
4490 Inset * inset = d->insetlist_.release(pos);
4491 /// does not honour change tracking!
4492 eraseChar(pos, false);
4497 Inset * Paragraph::getInset(pos_type pos)
4499 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4500 ? d->insetlist_.get(pos) : nullptr;
4504 Inset const * Paragraph::getInset(pos_type pos) const
4506 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4507 ? d->insetlist_.get(pos) : nullptr;
4511 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4512 pos_type & right, TextCase action)
4514 // process sequences of modified characters; in change
4515 // tracking mode, this approach results in much better
4516 // usability than changing case on a char-by-char basis
4517 // We also need to track the current font, since font
4518 // changes within sequences can occur.
4519 vector<pair<char_type, Font> > changes;
4521 bool const trackChanges = bparams.track_changes;
4523 bool capitalize = true;
4525 for (; pos < right; ++pos) {
4526 char_type oldChar = d->text_[pos];
4527 char_type newChar = oldChar;
4529 // ignore insets and don't play with deleted text!
4530 if (oldChar != META_INSET && !isDeleted(pos)) {
4532 case text_lowercase:
4533 newChar = lowercase(oldChar);
4535 case text_capitalization:
4537 newChar = uppercase(oldChar);
4541 case text_uppercase:
4542 newChar = uppercase(oldChar);
4547 if (isWordSeparator(pos) || isDeleted(pos)) {
4548 // permit capitalization again
4552 if (oldChar != newChar) {
4553 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4554 if (pos != right - 1)
4556 // step behind the changing area
4560 int erasePos = pos - changes.size();
4561 for (auto const & change : changes) {
4562 insertChar(pos, change.first, change.second, trackChanges);
4563 if (!eraseChar(erasePos, trackChanges)) {
4566 ++right; // expand selection
4574 int Paragraph::find(docstring const & str, bool cs, bool mw,
4575 pos_type start_pos, bool del) const
4577 pos_type pos = start_pos;
4578 int const strsize = str.length();
4580 pos_type const parsize = d->text_.size();
4581 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4582 // ignore deleted matter
4583 if (!del && isDeleted(pos)) {
4584 if (pos == parsize - 1)
4590 // Ignore "invisible" letters such as ligature breaks
4591 // and hyphenation chars while searching
4592 bool nonmatch = false;
4593 while (pos < parsize && isInset(pos)) {
4594 Inset const * inset = getInset(pos);
4595 if (!inset->isLetter() && !inset->isChar())
4597 odocstringstream os;
4598 inset->toString(os);
4599 docstring const insetstring = os.str();
4600 if (!insetstring.empty()) {
4601 int const insetstringsize = insetstring.length();
4602 for (int j = 0; j < insetstringsize && pos < parsize; ++i, ++j) {
4603 if ((cs && str[i] != insetstring[j])
4604 || (!cs && uppercase(str[i]) != uppercase(insetstring[j]))) {
4612 if (nonmatch || i == strsize)
4614 if (cs && str[i] != d->text_[pos])
4616 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
4623 // if necessary, check whether string matches word
4625 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4628 && !isWordSeparator(pos))
4632 return pos - start_pos;
4636 char_type Paragraph::getChar(pos_type pos) const
4638 return d->text_[pos];
4642 pos_type Paragraph::size() const
4644 return d->text_.size();
4648 bool Paragraph::empty() const
4650 return d->text_.empty();
4654 bool Paragraph::isInset(pos_type pos) const
4656 return d->text_[pos] == META_INSET;
4660 bool Paragraph::isSeparator(pos_type pos) const
4662 //FIXME: Are we sure this can be the only separator?
4663 return d->text_[pos] == ' ';
4667 void Paragraph::deregisterWords()
4669 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4670 Private::LangWordsMap::const_iterator ite = d->words_.end();
4671 for (; itl != ite; ++itl) {
4672 WordList & wl = theWordList(itl->first);
4673 Private::Words::const_iterator it = (itl->second).begin();
4674 Private::Words::const_iterator et = (itl->second).end();
4675 for (; it != et; ++it)
4682 void Paragraph::locateWord(pos_type & from, pos_type & to,
4683 word_location const loc, bool const ignore_deleted) const
4686 case WHOLE_WORD_STRICT:
4687 if (from == 0 || from == size()
4688 || isWordSeparator(from, ignore_deleted)
4689 || isWordSeparator(from - 1, ignore_deleted)) {
4696 // If we are already at the beginning of a word, do nothing
4697 if (!from || isWordSeparator(from - 1, ignore_deleted))
4702 // always move the cursor to the beginning of previous word
4703 while (from && !isWordSeparator(from - 1, ignore_deleted))
4707 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4710 // no need to move the 'from' cursor
4714 while (to < size() && !isWordSeparator(to, ignore_deleted))
4719 void Paragraph::collectWords()
4721 for (pos_type pos = 0; pos < size(); ++pos) {
4722 if (isWordSeparator(pos))
4724 pos_type from = pos;
4725 locateWord(from, pos, WHOLE_WORD);
4726 // Work around MSVC warning: The statement
4727 // if (pos < from + lyxrc.completion_minlength)
4728 // triggers a signed vs. unsigned warning.
