3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Kimberly Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
24 #include "BufferParams.h"
25 #include "BufferEncodings.h"
28 #include "InsetList.h"
30 #include "LaTeXFeatures.h"
35 #include "OutputParams.h"
36 #include "output_latex.h"
37 #include "output_xhtml.h"
38 #include "output_docbook.h"
39 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
42 #include "texstream.h"
45 #include "TextClass.h"
46 #include "WordLangTuple.h"
49 #include "frontends/alert.h"
51 #include "insets/InsetArgument.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
55 #include "insets/InsetText.h"
57 #include "mathed/InsetMathHull.h"
59 #include "support/debug.h"
60 #include "support/docstring_list.h"
61 #include "support/ExceptionMessage.h"
62 #include "support/gettext.h"
63 #include "support/lassert.h"
64 #include "support/lstrings.h"
65 #include "support/textutils.h"
72 using namespace lyx::support;
74 // OSX clang and msvc < 2015 do not support C++11 thread_local
75 #if defined(__APPLE__)
76 #define THREAD_LOCAL_STATIC static __thread
77 #elif defined(_MSC_VER) && (_MSC_VER < 1900)
78 #define THREAD_LOCAL_STATIC static __declspec(thread)
80 #define THREAD_LOCAL_STATIC thread_local static
85 /////////////////////////////////////////////////////////////////////
89 /////////////////////////////////////////////////////////////////////
91 class SpellResultRange {
93 SpellResultRange(FontSpan range, SpellChecker::Result result)
94 : range_(range), result_(result)
97 FontSpan const & range() const { return range_; }
99 void range(FontSpan const & r) { range_ = r; }
101 SpellChecker::Result result() const { return result_; }
103 void result(SpellChecker::Result r) { result_ = r; }
105 bool contains(pos_type pos) const { return range_.contains(pos); }
107 bool covered(FontSpan const & r) const
109 // 1. first of new range inside current range or
110 // 2. last of new range inside current range or
111 // 3. first of current range inside new range or
112 // 4. last of current range inside new range
113 //FIXME: is this the same as !range_.intersect(r).empty() ?
114 return range_.contains(r.first) || range_.contains(r.last) ||
115 r.contains(range_.first) || r.contains(range_.last);
118 void shift(pos_type pos, int offset)
120 if (range_.first > pos) {
121 range_.first += offset;
122 range_.last += offset;
123 } else if (range_.last >= pos) {
124 range_.last += offset;
129 SpellChecker::Result result_ ;
133 /////////////////////////////////////////////////////////////////////
137 /////////////////////////////////////////////////////////////////////
139 class SpellCheckerState {
143 needs_refresh_ = true;
144 current_change_number_ = 0;
147 void setRange(FontSpan const & fp, SpellChecker::Result state)
150 RangesIterator et = ranges_.end();
151 RangesIterator it = ranges_.begin();
152 for (; it != et; ++it) {
153 if (!it->covered(fp))
154 result.push_back(SpellResultRange(it->range(), it->result()));
155 else if (state == SpellChecker::WORD_OK) {
156 // trim or split the current misspelled range
157 // store misspelled ranges only
158 FontSpan range = it->range();
159 if (fp.first > range.first) {
160 // misspelled area in front of WORD_OK
161 range.last = fp.first - 1;
162 result.push_back(SpellResultRange(range, it->result()));
165 if (fp.last < range.last) {
166 // misspelled area after WORD_OK range
167 range.first = fp.last + 1;
168 result.push_back(SpellResultRange(range, it->result()));
173 if (state != SpellChecker::WORD_OK)
174 ranges_.push_back(SpellResultRange(fp, state));
177 void increasePosAfterPos(pos_type pos)
179 correctRangesAfterPos(pos, 1);
183 void decreasePosAfterPos(pos_type pos)
185 correctRangesAfterPos(pos, -1);
189 void refreshLast(pos_type pos)
191 if (pos < refresh_.last)
195 SpellChecker::Result getState(pos_type pos) const
197 SpellChecker::Result result = SpellChecker::WORD_OK;
198 RangesIterator et = ranges_.end();
199 RangesIterator it = ranges_.begin();
200 for (; it != et; ++it) {
201 if(it->contains(pos)) {
208 FontSpan const & getRange(pos_type pos) const
210 /// empty span to indicate mismatch
211 static FontSpan empty_;
212 RangesIterator et = ranges_.end();
213 RangesIterator it = ranges_.begin();
214 for (; it != et; ++it) {
215 if(it->contains(pos)) {
222 bool needsRefresh() const
224 return needs_refresh_;
227 SpellChecker::ChangeNumber currentChangeNumber() const
229 return current_change_number_;
232 void refreshRange(pos_type & first, pos_type & last) const
234 first = refresh_.first;
235 last = refresh_.last;
238 void needsRefresh(pos_type pos)
240 if (needs_refresh_ && pos != -1) {
241 if (pos < refresh_.first)
242 refresh_.first = pos;
243 if (pos > refresh_.last)
245 } else if (pos != -1) {
246 // init request check for neighbour positions too
247 refresh_.first = pos > 0 ? pos - 1 : 0;
248 // no need for special end of paragraph check
249 refresh_.last = pos + 1;
251 needs_refresh_ = pos != -1;
254 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
256 needs_refresh_ = true;
259 current_change_number_ = change_number;
262 typedef vector<SpellResultRange> Ranges;
263 typedef Ranges::const_iterator RangesIterator;
265 /// the area of the paragraph with pending spell check
267 /// spell state cache version number
268 SpellChecker::ChangeNumber current_change_number_;
272 void correctRangesAfterPos(pos_type pos, int offset)
274 RangesIterator et = ranges_.end();
275 Ranges::iterator it = ranges_.begin();
276 for (; it != et; ++it) {
277 it->shift(pos, offset);
283 /////////////////////////////////////////////////////////////////////
285 // Paragraph::Private
287 /////////////////////////////////////////////////////////////////////
289 class Paragraph::Private
291 // Enforce our own "copy" constructor
292 Private(Private const &) = delete;
293 Private & operator=(Private const &) = delete;
294 // Unique ID generator
295 static int make_id();
298 Private(Paragraph * owner, Layout const & layout);
299 /// "Copy constructor"
300 Private(Private const &, Paragraph * owner);
301 /// Copy constructor from \p beg to \p end
302 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
305 void insertChar(pos_type pos, char_type c, Change const & change);
307 /// Output the surrogate pair formed by \p c and \p next to \p os.
308 /// \return the number of characters written.
309 int latexSurrogatePair(BufferParams const &, otexstream & os,
310 char_type c, char_type next,
311 OutputParams const &) const;
313 /// Output a space in appropriate formatting (or a surrogate pair
314 /// if the next character is a combining character).
315 /// \return whether a surrogate pair was output.
316 bool simpleTeXBlanks(BufferParams const &,
317 OutputParams const &,
320 unsigned int & column,
322 Layout const & style) const;
324 /// This could go to ParagraphParameters if we want to.
325 int startTeXParParams(BufferParams const &, otexstream &,
326 OutputParams const &) const;
328 /// This could go to ParagraphParameters if we want to.
329 bool endTeXParParams(BufferParams const &, otexstream &,
330 OutputParams const &) const;
333 void latexInset(BufferParams const &,
338 Font const & outerfont,
340 Change & running_change,
341 Layout const & style,
343 unsigned int & column,
344 bool const fontswitch_inset,
345 bool const closeLanguage,
346 bool const lang_switched_at_inset) const;
349 void latexSpecialChar(
351 BufferParams const & bparams,
352 OutputParams const & runparams,
353 Font const & running_font,
354 string & alien_script,
355 Layout const & style,
358 unsigned int & column) const;
365 unsigned int & column) const;
371 unsigned int & column) const;
377 unsigned int & column) const;
380 void validate(LaTeXFeatures & features) const;
382 /// Checks if the paragraph contains only text and no inset or font change.
383 bool onlyText(Buffer const & buf, Font const & outerfont,
384 pos_type initial) const;
386 /// a vector of speller skip positions
387 typedef vector<FontSpan> SkipPositions;
388 typedef SkipPositions::const_iterator SkipPositionsIterator;
390 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
392 Language * getSpellLanguage(pos_type const from) const;
394 Language * locateSpellRange(pos_type & from, pos_type & to,
395 SkipPositions & skips) const;
397 bool hasSpellerChange() const
399 SpellChecker::ChangeNumber speller_change_number = 0;
400 if (theSpellChecker())
401 speller_change_number = theSpellChecker()->changeNumber();
402 return speller_change_number > speller_state_.currentChangeNumber();
405 bool ignoreWord(docstring const & word) const;
407 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
409 pos_type textsize = owner_->size();
410 // check for sane arguments
411 if (to <= from || from >= textsize)
413 FontSpan fp = FontSpan(from, to - 1);
414 speller_state_.setRange(fp, state);
417 void requestSpellCheck(pos_type pos)
420 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
422 speller_state_.needsRefresh(pos);
425 void readySpellCheck()
427 speller_state_.needsRefresh(-1);
430 bool needsSpellCheck() const
432 return speller_state_.needsRefresh();
435 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
437 speller_state_.refreshRange(first, last);
439 last = owner_->size();
442 pos_type endpos = last;
443 owner_->locateWord(first, endpos, WHOLE_WORD, true);
446 owner_->locateWord(last, endpos, WHOLE_WORD, true);
451 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
455 while (it != et && it->first < start) {
456 long skip = it->last - it->first + 1;
464 void markMisspelledWords(Language const * lang,
465 pos_type const & first, pos_type const & last,
466 SpellChecker::Result result,
467 docstring const & word,
468 SkipPositions const & skips);
470 InsetCode ownerCode() const
472 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
475 /// Which Paragraph owns us?
479 Inset const * inset_owner_;
485 ParagraphParameters params_;
487 /// for recording and looking up changes
491 InsetList insetlist_;
494 pos_type begin_of_body_;
496 typedef docstring TextContainer;
500 typedef set<docstring> Words;
501 typedef map<string, Words> LangWordsMap;
505 Layout const * layout_;
507 SpellCheckerState speller_state_;
513 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
514 : owner_(owner), inset_owner_(nullptr), begin_of_body_(0), layout_(&layout), id_(-1)
521 int Paragraph::Private::make_id()
523 // The id is unique per session across buffers because it is used in
524 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
527 static int next_id(0);
532 Paragraph::Private::Private(Private const & p, Paragraph * owner)
533 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
534 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
535 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
536 layout_(p.layout_), id_(make_id())
538 requestSpellCheck(p.text_.size());
542 Paragraph::Private::Private(Private const & p, Paragraph * owner,
543 pos_type beg, pos_type end)
544 : owner_(owner), inset_owner_(p.inset_owner_),
545 params_(p.params_), changes_(p.changes_),
546 insetlist_(p.insetlist_, beg, end),
547 begin_of_body_(p.begin_of_body_), words_(p.words_),
548 layout_(p.layout_), id_(make_id())
550 if (beg >= pos_type(p.text_.size()))
552 text_ = p.text_.substr(beg, end - beg);
554 FontList::const_iterator fcit = fontlist_.begin();
555 FontList::const_iterator fend = fontlist_.end();
556 for (; fcit != fend; ++fcit) {
557 if (fcit->pos() < beg)
559 if (fcit->pos() >= end) {
560 // Add last entry in the fontlist_.
561 fontlist_.set(text_.size() - 1, fcit->font());
564 // Add a new entry in the fontlist_.
565 fontlist_.set(fcit->pos() - beg, fcit->font());
567 requestSpellCheck(p.text_.size());
571 /////////////////////////////////////////////////////////////////////
575 /////////////////////////////////////////////////////////////////////
579 /** This helper class should be instantiated at the start of methods
580 * that can create or merge changes. If as a result the value of
581 * Paragraph::isChanged is modified, it makes sure that updateBuffer()
584 struct ChangesMonitor {
586 ChangesMonitor(Paragraph & par)
587 : par_(par), was_changed_(par.isChanged()) {}
591 /* We may need to run updateBuffer to check whether the buffer
592 * contains changes (and toggle the changes toolbar). We do it
594 * 1. the `changedness' of the paragraph has changed,
595 * 2. and we are not in the situation where the buffer has changes
596 * and new changes are added to the paragraph.
599 if (par_.isChanged() != was_changed_
600 && par_.inInset().isBufferValid()
601 && !(par_.inInset().buffer().areChangesPresent() && par_.isChanged()))
602 par_.inInset().buffer().forceUpdate();
603 } catch(support::ExceptionMessage const &) {}
608 Paragraph const & par_;
615 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
616 bool output_active, TocBackend & backend) const
618 d->changes_.addToToc(cdit, buf, output_active, backend);
622 bool Paragraph::isDeleted(pos_type start, pos_type end) const
624 LASSERT(start >= 0 && start <= size(), return false);
625 LASSERT(end > start && end <= size() + 1, return false);
627 return d->changes_.isDeleted(start, end);
631 bool Paragraph::isChanged(pos_type start, pos_type end) const
633 LASSERT(start >= 0 && start <= size(), return false);
634 LASSERT(end > start && end <= size() + 1, return false);
636 return d->changes_.isChanged(start, end);
639 // FIXME: Ideally the diverse isChanged() methods should account for that!
640 bool Paragraph::hasChangedInsets(pos_type start, pos_type end) const
642 LASSERT(start >= 0 && start <= size(), return false);
643 LASSERT(end > start && end <= size() + 1, return false);
645 for (auto const & icit : d->insetlist_) {
646 if (icit.pos < start)
650 if (icit.inset && icit.inset->isChanged())
656 bool Paragraph::isChanged() const
658 return d->changes_.isChanged();
662 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
664 // keep the logic here in sync with the logic of eraseChars()
668 Change const & change = d->changes_.lookup(size());
669 return change.inserted() && change.currentAuthor();
672 Change Paragraph::parEndChange() const
674 return d->changes_.lookup(size());
678 void Paragraph::setChange(Change const & change)
680 // Make sure that Buffer::hasChangesPresent is updated
681 ChangesMonitor cm(*this);
683 // beware of the imaginary end-of-par character!
684 d->changes_.set(change, 0, size() + 1);
687 * Propagate the change recursively - but not in case of DELETED!
689 * Imagine that your co-author makes changes in an existing inset. He
690 * sends your document to you and you come to the conclusion that the
691 * inset should go completely. If you erase it, LyX must not delete all
692 * text within the inset. Otherwise, the change tracked insertions of
693 * your co-author get lost and there is no way to restore them later.
