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 // babel with Xe/LuaTeX does not need a switch
1098 // and \L is not defined there.
1099 && (!runparams.isFullUnicode() || !runparams.use_babel)
1100 && running_font.isRightToLeft()
1101 // ERT is an exception, it should be output with no
1102 // decorations at all
1103 && inset->lyxCode() != ERT_CODE) {
1104 if (runparams.use_polyglossia)
1107 else if (running_font.language()->lang() == "farsi"
1108 || running_font.language()->lang() == "arabic_arabi")
1109 os << "\\textLR{" << termcmd;
1116 if (open_font && fontswitch_inset) {
1117 bool lang_closed = false;
1118 // Close language if needed
1119 if (closeLanguage && !lang_switched_at_inset) {
1120 // We need prev_font here as language changes directly at inset
1121 // will only be started inside the inset.
1122 Font const prev_font = (i > 0) ?
1123 owner_->getFont(bparams, i - 1, outerfont)
1125 Font tmpfont(basefont);
1126 tmpfont.setLanguage(prev_font.language());
1127 bool needPar = false;
1128 unsigned int count = tmpfont.latexWriteEndChanges(os, bparams, runparams,
1130 needPar, closeLanguage);
1132 lang_closed = count > 0;
1134 // Update the running_font, making sure, however,
1135 // to leave the language as it was.
1136 // FIXME: probably a better way to keep track of the old
1137 // language, than copying the entire font?
1138 Font const copy_font(running_font);
1139 basefont = owner_->getLayoutFont(bparams, outerfont);
1140 running_font = basefont;
1142 running_font.setLanguage(copy_font.language());
1143 OutputParams rp = runparams;
1144 rp.encoding = basefont.language()->encoding();
1145 // For these, we use switches, so they should be taken as
1146 // base inside the inset.
1147 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1148 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1149 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1150 // Now re-do font changes in a way needed here
1151 // (using switches with multi-par insets)
1152 InsetText const * textinset = inset->asInsetText();
1153 bool const cprotect = textinset
1154 ? textinset->hasCProtectContent(runparams.moving_arg)
1155 && !textinset->text().isMainText()
1156 && inset->lyxCode() != BRANCH_CODE
1158 unsigned int count2 = basefont.latexWriteStartChanges(os, bparams,
1164 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1168 runparams.local_font = &basefont;
1171 if (fontswitch_inset && !closeLanguage) {
1172 // The directionality has been switched at inset.
1173 // Force markup inside.
1174 runparams.local_font = &basefont;
1177 size_t const previous_row_count = os.texrow().rows();
1180 runparams.lastid = id_;
1181 runparams.lastpos = i;
1182 inset->latex(os, runparams);
1183 } catch (EncodingException & e) {
1184 // add location information and throw again.
1193 if (os.texrow().rows() > previous_row_count) {
1194 os.texrow().start(owner_->id(), i + 1);
1197 column += (unsigned int)(os.os().tellp() - len);
1200 if (owner_->isDeleted(i))
1201 --runparams.inDeletedInset;
1205 void Paragraph::Private::latexSpecialChar(otexstream & os,
1206 BufferParams const & bparams,
1207 OutputParams const & runparams,
1208 Font const & running_font,
1209 string & alien_script,
1210 Layout const & style,
1213 unsigned int & column) const
1215 char_type const c = owner_->getUChar(bparams, runparams, i);
1217 if (style.pass_thru || runparams.pass_thru || runparams.find_effective()
1218 || contains(style.pass_thru_chars, c)
1219 || contains(runparams.pass_thru_chars, c)) {
1220 if (runparams.find_effective()) {
1226 os << "\\braceleft ";
1229 os << "\\braceright ";
1232 os << "\\lyxdollar ";
1235 os << "\\lyxtilde ";
1245 else if (c != '\0') {
1246 Encoding const * const enc = runparams.encoding;
1247 if (enc && !enc->encodable(c))
1248 throw EncodingException(c);
1254 // TIPA uses its own T3 encoding
1255 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1257 // If T1 font encoding is used, use the special
1258 // characters it provides.
1259 // NOTE: Some languages reset the font encoding internally to a
1260 // non-standard font encoding. If we are using such a language,
1261 // we do not output special T1 chars.
1262 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1263 && !runparams.isFullUnicode() && runparams.main_fontenc == "T1"
1264 && latexSpecialT1(c, os, i, column))
1266 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1267 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1268 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1271 // Otherwise, we use what LaTeX provides us.
1274 os << "\\textbackslash" << termcmd;
1278 os << "\\textless" << termcmd;
1282 os << "\\textgreater" << termcmd;
1286 os << "\\textbar" << termcmd;
1291 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1292 (end_pos == -1 || i + 1 < end_pos) &&
1293 text_[i+1] == '-') {
1294 // Prevent "--" becoming an en dash and "---" an em dash.
1295 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1301 os << "\\textquotedbl" << termcmd;
1306 case '%': case '#': case '{':
1314 os << "\\textasciitilde" << termcmd;
1319 os << "\\textasciicircum" << termcmd;
1326 // avoid being mistaken for optional arguments
1334 // Blanks are printed before font switching.
1335 // Sure? I am not! (try nice-latex)
1336 // I am sure it's correct. LyX might be smarter
1337 // in the future, but for now, nothing wrong is
1343 // XeTeX's dash behaviour is determined via a global setting
1344 if (bparams.use_dash_ligatures
1345 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1347 // TODO #10961: && not in inset Flex Code
1348 // TODO #10961: && not in layout LyXCode
1349 && (!bparams.useNonTeXFonts || runparams.flavor != Flavor::XeTeX)) {
1366 Encoding const & encoding = *(runparams.encoding);
1367 char_type next = '\0';
1368 if (i + 1 < int(text_.size())) {
1369 next = text_[i + 1];
1370 if (Encodings::isCombiningChar(next)) {
1371 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1376 pair<docstring, bool> latex = encoding.latexChar(c);
1377 docstring nextlatex;
1378 bool nexttipas = false;
1379 string nexttipashortcut;
1380 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1381 nextlatex = encoding.latexChar(next).first;
1382 if (runparams.inIPA) {
1383 nexttipashortcut = Encodings::TIPAShortcut(next);
1384 nexttipas = !nexttipashortcut.empty();
1388 if (runparams.inIPA) {
1389 string const tipashortcut = Encodings::TIPAShortcut(c);
1390 if (!tipashortcut.empty()) {
1391 latex.first = from_ascii(tipashortcut);
1392 latex.second = false;
1396 // eventually close "script wrapper" command (see `Paragraph::latex`)
1397 if (!alien_script.empty()
1398 && alien_script != Encodings::isKnownScriptChar(next)) {
1399 column += latex.first.length();
1400 alien_script.clear();
1401 os << latex.first << "}";
1405 && ((!prefixIs(nextlatex, '\\')
1406 && !prefixIs(nextlatex, '{')
1407 && !prefixIs(nextlatex, '}'))
1409 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1411 // Prevent eating of a following space or command corruption by
1412 // following characters
1413 if (next == ' ' || next == '\0') {
1414 column += latex.first.length() + 1;
1415 os << latex.first << "{}";
1417 column += latex.first.length();
1418 os << latex.first << " ";
1421 column += latex.first.length() - 1;
1429 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1430 pos_type i, unsigned int & column) const
1436 // In T1 encoding, these characters exist
1437 // but we should avoid ligatures
1438 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1440 os << "\\textcompwordmark" << termcmd;
1447 // soul.sty breaks with \char`\"
1448 os << "\\textquotedbl" << termcmd;
1457 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1458 pos_type i, unsigned int & column) const
1460 // TU encoding is currently on par with T1.
1461 return latexSpecialT1(c, os, i, column);
1465 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1466 pos_type /*i*/, unsigned int & column) const
1476 os << "\\textvertline" << termcmd;
1485 void Paragraph::Private::validate(LaTeXFeatures & features) const
1487 if (layout_->inpreamble && inset_owner_) {
1488 // FIXME: Using a string stream here circumvents the encoding
1489 // switching machinery of odocstream. Therefore the
1490 // output is wrong if this paragraph contains content
1491 // that needs to switch encoding.
1492 Buffer const & buf = inset_owner_->buffer();
1493 otexstringstream os;
1494 os << layout_->preamble();
1495 size_t const length = os.length();
1496 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1497 buf.getParFromID(owner_->id()).pit(), os,
1498 features.runparams(), string(), 0, -1, true);
1499 if (os.length() > length)
1500 features.addPreambleSnippet(os.release(), true);
1503 if (features.runparams().flavor == Flavor::Html
1504 && layout_->htmltitle()) {
1505 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1508 // check the params.
1509 if (!params_.spacing().isDefault())
1510 features.require("setspace");
1513 features.useLayout(layout_->name());
1516 fontlist_.validate(features);
1518 // then the indentation
1519 if (!params_.leftIndent().zero())
1520 features.require("ParagraphLeftIndent");
1523 InsetList::const_iterator icit = insetlist_.begin();
1524 InsetList::const_iterator iend = insetlist_.end();
1525 for (; icit != iend; ++icit) {
1527 features.inDeletedInset(owner_->isDeleted(icit->pos));
1528 if (icit->inset->lyxCode() == FOOT_CODE) {
1529 // FIXME: an item inset would make things much easier.
1530 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1531 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1532 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1533 && (icit->pos < begin_of_body_
1534 || (icit->pos == begin_of_body_
1535 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1536 features.saveNoteEnv("description");
1538 icit->inset->validate(features);
1539 features.inDeletedInset(false);
1540 if (layout_->needprotect &&
1541 icit->inset->lyxCode() == FOOT_CODE)
1542 features.require("NeedLyXFootnoteCode");
1546 // then the contents
1547 BufferParams const bp = features.runparams().is_child
1548 ? features.buffer().masterParams() : features.buffer().params();
1549 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1550 char_type c = text_[i];
1551 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1553 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1554 features.require("textquotedblp");
1555 else if (features.runparams().main_fontenc != "T1"
1556 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1557 features.require("textquotedbl");
1558 } else if (ci.textFeature() && contains(ci.textPreamble(), '=')) {
1559 // features that depend on the font or input encoding
1560 string feats = ci.textPreamble();
1561 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1562 if (fontenc.empty())
1563 fontenc = features.runparams().main_fontenc;
1564 while (!feats.empty()) {
1566 feats = split(feats, feat, ',');
1567 if (contains(feat, "!=")) {
1568 // a feature that is required except for the spcified
1569 // font or input encodings
1571 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1572 // multiple encodings are separated by semicolon
1573 vector<string> context = getVectorFromString(contexts, ";");
1574 // require feature if the context matches neither current font
1575 // nor input encoding
1576 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1577 && std::find(context.begin(), context.end(),
1578 features.runparams().encoding->name()) == context.end())
1579 features.require(realfeature);
1580 } else if (contains(feat, '=')) {
1581 // a feature that is required only for the spcified
1582 // font or input encodings
1584 string const contexts = split(feat, realfeature, '=');
1585 // multiple encodings are separated by semicolon
1586 vector<string> context = getVectorFromString(contexts, ";");
1587 // require feature if the context matches either current font
1588 // or input encoding
1589 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1590 || std::find(context.begin(), context.end(),
1591 features.runparams().encoding->name()) != context.end())
1592 features.require(realfeature);
1595 } else if (!bp.use_dash_ligatures
1596 && (c == 0x2013 || c == 0x2014)
1597 && bp.useNonTeXFonts
1598 && features.runparams().flavor == Flavor::XeTeX)
1599 // XeTeX's dash behaviour is determined via a global setting
1600 features.require("xetexdashbreakstate");
1601 BufferEncodings::validate(c, features);
1605 /////////////////////////////////////////////////////////////////////
1609 /////////////////////////////////////////////////////////////////////
1612 Layout const emptyParagraphLayout;
1615 Paragraph::Paragraph()
1616 : d(new Paragraph::Private(this, emptyParagraphLayout))
1623 Paragraph::Paragraph(Paragraph const & par)
1624 : itemdepth(par.itemdepth),
1625 d(new Paragraph::Private(*par.d, this))
1631 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1632 : itemdepth(par.itemdepth),
1633 d(new Paragraph::Private(*par.d, this, beg, end))
1639 Paragraph & Paragraph::operator=(Paragraph const & par)
1641 // needed as we will destroy the private part before copying it
1643 itemdepth = par.itemdepth;
1647 d = new Private(*par.d, this);
1654 Paragraph::~Paragraph()
1663 // this shall be called just before every "os << ..." action.