4729 // I don't know why this happens, it could be a MSVC bug, or
4730 // related to LLP64 (windows) vs. LP64 (unix) programming
4731 // model, or the C++ standard might be ambigous in the section
4732 // defining the "usual arithmetic conversions". However, using
4733 // a temporary variable is safe and works on all compilers.
4734 pos_type const endpos = from + lyxrc.completion_minlength;
4737 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4738 if (cit == d->fontlist_.end())
4740 Language const * lang = cit->font().language();
4741 docstring const word = asString(from, pos, AS_STR_NONE);
4742 d->words_[lang->lang()].insert(word);
4747 void Paragraph::registerWords()
4749 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4750 Private::LangWordsMap::const_iterator ite = d->words_.end();
4751 for (; itl != ite; ++itl) {
4752 WordList & wl = theWordList(itl->first);
4753 Private::Words::const_iterator it = (itl->second).begin();
4754 Private::Words::const_iterator et = (itl->second).end();
4755 for (; it != et; ++it)
4761 void Paragraph::updateWords()
4769 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4771 SkipPositionsIterator begin = skips.begin();
4772 SkipPositions::iterator end = skips.end();
4773 if (pos > 0 && begin < end) {
4775 if (end->last == pos - 1) {
4780 skips.insert(end, FontSpan(pos, pos));
4784 Language * Paragraph::Private::locateSpellRange(
4785 pos_type & from, pos_type & to,
4786 SkipPositions & skips) const
4788 // skip leading white space
4789 while (from < to && owner_->isWordSeparator(from))
4791 // don't check empty range
4794 // get current language
4795 Language * lang = getSpellLanguage(from);
4796 pos_type last = from;
4797 bool samelang = true;
4798 bool sameinset = true;
4799 while (last < to && samelang && sameinset) {
4800 // hop to end of word
4801 while (last < to && !owner_->isWordSeparator(last)) {
4802 Inset const * inset = owner_->getInset(last);
4803 if (inset && dynamic_cast<const InsetSpecialChar *>(inset)) {
4804 // check for "invisible" letters such as ligature breaks
4805 odocstringstream os;
4806 inset->toString(os);
4807 if (os.str().length() != 0) {
4808 // avoid spell check of visible special char insets
4809 // stop the loop in front of the special char inset
4814 appendSkipPosition(skips, last);
4815 } else if (owner_->isDeleted(last)) {
4816 appendSkipPosition(skips, last);
4820 // hop to next word while checking for insets
4821 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4822 if (Inset const * inset = owner_->getInset(last))
4823 sameinset = inset->isChar() && inset->isLetter();
4824 if (sameinset && owner_->isDeleted(last)) {
4825 appendSkipPosition(skips, last);
4830 if (sameinset && last < to) {
4831 // now check for language change
4832 samelang = lang == getSpellLanguage(last);
4835 // if language change detected backstep is needed
4843 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4846 const_cast<Language *>(owner_->getFontSettings(
4847 inset_owner_->buffer().params(), from).language());
4848 if (lang == inset_owner_->buffer().params().language
4849 && !lyxrc.spellchecker_alt_lang.empty()) {
4851 string const lang_variety =
4852 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4853 lang->setCode(lang_code);
4854 lang->setVariety(lang_variety);
4860 void Paragraph::requestSpellCheck(pos_type pos)
4862 d->requestSpellCheck(pos);
4864 // Also request spellcheck within (text) insets
4865 for (auto const & insets : insetList()) {
4866 if (!insets.inset->asInsetText())
4868 ParagraphList & inset_pars =
4869 insets.inset->asInsetText()->paragraphs();
4870 ParagraphList::iterator pit = inset_pars.begin();
4871 ParagraphList::iterator pend = inset_pars.end();
4872 for (; pit != pend; ++pit)
4873 pit->requestSpellCheck();
4879 bool Paragraph::needsSpellCheck() const
4881 SpellChecker::ChangeNumber speller_change_number = 0;
4882 if (theSpellChecker())
4883 speller_change_number = theSpellChecker()->changeNumber();
4884 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4885 d->speller_state_.needsCompleteRefresh(speller_change_number);
4887 return d->needsSpellCheck();
4891 bool Paragraph::Private::ignoreWord(docstring const & word) const
4893 // Ignore words with digits
4894 // FIXME: make this customizable
4895 // (note that some checkers ignore words with digits by default)
4896 docstring::const_iterator cit = word.begin();
4897 docstring::const_iterator const end = word.end();
4898 for (; cit != end; ++cit) {
4899 if (isNumber((*cit)))
4906 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4907 WordLangTuple & wl, docstring_list & suggestions,
4908 bool do_suggestion, bool check_learned) const
4910 SpellChecker::Result result = SpellChecker::WORD_OK;
4911 SpellChecker * speller = theSpellChecker();
4915 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4918 locateWord(from, to, WHOLE_WORD, true);
4919 if (from == to || from >= size())