695 * Conclusion: An inset's content should remain untouched if you delete it
698 if (!change.deleted()) {
699 for (pos_type pos = 0; pos < size(); ++pos) {
700 if (Inset * inset = getInset(pos))
701 inset->setChange(change);
707 void Paragraph::setChange(pos_type pos, Change const & change)
709 // Make sure that Buffer::hasChangesPresent is updated
710 ChangesMonitor cm(*this);
712 LASSERT(pos >= 0 && pos <= size(), return);
713 d->changes_.set(change, pos);
715 // see comment in setChange(Change const &) above
716 if (!change.deleted() && pos < size())
717 if (Inset * inset = getInset(pos))
718 inset->setChange(change);
722 Change const & Paragraph::lookupChange(pos_type pos) const
724 LBUFERR(pos >= 0 && pos <= size());
725 return d->changes_.lookup(pos);
729 void Paragraph::acceptChanges(pos_type start, pos_type end)
731 // Make sure that Buffer::hasChangesPresent is updated
732 ChangesMonitor cm(*this);
734 LASSERT(start >= 0 && start <= size(), return);
735 LASSERT(end > start && end <= size() + 1, return);
737 for (pos_type pos = start; pos < end; ++pos) {
738 switch (lookupChange(pos).type) {
739 case Change::UNCHANGED:
740 // accept changes in nested inset
741 if (Inset * inset = getInset(pos))
742 inset->acceptChanges();
745 case Change::INSERTED:
746 d->changes_.set(Change(Change::UNCHANGED), pos);
747 // also accept changes in nested inset
748 if (Inset * inset = getInset(pos))
749 inset->acceptChanges();
752 case Change::DELETED:
753 // Suppress access to non-existent
754 // "end-of-paragraph char"
756 eraseChar(pos, false);
767 void Paragraph::rejectChanges(pos_type start, pos_type end)
769 LASSERT(start >= 0 && start <= size(), return);
770 LASSERT(end > start && end <= size() + 1, return);
772 // Make sure that Buffer::hasChangesPresent is updated
773 ChangesMonitor cm(*this);
775 for (pos_type pos = start; pos < end; ++pos) {
776 switch (lookupChange(pos).type) {
777 case Change::UNCHANGED:
778 // reject changes in nested inset
779 if (Inset * inset = getInset(pos))
780 inset->rejectChanges();
783 case Change::INSERTED:
784 // Suppress access to non-existent
785 // "end-of-paragraph char"
787 eraseChar(pos, false);
793 case Change::DELETED:
794 d->changes_.set(Change(Change::UNCHANGED), pos);
796 // Do NOT reject changes within a deleted inset!
797 // There may be insertions of a co-author inside of it!
805 void Paragraph::Private::insertChar(pos_type pos, char_type c,
806 Change const & change)
808 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
810 // Make sure that Buffer::hasChangesPresent is updated
811 ChangesMonitor cm(*owner_);
814 changes_.insert(change, pos);
816 // This is actually very common when parsing buffers (and
817 // maybe inserting ascii text)
818 if (pos == pos_type(text_.size())) {
819 // when appending characters, no need to update tables
821 // but we want spell checking
822 requestSpellCheck(pos);
826 text_.insert(text_.begin() + pos, c);
828 // Update the font table.
829 fontlist_.increasePosAfterPos(pos);
832 insetlist_.increasePosAfterPos(pos);
834 // Update list of misspelled positions
835 speller_state_.increasePosAfterPos(pos);
838 if (inset_owner_ && inset_owner_->isBufferValid())
839 theSession().bookmarks().adjustPosAfterPos(inset_owner_->buffer().fileName(),
844 bool Paragraph::insertInset(pos_type pos, Inset * inset,
845 Font const & font, Change const & change)
847 LASSERT(inset, return false);
848 LASSERT(pos >= 0 && pos <= size(), return false);
850 // Make sure that Buffer::hasChangesPresent is updated
851 ChangesMonitor cm(*this);
853 // Paragraph::insertInset() can be used in cut/copy/paste operation where
854 // d->inset_owner_ is not set yet.
855 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
858 d->insertChar(pos, META_INSET, change);
859 LASSERT(d->text_[pos] == META_INSET, return false);
861 // Add a new entry in the insetlist_.
862 d->insetlist_.insert(inset, pos);
864 // Some insets require run of spell checker
865 requestSpellCheck(pos);
871 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
873 LASSERT(pos >= 0 && pos <= size(), return false);
875 // Make sure that Buffer::hasChangesPresent is updated
876 ChangesMonitor cm(*this);
878 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
881 Change change = d->changes_.lookup(pos);
883 // set the character to DELETED if
884 // a) it was previously unchanged or
885 // b) it was inserted by a co-author
887 if (!change.changed() ||
888 (change.inserted() && !change.currentAuthor())) {
889 setChange(pos, Change(Change::DELETED));
890 // request run of spell checker
891 requestSpellCheck(pos);
895 if (change.deleted())
899 // Don't physically access the imaginary end-of-paragraph character.
900 // eraseChar() can only mark it as DELETED. A physical deletion of
901 // end-of-par must be handled externally.
907 d->changes_.erase(pos);
909 // if it is an inset, delete the inset entry
910 if (d->text_[pos] == META_INSET)
911 d->insetlist_.erase(pos);
913 d->text_.erase(d->text_.begin() + pos);
915 // Update the fontlist_
916 d->fontlist_.erase(pos);
918 // Update the insetlist_
919 d->insetlist_.decreasePosAfterPos(pos);
921 // Update list of misspelled positions
922 d->speller_state_.decreasePosAfterPos(pos);
923 d->speller_state_.refreshLast(size());
926 if (d->inset_owner_ && d->inset_owner_->isBufferValid())
927 theSession().bookmarks().adjustPosAfterPos(d->inset_owner_->buffer().fileName(),
934 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
936 LASSERT(start >= 0 && start <= size(), return 0);
937 LASSERT(end >= start && end <= size() + 1, return 0);
940 for (pos_type count = end - start; count; --count) {
941 if (!eraseChar(i, trackChanges))
948 // Handle combining characters
949 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
950 otexstream & os, char_type c, char_type next,
951 OutputParams const & runparams) const
953 // Writing next here may circumvent a possible font change between
954 // c and next. Since next is only output if it forms a surrogate pair
955 // with c we can ignore this:
956 // A font change inside a surrogate pair does not make sense and is
957 // hopefully impossible to input.
958 // FIXME: change tracking
959 // Is this correct WRT change tracking?
960 Encoding const & encoding = *(runparams.encoding);
961 docstring latex1 = encoding.latexChar(next).first;
962 if (runparams.inIPA) {
963 string const tipashortcut = Encodings::TIPAShortcut(next);
964 if (!tipashortcut.empty()) {
965 latex1 = from_ascii(tipashortcut);
968 docstring latex2 = encoding.latexChar(c).first;
970 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
971 // Encoding supports the combination:
972 // output as is (combining char after base char).
973 os << latex2 << latex1;
974 return latex1.length() + latex2.length();
977 os << latex1 << "{" << latex2 << "}";
978 return latex1.length() + latex2.length() + 2;
982 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
983 OutputParams const & runparams,
986 unsigned int & column,
988 Layout const & style) const
990 if (style.pass_thru || runparams.pass_thru)
993 if (i + 1 < int(text_.size())) {
994 char_type next = text_[i + 1];
995 if (Encodings::isCombiningChar(next)) {
996 // This space has an accent, so we must always output it.
997 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
1002 if (runparams.linelen > 0
1003 && column > runparams.linelen
1005 && text_[i - 1] != ' '
1006 && (i + 1 < int(text_.size()))
1007 // same in FreeSpacing mode
1008 && !owner_->isFreeSpacing()
1009 // In typewriter mode, we want to avoid
1010 // ! . ? : at the end of a line
1011 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
1012 && (text_[i - 1] == '.'
1013 || text_[i - 1] == '?'
1014 || text_[i - 1] == ':'
1015 || text_[i - 1] == '!'))) {
1017 os.texrow().start(owner_->id(), i + 1);
1019 } else if (style.free_spacing) {
1028 void Paragraph::Private::latexInset(BufferParams const & bparams,
1030 OutputParams & runparams,
1031 Font & running_font,
1033 Font const & outerfont,
1035 Change & running_change,
1036 Layout const & style,
1038 unsigned int & column,
1039 bool const fontswitch_inset,
1040 bool const closeLanguage,
1041 bool const lang_switched_at_inset) const
1043 Inset * inset = owner_->getInset(i);
1046 if (style.pass_thru) {
1047 odocstringstream ods;
1048 inset->plaintext(ods, runparams);
1053 // FIXME: move this to InsetNewline::latex
1054 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1055 // newlines are handled differently here than
1056 // the default in simpleTeXSpecialChars().
1057 if (!style.newline_allowed) {
1061 bool needPar = false;
1062 column += running_font.latexWriteEndChanges(
1063 os, bparams, runparams,
1064 basefont, basefont, needPar);
1068 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1071 basefont = owner_->getLayoutFont(bparams, outerfont);
1072 running_font = basefont;
1074 if (runparams.moving_arg)
1078 os.texrow().start(owner_->id(), i + 1);
1082 if (owner_->isDeleted(i)) {
1083 if( ++runparams.inDeletedInset == 1)
1084 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1087 if (inset->canTrackChanges()) {
1088 column += Changes::latexMarkChange(os, bparams, running_change,
1089 Change(Change::UNCHANGED), runparams);
1090 running_change = Change(Change::UNCHANGED);
1094 odocstream::pos_type const len = os.os().tellp();
1096 if (inset->forceLTR(runparams)
1097 && running_font.isRightToLeft()
1098 // ERT is an exception, it should be output with no
1099 // decorations at all
1100 && inset->lyxCode() != ERT_CODE) {
1101 if (runparams.use_polyglossia) {
1103 } else if (running_font.language()->lang() == "farsi"
1104 || running_font.language()->lang() == "arabic_arabi")
1105 os << "\\textLR{" << termcmd;
1111 if (open_font && fontswitch_inset) {
1112 bool lang_closed = false;
1113 // Close language if needed
1114 if (closeLanguage && !lang_switched_at_inset) {
1115 // We need prev_font here as language changes directly at inset
1116 // will only be started inside the inset.
1117 Font const prev_font = (i > 0) ?
1118 owner_->getFont(bparams, i - 1, outerfont)
1120 Font tmpfont(basefont);
1121 tmpfont.setLanguage(prev_font.language());
1122 bool needPar = false;
1123 unsigned int count = tmpfont.latexWriteEndChanges(os, bparams, runparams,
1125 needPar, closeLanguage);
1127 lang_closed = count > 0;
1129 // Update the running_font, making sure, however,
1130 // to leave the language as it was.
1131 // FIXME: probably a better way to keep track of the old
1132 // language, than copying the entire font?
1133 Font const copy_font(running_font);
1134 basefont = owner_->getLayoutFont(bparams, outerfont);
1135 running_font = basefont;
1137 running_font.setLanguage(copy_font.language());
1138 OutputParams rp = runparams;
1139 rp.encoding = basefont.language()->encoding();
1140 // For these, we use switches, so they should be taken as
1141 // base inside the inset.
1142 basefont.fontInfo().setSize(copy_font.fontInfo().size());
1143 basefont.fontInfo().setFamily(copy_font.fontInfo().family());
1144 basefont.fontInfo().setSeries(copy_font.fontInfo().series());
1145 // Now re-do font changes in a way needed here
1146 // (using switches with multi-par insets)
1147 InsetText const * textinset = inset->asInsetText();
1148 bool const cprotect = textinset
1149 ? textinset->hasCProtectContent(runparams.moving_arg)
1150 && !textinset->text().isMainText()
1151 && inset->lyxCode() != BRANCH_CODE
1153 unsigned int count2 = basefont.latexWriteStartChanges(os, bparams,
1159 if (count2 == 0 && (lang_closed || lang_switched_at_inset))
1163 runparams.local_font = &basefont;
1166 if (fontswitch_inset && !closeLanguage) {
1167 // The directionality has been switched at inset.
1168 // Force markup inside.
1169 runparams.local_font = &basefont;
1172 size_t const previous_row_count = os.texrow().rows();
1175 runparams.lastid = id_;
1176 runparams.lastpos = i;
1177 inset->latex(os, runparams);
1178 } catch (EncodingException & e) {
1179 // add location information and throw again.
1188 if (os.texrow().rows() > previous_row_count) {
1189 os.texrow().start(owner_->id(), i + 1);
1192 column += (unsigned int)(os.os().tellp() - len);
1195 if (owner_->isDeleted(i))
1196 --runparams.inDeletedInset;
1200 void Paragraph::Private::latexSpecialChar(otexstream & os,
1201 BufferParams const & bparams,
1202 OutputParams const & runparams,
1203 Font const & running_font,
1204 string & alien_script,
1205 Layout const & style,
1208 unsigned int & column) const
1210 char_type const c = owner_->getUChar(bparams, runparams, i);
1212 if (style.pass_thru || runparams.pass_thru || runparams.find_effective()
1213 || contains(style.pass_thru_chars, c)
1214 || contains(runparams.pass_thru_chars, c)) {
1215 if (runparams.find_effective()) {
1221 os << "\\braceleft ";
1224 os << "\\braceright ";
1227 os << "\\lyxdollar ";
1230 os << "\\lyxtilde ";
1240 else if (c != '\0') {
1241 Encoding const * const enc = runparams.encoding;
1242 if (enc && !enc->encodable(c))
1243 throw EncodingException(c);
1249 // TIPA uses its own T3 encoding
1250 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1252 // If T1 font encoding is used, use the special
1253 // characters it provides.
1254 // NOTE: Some languages reset the font encoding internally to a
1255 // non-standard font encoding. If we are using such a language,
1256 // we do not output special T1 chars.
1257 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1258 && !runparams.isFullUnicode() && runparams.main_fontenc == "T1"
1259 && latexSpecialT1(c, os, i, column))
1261 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1262 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1263 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1266 // Otherwise, we use what LaTeX provides us.
1269 os << "\\textbackslash" << termcmd;
1273 os << "\\textless" << termcmd;
1277 os << "\\textgreater" << termcmd;
1281 os << "\\textbar" << termcmd;
1286 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1287 (end_pos == -1 || i + 1 < end_pos) &&
1288 text_[i+1] == '-') {
1289 // Prevent "--" becoming an en dash and "---" an em dash.
1290 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1296 os << "\\textquotedbl" << termcmd;
1301 case '%': case '#': case '{':
1309 os << "\\textasciitilde" << termcmd;
1314 os << "\\textasciicircum" << termcmd;
1321 // avoid being mistaken for optional arguments
1329 // Blanks are printed before font switching.
1330 // Sure? I am not! (try nice-latex)
1331 // I am sure it's correct. LyX might be smarter
1332 // in the future, but for now, nothing wrong is
1338 // XeTeX's dash behaviour is determined via a global setting
1339 if (bparams.use_dash_ligatures
1340 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1342 // TODO #10961: && not in inset Flex Code
1343 // TODO #10961: && not in layout LyXCode
1344 && (!bparams.useNonTeXFonts || runparams.flavor != Flavor::XeTeX)) {
1361 Encoding const & encoding = *(runparams.encoding);
1362 char_type next = '\0';
1363 if (i + 1 < int(text_.size())) {
1364 next = text_[i + 1];
1365 if (Encodings::isCombiningChar(next)) {
1366 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1371 pair<docstring, bool> latex = encoding.latexChar(c);
1372 docstring nextlatex;
1373 bool nexttipas = false;
1374 string nexttipashortcut;
1375 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1376 nextlatex = encoding.latexChar(next).first;
1377 if (runparams.inIPA) {
1378 nexttipashortcut = Encodings::TIPAShortcut(next);
1379 nexttipas = !nexttipashortcut.empty();
1383 if (runparams.inIPA) {
1384 string const tipashortcut = Encodings::TIPAShortcut(c);
1385 if (!tipashortcut.empty()) {
1386 latex.first = from_ascii(tipashortcut);
1387 latex.second = false;
1391 // eventually close "script wrapper" command (see `Paragraph::latex`)
1392 if (!alien_script.empty()
1393 && alien_script != Encodings::isKnownScriptChar(next)) {
1394 column += latex.first.length();
1395 alien_script.clear();
1396 os << latex.first << "}";
1400 && ((!prefixIs(nextlatex, '\\')
1401 && !prefixIs(nextlatex, '{')
1402 && !prefixIs(nextlatex, '}'))
1404 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1406 // Prevent eating of a following space or command corruption by
1407 // following characters
1408 if (next == ' ' || next == '\0') {
1409 column += latex.first.length() + 1;
1410 os << latex.first << "{}";
1412 column += latex.first.length();
1413 os << latex.first << " ";
1416 column += latex.first.length() - 1;
1424 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1425 pos_type i, unsigned int & column) const
1431 // In T1 encoding, these characters exist
1432 // but we should avoid ligatures
1433 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1435 os << "\\textcompwordmark" << termcmd;
1442 // soul.sty breaks with \char`\"
1443 os << "\\textquotedbl" << termcmd;
1452 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1453 pos_type i, unsigned int & column) const
1455 // TU encoding is currently on par with T1.