1664 void flushString(ostream & os, docstring & s)
1673 void Paragraph::write(ostream & os, BufferParams const & bparams,
1674 depth_type & depth) const
1676 // The beginning or end of a deeper (i.e. nested) area?
1677 if (depth != d->params_.depth()) {
1678 if (d->params_.depth() > depth) {
1679 while (d->params_.depth() > depth) {
1680 os << "\n\\begin_deeper";
1684 while (d->params_.depth() < depth) {
1685 os << "\n\\end_deeper";
1691 // First write the layout
1692 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1694 d->params_.write(os);
1696 Font font1(inherit_font, bparams.language);
1698 Change running_change = Change(Change::UNCHANGED);
1700 // this string is used as a buffer to avoid repetitive calls
1701 // to to_utf8(), which turn out to be expensive (JMarc)
1702 docstring write_buffer;
1705 for (pos_type i = 0; i <= size(); ++i) {
1707 Change const & change = lookupChange(i);
1708 if (change != running_change)
1709 flushString(os, write_buffer);
1710 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1711 running_change = change;
1716 // Write font changes
1717 Font font2 = getFontSettings(bparams, i);
1718 if (font2 != font1) {
1719 flushString(os, write_buffer);
1720 font2.lyxWriteChanges(font1, os);
1725 char_type const c = d->text_[i];
1728 if (Inset const * inset = getInset(i)) {
1729 flushString(os, write_buffer);
1730 if (inset->directWrite()) {
1731 // international char, let it write
1732 // code directly so it's shorter in
1738 os << "\\begin_inset ";
1740 os << "\n\\end_inset\n\n";
1743 // FIXME This can be removed again once the mystery
1744 // crash has been resolved.
1749 flushString(os, write_buffer);
1750 os << "\n\\backslash\n";
1759 case 0x061F: // ؟ U+061F ARABIC QUESTION MARK
1760 case 0x061B: // ؛ U+061B ARABIC SEMICOLON
1761 case 0x060C: // ، U+060C ARABIC COMMA
1762 flushString(os, write_buffer);
1763 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1764 os << to_utf8(docstring(1, c)) << '\n';
1767 os << to_utf8(docstring(1, c));
1769 case 0x2014: // — U+2014 EM DASH
1770 case 0x3002: // 。 U+3002 IDEOGRAPHIC FULL STOP
1771 case 0xFF01: // ! U+FF01 FULLWIDTH EXCLAMATION MARK
1772 case 0xFF1F: // ? U+FF1F FULLWIDTH QUESTION MARK
1773 case 0xFF1A: // : U+FF1A FULLWIDTH COLON
1774 case 0xFF1B: // ; U+FF1B FULLWIDTH SEMICOLON
1775 case 0xFF0C: // , U+FF0C FULLWIDTH COMMA
1776 flushString(os, write_buffer);
1777 os << to_utf8(docstring(1, c)) << '\n';
1782 flushString(os, write_buffer);
1786 // this check is to amend a bug. LyX sometimes
1787 // inserts '\0' this could cause problems.
1789 write_buffer.push_back(c);
1791 LYXERR0("NUL char in structure.");
1797 flushString(os, write_buffer);
1798 os << "\n\\end_layout\n";
1799 // FIXME This can be removed again once the mystery
1800 // crash has been resolved.
1805 void Paragraph::validate(LaTeXFeatures & features) const
1807 d->validate(features);
1808 bool fragile = features.runparams().moving_arg;
1809 fragile |= layout().needprotect;
1810 if (inInset().getLayout().isNeedProtect())
1812 if (needsCProtection(fragile))
1813 features.require("cprotect");
1817 void Paragraph::insert(pos_type pos, docstring const & str,
1818 Font const & font, Change const & change)
1820 for (size_t i = 0, n = str.size(); i != n ; ++i)
1821 insertChar(pos + i, str[i], font, change);
1825 void Paragraph::appendChar(char_type c, Font const & font,
1826 Change const & change)
1828 // Make sure that Buffer::hasChangesPresent is updated
1829 ChangesMonitor cm(*this);
1832 d->changes_.insert(change, d->text_.size());
1833 // when appending characters, no need to update tables
1834 d->text_.push_back(c);
1835 setFont(d->text_.size() - 1, font);
1836 d->requestSpellCheck(d->text_.size() - 1);
1840 void Paragraph::appendString(docstring const & s, Font const & font,
1841 Change const & change)
1843 // Make sure that Buffer::hasChangesPresent is updated
1844 ChangesMonitor cm(*this);
1846 pos_type end = s.size();
1847 size_t oldsize = d->text_.size();
1848 size_t newsize = oldsize + end;
1849 size_t capacity = d->text_.capacity();
1850 if (newsize >= capacity)
1851 d->text_.reserve(max(capacity + 100, newsize));
1853 // when appending characters, no need to update tables
1856 // FIXME: Optimize this!
1857 for (size_t i = oldsize; i != newsize; ++i) {
1859 d->changes_.insert(change, i);
1860 d->requestSpellCheck(i);
1862 d->fontlist_.set(oldsize, font);
1863 d->fontlist_.set(newsize - 1, font);
1867 void Paragraph::insertChar(pos_type pos, char_type c,
1870 d->insertChar(pos, c, Change(trackChanges ?
1871 Change::INSERTED : Change::UNCHANGED));
1875 void Paragraph::insertChar(pos_type pos, char_type c,
1876 Font const & font, bool trackChanges)
1878 d->insertChar(pos, c, Change(trackChanges ?
1879 Change::INSERTED : Change::UNCHANGED));
1884 void Paragraph::insertChar(pos_type pos, char_type c,
1885 Font const & font, Change const & change)
1887 d->insertChar(pos, c, change);
1892 void Paragraph::resetFonts(Font const & font)
1894 d->fontlist_.clear();
1895 d->fontlist_.set(0, font);
1896 d->fontlist_.set(d->text_.size() - 1, font);
1899 // Gets uninstantiated font setting at position.
1900 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1904 LYXERR0("pos: " << pos << " size: " << size());
1908 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1909 if (cit != d->fontlist_.end())
1912 if (pos == size() && !empty())
1913 return getFontSettings(bparams, pos - 1);
1915 // Optimisation: avoid a full font instantiation if there is no
1916 // language change from previous call.
1917 static Font previous_font;
1918 static Language const * previous_lang = nullptr;
1919 Language const * lang = getParLanguage(bparams);
1920 if (lang != previous_lang) {
1921 previous_lang = lang;
1922 previous_font = Font(inherit_font, lang);
1924 return previous_font;
1928 FontSpan Paragraph::fontSpan(pos_type pos) const
1930 LBUFERR(pos <= size());
1933 return FontSpan(pos, pos);
1936 FontList::const_iterator cit = d->fontlist_.begin();
1937 FontList::const_iterator end = d->fontlist_.end();
1938 for (; cit != end; ++cit) {
1939 if (cit->pos() >= pos) {
1940 if (pos >= beginOfBody())
1941 return FontSpan(max(start, beginOfBody()),
1944 return FontSpan(start,
1945 min(beginOfBody() - 1,
1948 start = cit->pos() + 1;
1951 // This should not happen, but if so, we take no chances.
1952 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1953 LASSERT(false, /**/);
1954 return FontSpan(pos, pos);
1958 // Gets uninstantiated font setting at position 0
1959 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1961 if (!empty() && !d->fontlist_.empty())
1962 return d->fontlist_.begin()->font();
1964 // Optimisation: avoid a full font instantiation if there is no
1965 // language change from previous call.
1966 static Font previous_font;
1967 static Language const * previous_lang = nullptr;
1968 if (bparams.language != previous_lang) {
1969 previous_lang = bparams.language;
1970 previous_font = Font(inherit_font, bparams.language);
1973 return previous_font;
1977 // Gets the fully instantiated font at a given position in a paragraph
1978 // This is basically the same function as Text::GetFont() in text2.cpp.
1979 // The difference is that this one is used for generating the LaTeX file,
1980 // and thus cosmetic "improvements" are disallowed: This has to deliver
1981 // the true picture of the buffer. (Asger)
1982 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1983 Font const & outerfont) const
1987 Font font = getFontSettings(bparams, pos);
1989 pos_type const body_pos = beginOfBody();
1990 FontInfo & fi = font.fontInfo();
1992 fi.realize(d->layout_->labelfont);
1994 fi.realize(d->layout_->font);
1996 fi.realize(outerfont.fontInfo());
1997 fi.realize(bparams.getFont().fontInfo());
2003 Font const Paragraph::getLabelFont
2004 (BufferParams const & bparams, Font const & outerfont) const
2006 FontInfo tmpfont = d->layout_->labelfont;
2007 tmpfont.realize(outerfont.fontInfo());
2008 tmpfont.realize(bparams.getFont().fontInfo());
2009 return Font(tmpfont, getParLanguage(bparams));
2013 Font const Paragraph::getLayoutFont
2014 (BufferParams const & bparams, Font const & outerfont) const
2016 FontInfo tmpfont = d->layout_->font;
2017 tmpfont.realize(outerfont.fontInfo());
2018 tmpfont.realize(bparams.getFont().fontInfo());
2019 return Font(tmpfont, getParLanguage(bparams));
2023 char_type Paragraph::getUChar(BufferParams const & bparams,
2024 OutputParams const & rp,
2027 char_type c = d->text_[pos];
2029 // Return unchanged character in LTR languages
2030 // or if we use poylglossia/bidi (XeTeX)
2031 // or with babel and Xe/LuaTeX.
2032 if (!getFontSettings(bparams, pos).isRightToLeft()
2033 || rp.useBidiPackage()
2034 || (rp.use_babel && rp.isFullUnicode()))
2037 // Without polyglossia/bidi, we need to account for some special cases.
2038 // FIXME This needs to be audited!
2040 // * The input is as expected for all delimiters
2041 // => checked for Hebrew!
2042 // * The output matches the display in the LyX workarea
2043 // => checked for Hebrew!
2044 // * The special cases below are really necessary
2045 // => checked for Hebrew!
2046 // * In arabic_arabi, brackets are transformed to Arabic
2047 // Ornate Parentheses. Is this is really wanted?
2049 string const & lang = getFontSettings(bparams, pos).language()->lang();
2052 // 1. In the following languages, parentheses need to be reversed.
2053 // Also with polyglodia/luabidi
2054 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
2056 // 2. In the following languages, brackets don't need to be reversed.
2057 bool const reversebrackets = lang != "arabic_arabtex"
2058 && lang != "arabic_arabi"
2061 // Now swap delimiters if needed.