4922 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4923 Language * lang = d->getSpellLanguage(from);
4925 BufferParams const & bparams = d->inset_owner_->buffer().params();
4927 if (getFontSettings(bparams, from).fontInfo().nospellcheck() == FONT_ON)
4930 wl = WordLangTuple(word, lang);
4935 if (needsSpellCheck() || check_learned) {
4937 if (!d->ignoreWord(word)) {
4938 bool const trailing_dot = to < size() && d->text_[to] == '.';
4939 result = speller->check(wl, bparams.spellignore());
4940 if (SpellChecker::misspelled(result) && trailing_dot) {
4941 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4942 result = speller->check(wl, bparams.spellignore());
4943 if (!SpellChecker::misspelled(result)) {
4944 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4946 from << ".." << to << "]");
4948 // spell check with dot appended failed too
4949 // restore original word/lang value
4950 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4951 wl = WordLangTuple(word, lang);
4955 if (!SpellChecker::misspelled(result)) {
4956 // area up to the begin of the next word is not misspelled
4957 while (end < size() && isWordSeparator(end))
4960 d->setMisspelled(from, end, result);
4962 result = d->speller_state_.getState(from);
4966 suggestions.clear();
4968 if (SpellChecker::misspelled(result)) {
4969 LYXERR(Debug::GUI, "misspelled word: \"" <<
4971 from << ".." << to << "]");
4973 speller->suggest(wl, suggestions);
4979 void Paragraph::anonymize()
4981 // This is a very crude anonymization for now
4982 for (char_type & c : d->text_)
4983 if (isLetterChar(c) || isNumber(c))
4988 void Paragraph::Private::markMisspelledWords(
4989 Language const * lang,
4990 pos_type const & first, pos_type const & last,
4991 SpellChecker::Result result,
4992 docstring const & word,
4993 SkipPositions const & skips)
4995 if (!SpellChecker::misspelled(result)) {
4996 setMisspelled(first, last, SpellChecker::WORD_OK);
5000 SpellChecker * speller = theSpellChecker();
5001 // locate and enumerate the error positions
5002 int nerrors = speller->numMisspelledWords();
5004 SkipPositionsIterator it = skips.begin();
5005 SkipPositionsIterator et = skips.end();
5006 for (int index = 0; index < nerrors; ++index) {
5009 speller->misspelledWord(index, wstart, wlen);
5010 /// should not happen if speller supports range checks
5013 WordLangTuple const candidate(word.substr(wstart, wlen), lang);
5014 wstart += first + numskipped;
5015 if (snext < wstart) {
5016 /// mark the range of correct spelling
5017 numskipped += countSkips(it, et, wstart);
5018 setMisspelled(snext,
5019 wstart - 1, SpellChecker::WORD_OK);
5021 snext = wstart + wlen;
5022 // Check whether the candidate is in the document's local dict
5023 SpellChecker::Result actresult = result;
5024 if (inset_owner_->buffer().params().spellignored(candidate))
5025 actresult = SpellChecker::DOCUMENT_LEARNED_WORD;
5026 numskipped += countSkips(it, et, snext);
5027 /// mark the range of misspelling
5028 setMisspelled(wstart, snext, actresult);
5029 if (actresult == SpellChecker::DOCUMENT_LEARNED_WORD)
5030 LYXERR(Debug::GUI, "local dictionary word: \"" <<
5031 candidate.word() << "\" [" <<
5032 wstart << ".." << (snext-1) << "]");
5034 LYXERR(Debug::GUI, "misspelled word: \"" <<
5035 candidate.word() << "\" [" <<
5036 wstart << ".." << (snext-1) << "]");
5039 if (snext <= last) {
5040 /// mark the range of correct spelling at end
5041 setMisspelled(snext, last, SpellChecker::WORD_OK);
5046 void Paragraph::spellCheck() const
5048 SpellChecker * speller = theSpellChecker();
5049 if (!speller || empty() ||!needsSpellCheck())
5053 d->rangeOfSpellCheck(start, endpos);
5054 if (speller->canCheckParagraph()) {
5055 // loop until we leave the range
5056 for (pos_type first = start; first < endpos; ) {
5057 pos_type last = endpos;
5058 Private::SkipPositions skips;
5059 Language * lang = d->locateSpellRange(first, last, skips);
5060 if (first >= endpos)
5062 // start the spell checker on the unit of meaning
5063 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
5064 WordLangTuple wl = WordLangTuple(word, lang);
5065 BufferParams const & bparams = d->inset_owner_->buffer().params();
5066 SpellChecker::Result result = !word.empty() ?
5067 speller->check(wl, bparams.spellignore()) : SpellChecker::WORD_OK;
5068 d->markMisspelledWords(lang, first, last, result, word, skips);
5072 static docstring_list suggestions;
5073 pos_type to = endpos;
5074 while (start < endpos) {
5076 spellCheck(start, to, wl, suggestions, false);
5080 d->readySpellCheck();
5084 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
5086 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
5087 if (result || pos <= 0 || pos > size())
5089 if (check_boundary && (pos == size() || isWordSeparator(pos)))
5090 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
5095 string Paragraph::magicLabel() const
5098 ss << "magicparlabel-" << id();