1456 return latexSpecialT1(c, os, i, column);
1460 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1461 pos_type /*i*/, unsigned int & column) const
1471 os << "\\textvertline" << termcmd;
1480 void Paragraph::Private::validate(LaTeXFeatures & features) const
1482 if (layout_->inpreamble && inset_owner_) {
1483 // FIXME: Using a string stream here circumvents the encoding
1484 // switching machinery of odocstream. Therefore the
1485 // output is wrong if this paragraph contains content
1486 // that needs to switch encoding.
1487 Buffer const & buf = inset_owner_->buffer();
1488 otexstringstream os;
1489 os << layout_->preamble();
1490 size_t const length = os.length();
1491 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1492 buf.getParFromID(owner_->id()).pit(), os,
1493 features.runparams(), string(), 0, -1, true);
1494 if (os.length() > length)
1495 features.addPreambleSnippet(os.release(), true);
1498 if (features.runparams().flavor == Flavor::Html
1499 && layout_->htmltitle()) {
1500 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1503 // check the params.
1504 if (!params_.spacing().isDefault())
1505 features.require("setspace");
1508 features.useLayout(layout_->name());
1511 fontlist_.validate(features);
1513 // then the indentation
1514 if (!params_.leftIndent().zero())
1515 features.require("ParagraphLeftIndent");
1518 InsetList::const_iterator icit = insetlist_.begin();
1519 InsetList::const_iterator iend = insetlist_.end();
1520 for (; icit != iend; ++icit) {
1522 features.inDeletedInset(owner_->isDeleted(icit->pos));
1523 if (icit->inset->lyxCode() == FOOT_CODE) {
1524 // FIXME: an item inset would make things much easier.
1525 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1526 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1527 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1528 && (icit->pos < begin_of_body_
1529 || (icit->pos == begin_of_body_
1530 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1531 features.saveNoteEnv("description");
1533 icit->inset->validate(features);
1534 features.inDeletedInset(false);
1535 if (layout_->needprotect &&
1536 icit->inset->lyxCode() == FOOT_CODE)
1537 features.require("NeedLyXFootnoteCode");
1541 // then the contents
1542 BufferParams const bp = features.runparams().is_child
1543 ? features.buffer().masterParams() : features.buffer().params();
1544 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1545 char_type c = text_[i];
1546 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1548 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1549 features.require("textquotedblp");
1550 else if (features.runparams().main_fontenc != "T1"
1551 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1552 features.require("textquotedbl");
1553 } else if (ci.textFeature() && contains(ci.textPreamble(), '=')) {
1554 // features that depend on the font or input encoding
1555 string feats = ci.textPreamble();
1556 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1557 if (fontenc.empty())
1558 fontenc = features.runparams().main_fontenc;
1559 while (!feats.empty()) {
1561 feats = split(feats, feat, ',');
1562 if (contains(feat, "!=")) {
1563 // a feature that is required except for the spcified
1564 // font or input encodings
1566 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1567 // multiple encodings are separated by semicolon
1568 vector<string> context = getVectorFromString(contexts, ";");
1569 // require feature if the context matches neither current font
1570 // nor input encoding
1571 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1572 && std::find(context.begin(), context.end(),
1573 features.runparams().encoding->name()) == context.end())
1574 features.require(realfeature);
1575 } else if (contains(feat, '=')) {
1576 // a feature that is required only for the spcified
1577 // font or input encodings
1579 string const contexts = split(feat, realfeature, '=');
1580 // multiple encodings are separated by semicolon
1581 vector<string> context = getVectorFromString(contexts, ";");
1582 // require feature if the context matches either current font
1583 // or input encoding
1584 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1585 || std::find(context.begin(), context.end(),
1586 features.runparams().encoding->name()) != context.end())
1587 features.require(realfeature);
1590 } else if (!bp.use_dash_ligatures
1591 && (c == 0x2013 || c == 0x2014)
1592 && bp.useNonTeXFonts
1593 && features.runparams().flavor == Flavor::XeTeX)
1594 // XeTeX's dash behaviour is determined via a global setting
1595 features.require("xetexdashbreakstate");
1596 BufferEncodings::validate(c, features);
1600 /////////////////////////////////////////////////////////////////////
1604 /////////////////////////////////////////////////////////////////////
1607 Layout const emptyParagraphLayout;
1610 Paragraph::Paragraph()
1611 : d(new Paragraph::Private(this, emptyParagraphLayout))
1618 Paragraph::Paragraph(Paragraph const & par)
1619 : itemdepth(par.itemdepth),
1620 d(new Paragraph::Private(*par.d, this))
1626 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1627 : itemdepth(par.itemdepth),
1628 d(new Paragraph::Private(*par.d, this, beg, end))
1634 Paragraph & Paragraph::operator=(Paragraph const & par)
1636 // needed as we will destroy the private part before copying it
1638 itemdepth = par.itemdepth;
1642 d = new Private(*par.d, this);
1649 Paragraph::~Paragraph()
1658 // this shall be called just before every "os << ..." action.
1659 void flushString(ostream & os, docstring & s)
1668 void Paragraph::write(ostream & os, BufferParams const & bparams,
1669 depth_type & depth) const
1671 // The beginning or end of a deeper (i.e. nested) area?
1672 if (depth != d->params_.depth()) {
1673 if (d->params_.depth() > depth) {
1674 while (d->params_.depth() > depth) {
1675 os << "\n\\begin_deeper";
1679 while (d->params_.depth() < depth) {
1680 os << "\n\\end_deeper";
1686 // First write the layout
1687 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1689 d->params_.write(os);
1691 Font font1(inherit_font, bparams.language);
1693 Change running_change = Change(Change::UNCHANGED);
1695 // this string is used as a buffer to avoid repetitive calls
1696 // to to_utf8(), which turn out to be expensive (JMarc)
1697 docstring write_buffer;
1700 for (pos_type i = 0; i <= size(); ++i) {
1702 Change const & change = lookupChange(i);
1703 if (change != running_change)
1704 flushString(os, write_buffer);
1705 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1706 running_change = change;
1711 // Write font changes
1712 Font font2 = getFontSettings(bparams, i);
1713 if (font2 != font1) {
1714 flushString(os, write_buffer);
1715 font2.lyxWriteChanges(font1, os);
1720 char_type const c = d->text_[i];
1723 if (Inset const * inset = getInset(i)) {
1724 flushString(os, write_buffer);
1725 if (inset->directWrite()) {
1726 // international char, let it write
1727 // code directly so it's shorter in
1733 os << "\\begin_inset ";
1735 os << "\n\\end_inset\n\n";
1738 // FIXME This can be removed again once the mystery
1739 // crash has been resolved.
1744 flushString(os, write_buffer);
1745 os << "\n\\backslash\n";
1749 flushString(os, write_buffer);
1750 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1757 if ((column > 70 && c == ' ')
1759 flushString(os, write_buffer);
1763 // this check is to amend a bug. LyX sometimes
1764 // inserts '\0' this could cause problems.
1766 write_buffer.push_back(c);
1768 LYXERR0("NUL char in structure.");
1774 flushString(os, write_buffer);
1775 os << "\n\\end_layout\n";
1776 // FIXME This can be removed again once the mystery
1777 // crash has been resolved.
1782 void Paragraph::validate(LaTeXFeatures & features) const
1784 d->validate(features);
1785 bool fragile = features.runparams().moving_arg;
1786 fragile |= layout().needprotect;
1787 if (inInset().getLayout().isNeedProtect())
1789 if (needsCProtection(fragile))
1790 features.require("cprotect");
1794 void Paragraph::insert(pos_type pos, docstring const & str,
1795 Font const & font, Change const & change)
1797 for (size_t i = 0, n = str.size(); i != n ; ++i)
1798 insertChar(pos + i, str[i], font, change);
1802 void Paragraph::appendChar(char_type c, Font const & font,
1803 Change const & change)
1805 // Make sure that Buffer::hasChangesPresent is updated
1806 ChangesMonitor cm(*this);
1809 d->changes_.insert(change, d->text_.size());
1810 // when appending characters, no need to update tables
1811 d->text_.push_back(c);
1812 setFont(d->text_.size() - 1, font);
1813 d->requestSpellCheck(d->text_.size() - 1);
1817 void Paragraph::appendString(docstring const & s, Font const & font,
1818 Change const & change)
1820 // Make sure that Buffer::hasChangesPresent is updated
1821 ChangesMonitor cm(*this);
1823 pos_type end = s.size();
1824 size_t oldsize = d->text_.size();
1825 size_t newsize = oldsize + end;
1826 size_t capacity = d->text_.capacity();
1827 if (newsize >= capacity)
1828 d->text_.reserve(max(capacity + 100, newsize));
1830 // when appending characters, no need to update tables
1833 // FIXME: Optimize this!
1834 for (size_t i = oldsize; i != newsize; ++i) {
1836 d->changes_.insert(change, i);
1837 d->requestSpellCheck(i);
1839 d->fontlist_.set(oldsize, font);
1840 d->fontlist_.set(newsize - 1, font);
1844 void Paragraph::insertChar(pos_type pos, char_type c,
1847 d->insertChar(pos, c, Change(trackChanges ?
1848 Change::INSERTED : Change::UNCHANGED));
1852 void Paragraph::insertChar(pos_type pos, char_type c,
1853 Font const & font, bool trackChanges)
1855 d->insertChar(pos, c, Change(trackChanges ?
1856 Change::INSERTED : Change::UNCHANGED));
1861 void Paragraph::insertChar(pos_type pos, char_type c,
1862 Font const & font, Change const & change)
1864 d->insertChar(pos, c, change);
1869 void Paragraph::resetFonts(Font const & font)
1871 d->fontlist_.clear();
1872 d->fontlist_.set(0, font);
1873 d->fontlist_.set(d->text_.size() - 1, font);
1876 // Gets uninstantiated font setting at position.
1877 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1881 LYXERR0("pos: " << pos << " size: " << size());
1885 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1886 if (cit != d->fontlist_.end())
1889 if (pos == size() && !empty())
1890 return getFontSettings(bparams, pos - 1);
1892 // Optimisation: avoid a full font instantiation if there is no
1893 // language change from previous call.
1894 static Font previous_font;
1895 static Language const * previous_lang = nullptr;
1896 Language const * lang = getParLanguage(bparams);
1897 if (lang != previous_lang) {
1898 previous_lang = lang;
1899 previous_font = Font(inherit_font, lang);
1901 return previous_font;
1905 FontSpan Paragraph::fontSpan(pos_type pos) const
1907 LBUFERR(pos <= size());
1910 return FontSpan(pos, pos);
1913 FontList::const_iterator cit = d->fontlist_.begin();
1914 FontList::const_iterator end = d->fontlist_.end();
1915 for (; cit != end; ++cit) {
1916 if (cit->pos() >= pos) {
1917 if (pos >= beginOfBody())
1918 return FontSpan(max(start, beginOfBody()),
1921 return FontSpan(start,
1922 min(beginOfBody() - 1,
1925 start = cit->pos() + 1;
1928 // This should not happen, but if so, we take no chances.
1929 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1930 LASSERT(false, /**/);
1931 return FontSpan(pos, pos);
1935 // Gets uninstantiated font setting at position 0
1936 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1938 if (!empty() && !d->fontlist_.empty())
1939 return d->fontlist_.begin()->font();
1941 // Optimisation: avoid a full font instantiation if there is no
1942 // language change from previous call.
1943 static Font previous_font;
1944 static Language const * previous_lang = nullptr;
1945 if (bparams.language != previous_lang) {
1946 previous_lang = bparams.language;
1947 previous_font = Font(inherit_font, bparams.language);
1950 return previous_font;
1954 // Gets the fully instantiated font at a given position in a paragraph
1955 // This is basically the same function as Text::GetFont() in text2.cpp.
1956 // The difference is that this one is used for generating the LaTeX file,
1957 // and thus cosmetic "improvements" are disallowed: This has to deliver
1958 // the true picture of the buffer. (Asger)
1959 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1960 Font const & outerfont) const
1964 Font font = getFontSettings(bparams, pos);
1966 pos_type const body_pos = beginOfBody();
1967 FontInfo & fi = font.fontInfo();
1969 fi.realize(d->layout_->labelfont);
1971 fi.realize(d->layout_->font);
1973 fi.realize(outerfont.fontInfo());
1974 fi.realize(bparams.getFont().fontInfo());
1980 Font const Paragraph::getLabelFont
1981 (BufferParams const & bparams, Font const & outerfont) const
1983 FontInfo tmpfont = d->layout_->labelfont;
1984 tmpfont.realize(outerfont.fontInfo());
1985 tmpfont.realize(bparams.getFont().fontInfo());
1986 return Font(tmpfont, getParLanguage(bparams));
1990 Font const Paragraph::getLayoutFont
1991 (BufferParams const & bparams, Font const & outerfont) const
1993 FontInfo tmpfont = d->layout_->font;
1994 tmpfont.realize(outerfont.fontInfo());
1995 tmpfont.realize(bparams.getFont().fontInfo());
1996 return Font(tmpfont, getParLanguage(bparams));
2000 char_type Paragraph::getUChar(BufferParams const & bparams,
2001 OutputParams const & rp,
2004 char_type c = d->text_[pos];
2006 // Return unchanged character in LTR languages
2007 // or if we use poylglossia/bidi (XeTeX).
2008 if (rp.useBidiPackage()
2009 || !getFontSettings(bparams, pos).isRightToLeft())
2012 // Without polyglossia/bidi, we need to account for some special cases.
2013 // FIXME This needs to be audited!
2015 // * The input is as expected for all delimiters
2016 // => checked for Hebrew!
2017 // * The output matches the display in the LyX workarea
2018 // => checked for Hebrew!
2019 // * The special cases below are really necessary
2020 // => checked for Hebrew!
2021 // * In arabic_arabi, brackets are transformed to Arabic
2022 // Ornate Parentheses. Is this is really wanted?
2024 string const & lang = getFontSettings(bparams, pos).language()->lang();
2027 // 1. In the following languages, parentheses need to be reversed.
2028 // Also with polyglodia/luabidi
2029 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
2031 // 2. In the following languages, brackets don't need to be reversed.
2032 bool const reversebrackets = lang != "arabic_arabtex"
2033 && lang != "arabic_arabi"
2036 // Now swap delimiters if needed.