2072 if (reversebrackets)
2076 if (reversebrackets)
2097 void Paragraph::setFont(pos_type pos, Font const & font)
2099 LASSERT(pos <= size(), return);
2101 // First, reduce font against layout/label font
2102 // Update: The setCharFont() routine in text2.cpp already
2103 // reduces font, so we don't need to do that here. (Asger)
2105 d->fontlist_.set(pos, font);
2109 void Paragraph::makeSameLayout(Paragraph const & par)
2111 d->layout_ = par.d->layout_;
2112 d->params_ = par.d->params_;
2116 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2118 if (isFreeSpacing())
2124 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2125 if (eraseChar(pos, trackChanges))
2131 return count > 0 || pos > 0;
2135 bool Paragraph::hasSameLayout(Paragraph const & par) const
2137 return par.d->layout_ == d->layout_
2138 && d->params_.sameLayout(par.d->params_);
2142 depth_type Paragraph::getDepth() const
2144 return d->params_.depth();
2148 depth_type Paragraph::getMaxDepthAfter() const
2150 if (d->layout_->isEnvironment())
2151 return d->params_.depth() + 1;
2153 return d->params_.depth();
2157 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2159 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2160 return getDefaultAlign(bparams);
2162 return d->params_.align();
2166 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2168 LyXAlignment res = layout().align;
2169 if (isRTL(bparams)) {
2171 if (res == LYX_ALIGN_LEFT)
2172 res = LYX_ALIGN_RIGHT;
2173 else if (res == LYX_ALIGN_RIGHT)
2174 res = LYX_ALIGN_LEFT;
2180 docstring const & Paragraph::labelString() const
2182 return d->params_.labelString();
2186 // the next two functions are for the manual labels
2187 docstring const Paragraph::getLabelWidthString() const
2189 if (d->layout_->margintype == MARGIN_MANUAL
2190 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2191 return d->params_.labelWidthString();
2193 return _("Senseless with this layout!");
2197 void Paragraph::setLabelWidthString(docstring const & s)
2199 d->params_.labelWidthString(s);
2203 docstring Paragraph::expandLabel(Layout const & layout,
2204 BufferParams const & bparams) const
2206 return expandParagraphLabel(layout, bparams, true);
2210 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2211 BufferParams const & bparams, bool process_appendix) const
2213 DocumentClass const & tclass = bparams.documentClass();
2214 string const & lang = getParLanguage(bparams)->code();
2215 bool const in_appendix = process_appendix && d->params_.appendix();
2216 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2218 if (fmt.empty() && !layout.counter.empty())
2219 return tclass.counters().theCounter(layout.counter, lang);
2221 // handle 'inherited level parts' in 'fmt',
2222 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2223 size_t const i = fmt.find('@', 0);
2224 if (i != docstring::npos) {
2225 size_t const j = fmt.find('@', i + 1);
2226 if (j != docstring::npos) {
2227 docstring parent(fmt, i + 1, j - i - 1);
2228 docstring label = from_ascii("??");
2229 if (tclass.hasLayout(parent))
2230 label = expandParagraphLabel(tclass[parent], bparams,
2232 fmt = docstring(fmt, 0, i) + label
2233 + docstring(fmt, j + 1, docstring::npos);
2237 return tclass.counters().counterLabel(fmt, lang);
2241 void Paragraph::applyLayout(Layout const & new_layout)
2243 d->layout_ = &new_layout;
2244 LyXAlignment const oldAlign = d->params_.align();
2246 if (!(oldAlign & d->layout_->alignpossible)) {
2247 frontend::Alert::warning(_("Alignment not permitted"),
2248 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2249 d->params_.align(LYX_ALIGN_LAYOUT);
2254 pos_type Paragraph::beginOfBody() const
2256 return d->begin_of_body_;
2260 void Paragraph::setBeginOfBody()
2262 if (d->layout_->labeltype != LABEL_MANUAL) {
2263 d->begin_of_body_ = 0;
2267 // Unroll the first two cycles of the loop
2268 // and remember the previous character to
2269 // remove unnecessary getChar() calls
2271 pos_type end = size();
2272 bool prev_char_deleted = false;
2273 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2276 char_type previous_char = d->text_[i];
2277 if (!(isNewline(i) || isEnvSeparator(i))) {
2279 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2280 char_type temp = d->text_[i];
2281 prev_char_deleted = isDeleted(i);
2282 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2285 previous_char = temp;
2291 d->begin_of_body_ = i;
2295 bool Paragraph::allowParagraphCustomization() const
2297 return inInset().allowParagraphCustomization();
2301 bool Paragraph::usePlainLayout() const
2303 return inInset().usePlainLayout();
2307 bool Paragraph::isPassThru() const
2309 return inInset().isPassThru() || d->layout_->pass_thru;
2313 bool Paragraph::parbreakIsNewline() const
2315 return inInset().getLayout().parbreakIsNewline() || d->layout_->parbreak_is_newline;
2319 bool Paragraph::isPartOfTextSequence() const
2321 for (pos_type i = 0; i < size(); ++i) {
2322 if (!isInset(i) || getInset(i)->isPartOfTextSequence())
2330 // paragraphs inside floats need different alignment tags to avoid
2333 bool noTrivlistCentering(InsetCode code)
2335 return code == FLOAT_CODE
2336 || code == WRAP_CODE
2337 || code == CELL_CODE;
2341 string correction(string const & orig)
2343 if (orig == "flushleft")
2344 return "raggedright";
2345 if (orig == "flushright")
2346 return "raggedleft";
2347 if (orig == "center")
2353 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2354 InsetCode code, bool const lastpar, int & col)
2356 string macro = suffix + "{";
2357 if (noTrivlistCentering(code)) {
2359 // the last paragraph in non-trivlist-aligned
2360 // context is special (to avoid unwanted whitespace)
2361 if (suffix == "\\begin") {
2362 macro = "\\" + correction(env) + "{}";
2363 os << from_ascii(macro);
2364 col += macro.size();
2369 macro += correction(env);
2373 if (suffix == "\\par\\end") {
2377 os << from_ascii(macro);
2378 col += macro.size();
2379 if (suffix == "\\begin") {
2389 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2390 otexstream & os, OutputParams const & runparams) const
2395 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2396 (layout_->toggle_indent != ITOGGLE_NEVER) :
2397 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2399 LyXAlignment const curAlign = params_.align();
2401 // Do not output \\noindent for paragraphs
2402 // 1. that cannot have indentation or are indented always,
2403 // 2. that are not part of the immediate text sequence (e.g., contain only floats),
2404 // 3. that are PassThru,
2405 // 4. or that are centered.
2406 if (canindent && params_.noindent()
2407 && owner_->isPartOfTextSequence()
2408 && !layout_->pass_thru
2409 && curAlign != LYX_ALIGN_CENTER) {
2410 if (!owner_->empty()
2411 && owner_->getInset(0)
2412 && owner_->getInset(0)->lyxCode() == VSPACE_CODE)
2413 // If the paragraph starts with a vspace, the \\noindent
2414 // needs to come after that (as it leaves vmode).
2415 // If the paragraph consists only of the vspace,
2416 // \\noindent is not needed at all.
2417 runparams.need_noindent = owner_->size() > 1;
2419 os << "\\noindent" << termcmd;
2424 if (curAlign == layout_->align)
2428 case LYX_ALIGN_NONE:
2429 case LYX_ALIGN_BLOCK:
2430 case LYX_ALIGN_LAYOUT:
2431 case LYX_ALIGN_SPECIAL:
2432 case LYX_ALIGN_DECIMAL:
2434 case LYX_ALIGN_LEFT:
2435 case LYX_ALIGN_RIGHT:
2436 case LYX_ALIGN_CENTER:
2437 if (runparams.moving_arg) {
2444 string const begin_tag = "\\begin";
2445 InsetCode code = ownerCode();
2446 bool const lastpar = runparams.isLastPar;
2447 // RTL in classic (PDF)LaTeX (without the Bidi package)
2448 // Luabibdi (used by LuaTeX) behaves like classic
2449 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2450 && !runparams.useBidiPackage();
2453 case LYX_ALIGN_NONE:
2454 case LYX_ALIGN_BLOCK:
2455 case LYX_ALIGN_LAYOUT:
2456 case LYX_ALIGN_SPECIAL:
2457 case LYX_ALIGN_DECIMAL:
2459 case LYX_ALIGN_LEFT: {
2461 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2462 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2464 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2466 } case LYX_ALIGN_RIGHT: {
2468 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2469 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2471 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2473 } case LYX_ALIGN_CENTER: {
2474 corrected_env(os, begin_tag, "center", code, lastpar, column);
2483 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2484 otexstream & os, OutputParams const & runparams) const
2486 LyXAlignment const curAlign = params_.align();
2488 if (curAlign == layout_->align)
2492 case LYX_ALIGN_NONE:
2493 case LYX_ALIGN_BLOCK:
2494 case LYX_ALIGN_LAYOUT:
2495 case LYX_ALIGN_SPECIAL:
2496 case LYX_ALIGN_DECIMAL:
2498 case LYX_ALIGN_LEFT:
2499 case LYX_ALIGN_RIGHT:
2500 case LYX_ALIGN_CENTER:
2501 if (runparams.moving_arg)
2506 bool output = false;
2508 string const end_tag = "\\par\\end";
2509 InsetCode code = ownerCode();
2510 bool const lastpar = runparams.isLastPar;
2511 // RTL in classic (PDF)LaTeX (without the Bidi package)
2512 // Luabibdi (used by LuaTeX) behaves like classic
2513 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2514 && !runparams.useBidiPackage();
2517 case LYX_ALIGN_NONE:
2518 case LYX_ALIGN_BLOCK:
2519 case LYX_ALIGN_LAYOUT:
2520 case LYX_ALIGN_SPECIAL:
2521 case LYX_ALIGN_DECIMAL:
2523 case LYX_ALIGN_LEFT: {
2525 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2526 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2528 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2530 } case LYX_ALIGN_RIGHT: {
2532 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2533 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2535 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2537 } case LYX_ALIGN_CENTER: {
2538 corrected_env(os, end_tag, "center", code, lastpar, col);
2543 return output || lastpar;
2547 // This one spits out the text of the paragraph
2548 void Paragraph::latex(BufferParams const & bparams,
2549 Font const & outerfont,
2551 OutputParams const & runparams,
2552 int start_pos, int end_pos, bool force) const
2554 LYXERR(Debug::OUTFILE, "Paragraph::latex... " << this);
2556 // FIXME This check should not be needed. Perhaps issue an
2557 // error if it triggers.
2558 Layout const & style = inInset().forcePlainLayout() ?
2559 bparams.documentClass().plainLayout() : *d->layout_;
2561 if (!force && style.inpreamble)
2564 bool const allowcust = allowParagraphCustomization();
2566 // Current base font for all inherited font changes, without any
2567 // change caused by an individual character, except for the language:
2568 // It is set to the language of the first character.
2569 // As long as we are in the label, this font is the base font of the
2570 // label. Before the first body character it is set to the base font
2574 // If there is an open font-encoding changing command (script wrapper),
2575 // alien_script is set to its name
2576 string alien_script;
2579 // Maybe we have to create a optional argument.
2580 pos_type body_pos = beginOfBody();
2581 unsigned int column = 0;
2583 // If we are inside an non inheritFont() inset,
2584 // the outerfont is the buffer's main font
2585 Font const real_outerfont =
2586 inInset().inheritFont() ? outerfont : Font(bparams.getFont());
2589 // the optional argument is kept in curly brackets in
2590 // case it contains a ']'
2591 // This is not strictly needed, but if this is changed it
2592 // would be a file format change, and tex2lyx would need
2593 // to be adjusted, since it unconditionally removes the
2594 // braces when it parses \item.
2597 basefont = getLabelFont(bparams, real_outerfont);
2599 basefont = getLayoutFont(bparams, real_outerfont);
2602 // Which font is currently active?
2603 Font running_font(basefont);
2604 // Do we have an open font change?
2605 bool open_font = false;
2607 Change runningChange =
2608 runparams.inDeletedInset && !inInset().canTrackChanges()
2609 ? runparams.changeOfDeletedInset : Change(Change::UNCHANGED);
2611 Encoding const * const prev_encoding = runparams.encoding;
2613 os.texrow().start(id(), 0);
2615 // if the paragraph is empty, the loop will not be entered at all
2617 // For InTitle commands, we have already opened a group
2618 // in output_latex::TeXOnePar.
2619 if (style.isCommand() && !style.intitle) {
2623 if (!style.leftdelim().empty()) {
2624 os << style.leftdelim();
2625 column += style.leftdelim().size();
2628 column += d->startTeXParParams(bparams, os, runparams);
2631 // Whether a \par can be issued for insets typeset inline with text.
2632 // Yes if greater than 0. This has to be static.
2633 THREAD_LOCAL_STATIC int parInline = 0;
2635 for (pos_type i = 0; i < size(); ++i) {
2636 // First char in paragraph or after label?
2637 if (i == body_pos) {
2640 bool needPar = false;
2641 column += running_font.latexWriteEndChanges(
2642 os, bparams, runparams,
2643 basefont, basefont, needPar);
2646 basefont = getLayoutFont(bparams, real_outerfont);
2647 running_font = basefont;
2649 column += Changes::latexMarkChange(os, bparams,
2650 runningChange, Change(Change::UNCHANGED),
2652 runningChange = Change(Change::UNCHANGED);
2654 os << ((isEnvSeparator(i) && !runparams.find_effective()) ? "}]~" : "}] ");
2657 // For InTitle commands, we have already opened a group
2658 // in output_latex::TeXOnePar.