2047 if (reversebrackets)
2051 if (reversebrackets)
2072 void Paragraph::setFont(pos_type pos, Font const & font)
2074 LASSERT(pos <= size(), return);
2076 // First, reduce font against layout/label font
2077 // Update: The setCharFont() routine in text2.cpp already
2078 // reduces font, so we don't need to do that here. (Asger)
2080 d->fontlist_.set(pos, font);
2084 void Paragraph::makeSameLayout(Paragraph const & par)
2086 d->layout_ = par.d->layout_;
2087 d->params_ = par.d->params_;
2091 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2093 if (isFreeSpacing())
2099 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2100 if (eraseChar(pos, trackChanges))
2106 return count > 0 || pos > 0;
2110 bool Paragraph::hasSameLayout(Paragraph const & par) const
2112 return par.d->layout_ == d->layout_
2113 && d->params_.sameLayout(par.d->params_);
2117 depth_type Paragraph::getDepth() const
2119 return d->params_.depth();
2123 depth_type Paragraph::getMaxDepthAfter() const
2125 if (d->layout_->isEnvironment())
2126 return d->params_.depth() + 1;
2128 return d->params_.depth();
2132 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
2134 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2135 return getDefaultAlign(bparams);
2137 return d->params_.align();
2141 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2143 LyXAlignment res = layout().align;
2144 if (isRTL(bparams)) {
2146 if (res == LYX_ALIGN_LEFT)
2147 res = LYX_ALIGN_RIGHT;
2148 else if (res == LYX_ALIGN_RIGHT)
2149 res = LYX_ALIGN_LEFT;
2155 docstring const & Paragraph::labelString() const
2157 return d->params_.labelString();
2161 // the next two functions are for the manual labels
2162 docstring const Paragraph::getLabelWidthString() const
2164 if (d->layout_->margintype == MARGIN_MANUAL
2165 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2166 return d->params_.labelWidthString();
2168 return _("Senseless with this layout!");
2172 void Paragraph::setLabelWidthString(docstring const & s)
2174 d->params_.labelWidthString(s);
2178 docstring Paragraph::expandLabel(Layout const & layout,
2179 BufferParams const & bparams) const
2181 return expandParagraphLabel(layout, bparams, true);
2185 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2186 BufferParams const & bparams, bool process_appendix) const
2188 DocumentClass const & tclass = bparams.documentClass();
2189 string const & lang = getParLanguage(bparams)->code();
2190 bool const in_appendix = process_appendix && d->params_.appendix();
2191 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2193 if (fmt.empty() && !layout.counter.empty())
2194 return tclass.counters().theCounter(layout.counter, lang);
2196 // handle 'inherited level parts' in 'fmt',
2197 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2198 size_t const i = fmt.find('@', 0);
2199 if (i != docstring::npos) {
2200 size_t const j = fmt.find('@', i + 1);
2201 if (j != docstring::npos) {
2202 docstring parent(fmt, i + 1, j - i - 1);
2203 docstring label = from_ascii("??");
2204 if (tclass.hasLayout(parent))
2205 label = expandParagraphLabel(tclass[parent], bparams,
2207 fmt = docstring(fmt, 0, i) + label
2208 + docstring(fmt, j + 1, docstring::npos);
2212 return tclass.counters().counterLabel(fmt, lang);
2216 void Paragraph::applyLayout(Layout const & new_layout)
2218 d->layout_ = &new_layout;
2219 LyXAlignment const oldAlign = d->params_.align();
2221 if (!(oldAlign & d->layout_->alignpossible)) {
2222 frontend::Alert::warning(_("Alignment not permitted"),
2223 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2224 d->params_.align(LYX_ALIGN_LAYOUT);
2229 pos_type Paragraph::beginOfBody() const
2231 return d->begin_of_body_;
2235 void Paragraph::setBeginOfBody()
2237 if (d->layout_->labeltype != LABEL_MANUAL) {
2238 d->begin_of_body_ = 0;
2242 // Unroll the first two cycles of the loop
2243 // and remember the previous character to
2244 // remove unnecessary getChar() calls
2246 pos_type end = size();
2247 bool prev_char_deleted = false;
2248 if (i < end && (!(isNewline(i) || isEnvSeparator(i)) || isDeleted(i))) {
2251 char_type previous_char = d->text_[i];
2252 if (!(isNewline(i) || isEnvSeparator(i))) {
2254 while (i < end && (previous_char != ' ' || prev_char_deleted)) {
2255 char_type temp = d->text_[i];
2256 prev_char_deleted = isDeleted(i);
2257 if (!isDeleted(i) && (isNewline(i) || isEnvSeparator(i)))
2260 previous_char = temp;
2266 d->begin_of_body_ = i;
2270 bool Paragraph::allowParagraphCustomization() const
2272 return inInset().allowParagraphCustomization();
2276 bool Paragraph::usePlainLayout() const
2278 return inInset().usePlainLayout();
2282 bool Paragraph::isPassThru() const
2284 return inInset().isPassThru() || d->layout_->pass_thru;
2288 bool Paragraph::parbreakIsNewline() const
2290 return inInset().getLayout().parbreakIsNewline() || d->layout_->parbreak_is_newline;
2294 bool Paragraph::isPartOfTextSequence() const
2296 for (pos_type i = 0; i < size(); ++i) {
2297 if (!isInset(i) || getInset(i)->isPartOfTextSequence())
2305 // paragraphs inside floats need different alignment tags to avoid
2308 bool noTrivlistCentering(InsetCode code)
2310 return code == FLOAT_CODE
2311 || code == WRAP_CODE
2312 || code == CELL_CODE;
2316 string correction(string const & orig)
2318 if (orig == "flushleft")
2319 return "raggedright";
2320 if (orig == "flushright")
2321 return "raggedleft";
2322 if (orig == "center")
2328 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2329 InsetCode code, bool const lastpar, int & col)
2331 string macro = suffix + "{";
2332 if (noTrivlistCentering(code)) {
2334 // the last paragraph in non-trivlist-aligned
2335 // context is special (to avoid unwanted whitespace)
2336 if (suffix == "\\begin") {
2337 macro = "\\" + correction(env) + "{}";
2338 os << from_ascii(macro);
2339 col += macro.size();
2344 macro += correction(env);
2348 if (suffix == "\\par\\end") {
2352 os << from_ascii(macro);
2353 col += macro.size();
2354 if (suffix == "\\begin") {
2364 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2365 otexstream & os, OutputParams const & runparams) const
2370 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2371 (layout_->toggle_indent != ITOGGLE_NEVER) :
2372 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2374 LyXAlignment const curAlign = params_.align();
2376 // Do not output \\noindent for paragraphs
2377 // 1. that cannot have indentation or are indented always,
2378 // 2. that are not part of the immediate text sequence (e.g., contain only floats),
2379 // 3. that are PassThru,
2380 // 4. or that are centered.
2381 if (canindent && params_.noindent()
2382 && owner_->isPartOfTextSequence()
2383 && !layout_->pass_thru
2384 && curAlign != LYX_ALIGN_CENTER) {
2385 if (!owner_->empty()
2386 && owner_->getInset(0)
2387 && owner_->getInset(0)->lyxCode() == VSPACE_CODE)
2388 // If the paragraph starts with a vspace, the \\noindent
2389 // needs to come after that (as it leaves vmode).
2390 // If the paragraph consists only of the vspace,
2391 // \\noindent is not needed at all.
2392 runparams.need_noindent = owner_->size() > 1;
2394 os << "\\noindent" << termcmd;
2399 if (curAlign == layout_->align)
2403 case LYX_ALIGN_NONE:
2404 case LYX_ALIGN_BLOCK:
2405 case LYX_ALIGN_LAYOUT:
2406 case LYX_ALIGN_SPECIAL:
2407 case LYX_ALIGN_DECIMAL:
2409 case LYX_ALIGN_LEFT:
2410 case LYX_ALIGN_RIGHT:
2411 case LYX_ALIGN_CENTER:
2412 if (runparams.moving_arg) {
2419 string const begin_tag = "\\begin";
2420 InsetCode code = ownerCode();
2421 bool const lastpar = runparams.isLastPar;
2422 // RTL in classic (PDF)LaTeX (without the Bidi package)
2423 // Luabibdi (used by LuaTeX) behaves like classic
2424 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2425 && !runparams.useBidiPackage();
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: {
2436 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2437 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2439 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2441 } case LYX_ALIGN_RIGHT: {
2443 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2444 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2446 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2448 } case LYX_ALIGN_CENTER: {
2449 corrected_env(os, begin_tag, "center", code, lastpar, column);
2458 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2459 otexstream & os, OutputParams const & runparams) const
2461 LyXAlignment const curAlign = params_.align();
2463 if (curAlign == layout_->align)
2467 case LYX_ALIGN_NONE:
2468 case LYX_ALIGN_BLOCK:
2469 case LYX_ALIGN_LAYOUT:
2470 case LYX_ALIGN_SPECIAL:
2471 case LYX_ALIGN_DECIMAL:
2473 case LYX_ALIGN_LEFT:
2474 case LYX_ALIGN_RIGHT:
2475 case LYX_ALIGN_CENTER:
2476 if (runparams.moving_arg)
2481 bool output = false;
2483 string const end_tag = "\\par\\end";
2484 InsetCode code = ownerCode();
2485 bool const lastpar = runparams.isLastPar;
2486 // RTL in classic (PDF)LaTeX (without the Bidi package)
2487 // Luabibdi (used by LuaTeX) behaves like classic
2488 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2489 && !runparams.useBidiPackage();
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: {
2500 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2501 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2503 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2505 } case LYX_ALIGN_RIGHT: {
2507 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2508 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2510 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2512 } case LYX_ALIGN_CENTER: {
2513 corrected_env(os, end_tag, "center", code, lastpar, col);
2518 return output || lastpar;
2522 // This one spits out the text of the paragraph
2523 void Paragraph::latex(BufferParams const & bparams,
2524 Font const & outerfont,
2526 OutputParams const & runparams,
2527 int start_pos, int end_pos, bool force) const
2529 LYXERR(Debug::OUTFILE, "Paragraph::latex... " << this);
2531 // FIXME This check should not be needed. Perhaps issue an
2532 // error if it triggers.
2533 Layout const & style = inInset().forcePlainLayout() ?
2534 bparams.documentClass().plainLayout() : *d->layout_;
2536 if (!force && style.inpreamble)
2539 bool const allowcust = allowParagraphCustomization();
2541 // Current base font for all inherited font changes, without any
2542 // change caused by an individual character, except for the language:
2543 // It is set to the language of the first character.
2544 // As long as we are in the label, this font is the base font of the
2545 // label. Before the first body character it is set to the base font
2549 // If there is an open font-encoding changing command (script wrapper),
2550 // alien_script is set to its name
2551 string alien_script;
2554 // Maybe we have to create a optional argument.
2555 pos_type body_pos = beginOfBody();
2556 unsigned int column = 0;
2558 // If we are inside an non inheritFont() inset,
2559 // the outerfont is the buffer's main font
2560 Font const real_outerfont =
2561 inInset().inheritFont() ? outerfont : Font(bparams.getFont());
2564 // the optional argument is kept in curly brackets in
2565 // case it contains a ']'
2566 // This is not strictly needed, but if this is changed it
2567 // would be a file format change, and tex2lyx would need
2568 // to be adjusted, since it unconditionally removes the
2569 // braces when it parses \item.
2572 basefont = getLabelFont(bparams, real_outerfont);
2574 basefont = getLayoutFont(bparams, real_outerfont);
2577 // Which font is currently active?
2578 Font running_font(basefont);
2579 // Do we have an open font change?
2580 bool open_font = false;
2582 Change runningChange =
2583 runparams.inDeletedInset && !inInset().canTrackChanges()
2584 ? runparams.changeOfDeletedInset : Change(Change::UNCHANGED);
2586 Encoding const * const prev_encoding = runparams.encoding;
2588 os.texrow().start(id(), 0);
2590 // if the paragraph is empty, the loop will not be entered at all
2592 // For InTitle commands, we have already opened a group
2593 // in output_latex::TeXOnePar.
2594 if (style.isCommand() && !style.intitle) {
2598 if (!style.leftdelim().empty()) {
2599 os << style.leftdelim();
2600 column += style.leftdelim().size();
2603 column += d->startTeXParParams(bparams, os, runparams);
2606 // Whether a \par can be issued for insets typeset inline with text.
2607 // Yes if greater than 0. This has to be static.
2608 THREAD_LOCAL_STATIC int parInline = 0;
2610 for (pos_type i = 0; i < size(); ++i) {
2611 // First char in paragraph or after label?
2612 if (i == body_pos) {
2615 bool needPar = false;
2616 column += running_font.latexWriteEndChanges(
2617 os, bparams, runparams,
2618 basefont, basefont, needPar);
2621 basefont = getLayoutFont(bparams, real_outerfont);
2622 running_font = basefont;
2624 column += Changes::latexMarkChange(os, bparams,
2625 runningChange, Change(Change::UNCHANGED),
2627 runningChange = Change(Change::UNCHANGED);
2629 os << ((isEnvSeparator(i) && !runparams.find_effective()) ? "}]~" : "}] ");
2632 // For InTitle commands, we have already opened a group
2633 // in output_latex::TeXOnePar.
2634 if (style.isCommand() && !style.intitle) {
2639 if (!style.leftdelim().empty()) {
2640 os << style.leftdelim();
2641 column += style.leftdelim().size();
2645 column += d->startTeXParParams(bparams, os,
2649 runparams.wasDisplayMath = runparams.inDisplayMath;
2650 runparams.inDisplayMath = false;
2651 bool deleted_display_math = false;
2652 Change const & change = runparams.inDeletedInset
2653 ? runparams.changeOfDeletedInset : lookupChange(i);
2655 char_type const c = d->text_[i];
2657 // Check whether a display math inset follows
2658 bool output_changes;
2659 if (!runparams.find_effective())
2660 output_changes = bparams.output_changes;
2662 output_changes = runparams.find_with_deleted();
2664 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2666 runparams.ctObject = getInset(i)->getCtObject(runparams);
2668 InsetMath const * im = getInset(i)->asInsetMath();
2669 if (im && im->asHullInset()
2670 && im->asHullInset()->outerDisplay()) {
2671 runparams.inDisplayMath = true;
2672 // runparams.inDeletedInset will be set by
2673 // latexInset later, but we need this info
2674 // before it is called. On the other hand, we
2675 // cannot set it here because it is a counter.
2676 deleted_display_math = isDeleted(i);
2678 if (output_changes && deleted_display_math
2679 && runningChange == change
2680 && change.type == Change::DELETED
2681 && !os.afterParbreak()) {
2682 // A display math in the same paragraph follows.
2683 // We have to close and then reopen \lyxdeleted,
2684 // otherwise the math will be shifted up.