2659 if (style.isCommand() && !style.intitle) {
2664 if (!style.leftdelim().empty()) {
2665 os << style.leftdelim();
2666 column += style.leftdelim().size();
2670 column += d->startTeXParParams(bparams, os,
2674 runparams.wasDisplayMath = runparams.inDisplayMath;
2675 runparams.inDisplayMath = false;
2676 bool deleted_display_math = false;
2677 Change const & change = runparams.inDeletedInset
2678 ? runparams.changeOfDeletedInset : lookupChange(i);
2680 char_type const c = d->text_[i];
2682 // Check whether a display math inset follows
2683 bool output_changes;
2684 if (!runparams.find_effective())
2685 output_changes = bparams.output_changes;
2687 output_changes = runparams.find_with_deleted();
2689 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2691 runparams.ctObject = getInset(i)->getCtObject(runparams);
2693 InsetMath const * im = getInset(i)->asInsetMath();
2694 if (im && im->asHullInset()
2695 && im->asHullInset()->outerDisplay()) {
2696 runparams.inDisplayMath = true;
2697 // runparams.inDeletedInset will be set by
2698 // latexInset later, but we need this info
2699 // before it is called. On the other hand, we
2700 // cannot set it here because it is a counter.
2701 deleted_display_math = isDeleted(i);
2703 if (output_changes && deleted_display_math
2704 && runningChange == change
2705 && change.type == Change::DELETED
2706 && !os.afterParbreak()) {
2707 // A display math in the same paragraph follows.
2708 // We have to close and then reopen \lyxdeleted,
2709 // otherwise the math will be shifted up.
2710 OutputParams rp = runparams;
2712 bool needPar = false;
2713 column += running_font.latexWriteEndChanges(
2714 os, bparams, rp, basefont,
2718 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2719 : getLayoutFont(bparams, real_outerfont);
2720 running_font = basefont;
2721 column += Changes::latexMarkChange(os, bparams,
2722 Change(Change::INSERTED), change, rp);
2726 if (output_changes && runningChange != change) {
2727 if (!alien_script.empty()) {
2730 alien_script.clear();
2733 bool needPar = false;
2734 column += running_font.latexWriteEndChanges(
2735 os, bparams, runparams,
2736 basefont, basefont, needPar);
2739 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2740 : getLayoutFont(bparams, real_outerfont);
2741 running_font = basefont;
2742 column += Changes::latexMarkChange(os, bparams, runningChange,
2744 runningChange = change;
2747 // do not output text which is marked deleted
2748 // if change tracking output is disabled
2749 if (!output_changes && change.deleted()) {
2755 // Fully instantiated font
2756 Font current_font = getFont(bparams, i, outerfont);
2758 Font const prev_font = (i > 0) ?
2759 getFont(bparams, i - 1, outerfont)
2762 Font const last_font = running_font;
2763 bool const in_ct_deletion = (output_changes
2764 && runningChange == change
2765 && change.type == Change::DELETED
2766 && !os.afterParbreak());
2767 // Insets where font switches are used (rather than font commands)
2768 bool const fontswitch_inset =
2771 && getInset(i)->allowMultiPar()
2772 && getInset(i)->lyxCode() != ERT_CODE
2773 && (getInset(i)->producesOutput()
2774 // FIXME Something more general?
2775 // Comments do not "produce output" but are still
2776 // part of the TeX source and require font switches
2777 // to be closed (otherwise LaTeX fails).
2778 || getInset(i)->layoutName() == "Note:Comment");
2780 bool closeLanguage = false;
2781 bool lang_switched_at_inset = false;
2782 if (fontswitch_inset) {
2783 // Some insets cannot be inside a font change command.
2784 // However, even such insets *can* be placed in \L or \R
2785 // or their equivalents (for RTL language switches),
2786 // so we don't close the language in those cases
2787 // (= differing isRightToLeft()).
2788 // ArabTeX, though, doesn't seem to handle this special behavior.
2789 closeLanguage = basefont.isRightToLeft() == current_font.isRightToLeft()
2790 || basefont.language()->lang() == "arabic_arabtex"
2791 || current_font.language()->lang() == "arabic_arabtex";
2792 // We need to check prev_font as language changes directly at inset
2793 // will only be started inside the inset.
2794 lang_switched_at_inset = prev_font.language() != current_font.language();
2797 // Do we need to close the previous font?
2798 bool langClosed = false;
2800 ((current_font != running_font
2801 || current_font.language() != running_font.language())
2802 || (fontswitch_inset
2803 && (current_font == prev_font))))
2805 // ensure there is no open script-wrapper
2806 if (!alien_script.empty()) {
2809 alien_script.clear();
2811 if (in_ct_deletion) {
2812 // We have to close and then reopen \lyxdeleted,
2813 // as strikeout needs to be on lowest level.
2817 if (closeLanguage) {
2818 // Force language closing
2819 current_font.setLanguage(basefont.language());
2822 Font const nextfont = (i == body_pos-1) ? basefont : current_font;
2823 bool needPar = false;
2824 column += running_font.latexWriteEndChanges(
2825 os, bparams, runparams, basefont,
2827 if (in_ct_deletion) {
2828 // We have to close and then reopen \lyxdeleted,
2829 // as strikeout needs to be on lowest level.
2830 OutputParams rp = runparams;
2831 column += Changes::latexMarkChange(os, bparams,
2832 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2834 // Has the language been closed in the latexWriteEndChanges() call above?
2835 langClosed |= running_font.language() != basefont.language()
2836 && running_font.language() != nextfont.language()
2837 && (running_font.language()->encoding()->package() != Encoding::CJK);
2838 running_font = basefont;
2839 open_font &= !langClosed;
2842 // if necessary, close language environment before opening CJK
2843 string const running_lang = running_font.language()->babel();
2844 string const lang_end_command = lyxrc.language_command_end;
2845 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2846 && !running_lang.empty()
2847 && running_lang == openLanguageName()
2848 && current_font.language()->encoding()->package() == Encoding::CJK) {
2849 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2850 os << from_ascii(end_tag);
2851 column += end_tag.length();
2852 if (!languageStackEmpty())
2856 // Switch file encoding if necessary (and allowed)
2857 if ((!fontswitch_inset || closeLanguage)
2858 && !runparams.pass_thru && !style.pass_thru &&
2859 runparams.encoding->package() != Encoding::none &&
2860 current_font.language()->encoding()->package() != Encoding::none) {
2861 pair<bool, int> const enc_switch =
2862 switchEncoding(os.os(), bparams, runparams,
2863 *(current_font.language()->encoding()));
2864 if (enc_switch.first) {
2865 column += enc_switch.second;
2866 runparams.encoding = current_font.language()->encoding();
2870 // A display math inset inside an ulem command will be output
2871 // as a box of width \linewidth, so we have to either disable
2872 // indentation if the inset starts a paragraph, or start a new
2873 // line to accommodate such box. This has to be done before
2874 // writing any font changing commands.
2875 if (runparams.inDisplayMath && !deleted_display_math
2876 && runparams.inulemcmd) {
2877 if (os.afterParbreak())
2883 // Do we need to change font?
2884 if ((current_font != running_font ||
2885 current_font.language() != running_font.language())
2886 && i != body_pos - 1)
2888 if (!fontswitch_inset) {
2889 if (in_ct_deletion) {
2890 // We have to close and then reopen \lyxdeleted,
2891 // as strikeout needs to be on lowest level.
2892 OutputParams rp = runparams;
2893 bool needPar = false;
2894 column += running_font.latexWriteEndChanges(
2895 os, bparams, rp, basefont,
2900 otexstringstream ots;
2901 InsetText const * textinset = inInset().asInsetText();
2902 bool const cprotect = textinset
2903 ? textinset->hasCProtectContent(runparams.moving_arg)
2904 && !textinset->text().isMainText()
2905 && inInset().lyxCode() != BRANCH_CODE
2907 column += current_font.latexWriteStartChanges(ots, bparams,
2908 runparams, basefont, last_font, false,
2910 // Check again for display math in ulem commands as a
2911 // font change may also occur just before a math inset.
2912 if (runparams.inDisplayMath && !deleted_display_math
2913 && runparams.inulemcmd) {
2914 if (os.afterParbreak())
2919 running_font = current_font;
2921 docstring fontchange = ots.str();
2923 // check whether the fontchange ends with a \\textcolor
2924 // modifier and the text starts with a space. If so we
2925 // need to add } in order to prevent \\textcolor from gobbling
2926 // the space (bug 4473).
2927 docstring const last_modifier = rsplit(fontchange, '\\');
2928 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2929 os << from_ascii("{}");
2930 else if (ots.terminateCommand())
2932 if (in_ct_deletion) {
2933 // We have to close and then reopen \lyxdeleted,
2934 // as strikeout needs to be on lowest level.
2935 OutputParams rp = runparams;
2936 column += Changes::latexMarkChange(os, bparams,
2937 Change(Change::UNCHANGED), change, rp);
2939 } else {// if fontswitch_inset
2940 if (current_font != running_font || !langClosed)
2941 // font is still open in fontswitch_insets if we have
2942 // a non-lang font difference or if the language
2943 // is the only difference but has not been forcedly
2946 running_font = current_font;
2950 // FIXME: think about end_pos implementation...
2951 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2952 // FIXME: integrate this case in latexSpecialChar
2953 // Do not print the separation of the optional argument
2954 // if style.pass_thru is false. This works because
2955 // latexSpecialChar ignores spaces if
2956 // style.pass_thru is false.
2957 if (i != body_pos - 1) {
2958 if (d->simpleTeXBlanks(bparams, runparams, os,
2959 i, column, current_font, style)) {
2960 // A surrogate pair was output. We
2961 // must not call latexSpecialChar
2962 // in this iteration, since it would output
2963 // the combining character again.
2970 OutputParams rp = runparams;
2971 rp.free_spacing = style.free_spacing;
2972 rp.local_font = ¤t_font;
2973 rp.intitle = style.intitle;
2975 // Two major modes: LaTeX or plain
2976 // Handle here those cases common to both modes
2977 // and then split to handle the two modes separately.
2978 if (c == META_INSET) {
2979 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2980 // Greyedout notes and, in general, all insets
2981 // with InsetLayout::isDisplay() == false,
2982 // are typeset inline with the text. So, we
2983 // can add a \par to the last paragraph of
2984 // such insets only if nothing else follows.
2985 bool incremented = false;
2986 Inset const * inset = getInset(i);
2987 InsetText const * textinset = inset
2988 ? inset->asInsetText()
2990 if (i + 1 == size() && textinset
2991 && !inset->getLayout().isDisplay()) {
2992 ParagraphList const & pars =
2993 textinset->text().paragraphs();
2994 pit_type const pit = pars.size() - 1;
2995 Font const lastfont =
2996 pit < 0 || pars[pit].empty()
2997 ? pars[pit].getLayoutFont(
3000 : pars[pit].getFont(bparams,
3001 pars[pit].size() - 1,
3003 if (lastfont.fontInfo().size() !=
3004 basefont.fontInfo().size()) {
3009 // We need to restore parts of this after insets with
3010 // allowMultiPar() true
3011 Font const save_basefont = basefont;
3012 d->latexInset(bparams, os, rp, running_font,
3013 basefont, real_outerfont, open_font,
3014 runningChange, style, i, column, fontswitch_inset,
3015 closeLanguage, (lang_switched_at_inset || langClosed));
3016 if (fontswitch_inset) {
3018 bool needPar = false;
3019 column += running_font.latexWriteEndChanges(
3020 os, bparams, runparams,
3021 basefont, basefont, needPar);
3024 basefont.fontInfo().setSize(save_basefont.fontInfo().size());
3025 basefont.fontInfo().setFamily(save_basefont.fontInfo().family());
3026 basefont.fontInfo().setSeries(save_basefont.fontInfo().series());
3031 if (runparams.ctObject == CtObject::DisplayObject
3032 || runparams.ctObject == CtObject::UDisplayObject) {
3033 // Close \lyx*deleted and force its
3034 // reopening (if needed)
3037 runningChange = Change(Change::UNCHANGED);
3038 runparams.ctObject = CtObject::Normal;
3041 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
3042 if (!bparams.useNonTeXFonts)
3043 script = Encodings::isKnownScriptChar(c);
3044 if (script != alien_script) {
3045 if (!alien_script.empty()) {
3047 alien_script.clear();
3049 string fontenc = running_font.language()->fontenc(bparams);
3051 && !Encodings::fontencSupportsScript(fontenc, script)) {
3052 column += script.length() + 2;
3053 os << "\\" << script << "{";
3054 alien_script = script;
3058 d->latexSpecialChar(os, bparams, rp, running_font,
3059 alien_script, style, i, end_pos, column);
3060 } catch (EncodingException & e) {
3061 if (runparams.dryrun) {
3062 os << "<" << _("LyX Warning: ")
3063 << _("uncodable character") << " '";
3067 // add location information and throw again.
3075 // Set the encoding to that returned from latexSpecialChar (see
3076 // comment for encoding member in OutputParams.h)
3077 runparams.encoding = rp.encoding;
3079 // Also carry on the info on a closed ulem command for insets
3080 // such as Note that do not produce any output, so that no
3081 // command is ever executed but its opening was recorded.
3082 runparams.inulemcmd = rp.inulemcmd;
3084 // These need to be passed upstream as well
3085 runparams.need_maketitle = rp.need_maketitle;
3086 runparams.have_maketitle = rp.have_maketitle;
3088 // And finally, pass the post_macros upstream
3089 runparams.post_macro = rp.post_macro;
3092 // Close wrapper for alien script
3093 if (!alien_script.empty()) {
3095 alien_script.clear();
3098 Font const font = empty()
3099 ? getLayoutFont(bparams, real_outerfont)
3100 : getFont(bparams, size() - 1, real_outerfont);
3102 InsetText const * textinset = inInset().asInsetText();
3104 bool const maintext = textinset
3105 ? textinset->text().isMainText()
3108 size_t const numpars = textinset
3109 ? textinset->text().paragraphs().size()
3112 bool needPar = false;
3114 if (style.resfont.size() != font.fontInfo().size()
3115 && (!runparams.isLastPar || maintext
3116 || (numpars > 1 && d->ownerCode() != CELL_CODE
3117 && (inInset().getLayout().isDisplay()
3119 && !style.isCommand()) {
3123 // If we have an open font definition, we have to close it
3125 // Make sure that \\par is done with the font of the last
3126 // character if this has another size as the default.
3127 // This is necessary because LaTeX (and LyX on the screen)
3128 // calculates the space between the baselines according
3129 // to this font. (Matthias)
3131 // We must not change the font for the last paragraph
3132 // of non-multipar insets, tabular cells or commands,
3133 // since this produces unwanted whitespace.
3134 #ifdef FIXED_LANGUAGE_END_DETECTION
3136 running_font.latexWriteEndChanges(os, bparams,
3137 runparams, basefont,
3138 next_->getFont(bparams, 0, outerfont),
3141 running_font.latexWriteEndChanges(os, bparams,
3142 runparams, basefont, basefont, needPar);
3145 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
3146 //FIXME: there as we start another \selectlanguage with the next paragraph if
3147 //FIXME: we are in need of this. This should be fixed sometime (Jug)
3148 running_font.latexWriteEndChanges(os, bparams, runparams,
3149 basefont, basefont, needPar);
3153 // The \par could not be inserted at the same nesting
3154 // level of the font size change, so do it now.
3155 os << "{\\" << font.latexSize() << "\\par}";
3158 if (!runparams.inDeletedInset || inInset().canTrackChanges())
3159 column += Changes::latexMarkChange(os, bparams, runningChange,
3160 Change(Change::UNCHANGED), runparams);
3162 // Needed if there is an optional argument but no contents.
3163 if (body_pos > 0 && body_pos == size()) {
3167 if (!style.rightdelim().empty()) {
3168 os << style.rightdelim();
3169 column += style.rightdelim().size();
3172 if (allowcust && d->endTeXParParams(bparams, os, runparams)
3173 && runparams.encoding != prev_encoding) {
3174 runparams.encoding = prev_encoding;
3175 os << setEncoding(prev_encoding->iconvName());
3178 LYXERR(Debug::OUTFILE, "Paragraph::latex... done " << this);
3182 bool Paragraph::emptyTag() const
3184 for (pos_type i = 0; i < size(); ++i) {
3185 if (Inset const * inset = getInset(i)) {
3186 InsetCode lyx_code = inset->lyxCode();
3187 // FIXME testing like that is wrong. What is
3189 if (lyx_code != TOC_CODE &&
3190 lyx_code != INCLUDE_CODE &&
3191 lyx_code != GRAPHICS_CODE &&
3192 lyx_code != ERT_CODE &&
3193 lyx_code != LISTINGS_CODE &&
3194 lyx_code != FLOAT_CODE &&
3195 lyx_code != TABULAR_CODE) {
3199 char_type c = d->text_[i];
3200 if (c != ' ' && c != '\t')
3208 string Paragraph::getID(Buffer const &, OutputParams const &)
3211 for (pos_type i = 0; i < size(); ++i) {
3212 if (Inset const * inset = getInset(i)) {
3213 InsetCode lyx_code = inset->lyxCode();
3214 if (lyx_code == LABEL_CODE) {
3215 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
3216 docstring const & id = il->getParam("name");
3217 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3225 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3228 for (i = 0; i < size(); ++i) {
3229 if (Inset const * inset = getInset(i)) {
3230 inset->docbook(xs, runparams);
3232 char_type c = d->text_[i];
3242 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3246 for (i = 0; i < size(); ++i) {
3247 if (Inset const * inset = getInset(i)) {
3248 inset->xhtml(xs, runparams);
3250 char_type c = d->text_[i];
3260 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3263 pos_type size = text_.size();
3264 for (pos_type i = initial; i < size; ++i) {
3265 Font font = owner_->getFont(buf.params(), i, outerfont);
3266 if (text_[i] == META_INSET)
3268 if (i != initial && font != font_old)
3279 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3280 vector<xml::EndFontTag> & tagsToClose,
3281 bool & flag, FontState curstate, xml::FontTypes type)
3283 if (curstate == FONT_ON) {
3284 tagsToOpen.push_back(docbookStartFontTag(type));
3287 tagsToClose.push_back(docbookEndFontTag(type));
3292 class OptionalFontType {
3297 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3298 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3301 OptionalFontType fontShapeToXml(FontShape fs)
3305 return {xml::FT_ITALIC};
3307 return {xml::FT_SLANTED};
3308 case SMALLCAPS_SHAPE:
3309 return {xml::FT_SMALLCAPS};
3314 // the other tags are for internal use
3320 OptionalFontType fontFamilyToXml(FontFamily fm)
3324 return {xml::FT_ROMAN};
3326 return {xml::FT_SANS};
3327 case TYPEWRITER_FAMILY:
3328 return {xml::FT_TYPE};
3329 case INHERIT_FAMILY:
3332 // the other tags are for internal use
3338 OptionalFontType fontSizeToXml(FontSize fs)
3342 return {xml::FT_SIZE_TINY};
3344 return {xml::FT_SIZE_SCRIPT};
3346 return {xml::FT_SIZE_FOOTNOTE};
3348 return {xml::FT_SIZE_SMALL};
3350 return {xml::FT_SIZE_LARGE};
3352 return {xml::FT_SIZE_LARGER};
3354 return {xml::FT_SIZE_LARGEST};
3356 return {xml::FT_SIZE_HUGE};
3358 return {xml::FT_SIZE_HUGER};
3360 return {xml::FT_SIZE_INCREASE};
3362 return {xml::FT_SIZE_DECREASE};
3367 // the other tags are for internal use
3373 struct DocBookFontState
3375 FontShape curr_fs = INHERIT_SHAPE;
3376 FontFamily curr_fam = INHERIT_FAMILY;
3377 FontSize curr_size = INHERIT_SIZE;
3379 // track whether we have opened these tags
3380 bool emph_flag = false;
3381 bool bold_flag = false;
3382 bool noun_flag = false;
3383 bool ubar_flag = false;
3384 bool dbar_flag = false;
3385 bool sout_flag = false;
3386 bool xout_flag = false;
3387 bool wave_flag = false;
3389 bool shap_flag = false;
3391 bool faml_flag = false;
3393 bool size_flag = false;
3396 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3398 std::string const & default_family,
3399 DocBookFontState & fs)
3401 vector<xml::FontTag> tagsToOpen;
3402 vector<xml::EndFontTag> tagsToClose;
3405 FontState curstate = font.fontInfo().emph();
3406 if (font_old.emph() != curstate)
3407 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3410 curstate = font.fontInfo().noun();
3411 if (font_old.noun() != curstate)
3412 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3415 curstate = font.fontInfo().underbar();
3416 if (font_old.underbar() != curstate)
3417 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3420 curstate = font.fontInfo().strikeout();
3421 if (font_old.strikeout() != curstate)
3422 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3425 curstate = font.fontInfo().xout();
3426 if (font_old.xout() != curstate)
3427 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3430 curstate = font.fontInfo().uuline();
3431 if (font_old.uuline() != curstate)
3432 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3435 curstate = font.fontInfo().uwave();
3436 if (font_old.uwave() != curstate)
3437 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3440 // a little hackish, but allows us to reuse what we have.
3441 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3442 if (font_old.series() != font.fontInfo().series())
3443 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3446 fs.curr_fs = font.fontInfo().shape();
3447 FontShape old_fs = font_old.shape();
3448 if (old_fs != fs.curr_fs) {
3450 OptionalFontType tag = fontShapeToXml(old_fs);
3452 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3453 fs.shap_flag = false;
3456 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3458 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3462 fs.curr_fam = font.fontInfo().family();
3463 FontFamily old_fam = font_old.family();
3464 if (old_fam != fs.curr_fam) {
3466 OptionalFontType tag = fontFamilyToXml(old_fam);
3468 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3469 fs.faml_flag = false;
3471 switch (fs.curr_fam) {
3473 // we will treat a "default" font family as roman, since we have
3474 // no other idea what to do.
3475 if (default_family != "rmdefault" && default_family != "default") {
3476 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3477 fs.faml_flag = true;
3481 if (default_family != "sfdefault") {
3482 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3483 fs.faml_flag = true;
3486 case TYPEWRITER_FAMILY:
3487 if (default_family != "ttdefault") {
3488 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3489 fs.faml_flag = true;
3492 case INHERIT_FAMILY:
3495 // the other tags are for internal use
3502 fs.curr_size = font.fontInfo().size();
3503 FontSize old_size = font_old.size();
3504 if (old_size != fs.curr_size) {
3506 OptionalFontType tag = fontSizeToXml(old_size);
3508 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3509 fs.size_flag = false;
3512 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3513 if (tag.has_value) {
3514 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3515 fs.size_flag = true;
3519 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3522 } // anonymous namespace
3525 std::tuple<std::vector<docstring>, std::vector<docstring>, std::vector<docstring>>
3526 Paragraph::simpleDocBookOnePar(Buffer const & buf,
3527 OutputParams const & runparams,
3528 Font const & outerfont,
3531 bool ignore_fonts) const
3533 // Return values: segregation of the content of this paragraph.
3534 std::vector<docstring> prependedParagraphs; // Anything that must be output before the main tag of this paragraph.
3535 std::vector<docstring> generatedParagraphs; // The main content of the paragraph.
3536 std::vector<docstring> appendedParagraphs; // Anything that must be output after the main tag of this paragraph.
3538 // Internal string stream to store the output before being added to one of the previous lists.
3539 odocstringstream os;
3541 // If there is an argument that must be output before the main tag, do it before handling the rest of the paragraph.
3542 // Also tag all arguments that shouldn't go in the main content right now, so that they are never generated at the
3544 OutputParams rp = runparams;
3545 for (pos_type i = initial; i < size(); ++i) {
3546 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3547 const InsetArgument * arg = getInset(i)->asInsetArgument();
3548 if (arg->docbookargumentbeforemaintag()) {
3549 auto xs_local = XMLStream(os);
3550 arg->docbook(xs_local, rp);
3552 prependedParagraphs.push_back(os.str());
3553 os.str(from_ascii(""));
3555 rp.docbook_prepended_arguments.insert(arg);
3556 } else if (arg->docbookargumentaftermaintag()) {
3557 rp.docbook_appended_arguments.insert(arg);
3562 // State variables for the main loop.