2685 OutputParams rp = runparams;
2687 bool needPar = false;
2688 column += running_font.latexWriteEndChanges(
2689 os, bparams, rp, basefont,
2693 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2694 : getLayoutFont(bparams, real_outerfont);
2695 running_font = basefont;
2696 column += Changes::latexMarkChange(os, bparams,
2697 Change(Change::INSERTED), change, rp);
2701 if (output_changes && runningChange != change) {
2702 if (!alien_script.empty()) {
2705 alien_script.clear();
2708 bool needPar = false;
2709 column += running_font.latexWriteEndChanges(
2710 os, bparams, runparams,
2711 basefont, basefont, needPar);
2714 basefont = (body_pos > i) ? getLabelFont(bparams, real_outerfont)
2715 : getLayoutFont(bparams, real_outerfont);
2716 running_font = basefont;
2717 column += Changes::latexMarkChange(os, bparams, runningChange,
2719 runningChange = change;
2722 // do not output text which is marked deleted
2723 // if change tracking output is disabled
2724 if (!output_changes && change.deleted()) {
2730 // Fully instantiated font
2731 Font current_font = getFont(bparams, i, outerfont);
2733 Font const prev_font = (i > 0) ?
2734 getFont(bparams, i - 1, outerfont)
2737 Font const last_font = running_font;
2738 bool const in_ct_deletion = (output_changes
2739 && runningChange == change
2740 && change.type == Change::DELETED
2741 && !os.afterParbreak());
2742 // Insets where font switches are used (rather than font commands)
2743 bool const fontswitch_inset =
2746 && getInset(i)->allowMultiPar()
2747 && getInset(i)->lyxCode() != ERT_CODE
2748 && (getInset(i)->producesOutput()
2749 // FIXME Something more general?
2750 // Comments do not "produce output" but are still
2751 // part of the TeX source and require font switches
2752 // to be closed (otherwise LaTeX fails).
2753 || getInset(i)->layoutName() == "Note:Comment");
2755 bool closeLanguage = false;
2756 bool lang_switched_at_inset = false;
2757 if (fontswitch_inset) {
2758 // Some insets cannot be inside a font change command.
2759 // However, even such insets *can* be placed in \L or \R
2760 // or their equivalents (for RTL language switches),
2761 // so we don't close the language in those cases
2762 // (= differing isRightToLeft()).
2763 // ArabTeX, though, doesn't seem to handle this special behavior.
2764 closeLanguage = basefont.isRightToLeft() == current_font.isRightToLeft()
2765 || basefont.language()->lang() == "arabic_arabtex"
2766 || current_font.language()->lang() == "arabic_arabtex";
2767 // We need to check prev_font as language changes directly at inset
2768 // will only be started inside the inset.
2769 lang_switched_at_inset = prev_font.language() != current_font.language();
2772 // Do we need to close the previous font?
2773 bool langClosed = false;
2775 ((current_font != running_font
2776 || current_font.language() != running_font.language())
2777 || (fontswitch_inset
2778 && (current_font == prev_font))))
2780 // ensure there is no open script-wrapper
2781 if (!alien_script.empty()) {
2784 alien_script.clear();
2786 if (in_ct_deletion) {
2787 // We have to close and then reopen \lyxdeleted,
2788 // as strikeout needs to be on lowest level.
2792 if (closeLanguage) {
2793 // Force language closing
2794 current_font.setLanguage(basefont.language());
2797 Font const nextfont = (i == body_pos-1) ? basefont : current_font;
2798 bool needPar = false;
2799 column += running_font.latexWriteEndChanges(
2800 os, bparams, runparams, basefont,
2802 if (in_ct_deletion) {
2803 // We have to close and then reopen \lyxdeleted,
2804 // as strikeout needs to be on lowest level.
2805 OutputParams rp = runparams;
2806 column += Changes::latexMarkChange(os, bparams,
2807 Change(Change::UNCHANGED), Change(Change::DELETED), rp);
2809 // Has the language been closed in the latexWriteEndChanges() call above?
2810 langClosed |= running_font.language() != basefont.language()
2811 && running_font.language() != nextfont.language()
2812 && (running_font.language()->encoding()->package() != Encoding::CJK);
2813 running_font = basefont;
2814 open_font &= !langClosed;
2817 // if necessary, close language environment before opening CJK
2818 string const running_lang = running_font.language()->babel();
2819 string const lang_end_command = lyxrc.language_command_end;
2820 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2821 && !running_lang.empty()
2822 && running_lang == openLanguageName()
2823 && current_font.language()->encoding()->package() == Encoding::CJK) {
2824 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2825 os << from_ascii(end_tag);
2826 column += end_tag.length();
2827 if (!languageStackEmpty())
2831 // Switch file encoding if necessary (and allowed)
2832 if ((!fontswitch_inset || closeLanguage)
2833 && !runparams.pass_thru && !style.pass_thru &&
2834 runparams.encoding->package() != Encoding::none &&
2835 current_font.language()->encoding()->package() != Encoding::none) {
2836 pair<bool, int> const enc_switch =
2837 switchEncoding(os.os(), bparams, runparams,
2838 *(current_font.language()->encoding()));
2839 if (enc_switch.first) {
2840 column += enc_switch.second;
2841 runparams.encoding = current_font.language()->encoding();
2845 // A display math inset inside an ulem command will be output
2846 // as a box of width \linewidth, so we have to either disable
2847 // indentation if the inset starts a paragraph, or start a new
2848 // line to accommodate such box. This has to be done before
2849 // writing any font changing commands.
2850 if (runparams.inDisplayMath && !deleted_display_math
2851 && runparams.inulemcmd) {
2852 if (os.afterParbreak())
2858 // Do we need to change font?
2859 if ((current_font != running_font ||
2860 current_font.language() != running_font.language())
2861 && i != body_pos - 1)
2863 if (!fontswitch_inset) {
2864 if (in_ct_deletion) {
2865 // We have to close and then reopen \lyxdeleted,
2866 // as strikeout needs to be on lowest level.
2867 OutputParams rp = runparams;
2868 bool needPar = false;
2869 column += running_font.latexWriteEndChanges(
2870 os, bparams, rp, basefont,
2875 otexstringstream ots;
2876 InsetText const * textinset = inInset().asInsetText();
2877 bool const cprotect = textinset
2878 ? textinset->hasCProtectContent(runparams.moving_arg)
2879 && !textinset->text().isMainText()
2880 && inInset().lyxCode() != BRANCH_CODE
2882 column += current_font.latexWriteStartChanges(ots, bparams,
2883 runparams, basefont, last_font, false,
2885 // Check again for display math in ulem commands as a
2886 // font change may also occur just before a math inset.
2887 if (runparams.inDisplayMath && !deleted_display_math
2888 && runparams.inulemcmd) {
2889 if (os.afterParbreak())
2894 running_font = current_font;
2896 docstring fontchange = ots.str();
2898 // check whether the fontchange ends with a \\textcolor
2899 // modifier and the text starts with a space. If so we
2900 // need to add } in order to prevent \\textcolor from gobbling
2901 // the space (bug 4473).
2902 docstring const last_modifier = rsplit(fontchange, '\\');
2903 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2904 os << from_ascii("{}");
2905 else if (ots.terminateCommand())
2907 if (in_ct_deletion) {
2908 // We have to close and then reopen \lyxdeleted,
2909 // as strikeout needs to be on lowest level.
2910 OutputParams rp = runparams;
2911 column += Changes::latexMarkChange(os, bparams,
2912 Change(Change::UNCHANGED), change, rp);
2914 } else {// if fontswitch_inset
2915 if (current_font != running_font || !langClosed)
2916 // font is still open in fontswitch_insets if we have
2917 // a non-lang font difference or if the language
2918 // is the only difference but has not been forcedly
2921 running_font = current_font;
2925 // FIXME: think about end_pos implementation...
2926 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2927 // FIXME: integrate this case in latexSpecialChar
2928 // Do not print the separation of the optional argument
2929 // if style.pass_thru is false. This works because
2930 // latexSpecialChar ignores spaces if
2931 // style.pass_thru is false.
2932 if (i != body_pos - 1) {
2933 if (d->simpleTeXBlanks(bparams, runparams, os,
2934 i, column, current_font, style)) {
2935 // A surrogate pair was output. We
2936 // must not call latexSpecialChar
2937 // in this iteration, since it would output
2938 // the combining character again.
2945 OutputParams rp = runparams;
2946 rp.free_spacing = style.free_spacing;
2947 rp.local_font = ¤t_font;
2948 rp.intitle = style.intitle;
2950 // Two major modes: LaTeX or plain
2951 // Handle here those cases common to both modes
2952 // and then split to handle the two modes separately.
2953 if (c == META_INSET) {
2954 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2955 // Greyedout notes and, in general, all insets
2956 // with InsetLayout::isDisplay() == false,
2957 // are typeset inline with the text. So, we
2958 // can add a \par to the last paragraph of
2959 // such insets only if nothing else follows.
2960 bool incremented = false;
2961 Inset const * inset = getInset(i);
2962 InsetText const * textinset = inset
2963 ? inset->asInsetText()
2965 if (i + 1 == size() && textinset
2966 && !inset->getLayout().isDisplay()) {
2967 ParagraphList const & pars =
2968 textinset->text().paragraphs();
2969 pit_type const pit = pars.size() - 1;
2970 Font const lastfont =
2971 pit < 0 || pars[pit].empty()
2972 ? pars[pit].getLayoutFont(
2975 : pars[pit].getFont(bparams,
2976 pars[pit].size() - 1,
2978 if (lastfont.fontInfo().size() !=
2979 basefont.fontInfo().size()) {
2984 // We need to restore parts of this after insets with
2985 // allowMultiPar() true
2986 Font const save_basefont = basefont;
2987 d->latexInset(bparams, os, rp, running_font,
2988 basefont, real_outerfont, open_font,
2989 runningChange, style, i, column, fontswitch_inset,
2990 closeLanguage, (lang_switched_at_inset || langClosed));
2991 if (fontswitch_inset) {
2993 bool needPar = false;
2994 column += running_font.latexWriteEndChanges(
2995 os, bparams, runparams,
2996 basefont, basefont, needPar);
2999 basefont.fontInfo().setSize(save_basefont.fontInfo().size());
3000 basefont.fontInfo().setFamily(save_basefont.fontInfo().family());
3001 basefont.fontInfo().setSeries(save_basefont.fontInfo().series());
3006 if (runparams.ctObject == CtObject::DisplayObject
3007 || runparams.ctObject == CtObject::UDisplayObject) {
3008 // Close \lyx*deleted and force its
3009 // reopening (if needed)
3012 runningChange = Change(Change::UNCHANGED);
3013 runparams.ctObject = CtObject::Normal;
3016 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
3017 if (!bparams.useNonTeXFonts)
3018 script = Encodings::isKnownScriptChar(c);
3019 if (script != alien_script) {
3020 if (!alien_script.empty()) {
3022 alien_script.clear();
3024 string fontenc = running_font.language()->fontenc(bparams);
3026 && !Encodings::fontencSupportsScript(fontenc, script)) {
3027 column += script.length() + 2;
3028 os << "\\" << script << "{";
3029 alien_script = script;
3033 d->latexSpecialChar(os, bparams, rp, running_font,
3034 alien_script, style, i, end_pos, column);
3035 } catch (EncodingException & e) {
3036 if (runparams.dryrun) {
3037 os << "<" << _("LyX Warning: ")
3038 << _("uncodable character") << " '";
3042 // add location information and throw again.
3050 // Set the encoding to that returned from latexSpecialChar (see
3051 // comment for encoding member in OutputParams.h)
3052 runparams.encoding = rp.encoding;
3054 // Also carry on the info on a closed ulem command for insets
3055 // such as Note that do not produce any output, so that no
3056 // command is ever executed but its opening was recorded.
3057 runparams.inulemcmd = rp.inulemcmd;
3059 // These need to be passed upstream as well
3060 runparams.need_maketitle = rp.need_maketitle;
3061 runparams.have_maketitle = rp.have_maketitle;
3063 // And finally, pass the post_macros upstream
3064 runparams.post_macro = rp.post_macro;
3067 // Close wrapper for alien script
3068 if (!alien_script.empty()) {
3070 alien_script.clear();
3073 Font const font = empty()
3074 ? getLayoutFont(bparams, real_outerfont)
3075 : getFont(bparams, size() - 1, real_outerfont);
3077 InsetText const * textinset = inInset().asInsetText();
3079 bool const maintext = textinset
3080 ? textinset->text().isMainText()
3083 size_t const numpars = textinset
3084 ? textinset->text().paragraphs().size()
3087 bool needPar = false;
3089 if (style.resfont.size() != font.fontInfo().size()
3090 && (!runparams.isLastPar || maintext
3091 || (numpars > 1 && d->ownerCode() != CELL_CODE
3092 && (inInset().getLayout().isDisplay()
3094 && !style.isCommand()) {
3098 // If we have an open font definition, we have to close it
3100 // Make sure that \\par is done with the font of the last
3101 // character if this has another size as the default.
3102 // This is necessary because LaTeX (and LyX on the screen)
3103 // calculates the space between the baselines according
3104 // to this font. (Matthias)
3106 // We must not change the font for the last paragraph
3107 // of non-multipar insets, tabular cells or commands,
3108 // since this produces unwanted whitespace.
3109 #ifdef FIXED_LANGUAGE_END_DETECTION
3111 running_font.latexWriteEndChanges(os, bparams,
3112 runparams, basefont,
3113 next_->getFont(bparams, 0, outerfont),
3116 running_font.latexWriteEndChanges(os, bparams,
3117 runparams, basefont, basefont, needPar);
3120 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
3121 //FIXME: there as we start another \selectlanguage with the next paragraph if
3122 //FIXME: we are in need of this. This should be fixed sometime (Jug)
3123 running_font.latexWriteEndChanges(os, bparams, runparams,
3124 basefont, basefont, needPar);
3128 // The \par could not be inserted at the same nesting
3129 // level of the font size change, so do it now.
3130 os << "{\\" << font.latexSize() << "\\par}";
3133 if (!runparams.inDeletedInset || inInset().canTrackChanges())
3134 column += Changes::latexMarkChange(os, bparams, runningChange,
3135 Change(Change::UNCHANGED), runparams);
3137 // Needed if there is an optional argument but no contents.
3138 if (body_pos > 0 && body_pos == size()) {
3142 if (!style.rightdelim().empty()) {
3143 os << style.rightdelim();
3144 column += style.rightdelim().size();
3147 if (allowcust && d->endTeXParParams(bparams, os, runparams)
3148 && runparams.encoding != prev_encoding) {
3149 runparams.encoding = prev_encoding;
3150 os << setEncoding(prev_encoding->iconvName());
3153 LYXERR(Debug::OUTFILE, "Paragraph::latex... done " << this);
3157 bool Paragraph::emptyTag() const
3159 for (pos_type i = 0; i < size(); ++i) {
3160 if (Inset const * inset = getInset(i)) {
3161 InsetCode lyx_code = inset->lyxCode();
3162 // FIXME testing like that is wrong. What is
3164 if (lyx_code != TOC_CODE &&
3165 lyx_code != INCLUDE_CODE &&
3166 lyx_code != GRAPHICS_CODE &&
3167 lyx_code != ERT_CODE &&
3168 lyx_code != LISTINGS_CODE &&
3169 lyx_code != FLOAT_CODE &&
3170 lyx_code != TABULAR_CODE) {
3174 char_type c = d->text_[i];
3175 if (c != ' ' && c != '\t')
3183 string Paragraph::getID(Buffer const &, OutputParams const &)
3186 for (pos_type i = 0; i < size(); ++i) {
3187 if (Inset const * inset = getInset(i)) {
3188 InsetCode lyx_code = inset->lyxCode();
3189 if (lyx_code == LABEL_CODE) {
3190 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
3191 docstring const & id = il->getParam("name");
3192 return "id='" + to_utf8(xml::cleanID(id)) + "'";
3200 pos_type Paragraph::firstWordDocBook(XMLStream & xs, OutputParams const & runparams) const
3203 for (i = 0; i < size(); ++i) {
3204 if (Inset const * inset = getInset(i)) {
3205 inset->docbook(xs, runparams);
3207 char_type c = d->text_[i];
3217 pos_type Paragraph::firstWordLyXHTML(XMLStream & xs, OutputParams const & runparams)
3221 for (i = 0; i < size(); ++i) {
3222 if (Inset const * inset = getInset(i)) {
3223 inset->xhtml(xs, runparams);
3225 char_type c = d->text_[i];
3235 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3238 pos_type size = text_.size();
3239 for (pos_type i = initial; i < size; ++i) {
3240 Font font = owner_->getFont(buf.params(), i, outerfont);
3241 if (text_[i] == META_INSET)
3243 if (i != initial && font != font_old)
3254 void doFontSwitchDocBook(vector<xml::FontTag> & tagsToOpen,
3255 vector<xml::EndFontTag> & tagsToClose,
3256 bool & flag, FontState curstate, xml::FontTypes type)
3258 if (curstate == FONT_ON) {
3259 tagsToOpen.push_back(docbookStartFontTag(type));
3262 tagsToClose.push_back(docbookEndFontTag(type));
3267 class OptionalFontType {
3272 OptionalFontType(): ft(xml::FT_EMPH), has_value(false) {} // A possible value at random for ft.
3273 OptionalFontType(xml::FontTypes ft): ft(ft), has_value(true) {}
3276 OptionalFontType fontShapeToXml(FontShape fs)
3280 return {xml::FT_ITALIC};
3282 return {xml::FT_SLANTED};
3283 case SMALLCAPS_SHAPE:
3284 return {xml::FT_SMALLCAPS};
3289 // the other tags are for internal use
3295 OptionalFontType fontFamilyToXml(FontFamily fm)
3299 return {xml::FT_ROMAN};
3301 return {xml::FT_SANS};
3302 case TYPEWRITER_FAMILY:
3303 return {xml::FT_TYPE};
3304 case INHERIT_FAMILY:
3307 // the other tags are for internal use
3313 OptionalFontType fontSizeToXml(FontSize fs)
3317 return {xml::FT_SIZE_TINY};
3319 return {xml::FT_SIZE_SCRIPT};
3321 return {xml::FT_SIZE_FOOTNOTE};
3323 return {xml::FT_SIZE_SMALL};
3325 return {xml::FT_SIZE_LARGE};
3327 return {xml::FT_SIZE_LARGER};
3329 return {xml::FT_SIZE_LARGEST};
3331 return {xml::FT_SIZE_HUGE};
3333 return {xml::FT_SIZE_HUGER};
3335 return {xml::FT_SIZE_INCREASE};
3337 return {xml::FT_SIZE_DECREASE};
3342 // the other tags are for internal use
3348 struct DocBookFontState
3350 FontShape curr_fs = INHERIT_SHAPE;
3351 FontFamily curr_fam = INHERIT_FAMILY;
3352 FontSize curr_size = INHERIT_SIZE;
3354 // track whether we have opened these tags
3355 bool emph_flag = false;
3356 bool bold_flag = false;
3357 bool noun_flag = false;
3358 bool ubar_flag = false;
3359 bool dbar_flag = false;
3360 bool sout_flag = false;
3361 bool xout_flag = false;
3362 bool wave_flag = false;
3364 bool shap_flag = false;
3366 bool faml_flag = false;
3368 bool size_flag = false;
3371 std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>> computeDocBookFontSwitch(FontInfo const & font_old,
3373 std::string const & default_family,
3374 DocBookFontState & fs)
3376 vector<xml::FontTag> tagsToOpen;
3377 vector<xml::EndFontTag> tagsToClose;
3380 FontState curstate = font.fontInfo().emph();
3381 if (font_old.emph() != curstate)
3382 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.emph_flag, curstate, xml::FT_EMPH);
3385 curstate = font.fontInfo().noun();
3386 if (font_old.noun() != curstate)
3387 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.noun_flag, curstate, xml::FT_NOUN);
3390 curstate = font.fontInfo().underbar();
3391 if (font_old.underbar() != curstate)
3392 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.ubar_flag, curstate, xml::FT_UBAR);
3395 curstate = font.fontInfo().strikeout();
3396 if (font_old.strikeout() != curstate)
3397 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.sout_flag, curstate, xml::FT_SOUT);
3400 curstate = font.fontInfo().xout();
3401 if (font_old.xout() != curstate)
3402 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.xout_flag, curstate, xml::FT_XOUT);
3405 curstate = font.fontInfo().uuline();
3406 if (font_old.uuline() != curstate)
3407 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.dbar_flag, curstate, xml::FT_DBAR);
3410 curstate = font.fontInfo().uwave();
3411 if (font_old.uwave() != curstate)
3412 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.wave_flag, curstate, xml::FT_WAVE);
3415 // a little hackish, but allows us to reuse what we have.
3416 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3417 if (font_old.series() != font.fontInfo().series())
3418 doFontSwitchDocBook(tagsToOpen, tagsToClose, fs.bold_flag, curstate, xml::FT_BOLD);
3421 fs.curr_fs = font.fontInfo().shape();
3422 FontShape old_fs = font_old.shape();
3423 if (old_fs != fs.curr_fs) {
3425 OptionalFontType tag = fontShapeToXml(old_fs);
3427 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3428 fs.shap_flag = false;
3431 OptionalFontType tag = fontShapeToXml(fs.curr_fs);
3433 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3437 fs.curr_fam = font.fontInfo().family();
3438 FontFamily old_fam = font_old.family();
3439 if (old_fam != fs.curr_fam) {
3441 OptionalFontType tag = fontFamilyToXml(old_fam);
3443 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3444 fs.faml_flag = false;
3446 switch (fs.curr_fam) {
3448 // we will treat a "default" font family as roman, since we have
3449 // no other idea what to do.
3450 if (default_family != "rmdefault" && default_family != "default") {
3451 tagsToOpen.push_back(docbookStartFontTag(xml::FT_ROMAN));
3452 fs.faml_flag = true;
3456 if (default_family != "sfdefault") {
3457 tagsToOpen.push_back(docbookStartFontTag(xml::FT_SANS));
3458 fs.faml_flag = true;
3461 case TYPEWRITER_FAMILY:
3462 if (default_family != "ttdefault") {
3463 tagsToOpen.push_back(docbookStartFontTag(xml::FT_TYPE));
3464 fs.faml_flag = true;
3467 case INHERIT_FAMILY:
3470 // the other tags are for internal use
3477 fs.curr_size = font.fontInfo().size();
3478 FontSize old_size = font_old.size();
3479 if (old_size != fs.curr_size) {
3481 OptionalFontType tag = fontSizeToXml(old_size);
3483 tagsToClose.push_back(docbookEndFontTag(tag.ft));
3484 fs.size_flag = false;
3487 OptionalFontType tag = fontSizeToXml(fs.curr_size);
3488 if (tag.has_value) {
3489 tagsToOpen.push_back(docbookStartFontTag(tag.ft));
3490 fs.size_flag = true;
3494 return std::tuple<vector<xml::FontTag>, vector<xml::EndFontTag>>(tagsToOpen, tagsToClose);
3497 } // anonymous namespace
3500 std::tuple<std::vector<docstring>, std::vector<docstring>, std::vector<docstring>>
3501 Paragraph::simpleDocBookOnePar(Buffer const & buf,
3502 OutputParams const & runparams,
3503 Font const & outerfont,
3506 bool ignore_fonts) const
3508 std::vector<docstring> prependedParagraphs;
3509 std::vector<docstring> generatedParagraphs;
3510 std::vector<docstring> appendedParagraphs;
3511 odocstringstream os;
3513 // If there is an argument that must be output before the main tag, do it before handling the rest of the paragraph.
3514 // Also tag all arguments that shouldn't go in the main content right now, so that they are never generated at the
3516 OutputParams rp = runparams;
3517 for (pos_type i = initial; i < size(); ++i) {
3518 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3519 const InsetArgument * arg = getInset(i)->asInsetArgument();
3520 if (arg->docbookargumentbeforemaintag()) {
3521 auto xs_local = XMLStream(os);
3522 arg->docbook(xs_local, rp);
3524 prependedParagraphs.push_back(os.str());
3525 os.str(from_ascii(""));
3527 rp.docbook_prepended_arguments.insert(arg);
3528 } else if (arg->docbookargumentaftermaintag()) {
3529 rp.docbook_appended_arguments.insert(arg);
3534 // State variables for the main loop.
3535 auto xs = new XMLStream(os); // XMLStream has no copy constructor: to create a new object, the only solution
3536 // is to hold a pointer to the XMLStream (xs = XMLStream(os) is not allowed once the first object is built).
3537 std::vector<char_type> delayedChars; // When a font tag ends with a space, output it after the closing font tag.
3538 // This requires to store delayed characters at some point.
3540 DocBookFontState fs; // Track whether we have opened font tags
3541 DocBookFontState old_fs = fs;
3543 Layout const & style = *d->layout_;
3544 FontInfo font_old = style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3545 string const default_family = buf.masterBuffer()->params().fonts_default_family;
3547 vector<xml::FontTag> tagsToOpen;
3548 vector<xml::EndFontTag> tagsToClose;
3550 // Parsing main loop.
3551 for (pos_type i = initial; i < size(); ++i) {
3552 bool ignore_fonts_i = ignore_fonts
3553 || style.docbooknofontinside()
3554 || (getInset(i) && getInset(i)->getLayout().docbooknofontinside());
3556 // Don't show deleted material in the output.
3560 // If this is an InsetNewline, generate a new paragraph. Also reset the fonts, so that tags are closed in
3562 if (getInset(i) && getInset(i)->lyxCode() == NEWLINE_CODE) {
3563 if (!ignore_fonts_i)
3564 xs->closeFontTags();
3566 // Output one paragraph (i.e. one string entry in generatedParagraphs).
3567 generatedParagraphs.push_back(os.str());
3569 // Create a new XMLStream for the new paragraph, completely independent from the previous one. This implies
3570 // that the string stream must be reset.
3571 os.str(from_ascii(""));
3573 xs = new XMLStream(os);
3575 // Restore the fonts for the new paragraph, so that the right tags are opened for the new entry.
3576 if (!ignore_fonts_i) {
3577 font_old = outerfont.fontInfo();
3582 // Determine which tags should be opened or closed regarding fonts.
3583 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3584 tie(tagsToOpen, tagsToClose) = computeDocBookFontSwitch(font_old, font, default_family, fs);
3586 if (!ignore_fonts_i) {
3587 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3588 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3589 for (; cit != cen; ++cit)
3593 // Deal with the delayed characters *after* closing font tags.
3594 if (!delayedChars.empty()) {
3595 for (char_type c: delayedChars)
3597 delayedChars.clear();
3600 if (!ignore_fonts_i) {
3601 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
3602 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
3603 for (; sit != sen; ++sit)
3606 tagsToClose.clear();
3610 // Finally, write the next character or inset.
3611 if (Inset const * inset = getInset(i)) {
3612 bool inset_is_argument_elsewhere = getInset(i)->asInsetArgument() &&
3613 rp.docbook_appended_arguments.find(inset->asInsetArgument()) != rp.docbook_appended_arguments.end() &&
3614 rp.docbook_prepended_arguments.find(inset->asInsetArgument()) != rp.docbook_prepended_arguments.end();
3616 if ((!rp.for_toc || inset->isInToc()) && !inset_is_argument_elsewhere) {
3617 // Arguments may need to be output
3618 OutputParams np = rp;
3619 np.local_font = &font;
3621 // TODO: special case will bite here.
3622 np.docbook_in_par = true;
3623 inset->docbook(*xs, np);
3626 char_type c = getUChar(buf.masterBuffer()->params(), rp, i);
3627 if (lyx::isSpace(c) && !ignore_fonts)
3628 delayedChars.push_back(c);
3632 font_old = font.fontInfo();
3635 // FIXME, this code is just imported from XHTML
3636 // I'm worried about what happens if a branch, say, is itself
3637 // wrapped in some font stuff. I think that will not work.
3639 xs->closeFontTags();
3641 // Deal with the delayed characters *after* closing font tags.
3642 if (!delayedChars.empty())
3643 for (char_type c: delayedChars)
3646 // In listings, new lines (i.e. \n characters in the output) are very important. Avoid generating one for the
3647 // last line to get a clean output.
3648 if (rp.docbook_in_listing && !is_last_par)
3651 // Finalise the last (and most likely only) paragraph.
3652 generatedParagraphs.push_back(os.str());
3653 os.str(from_ascii(""));
3656 // If there is an argument that must be output after the main tag, do it after handling the rest of the paragraph.
3657 for (pos_type i = initial; i < size(); ++i) {
3658 if (getInset(i) && getInset(i)->lyxCode() == ARG_CODE) {
3659 const InsetArgument * arg = getInset(i)->asInsetArgument();
3660 if (arg->docbookargumentaftermaintag()) {
3661 // Don't use rp, as this argument would not generate anything.
3662 auto xs_local = XMLStream(os);
3663 arg->docbook(xs_local, runparams);
3665 appendedParagraphs.push_back(os.str());
3666 os.str(from_ascii(""));
3671 return std::make_tuple(prependedParagraphs, generatedParagraphs, appendedParagraphs);
3677 void doFontSwitchXHTML(vector<xml::FontTag> & tagsToOpen,
3678 vector<xml::EndFontTag> & tagsToClose,
3679 bool & flag, FontState curstate, xml::FontTypes type)
3681 if (curstate == FONT_ON) {
3682 tagsToOpen.push_back(xhtmlStartFontTag(type));
3685 tagsToClose.push_back(xhtmlEndFontTag(type));
3690 } // anonymous namespace
3693 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3695 OutputParams const & runparams,
3696 Font const & outerfont,
3697 bool start_paragraph, bool close_paragraph,
3698 pos_type initial) const
3702 // track whether we have opened these tags
3703 bool emph_flag = false;
3704 bool bold_flag = false;
3705 bool noun_flag = false;
3706 bool ubar_flag = false;
3707 bool dbar_flag = false;
3708 bool sout_flag = false;
3709 bool xout_flag = false;
3710 bool wave_flag = false;
3712 bool shap_flag = false;
3714 bool faml_flag = false;
3716 bool size_flag = false;
3718 Layout const & style = *d->layout_;
3720 if (start_paragraph)
3721 xs.startDivision(allowEmpty());
3724 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3726 FontShape curr_fs = INHERIT_SHAPE;
3727 FontFamily curr_fam = INHERIT_FAMILY;
3728 FontSize curr_size = INHERIT_SIZE;
3730 string const default_family =
3731 buf.masterBuffer()->params().fonts_default_family;
3733 vector<xml::FontTag> tagsToOpen;
3734 vector<xml::EndFontTag> tagsToClose;
3736 // parsing main loop
3737 for (pos_type i = initial; i < size(); ++i) {
3738 // let's not show deleted material in the output
3742 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3745 FontState curstate = font.fontInfo().emph();
3746 if (font_old.emph() != curstate)
3747 doFontSwitchXHTML(tagsToOpen, tagsToClose, emph_flag, curstate, xml::FT_EMPH);
3750 curstate = font.fontInfo().noun();
3751 if (font_old.noun() != curstate)
3752 doFontSwitchXHTML(tagsToOpen, tagsToClose, noun_flag, curstate, xml::FT_NOUN);
3755 curstate = font.fontInfo().underbar();
3756 if (font_old.underbar() != curstate)
3757 doFontSwitchXHTML(tagsToOpen, tagsToClose, ubar_flag, curstate, xml::FT_UBAR);
3760 curstate = font.fontInfo().strikeout();
3761 if (font_old.strikeout() != curstate)
3762 doFontSwitchXHTML(tagsToOpen, tagsToClose, sout_flag, curstate, xml::FT_SOUT);
3765 curstate = font.fontInfo().xout();
3766 if (font_old.xout() != curstate)
3767 doFontSwitchXHTML(tagsToOpen, tagsToClose, xout_flag, curstate, xml::FT_XOUT);
3770 curstate = font.fontInfo().uuline();
3771 if (font_old.uuline() != curstate)
3772 doFontSwitchXHTML(tagsToOpen, tagsToClose, dbar_flag, curstate, xml::FT_DBAR);
3775 curstate = font.fontInfo().uwave();
3776 if (font_old.uwave() != curstate)
3777 doFontSwitchXHTML(tagsToOpen, tagsToClose, wave_flag, curstate, xml::FT_WAVE);
3780 // a little hackish, but allows us to reuse what we have.