3563 auto xs = new XMLStream(os); // XMLStream has no copy constructor: to create a new object, the only solution
3564 // is to hold a pointer to the XMLStream (xs = XMLStream(os) is not allowed once the first object is built).
3565 std::vector<docstring> delayedChars; // When a font tag ends with a space, output it after the closing font tag.
3566 // This requires to store delayed characters at some point.
3568 // Track whether we have opened font tags
3569 DocBookFontState fs;
3570 DocBookFontState old_fs = fs;
3572 Layout const & style = *d->layout_;
3574 // Conversion of the font opening/closing into DocBook tags.
3575 vector<xml::FontTag> tagsToOpen;
3576 vector<xml::EndFontTag> tagsToClose;
3578 // Parsing main loop.
3579 for (pos_type i = initial; i < size(); ++i) {
3580 bool ignore_fonts_i = ignore_fonts
3581 || style.docbooknofontinside()
3582 || (getInset(i) && getInset(i)->getLayout().docbooknofontinside());
3584 // Don't show deleted material in the output.
3588 // If this is an InsetNewline, generate a new paragraph. Also reset the fonts, so that tags are closed in
3590 if (getInset(i) && getInset(i)->lyxCode() == NEWLINE_CODE) {
3591 if (!ignore_fonts_i)
3592 xs->closeFontTags();
3594 // Output one paragraph (i.e. one string entry in generatedParagraphs).
3595 generatedParagraphs.push_back(os.str());
3597 // Create a new XMLStream for the new paragraph, completely independent of the previous one. This implies
3598 // that the string stream must be reset.
3599 os.str(from_ascii(""));
3601 xs = new XMLStream(os);
3603 // Restore the fonts for the new paragraph, so that the right tags are opened for the new entry.
3604 if (!ignore_fonts_i) {
3609 // Determine which tags should be opened or closed regarding fonts.
3610 FontInfo const font_old = (i == 0 ?
3611 (style.labeltype == LABEL_MANUAL ? style.labelfont : style.font) :
3612 getFont(buf.masterBuffer()->params(), i - 1, outerfont).fontInfo());
3613 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3614 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(
3615 font_old, font, buf.masterBuffer()->params().fonts_default_family, fs);
3617 if (!ignore_fonts_i) {
3618 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3619 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3620 for (; cit != cen; ++cit)
3624 // Deal with the delayed characters *after* closing font tags.
3625 if (!delayedChars.empty()) {
3626 for (const docstring& c: delayedChars)
3627 *xs << XMLStream::ESCAPE_NONE << c;
3628 delayedChars.clear();
3631 if (!ignore_fonts_i) {
3632 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3633 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3634 for (; sit != sen; ++sit)
3637 tagsToClose.clear();
3641 // Finally, write the next character or inset.
3642 if (Inset const * inset = getInset(i)) {
3643 bool inset_is_argument_elsewhere = getInset(i)->asInsetArgument() &&
3644 rp.docbook_appended_arguments.find(inset->asInsetArgument()) != rp.docbook_appended_arguments.end() &&
3645 rp.docbook_prepended_arguments.find(inset->asInsetArgument()) != rp.docbook_prepended_arguments.end();
3647 if ((!rp.for_toc || inset->isInToc()) && !inset_is_argument_elsewhere) {
3648 // Arguments may need to be output
3649 OutputParams np = rp;
3650 np.local_font = &font;
3652 // TODO: special case will bite here.
3653 np.docbook_in_par = true;
3654 inset->docbook(*xs, np);
3657 char_type c = getUChar(buf.masterBuffer()->params(), rp, i);
3658 if (lyx::isSpace(c) && !ignore_fonts) { // Delay spaces *after* the font-tag closure for cleaner output.
3659 if (c == ' ' && (style.free_spacing || rp.free_spacing)) {
3660 delayedChars.push_back(from_ascii(" "));
3662 delayedChars.emplace_back(1, c);
3664 } else { // No need to delay the character.
3665 if (c == '\'' && !ignore_fonts)
3666 *xs << XMLStream::ESCAPE_NONE << "’";
3673 // FIXME, this code is just imported from XHTML
3674 // I'm worried about what happens if a branch, say, is itself
3675 // wrapped in some font stuff. I think that will not work.
3677 xs->closeFontTags();
3679 // Deal with the delayed characters *after* closing font tags.
3680 if (!delayedChars.empty()) {
3681 for (const docstring &c: delayedChars)
3682 *xs << XMLStream::ESCAPE_NONE << c;
3683 delayedChars.clear();
3686 // In listings, new lines (i.e. \n characters in the output) are very important. Avoid generating one for the
3687 // last line to get a clean output.
3688 if (rp.docbook_in_listing && !is_last_par)
3691 // Finalise the last (and most likely only) paragraph.
3692 generatedParagraphs.push_back(os.str());
3693 os.str(from_ascii(""));
3696 // If there is an argument that must be output after the main tag, do it after handling the rest of the paragraph.
3697 for (pos_type i = initial; i < size(); ++i) {
3698 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3699 const InsetArgument * arg = getInset(i)->asInsetArgument();
3700 if (arg->docbookargumentaftermaintag()) {
3701 // Don't use rp, as this argument would not generate anything.
3702 auto xs_local = XMLStream(os);
3703 arg->docbook(xs_local, runparams);
3705 appendedParagraphs.push_back(os.str());
3706 os.str(from_ascii(""));
3711 return std::make_tuple(prependedParagraphs, generatedParagraphs, appendedParagraphs);
3717 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3718 vector<xml::EndFontTag> & tagsToClose,
3719 bool & flag, FontState curstate, xml::FontTypes type)
3721 if (curstate == FONT_ON) {
3722 tagsToOpen.push_back(xhtmlStartFontTag(type));
3725 tagsToClose.push_back(xhtmlEndFontTag(type));
3730 } // anonymous namespace
3733 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3735 OutputParams const & runparams,
3736 Font const & outerfont,
3737 bool start_paragraph, bool close_paragraph,
3738 pos_type initial) const
3742 // track whether we have opened these tags
3743 bool emph_flag = false;
3744 bool bold_flag = false;
3745 bool noun_flag = false;
3746 bool ubar_flag = false;
3747 bool dbar_flag = false;
3748 bool sout_flag = false;
3749 bool xout_flag = false;
3750 bool wave_flag = false;
3752 bool shap_flag = false;
3754 bool faml_flag = false;
3756 bool size_flag = false;
3758 Layout const & style = *d->layout_;
3760 if (start_paragraph)
3761 xs.startDivision(allowEmpty());
3764 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3766 FontShape curr_fs = INHERIT_SHAPE;
3767 FontFamily curr_fam = INHERIT_FAMILY;
3768 FontSize curr_size = INHERIT_SIZE;
3770 string const default_family =
3771 buf.masterBuffer()->params().fonts_default_family;
3773 vector<xml::FontTag> tagsToOpen;
3774 vector<xml::EndFontTag> tagsToClose;
3776 // parsing main loop
3777 for (pos_type i = initial; i < size(); ++i) {
3778 // let's not show deleted material in the output
3782 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3785 FontState curstate = font.fontInfo().emph();
3786 if (font_old.emph() != curstate)
3787 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3790 curstate = font.fontInfo().noun();
3791 if (font_old.noun() != curstate)
3792 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3795 curstate = font.fontInfo().underbar();
3796 if (font_old.underbar() != curstate)
3797 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3800 curstate = font.fontInfo().strikeout();
3801 if (font_old.strikeout() != curstate)
3802 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3805 curstate = font.fontInfo().xout();
3806 if (font_old.xout() != curstate)
3807 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3810 curstate = font.fontInfo().uuline();
3811 if (font_old.uuline() != curstate)
3812 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3815 curstate = font.fontInfo().uwave();
3816 if (font_old.uwave() != curstate)
3817 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3820 // a little hackish, but allows us to reuse what we have.
3821 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3822 if (font_old.series() != font.fontInfo().series())
3823 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3826 curr_fs = font.fontInfo().shape();
3827 FontShape old_fs = font_old.shape();
3828 if (old_fs != curr_fs) {
3832 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3835 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3837 case SMALLCAPS_SHAPE:
3838 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3844 // the other tags are for internal use
3852 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3856 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3859 case SMALLCAPS_SHAPE:
3860 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3867 // the other tags are for internal use
3874 curr_fam = font.fontInfo().family();
3875 FontFamily old_fam = font_old.family();
3876 if (old_fam != curr_fam) {
3880 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3883 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3885 case TYPEWRITER_FAMILY:
3886 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3888 case INHERIT_FAMILY:
3891 // the other tags are for internal use
3899 // we will treat a "default" font family as roman, since we have
3900 // no other idea what to do.
3901 if (default_family != "rmdefault" && default_family != "default") {
3902 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3907 if (default_family != "sfdefault") {
3908 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3912 case TYPEWRITER_FAMILY:
3913 if (default_family != "ttdefault") {
3914 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3918 case INHERIT_FAMILY:
3921 // the other tags are for internal use
3928 curr_size = font.fontInfo().size();
3929 FontSize old_size = font_old.size();
3930 if (old_size != curr_size) {
3934 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3937 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3940 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3943 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3946 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3949 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3952 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3955 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3958 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3961 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3964 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3970 // the other tags are for internal use
3976 switch (curr_size) {
3978 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3982 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3986 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3990 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3994 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3998 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
4002 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
4006 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
4010 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
4014 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
4018 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
4025 // the other tags are for internal use
4032 // Other such tags? What about the other text ranges?
4034 for (auto const & t : tagsToClose)
4036 for (auto const & t : tagsToOpen)
4038 tagsToClose.clear();
4041 Inset const * inset = getInset(i);
4043 if (!runparams.for_toc || inset->isInToc()) {
4044 OutputParams np = runparams;
4045 np.local_font = &font;
4046 // If the paragraph has size 1, then we are in the "special
4047 // case" where we do not output the containing paragraph info
4048 if (!inset->getLayout().htmlisblock() && size() != 1)
4049 np.html_in_par = true;
4050 retval += inset->xhtml(xs, np);
4053 char_type c = getUChar(buf.masterBuffer()->params(),
4055 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
4056 xs << XMLStream::ESCAPE_NONE << " ";
4058 xs << XMLStream::ESCAPE_NONE << "’";
4062 font_old = font.fontInfo();
4066 // I'm worried about what happens if a branch, say, is itself
4067 // wrapped in some font stuff. I think that will not work.
4069 if (close_paragraph)
4076 bool Paragraph::isHfill(pos_type pos) const
4078 Inset const * inset = getInset(pos);
4079 return inset && inset->isHfill();
4083 bool Paragraph::isNewline(pos_type pos) const
4085 // U+2028 LINE SEPARATOR
4086 // U+2029 PARAGRAPH SEPARATOR
4087 char_type const c = d->text_[pos];
4088 if (c == 0x2028 || c == 0x2029)
4090 Inset const * inset = getInset(pos);
4091 return inset && inset->lyxCode() == NEWLINE_CODE;
4095 bool Paragraph::isEnvSeparator(pos_type pos) const
4097 Inset const * inset = getInset(pos);
4098 return inset && inset->lyxCode() == SEPARATOR_CODE;
4102 bool Paragraph::isLineSeparator(pos_type pos) const
4104 char_type const c = d->text_[pos];
4105 if (isLineSeparatorChar(c))
4107 Inset const * inset = getInset(pos);
4108 return inset && inset->isLineSeparator();
4112 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
4116 if (ignore_deleted && isDeleted(pos))
4118 if (Inset const * inset = getInset(pos))
4119 return !inset->isLetter();
4120 // if we have a hard hyphen (no en- or emdash) or apostrophe
4121 // we pass this to the spell checker
4122 // FIXME: this method is subject to change, visit
4123 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
4124 // to get an impression how complex this is.