3781 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3782 if (font_old.series() != font.fontInfo().series())
3783 doFontSwitchXHTML(tagsToOpen, tagsToClose, bold_flag, curstate, xml::FT_BOLD);
3786 curr_fs = font.fontInfo().shape();
3787 FontShape old_fs = font_old.shape();
3788 if (old_fs != curr_fs) {
3792 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ITALIC));
3795 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SLANTED));
3797 case SMALLCAPS_SHAPE:
3798 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SMALLCAPS));
3804 // the other tags are for internal use
3812 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ITALIC));
3816 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SLANTED));
3819 case SMALLCAPS_SHAPE:
3820 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SMALLCAPS));
3827 // the other tags are for internal use
3834 curr_fam = font.fontInfo().family();
3835 FontFamily old_fam = font_old.family();
3836 if (old_fam != curr_fam) {
3840 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_ROMAN));
3843 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SANS));
3845 case TYPEWRITER_FAMILY:
3846 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_TYPE));
3848 case INHERIT_FAMILY:
3851 // the other tags are for internal use
3859 // we will treat a "default" font family as roman, since we have
3860 // no other idea what to do.
3861 if (default_family != "rmdefault" && default_family != "default") {
3862 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_ROMAN));
3867 if (default_family != "sfdefault") {
3868 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SANS));
3872 case TYPEWRITER_FAMILY:
3873 if (default_family != "ttdefault") {
3874 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_TYPE));
3878 case INHERIT_FAMILY:
3881 // the other tags are for internal use
3888 curr_size = font.fontInfo().size();
3889 FontSize old_size = font_old.size();
3890 if (old_size != curr_size) {
3894 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_TINY));
3897 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SCRIPT));
3900 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_FOOTNOTE));
3903 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_SMALL));
3906 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGE));
3909 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGER));
3912 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_LARGEST));
3915 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGE));
3918 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_HUGER));
3921 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_INCREASE));
3924 tagsToClose.emplace_back(xhtmlEndFontTag(xml::FT_SIZE_DECREASE));
3930 // the other tags are for internal use
3936 switch (curr_size) {
3938 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_TINY));
3942 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SCRIPT));
3946 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_FOOTNOTE));
3950 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_SMALL));
3954 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGE));
3958 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGER));
3962 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_LARGEST));
3966 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGE));
3970 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_HUGER));
3974 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_INCREASE));
3978 tagsToOpen.emplace_back(xhtmlStartFontTag(xml::FT_SIZE_DECREASE));
3985 // the other tags are for internal use
3992 // Other such tags? What about the other text ranges?
3994 vector<xml::EndFontTag>::const_iterator cit = tagsToClose.begin();
3995 vector<xml::EndFontTag>::const_iterator cen = tagsToClose.end();
3996 for (; cit != cen; ++cit)
3999 vector<xml::FontTag>::const_iterator sit = tagsToOpen.begin();
4000 vector<xml::FontTag>::const_iterator sen = tagsToOpen.end();
4001 for (; sit != sen; ++sit)
4004 tagsToClose.clear();
4007 Inset const * inset = getInset(i);
4009 if (!runparams.for_toc || inset->isInToc()) {
4010 OutputParams np = runparams;
4011 np.local_font = &font;
4012 // If the paragraph has size 1, then we are in the "special
4013 // case" where we do not output the containing paragraph info
4014 if (!inset->getLayout().htmlisblock() && size() != 1)
4015 np.html_in_par = true;
4016 retval += inset->xhtml(xs, np);
4019 char_type c = getUChar(buf.masterBuffer()->params(),
4021 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
4022 xs << XMLStream::ESCAPE_NONE << " ";
4024 xs << XMLStream::ESCAPE_NONE << "’";
4028 font_old = font.fontInfo();
4032 // I'm worried about what happens if a branch, say, is itself
4033 // wrapped in some font stuff. I think that will not work.
4035 if (close_paragraph)
4042 bool Paragraph::isHfill(pos_type pos) const
4044 Inset const * inset = getInset(pos);
4045 return inset && inset->isHfill();
4049 bool Paragraph::isNewline(pos_type pos) const
4051 // U+2028 LINE SEPARATOR
4052 // U+2029 PARAGRAPH SEPARATOR
4053 char_type const c = d->text_[pos];
4054 if (c == 0x2028 || c == 0x2029)
4056 Inset const * inset = getInset(pos);
4057 return inset && inset->lyxCode() == NEWLINE_CODE;
4061 bool Paragraph::isEnvSeparator(pos_type pos) const
4063 Inset const * inset = getInset(pos);
4064 return inset && inset->lyxCode() == SEPARATOR_CODE;
4068 bool Paragraph::isLineSeparator(pos_type pos) const
4070 char_type const c = d->text_[pos];
4071 if (isLineSeparatorChar(c))
4073 Inset const * inset = getInset(pos);
4074 return inset && inset->isLineSeparator();
4078 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
4082 if (ignore_deleted && isDeleted(pos))
4084 if (Inset const * inset = getInset(pos))
4085 return !inset->isLetter();
4086 // if we have a hard hyphen (no en- or emdash) or apostrophe
4087 // we pass this to the spell checker
4088 // FIXME: this method is subject to change, visit
4089 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
4090 // to get an impression how complex this is.
4091 if (isHardHyphenOrApostrophe(pos))
4093 char_type const c = d->text_[pos];
4094 // We want to pass the escape chars to the spellchecker
4095 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
4096 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
4100 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
4102 pos_type const psize = size();
4105 char_type const c = d->text_[pos];
4106 if (c != '-' && c != '\'')
4108 pos_type nextpos = pos + 1;
4109 pos_type prevpos = pos > 0 ? pos - 1 : 0;
4110 if ((nextpos == psize || isSpace(nextpos))
4111 && (pos == 0 || isSpace(prevpos)))
4117 bool Paragraph::needsCProtection(bool const fragile) const
4119 // first check the layout of the paragraph, but only in insets
4120 InsetText const * textinset = inInset().asInsetText();
4121 bool const maintext = textinset
4122 ? textinset->text().isMainText()
4125 if (!maintext && layout().needcprotect) {
4126 // Environments need cprotection regardless the content
4127 if (layout().latextype == LATEX_ENVIRONMENT)
4130 // Commands need cprotection if they contain specific chars
4131 int const nchars_escape = 9;
4132 static char_type const chars_escape[nchars_escape] = {
4133 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
4135 docstring const pars = asString();
4136 for (int k = 0; k < nchars_escape; k++) {
4137 if (contains(pars, chars_escape[k]))
4142 // now check whether we have insets that need cprotection
4143 for (auto const & icit : d->insetlist_) {
4144 Inset const * ins = icit.inset;
4147 if (ins->needsCProtection(maintext, fragile))
4155 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
4157 return d->speller_state_.getRange(pos);
4161 bool Paragraph::isChar(pos_type pos) const
4163 if (Inset const * inset = getInset(pos))
4164 return inset->isChar();
4165 char_type const c = d->text_[pos];
4166 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
4170 bool Paragraph::isSpace(pos_type pos) const
4172 if (Inset const * inset = getInset(pos))
4173 return inset->isSpace();
4174 char_type const c = d->text_[pos];
4175 return lyx::isSpace(c);
4180 Paragraph::getParLanguage(BufferParams const & bparams) const
4183 return getFirstFontSettings(bparams).language();
4184 // FIXME: we should check the prev par as well (Lgb)
4185 return bparams.language;
4189 bool Paragraph::isRTL(BufferParams const & bparams) const
4191 return getParLanguage(bparams)->rightToLeft()
4192 && !inInset().getLayout().forceLTR();
4196 void Paragraph::changeLanguage(BufferParams const & bparams,
4197 Language const * from, Language const * to)
4199 // change language including dummy font change at the end
4200 for (pos_type i = 0; i <= size(); ++i) {
4201 Font font = getFontSettings(bparams, i);
4202 if (font.language() == from) {
4203 font.setLanguage(to);
4205 d->requestSpellCheck(i);
4211 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
4213 Language const * doc_language = bparams.language;
4214 for (auto const & f : d->fontlist_)
4215 if (f.font().language() != ignore_language &&
4216 f.font().language() != latex_language &&
4217 f.font().language() != doc_language)
4223 void Paragraph::getLanguages(std::set<Language const *> & langs) const
4225 for (auto const & f : d->fontlist_) {
4226 Language const * lang = f.font().language();
4227 if (lang != ignore_language &&
4228 lang != latex_language)
4234 docstring Paragraph::asString(int options) const
4236 return asString(0, size(), options);
4240 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
4242 odocstringstream os;
4245 && options & AS_STR_LABEL
4246 && !d->params_.labelString().empty())
4247 os << d->params_.labelString() << ' ';
4249 for (pos_type i = beg; i < end; ++i) {
4250 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
4252 char_type const c = d->text_[i];
4253 if (isPrintable(c) || c == '\t'
4254 || (c == '\n' && (options & AS_STR_NEWLINES)))
4256 else if (c == META_INSET && (options & AS_STR_INSETS)) {
4257 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
4258 LASSERT(runparams != nullptr, return docstring());
4259 if (runparams->find_effective())
4260 getInset(i)->toString(os);
4262 getInset(i)->plaintext(os, *runparams);
4263 } else if (c == META_INSET && (options & AS_STR_MATHED)
4264 && getInset(i)->lyxCode() == REF_CODE) {
4265 Buffer const & buf = getInset(i)->buffer();
4266 OutputParams rp(&buf.params().encoding());
4267 Font const font(inherit_font, buf.params().language);
4268 rp.local_font = &font;
4270 getInset(i)->latex(ots, rp);
4272 getInset(i)->toString(os);
4281 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
4282 bool const shorten, bool const label) const
4284 size_t tmplen = shorten ? maxlen + 1 : maxlen;
4285 if (label && !labelString().empty())
4286 os += labelString() + ' ';
4287 if (!layout().isTocCaption())
4289 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
4292 char_type const c = d->text_[i];
4295 else if (c == META_INSET)
4296 getInset(i)->forOutliner(os, tmplen, false);
4299 Text::shortenForOutliner(os, maxlen);
4303 void Paragraph::setInsetOwner(Inset const * inset)
4305 d->inset_owner_ = inset;
4309 int Paragraph::id() const
4315 void Paragraph::setId(int id)
4321 Layout const & Paragraph::layout() const
4327 void Paragraph::setLayout(Layout const & layout)
4329 d->layout_ = &layout;
4333 void Paragraph::setDefaultLayout(DocumentClass const & tc)
4335 setLayout(tc.defaultLayout());
4339 void Paragraph::setPlainLayout(DocumentClass const & tc)
4341 setLayout(tc.plainLayout());
4345 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tc)
4347 if (usePlainLayout())
4350 setDefaultLayout(tc);
4354 Inset const & Paragraph::inInset() const
4356 LBUFERR(d->inset_owner_);
4357 return *d->inset_owner_;
4361 ParagraphParameters & Paragraph::params()
4367 ParagraphParameters const & Paragraph::params() const
4373 bool Paragraph::isFreeSpacing() const
4375 if (d->layout_->free_spacing)
4377 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
4381 bool Paragraph::allowEmpty() const
4383 if (d->layout_->keepempty)
4385 return d->inset_owner_ && d->inset_owner_->allowEmpty();
4389 bool Paragraph::brokenBiblio() const
4391 // There is a problem if there is no bibitem at position 0 in
4392 // paragraphs that need one, if there is another bibitem in the
4393 // paragraph or if this paragraph is not supposed to have
4394 // a bibitem inset at all.
4395 return ((d->layout_->labeltype == LABEL_BIBLIO
4396 && (d->insetlist_.find(BIBITEM_CODE) != 0
4397 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
4398 || (d->layout_->labeltype != LABEL_BIBLIO
4399 && d->insetlist_.find(BIBITEM_CODE) != -1));
4403 int Paragraph::fixBiblio(Buffer const & buffer)
4405 // FIXME: when there was already an inset at 0, the return value is 1,
4406 // which does not tell whether another inset has been removed; the
4407 // cursor cannot be correctly updated.
4409 bool const track_changes = buffer.params().track_changes;
4410 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
4412 // The case where paragraph is not BIBLIO
4413 if (d->layout_->labeltype != LABEL_BIBLIO) {
4414 if (bibitem_pos == -1)
4415 // No InsetBibitem => OK
4417 // There is an InsetBibitem: remove it!
4418 d->insetlist_.release(bibitem_pos);
4419 eraseChar(bibitem_pos, track_changes);
4420 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
4423 bool const hasbibitem0 = bibitem_pos == 0;
4425 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
4426 // There was an InsetBibitem at pos 0,
4427 // and no other one => OK
4428 if (bibitem_pos == -1)
4430 // there is a bibitem at the 0 position, but since
4431 // there is a second one, we copy the second on the
4432 // first. We're assuming there are at most two of
4433 // these, which there should be.
4434 // FIXME: why does it make sense to do that rather
4435 // than keep the first? (JMarc)
4436 Inset * inset = releaseInset(bibitem_pos);
4437 d->insetlist_.begin()->inset = inset;
4438 // This needs to be done to update the counter (#8499)
4439 buffer.updateBuffer();
4440 return -bibitem_pos;
4443 // We need to create an inset at the beginning
4444 Inset * inset = nullptr;
4445 if (bibitem_pos > 0) {
4446 // there was one somewhere in the paragraph, let's move it
4447 inset = d->insetlist_.release(bibitem_pos);
4448 eraseChar(bibitem_pos, track_changes);
4451 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
4452 InsetCommandParams(BIBITEM_CODE));
4454 Font font(inherit_font, buffer.params().language);
4455 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
4456 : Change::UNCHANGED));
4458 // This is needed to get the counters right
4459 buffer.updateBuffer();
4464 void Paragraph::checkAuthors(AuthorList const & authorList)
4466 d->changes_.checkAuthors(authorList);
4470 bool Paragraph::isChanged(pos_type pos) const
4472 return lookupChange(pos).changed();
4476 bool Paragraph::isInserted(pos_type pos) const
4478 return lookupChange(pos).inserted();
4482 bool Paragraph::isDeleted(pos_type pos) const
4484 return lookupChange(pos).deleted();
4488 InsetList const & Paragraph::insetList() const
4490 return d->insetlist_;
4494 void Paragraph::setInsetBuffers(Buffer & b)
4496 d->insetlist_.setBuffer(b);
4500 void Paragraph::resetBuffer()
4502 d->insetlist_.resetBuffer();
4506 Inset * Paragraph::releaseInset(pos_type pos)
4508 Inset * inset = d->insetlist_.release(pos);
4509 /// does not honour change tracking!
4510 eraseChar(pos, false);
4515 Inset * Paragraph::getInset(pos_type pos)
4517 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4518 ? d->insetlist_.get(pos) : nullptr;
4522 Inset const * Paragraph::getInset(pos_type pos) const
4524 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
4525 ? d->insetlist_.get(pos) : nullptr;
4529 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
4530 pos_type & right, TextCase action)
4532 // process sequences of modified characters; in change
4533 // tracking mode, this approach results in much better
4534 // usability than changing case on a char-by-char basis
4535 // We also need to track the current font, since font
4536 // changes within sequences can occur.
4537 vector<pair<char_type, Font> > changes;
4539 bool const trackChanges = bparams.track_changes;
4541 bool capitalize = true;
4543 for (; pos < right; ++pos) {
4544 char_type oldChar = d->text_[pos];
4545 char_type newChar = oldChar;
4547 // ignore insets and don't play with deleted text!
4548 if (oldChar != META_INSET && !isDeleted(pos)) {
4550 case text_lowercase:
4551 newChar = lowercase(oldChar);
4553 case text_capitalization:
4555 newChar = uppercase(oldChar);
4559 case text_uppercase:
4560 newChar = uppercase(oldChar);
4565 if (isWordSeparator(pos) || isDeleted(pos)) {
4566 // permit capitalization again
4570 if (oldChar != newChar) {
4571 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
4572 if (pos != right - 1)
4574 // step behind the changing area
4578 int erasePos = pos - changes.size();
4579 for (auto const & change : changes) {
4580 insertChar(pos, change.first, change.second, trackChanges);
4581 if (!eraseChar(erasePos, trackChanges)) {
4584 ++right; // expand selection
4591 int Paragraph::find(docstring const & str, bool cs, bool mw,
4592 pos_type start_pos, bool del) const
4594 pos_type pos = start_pos;
4595 int const strsize = str.length();
4597 pos_type const parsize = d->text_.size();
4598 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
4599 // ignore deleted matter
4600 if (!del && isDeleted(pos)) {
4601 if (pos == parsize - 1)
4607 // Ignore "invisible" letters such as ligature breaks
4608 // and hyphenation chars while searching
4609 bool nonmatch = false;
4610 while (pos < parsize && isInset(pos)) {
4611 Inset const * inset = getInset(pos);
4612 if (!inset->isLetter() && !inset->isChar())
4614 odocstringstream os;
4615 if (inset->lyxCode() == lyx::QUOTE_CODE) {
4617 op.find_set_feature(OutputParams::SearchQuick);
4618 inset->plaintext(os, op);
4621 inset->toString(os);
4623 docstring const insetstring = os.str();
4624 if (!insetstring.empty()) {
4625 int const insetstringsize = insetstring.length();
4626 for (int j = 0; j < insetstringsize && pos < parsize; ++i, ++j) {
4627 if ((cs && str[i] != insetstring[j])
4628 || (!cs && uppercase(str[i]) != uppercase(insetstring[j]))) {
4636 if (nonmatch || i == strsize)
4638 char_type dp = d->text_[pos];
4639 if (cs && str[i] != dp)
4641 if (!cs && uppercase(str[i]) != uppercase(dp))
4648 // if necessary, check whether string matches word
4650 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
4653 && !isWordSeparator(pos))
4657 return pos - start_pos;
4661 char_type Paragraph::getChar(pos_type pos) const
4663 return d->text_[pos];
4667 pos_type Paragraph::size() const
4669 return d->text_.size();
4673 bool Paragraph::empty() const
4675 return d->text_.empty();
4679 bool Paragraph::isInset(pos_type pos) const
4681 return d->text_[pos] == META_INSET;
4685 bool Paragraph::isSeparator(pos_type pos) const
4687 //FIXME: Are we sure this can be the only separator?
4688 return d->text_[pos] == ' ';
4692 void Paragraph::deregisterWords()
4694 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4695 Private::LangWordsMap::const_iterator ite = d->words_.end();
4696 for (; itl != ite; ++itl) {
4697 WordList & wl = theWordList(itl->first);
4698 Private::Words::const_iterator it = (itl->second).begin();
4699 Private::Words::const_iterator et = (itl->second).end();
4700 for (; it != et; ++it)
4707 void Paragraph::locateWord(pos_type & from, pos_type & to,
4708 word_location const loc, bool const ignore_deleted) const
4711 case WHOLE_WORD_STRICT:
4712 if (from == 0 || from == size()
4713 || isWordSeparator(from, ignore_deleted)
4714 || isWordSeparator(from - 1, ignore_deleted)) {
4721 // If we are already at the beginning of a word, do nothing
4722 if (!from || isWordSeparator(from - 1, ignore_deleted))
4727 // always move the cursor to the beginning of previous word
4728 while (from && !isWordSeparator(from - 1, ignore_deleted))
4732 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4735 // no need to move the 'from' cursor
4739 while (to < size() && !isWordSeparator(to, ignore_deleted))
4744 void Paragraph::collectWords()
4746 for (pos_type pos = 0; pos < size(); ++pos) {
4747 if (isWordSeparator(pos))
4749 pos_type from = pos;
4750 locateWord(from, pos, WHOLE_WORD);
4751 // Work around MSVC warning: The statement
4752 // if (pos < from + lyxrc.completion_minlength)
4753 // triggers a signed vs. unsigned warning.
4754 // I don't know why this happens, it could be a MSVC bug, or
4755 // related to LLP64 (windows) vs. LP64 (unix) programming
4756 // model, or the C++ standard might be ambigous in the section
4757 // defining the "usual arithmetic conversions". However, using
4758 // a temporary variable is safe and works on all compilers.
4759 pos_type const endpos = from + lyxrc.completion_minlength;
4762 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4763 if (cit == d->fontlist_.end())
4765 Language const * lang = cit->font().language();
4766 docstring const word = asString(from, pos, AS_STR_NONE);
4767 d->words_[lang->lang()].insert(word);
4772 void Paragraph::registerWords()
4774 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4775 Private::LangWordsMap::const_iterator ite = d->words_.end();
4776 for (; itl != ite; ++itl) {
4777 WordList & wl = theWordList(itl->first);
4778 Private::Words::const_iterator it = (itl->second).begin();
4779 Private::Words::const_iterator et = (itl->second).end();
4780 for (; it != et; ++it)
4786 void Paragraph::updateWords()
4794 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4796 SkipPositionsIterator begin = skips.begin();
4797 SkipPositions::iterator end = skips.end();
4798 if (pos > 0 && begin < end) {
4800 if (end->last == pos - 1) {
4805 skips.insert(end, FontSpan(pos, pos));
4809 Language * Paragraph::Private::locateSpellRange(
4810 pos_type & from, pos_type & to,
4811 SkipPositions & skips) const
4813 // skip leading white space
4814 while (from < to && owner_->isWordSeparator(from))
4816 // don't check empty range
4819 // get current language
4820 Language * lang = getSpellLanguage(from);
4821 pos_type last = from;
4822 bool samelang = true;
4823 bool sameinset = true;
4824 while (last < to && samelang && sameinset) {
4825 // hop to end of word
4826 while (last < to && !owner_->isWordSeparator(last)) {
4827 Inset const * inset = owner_->getInset(last);
4828 if (inset && dynamic_cast<const InsetSpecialChar *>(inset)) {
4829 // check for "invisible" letters such as ligature breaks
4830 odocstringstream os;
4831 inset->toString(os);
4832 if (os.str().length() != 0) {
4833 // avoid spell check of visible special char insets
4834 // stop the loop in front of the special char inset
4839 appendSkipPosition(skips, last);
4840 } else if (owner_->isDeleted(last)) {
4841 appendSkipPosition(skips, last);
4845 // hop to next word while checking for insets
4846 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4847 if (Inset const * inset = owner_->getInset(last))
4848 sameinset = inset->isChar() && inset->isLetter();
4849 if (sameinset && owner_->isDeleted(last)) {
4850 appendSkipPosition(skips, last);
4855 if (sameinset && last < to) {
4856 // now check for language change
4857 samelang = lang == getSpellLanguage(last);
4860 // if language change detected backstep is needed
4868 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4871 const_cast<Language *>(owner_->getFontSettings(
4872 inset_owner_->buffer().params(), from).language());
4873 if (lang == inset_owner_->buffer().params().language
4874 && !lyxrc.spellchecker_alt_lang.empty()) {
4876 string const lang_variety =
4877 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4878 lang->setCode(lang_code);
4879 lang->setVariety(lang_variety);
4885 void Paragraph::requestSpellCheck(pos_type pos)
4887 d->requestSpellCheck(pos);
4889 // Also request spellcheck within (text) insets
4890 for (auto const & insets : insetList()) {
4891 if (!insets.inset->asInsetText())
4893 ParagraphList & inset_pars =
4894 insets.inset->asInsetText()->paragraphs();
4895 ParagraphList::iterator pit = inset_pars.begin();
4896 ParagraphList::iterator pend = inset_pars.end();
4897 for (; pit != pend; ++pit)
4898 pit->requestSpellCheck();
4904 bool Paragraph::needsSpellCheck() const
4906 SpellChecker::ChangeNumber speller_change_number = 0;
4907 if (theSpellChecker())
4908 speller_change_number = theSpellChecker()->changeNumber();
4909 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4910 d->speller_state_.needsCompleteRefresh(speller_change_number);
4912 return d->needsSpellCheck();
4916 bool Paragraph::Private::ignoreWord(docstring const & word) const
4918 // Ignore words with digits
4919 // FIXME: make this customizable
4920 // (note that some checkers ignore words with digits by default)
4921 docstring::const_iterator cit = word.begin();
4922 docstring::const_iterator const end = word.end();
4923 for (; cit != end; ++cit) {
4924 if (isNumber((*cit)))
4931 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4932 WordLangTuple & wl, docstring_list & suggestions,
4933 bool do_suggestion, bool check_learned) const
4935 SpellChecker::Result result = SpellChecker::WORD_OK;
4936 SpellChecker * speller = theSpellChecker();
4940 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4943 locateWord(from, to, WHOLE_WORD, true);
4944 if (from == to || from >= size())
4947 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4948 Language * lang = d->getSpellLanguage(from);
4950 BufferParams const & bparams = d->inset_owner_->buffer().params();
4952 if (getFontSettings(bparams, from).fontInfo().nospellcheck() == FONT_ON)
4955 wl = WordLangTuple(word, lang);
4960 if (needsSpellCheck() || check_learned) {
4962 if (!d->ignoreWord(word)) {
4963 bool const trailing_dot = to < size() && d->text_[to] == '.';
4964 result = speller->check(wl, bparams.spellignore());
4965 if (SpellChecker::misspelled(result) && trailing_dot) {
4966 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4967 result = speller->check(wl, bparams.spellignore());
4968 if (!SpellChecker::misspelled(result)) {
4969 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4971 from << ".." << to << "]");
4973 // spell check with dot appended failed too
4974 // restore original word/lang value
4975 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4976 wl = WordLangTuple(word, lang);
4980 if (!SpellChecker::misspelled(result)) {
4981 // area up to the begin of the next word is not misspelled
4982 while (end < size() && isWordSeparator(end))
4985 d->setMisspelled(from, end, result);
4987 result = d->speller_state_.getState(from);
4991 suggestions.clear();
4993 if (SpellChecker::misspelled(result)) {
4994 LYXERR(Debug::GUI, "misspelled word: \"" <<
4996 from << ".." << to << "]");
4998 speller->suggest(wl, suggestions);
5004 void Paragraph::anonymize()
5006 // This is a very crude anonymization for now
5007 for (char_type & c : d->text_)
5008 if (isLetterChar(c) || isNumber(c))
5013 void Paragraph::Private::markMisspelledWords(
5014 Language const * lang,
5015 pos_type const & first, pos_type const & last,
5016 SpellChecker::Result result,
5017 docstring const & word,
5018 SkipPositions const & skips)
5020 if (!SpellChecker::misspelled(result)) {
5021 setMisspelled(first, last, SpellChecker::WORD_OK);
5025 SpellChecker * speller = theSpellChecker();
5026 // locate and enumerate the error positions
5027 int nerrors = speller->numMisspelledWords();
5029 SkipPositionsIterator it = skips.begin();
5030 SkipPositionsIterator et = skips.end();
5031 for (int index = 0; index < nerrors; ++index) {
5034 speller->misspelledWord(index, wstart, wlen);
5035 /// should not happen if speller supports range checks
5038 WordLangTuple const candidate(word.substr(wstart, wlen), lang);
5039 wstart += first + numskipped;
5040 if (snext < wstart) {
5041 /// mark the range of correct spelling
5042 numskipped += countSkips(it, et, wstart);
5043 setMisspelled(snext,
5044 wstart - 1, SpellChecker::WORD_OK);
5046 snext = wstart + wlen;
5047 // Check whether the candidate is in the document's local dict
5048 SpellChecker::Result actresult = result;
5049 if (inset_owner_->buffer().params().spellignored(candidate))
5050 actresult = SpellChecker::DOCUMENT_LEARNED_WORD;
5051 numskipped += countSkips(it, et, snext);
5052 /// mark the range of misspelling
5053 setMisspelled(wstart, snext, actresult);
5054 if (actresult == SpellChecker::DOCUMENT_LEARNED_WORD)
5055 LYXERR(Debug::GUI, "local dictionary word: \"" <<
5056 candidate.word() << "\" [" <<
5057 wstart << ".." << (snext-1) << "]");
5059 LYXERR(Debug::GUI, "misspelled word: \"" <<
5060 candidate.word() << "\" [" <<
5061 wstart << ".." << (snext-1) << "]");
5064 if (snext <= last) {
5065 /// mark the range of correct spelling at end
5066 setMisspelled(snext, last, SpellChecker::WORD_OK);
5071 void Paragraph::spellCheck() const
5073 SpellChecker * speller = theSpellChecker();
5074 if (!speller || empty() ||!needsSpellCheck())
5078 d->rangeOfSpellCheck(start, endpos);
5079 if (speller->canCheckParagraph()) {
5080 // loop until we leave the range
5081 for (pos_type first = start; first < endpos; ) {
5082 pos_type last = endpos;
5083 Private::SkipPositions skips;
5084 Language * lang = d->locateSpellRange(first, last, skips);
5085 if (first >= endpos)
5087 // start the spell checker on the unit of meaning
5088 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
5089 WordLangTuple wl = WordLangTuple(word, lang);
5090 BufferParams const & bparams = d->inset_owner_->buffer().params();
5091 SpellChecker::Result result = !word.empty() ?
5092 speller->check(wl, bparams.spellignore()) : SpellChecker::WORD_OK;
5093 d->markMisspelledWords(lang, first, last, result, word, skips);
5097 static docstring_list suggestions;
5098 pos_type to = endpos;
5099 while (start < endpos) {
5101 spellCheck(start, to, wl, suggestions, false);
5105 d->readySpellCheck();
5109 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
5111 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
5112 if (result || pos <= 0 || pos > size())
5114 if (check_boundary && (pos == size() || isWordSeparator(pos)))
5115 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
5120 string Paragraph::magicLabel() const
5123 ss << "magicparlabel-" << id();