4125 if (isHardHyphenOrApostrophe(pos))
4127 char_type const c = d->text_[pos];
4128 // We want to pass the escape chars to the spellchecker
4129 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
4130 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
4134 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
4136 pos_type const psize = size();
4139 char_type const c = d->text_[pos];
4140 if (c != '-' && c != '\'')
4142 pos_type nextpos = pos + 1;
4143 pos_type prevpos = pos > 0 ? pos - 1 : 0;
4144 if ((nextpos == psize || isSpace(nextpos))
4145 && (pos == 0 || isSpace(prevpos)))
4151 bool Paragraph::needsCProtection(bool const fragile) const
4153 // first check the layout of the paragraph, but only in insets
4154 InsetText const * textinset = inInset().asInsetText();
4155 bool const maintext = textinset
4156 ? textinset->text().isMainText()
4159 if (!maintext && layout().needcprotect) {
4160 // Environments need cprotection regardless the content
4161 if (layout().latextype == LATEX_ENVIRONMENT)
4164 // Commands need cprotection if they contain specific chars
4165 int const nchars_escape = 9;
4166 static char_type const chars_escape[nchars_escape] = {
4167 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
4169 docstring const pars = asString();
4170 for (int k = 0; k < nchars_escape; k++) {
4171 if (contains(pars, chars_escape[k]))
4176 // now check whether we have insets that need cprotection
4177 for (auto const & icit : d->insetlist_) {
4178 Inset const * ins = icit.inset;
4181 if (ins->needsCProtection(maintext, fragile))
4189 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
4191 return d->speller_state_.getRange(pos);
4195 bool Paragraph::isChar(pos_type pos) const
4197 if (Inset const * inset = getInset(pos))
4198 return inset->isChar();
4199 char_type const c = d->text_[pos];
4200 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
4204 bool Paragraph::isSpace(pos_type pos) const
4206 if (Inset const * inset = getInset(pos))
4207 return inset->isSpace();
4208 char_type const c = d->text_[pos];
4209 return lyx::isSpace(c);
4214 Paragraph::getParLanguage(BufferParams const & bparams) const
4217 return getFirstFontSettings(bparams).language();
4218 // FIXME: we should check the prev par as well (Lgb)
4219 return bparams.language;
4223 bool Paragraph::isRTL(BufferParams const & bparams) const
4225 return getParLanguage(bparams)->rightToLeft()
4226 && !inInset().getLayout().forceLTR();
4230 void Paragraph::changeLanguage(BufferParams const & bparams,
4231 Language const * from, Language const * to)
4233 // change language including dummy font change at the end
4234 for (pos_type i = 0; i <= size(); ++i) {
4235 Font font = getFontSettings(bparams, i);
4236 if (font.language() == from) {
4237 font.setLanguage(to);
4239 d->requestSpellCheck(i);
4245 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
4247 Language const * doc_language = bparams.language;
4248 for (auto const & f : d->fontlist_)
4249 if (f.font().language() != ignore_language &&
4250 f.font().language() != latex_language &&
4251 f.font().language() != doc_language)
4257 void Paragraph::getLanguages(std::set<Language const *> & langs) const
4259 for (auto const & f : d->fontlist_) {
4260 Language const * lang = f.font().language();
4261 if (lang != ignore_language &&
4262 lang != latex_language)
4268 docstring Paragraph::asString(int options) const
4270 return asString(0, size(), options);
4274 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
4276 odocstringstream os;
4279 && options & AS_STR_LABEL
4280 && !d->params_.labelString().empty())
4281 os << d->params_.labelString() << ' ';
4283 for (pos_type i = beg; i < end; ++i) {
4284 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
4286 char_type const c = d->text_[i];
4287 if (isPrintable(c) || c == '\t'
4288 || (c == '\n' && (options & AS_STR_NEWLINES)))
4290 else if (c == META_INSET && (options & AS_STR_INSETS)) {
4291 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
4292 LASSERT(runparams != nullptr, return docstring());
4293 if (runparams->find_effective())
4294 getInset(i)->toString(os);
4296 getInset(i)->plaintext(os, *runparams);
4297 } else if (c == META_INSET && (options & AS_STR_MATHED)
4298 && getInset(i)->lyxCode() == REF_CODE) {
4299 Buffer const & buf = getInset(i)->buffer();
4300 OutputParams rp(&buf.params().encoding());
4301 Font const font(inherit_font, buf.params().language);
4302 rp.local_font = &font;
4304 getInset(i)->latex(ots, rp);
4306 getInset(i)->toString(os);
4315 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
4316 bool const shorten, bool const label) const
4318 size_t tmplen = shorten ? maxlen + 1 : maxlen;
4319 if (label && !labelString().empty())
4320 os += labelString() + ' ';
4321 if (!layout().isTocCaption())
4323 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4326 char_type const c = d->text_[i];
4329 else if (c == META_INSET)
4330 getInset(i)->forOutliner(os, tmplen, false);
4333 Text::shortenForOutliner(os, maxlen);
4337 void Paragraph::setInsetOwner(Inset const * inset)
4339 d->inset_owner_ = inset;
4343 int Paragraph::id() const
4349 void Paragraph::setId(int id)
4355 Layout const & Paragraph::layout() const
4361 void Paragraph::setLayout(Layout const & layout)
4363 d->layout_ = &layout;
4367 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4369 setLayout(tc.defaultLayout());
4373 void Paragraph::setPlainLayout(DocumentClass const & tc)
4375 setLayout(tc.plainLayout());
4379 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tc)
4381 if (usePlainLayout())
4384 setDefaultLayout(tc);
4388 Inset const & Paragraph::inInset() const
4390 LBUFERR(d->inset_owner_);
4391 return *d->inset_owner_;
4395 ParagraphParameters & Paragraph::params()
4401 ParagraphParameters const & Paragraph::params() const
4407 bool Paragraph::isFreeSpacing() const
4409 if (d->layout_->free_spacing)
4411 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4415 bool Paragraph::allowEmpty() const
4417 if (d->layout_->keepempty)
4419 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4423 bool Paragraph::brokenBiblio() const
4425 // There is a problem if there is no bibitem at position 0 in
4426 // paragraphs that need one, if there is another bibitem in the
4427 // paragraph or if this paragraph is not supposed to have
4428 // a bibitem inset at all.
4429 return ((d->layout_->labeltype == LABEL_BIBLIO
4430 && (d->insetlist_.find(BIBITEM_CODE) != 0
4431 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4432 || (d->layout_->labeltype != LABEL_BIBLIO
4433 && d->insetlist_.find(BIBITEM_CODE) != -1));
4437 int Paragraph::fixBiblio(Buffer const & buffer)
4439 // FIXME: when there was already an inset at 0, the return value is 1,
4440 // which does not tell whether another inset has been removed; the
4441 // cursor cannot be correctly updated.
4443 bool const track_changes = buffer.params().track_changes;
4444 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4446 // The case where paragraph is not BIBLIO
4447 if (d->layout_->labeltype != LABEL_BIBLIO) {
4448 if (bibitem_pos == -1)
4449 // No InsetBibitem => OK
4451 // There is an InsetBibitem: remove it!
4452 d->insetlist_.release(bibitem_pos);
4453 eraseChar(bibitem_pos, track_changes);
4454 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4457 bool const hasbibitem0 = bibitem_pos == 0;
4459 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4460 // There was an InsetBibitem at pos 0,
4461 // and no other one => OK
4462 if (bibitem_pos == -1)
4464 // there is a bibitem at the 0 position, but since
4465 // there is a second one, we copy the second on the
4466 // first. We're assuming there are at most two of
4467 // these, which there should be.
4468 // FIXME: why does it make sense to do that rather
4469 // than keep the first? (JMarc)
4470 Inset * inset = releaseInset(bibitem_pos);
4471 d->insetlist_.begin()->inset = inset;
4472 // This needs to be done to update the counter (#8499)
4473 buffer.updateBuffer();
4474 return -bibitem_pos;
4477 // We need to create an inset at the beginning
4478 Inset * inset = nullptr;
4479 if (bibitem_pos > 0) {
4480 // there was one somewhere in the paragraph, let's move it
4481 inset = d->insetlist_.release(bibitem_pos);
4482 eraseChar(bibitem_pos, track_changes);
4485 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4486 InsetCommandParams(BIBITEM_CODE));
4488 Font font(inherit_font, buffer.params().language);
4489 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4490 : Change::UNCHANGED));
4492 // This is needed to get the counters right
4493 buffer.updateBuffer();
4498 void Paragraph::checkAuthors(AuthorList const & authorList)
4500 d->changes_.checkAuthors(authorList);
4504 bool Paragraph::isChanged(pos_type pos) const
4506 return lookupChange(pos).changed();
4510 bool Paragraph::isInserted(pos_type pos) const
4512 return lookupChange(pos).inserted();
4516 bool Paragraph::isDeleted(pos_type pos) const
4518 return lookupChange(pos).deleted();
4522 InsetList const & Paragraph::insetList() const
4524 return d->insetlist_;
4528 void Paragraph::setInsetBuffers(Buffer & b)
4530 d->insetlist_.setBuffer(b);
4534 void Paragraph::resetBuffer()
4536 d->insetlist_.resetBuffer();
4540 Inset * Paragraph::releaseInset(pos_type pos)
4542 Inset * inset = d->insetlist_.release(pos);
4543 /// does not honour change tracking!
4544 eraseChar(pos, false);
4549 Inset * Paragraph::getInset(pos_type pos)
4551 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4552 ? d->insetlist_.get(pos) : nullptr;
4556 Inset const * Paragraph::getInset(pos_type pos) const
4558 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4559 ? d->insetlist_.get(pos) : nullptr;
4563 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4564 pos_type & right, TextCase action)
4566 // process sequences of modified characters; in change
4567 // tracking mode, this approach results in much better
4568 // usability than changing case on a char-by-char basis
4569 // We also need to track the current font, since font
4570 // changes within sequences can occur.
4571 vector<pair<char_type, Font> > changes;
4573 bool const trackChanges = bparams.track_changes;
4575 bool capitalize = true;
4577 for (; pos < right; ++pos) {
4578 char_type oldChar = d->text_[pos];
4579 char_type newChar = oldChar;
4581 // ignore insets and don't play with deleted text!
4582 if (oldChar != META_INSET && !isDeleted(pos)) {
4584 case text_lowercase:
4585 newChar = lowercase(oldChar);
4587 case text_capitalization:
4589 newChar = uppercase(oldChar);
4593 case text_uppercase:
4594 newChar = uppercase(oldChar);
4599 if (isWordSeparator(pos) || isDeleted(pos)) {
4600 // permit capitalization again
4604 if (oldChar != newChar) {
4605 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4606 if (pos != right - 1)
4608 // step behind the changing area
4612 int erasePos = pos - changes.size();
4613 for (auto const & change : changes) {
4614 insertChar(pos, change.first, change.second, trackChanges);
4615 if (!eraseChar(erasePos, trackChanges)) {
4618 ++right; // expand selection
4625 int Paragraph::find(docstring const & str, bool cs, bool mw,
4626 pos_type start_pos, bool del) const
4628 pos_type pos = start_pos;
4629 int const strsize = str.length();
4631 pos_type const parsize = d->text_.size();
4632 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4633 // ignore deleted matter
4634 if (!del && isDeleted(pos)) {
4635 if (pos == parsize - 1)
4641 // Ignore "invisible" letters such as ligature breaks
4642 // and hyphenation chars while searching
4643 bool nonmatch = false;
4644 while (pos < parsize && isInset(pos)) {
4645 Inset const * inset = getInset(pos);
4646 if (!inset->isLetter() && !inset->isChar())
4648 odocstringstream os;
4649 if (inset->lyxCode() == lyx::QUOTE_CODE) {
4651 op.find_set_feature(OutputParams::SearchQuick);
4652 inset->plaintext(os, op);
4655 inset->toString(os);
4657 docstring const insetstring = os.str();
4658 if (!insetstring.empty()) {
4659 int const insetstringsize = insetstring.length();
4660 for (int j = 0; j < insetstringsize && pos < parsize; ++i, ++j) {
4661 if ((cs && str[i] != insetstring[j])
4662 || (!cs && uppercase(str[i]) != uppercase(insetstring[j]))) {
4670 if (nonmatch || i == strsize)
4672 char_type dp = d->text_[pos];
4673 if (cs && str[i] != dp)
4675 if (!cs && uppercase(str[i]) != uppercase(dp))
4682 // if necessary, check whether string matches word
4684 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4687 && !isWordSeparator(pos))
4691 return pos - start_pos;
4695 char_type Paragraph::getChar(pos_type pos) const
4697 return d->text_[pos];
4701 pos_type Paragraph::size() const
4703 return d->text_.size();
4707 bool Paragraph::empty() const
4709 return d->text_.empty();
4713 bool Paragraph::isInset(pos_type pos) const
4715 return d->text_[pos] == META_INSET;
4719 bool Paragraph::isSeparator(pos_type pos) const
4721 //FIXME: Are we sure this can be the only separator?
4722 return d->text_[pos] == ' ';
4726 void Paragraph::deregisterWords()
4728 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4729 Private::LangWordsMap::const_iterator ite = d->words_.end();
4730 for (; itl != ite; ++itl) {
4731 WordList & wl = theWordList(itl->first);
4732 Private::Words::const_iterator it = (itl->second).begin();
4733 Private::Words::const_iterator et = (itl->second).end();
4734 for (; it != et; ++it)
4741 void Paragraph::locateWord(pos_type & from, pos_type & to,
4742 word_location const loc, bool const ignore_deleted) const
4745 case WHOLE_WORD_STRICT:
4746 if (from == 0 || from == size()
4747 || isWordSeparator(from, ignore_deleted)
4748 || isWordSeparator(from - 1, ignore_deleted)) {
4755 // If we are already at the beginning of a word, do nothing
4756 if (!from || isWordSeparator(from - 1, ignore_deleted))
4761 // always move the cursor to the beginning of previous word
4762 while (from && !isWordSeparator(from - 1, ignore_deleted))
4766 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4769 // no need to move the 'from' cursor
4773 while (to < size() && !isWordSeparator(to, ignore_deleted))
4778 void Paragraph::collectWords()
4780 for (pos_type pos = 0; pos < size(); ++pos) {
4781 if (isWordSeparator(pos))
4783 pos_type from = pos;
4784 locateWord(from, pos, WHOLE_WORD);
4785 // Work around MSVC warning: The statement
4786 // if (pos < from + lyxrc.completion_minlength)
4787 // triggers a signed vs. unsigned warning.
4788 // I don't know why this happens, it could be a MSVC bug, or
4789 // related to LLP64 (windows) vs. LP64 (unix) programming
4790 // model, or the C++ standard might be ambigous in the section
4791 // defining the "usual arithmetic conversions". However, using
4792 // a temporary variable is safe and works on all compilers.
4793 pos_type const endpos = from + lyxrc.completion_minlength;
4796 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4797 if (cit == d->fontlist_.end())
4799 Language const * lang = cit->font().language();
4800 docstring const word = asString(from, pos, AS_STR_NONE);
4801 d->words_[lang->lang()].insert(word);
4806 void Paragraph::registerWords()
4808 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4809 Private::LangWordsMap::const_iterator ite = d->words_.end();
4810 for (; itl != ite; ++itl) {
4811 WordList & wl = theWordList(itl->first);
4812 Private::Words::const_iterator it = (itl->second).begin();
4813 Private::Words::const_iterator et = (itl->second).end();
4814 for (; it != et; ++it)
4820 void Paragraph::updateWords()
4828 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4830 SkipPositionsIterator begin = skips.begin();
4831 SkipPositions::iterator end = skips.end();
4832 if (pos > 0 && begin < end) {
4834 if (end->last == pos - 1) {
4839 skips.insert(end, FontSpan(pos, pos));
4843 Language * Paragraph::Private::locateSpellRange(
4844 pos_type & from, pos_type & to,
4845 SkipPositions & skips) const
4847 // skip leading white space
4848 while (from < to && owner_->isWordSeparator(from))
4850 // don't check empty range
4853 // get current language
4854 Language * lang = getSpellLanguage(from);
4855 pos_type last = from;
4856 bool samelang = true;
4857 bool sameinset = true;
4858 while (last < to && samelang && sameinset) {
4859 // hop to end of word
4860 while (last < to && !owner_->isWordSeparator(last)) {
4861 Inset const * inset = owner_->getInset(last);
4862 if (inset && dynamic_cast<const InsetSpecialChar *>(inset)) {
4863 // check for "invisible" letters such as ligature breaks
4864 odocstringstream os;
4865 inset->toString(os);
4866 if (os.str().length() != 0) {
4867 // avoid spell check of visible special char insets
4868 // stop the loop in front of the special char inset
4873 appendSkipPosition(skips, last);
4874 } else if (owner_->isDeleted(last)) {
4875 appendSkipPosition(skips, last);
4879 // hop to next word while checking for insets
4880 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4881 if (Inset const * inset = owner_->getInset(last))
4882 sameinset = inset->isChar() && inset->isLetter();
4883 if (sameinset && owner_->isDeleted(last)) {
4884 appendSkipPosition(skips, last);
4889 if (sameinset && last < to) {
4890 // now check for language change
4891 samelang = lang == getSpellLanguage(last);
4894 // if language change detected backstep is needed
4902 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4905 const_cast<Language *>(owner_->getFontSettings(
4906 inset_owner_->buffer().params(), from).language());
4907 if (lang == inset_owner_->buffer().params().language
4908 && !lyxrc.spellchecker_alt_lang.empty()) {
4910 string const lang_variety =
4911 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4912 lang->setCode(lang_code);
4913 lang->setVariety(lang_variety);
4919 void Paragraph::requestSpellCheck(pos_type pos)
4921 d->requestSpellCheck(pos);
4923 // Also request spellcheck within (text) insets
4924 for (auto const & insets : insetList()) {
4925 if (!insets.inset->asInsetText())
4927 ParagraphList & inset_pars =
4928 insets.inset->asInsetText()->paragraphs();
4929 ParagraphList::iterator pit = inset_pars.begin();
4930 ParagraphList::iterator pend = inset_pars.end();
4931 for (; pit != pend; ++pit)
4932 pit->requestSpellCheck();
4938 bool Paragraph::needsSpellCheck() const
4940 SpellChecker::ChangeNumber speller_change_number = 0;
4941 if (theSpellChecker())
4942 speller_change_number = theSpellChecker()->changeNumber();
4943 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4944 d->speller_state_.needsCompleteRefresh(speller_change_number);
4946 return d->needsSpellCheck();
4950 bool Paragraph::Private::ignoreWord(docstring const & word) const
4952 // Ignore words with digits
4953 // FIXME: make this customizable
4954 // (note that some checkers ignore words with digits by default)
4955 docstring::const_iterator cit = word.begin();
4956 docstring::const_iterator const end = word.end();
4957 for (; cit != end; ++cit) {
4958 if (isNumber((*cit)))
4965 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4966 WordLangTuple & wl, docstring_list & suggestions,
4967 bool do_suggestion, bool check_learned) const
4969 SpellChecker::Result result = SpellChecker::WORD_OK;
4970 SpellChecker * speller = theSpellChecker();
4974 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4977 locateWord(from, to, WHOLE_WORD, true);
4978 if (from == to || from >= size())
4981 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4982 Language * lang = d->getSpellLanguage(from);
4984 BufferParams const & bparams = d->inset_owner_->buffer().params();
4986 if (getFontSettings(bparams, from).fontInfo().nospellcheck() == FONT_ON)
4989 wl = WordLangTuple(word, lang);
4994 if (needsSpellCheck() || check_learned) {
4996 if (!d->ignoreWord(word)) {
4997 bool const trailing_dot = to < size() && d->text_[to] == '.';
4998 result = speller->check(wl, bparams.spellignore());
4999 if (SpellChecker::misspelled(result) && trailing_dot) {
5000 wl = WordLangTuple(word.append(from_ascii(".")), lang);
5001 result = speller->check(wl, bparams.spellignore());
5002 if (!SpellChecker::misspelled(result)) {
5003 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
5005 from << ".." << to << "]");
5007 // spell check with dot appended failed too
5008 // restore original word/lang value
5009 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
5010 wl = WordLangTuple(word, lang);
5014 if (!SpellChecker::misspelled(result)) {
5015 // area up to the begin of the next word is not misspelled
5016 while (end < size() && isWordSeparator(end))
5019 d->setMisspelled(from, end, result);
5021 result = d->speller_state_.getState(from);
5025 suggestions.clear();
5027 if (SpellChecker::misspelled(result)) {
5028 LYXERR(Debug::GUI, "misspelled word: \"" <<
5030 from << ".." << to << "]");
5032 speller->suggest(wl, suggestions);
5038 void Paragraph::anonymize()
5040 // This is a very crude anonymization for now
5041 for (char_type & c : d->text_)
5042 if (isLetterChar(c) || isNumber(c))
5047 void Paragraph::Private::markMisspelledWords(
5048 Language const * lang,
5049 pos_type const & first, pos_type const & last,
5050 SpellChecker::Result result,
5051 docstring const & word,
5052 SkipPositions const & skips)
5054 if (!SpellChecker::misspelled(result)) {
5055 setMisspelled(first, last, SpellChecker::WORD_OK);
5059 SpellChecker * speller = theSpellChecker();
5060 // locate and enumerate the error positions
5061 int nerrors = speller->numMisspelledWords();
5063 SkipPositionsIterator it = skips.begin();
5064 SkipPositionsIterator et = skips.end();
5065 for (int index = 0; index < nerrors; ++index) {
5068 speller->misspelledWord(index, wstart, wlen);
5069 /// should not happen if speller supports range checks
5072 WordLangTuple const candidate(word.substr(wstart, wlen), lang);
5073 wstart += first + numskipped;
5074 if (snext < wstart) {
5075 /// mark the range of correct spelling
5076 numskipped += countSkips(it, et, wstart);
5077 setMisspelled(snext,
5078 wstart - 1, SpellChecker::WORD_OK);
5080 snext = wstart + wlen;
5081 // Check whether the candidate is in the document's local dict
5082 SpellChecker::Result actresult = result;
5083 if (inset_owner_->buffer().params().spellignored(candidate))
5084 actresult = SpellChecker::DOCUMENT_LEARNED_WORD;
5085 numskipped += countSkips(it, et, snext);
5086 /// mark the range of misspelling
5087 setMisspelled(wstart, snext, actresult);
5088 if (actresult == SpellChecker::DOCUMENT_LEARNED_WORD)
5089 LYXERR(Debug::GUI, "local dictionary word: \"" <<
5090 candidate.word() << "\" [" <<
5091 wstart << ".." << (snext-1) << "]");
5093 LYXERR(Debug::GUI, "misspelled word: \"" <<
5094 candidate.word() << "\" [" <<
5095 wstart << ".." << (snext-1) << "]");
5098 if (snext <= last) {
5099 /// mark the range of correct spelling at end
5100 setMisspelled(snext, last, SpellChecker::WORD_OK);
5105 void Paragraph::spellCheck() const
5107 SpellChecker * speller = theSpellChecker();
5108 if (!speller || empty() ||!needsSpellCheck())
5112 d->rangeOfSpellCheck(start, endpos);
5113 if (speller->canCheckParagraph()) {
5114 // loop until we leave the range
5115 for (pos_type first = start; first < endpos; ) {
5116 pos_type last = endpos;
5117 Private::SkipPositions skips;
5118 Language * lang = d->locateSpellRange(first, last, skips);
5119 if (first >= endpos)
5121 // start the spell checker on the unit of meaning
5122 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
5123 WordLangTuple wl = WordLangTuple(word, lang);
5124 BufferParams const & bparams = d->inset_owner_->buffer().params();
5125 SpellChecker::Result result = !word.empty() ?
5126 speller->check(wl, bparams.spellignore()) : SpellChecker::WORD_OK;
5127 d->markMisspelledWords(lang, first, last, result, word, skips);
5131 static docstring_list suggestions;
5132 pos_type to = endpos;
5133 while (start < endpos) {
5135 spellCheck(start, to, wl, suggestions, false);
5139 d->readySpellCheck();
5143 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
5145 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
5146 if (result || pos <= 0 || pos > size())
5148 if (check_boundary && (pos == size() || isWordSeparator(pos)))
5149 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
5154 string Paragraph::magicLabel() const
5157 ss << "magicparlabel-" << id();