3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Asger Alstrup
7 * \author Lars Gullik Bjønnes
8 * \author Richard Heck (XHTML output)
9 * \author Jean-Marc Lasgouttes
10 * \author Angus Leeming
12 * \author André Pönitz
14 * \author Jürgen Vigna
16 * Full author contact details are available in file CREDITS.
21 #include "Paragraph.h"
23 #include "LayoutFile.h"
25 #include "BufferParams.h"
28 #include "BufferEncodings.h"
29 #include "InsetList.h"
31 #include "LaTeXFeatures.h"
37 #include "OutputParams.h"
38 #include "output_latex.h"
39 #include "output_xhtml.h"
40 #include "ParagraphParameters.h"
41 #include "SpellChecker.h"
43 #include "texstream.h"
44 #include "TextClass.h"
47 #include "WordLangTuple.h"
50 #include "frontends/alert.h"
52 #include "insets/InsetBibitem.h"
53 #include "insets/InsetLabel.h"
54 #include "insets/InsetSpecialChar.h"
55 #include "insets/InsetText.h"
57 #include "mathed/InsetMathHull.h"
59 #include "support/debug.h"
60 #include "support/docstring_list.h"
61 #include "support/ExceptionMessage.h"
62 #include "support/gettext.h"
63 #include "support/lassert.h"
64 #include "support/lstrings.h"
65 #include "support/textutils.h"
72 using namespace lyx::support;
74 // OSX clang, gcc < 4.8.0, and msvc < 2015 do not support C++11 thread_local
75 #if defined(__APPLE__) || (defined(__GNUC__) && __GNUC__ == 4 && __GNUC_MINOR__ < 8)
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
87 /// Inset identifier (above 0x10ffff, for ucs-4)
88 char_type const META_INSET = 0x200001;
93 /////////////////////////////////////////////////////////////////////
97 /////////////////////////////////////////////////////////////////////
99 class SpellResultRange {
101 SpellResultRange(FontSpan range, SpellChecker::Result result)
102 : range_(range), result_(result)
105 FontSpan const & range() const { return range_; }
107 void range(FontSpan const & r) { range_ = r; }
109 SpellChecker::Result result() const { return result_; }
111 void result(SpellChecker::Result r) { result_ = r; }
113 bool contains(pos_type pos) const { return range_.contains(pos); }
115 bool covered(FontSpan const & r) const
117 // 1. first of new range inside current range or
118 // 2. last of new range inside current range or
119 // 3. first of current range inside new range or
120 // 4. last of current range inside new range
121 //FIXME: is this the same as !range_.intersect(r).empty() ?
122 return range_.contains(r.first) || range_.contains(r.last) ||
123 r.contains(range_.first) || r.contains(range_.last);
126 void shift(pos_type pos, int offset)
128 if (range_.first > pos) {
129 range_.first += offset;
130 range_.last += offset;
131 } else if (range_.last >= pos) {
132 range_.last += offset;
137 SpellChecker::Result result_ ;
141 /////////////////////////////////////////////////////////////////////
145 /////////////////////////////////////////////////////////////////////
147 class SpellCheckerState {
151 needs_refresh_ = true;
152 current_change_number_ = 0;
155 void setRange(FontSpan const & fp, SpellChecker::Result state)
158 RangesIterator et = ranges_.end();
159 RangesIterator it = ranges_.begin();
160 for (; it != et; ++it) {
161 if (!it->covered(fp))
162 result.push_back(SpellResultRange(it->range(), it->result()));
163 else if (state == SpellChecker::WORD_OK) {
164 // trim or split the current misspelled range
165 // store misspelled ranges only
166 FontSpan range = it->range();
167 if (fp.first > range.first) {
168 // misspelled area in front of WORD_OK
169 range.last = fp.first - 1;
170 result.push_back(SpellResultRange(range, it->result()));
173 if (fp.last < range.last) {
174 // misspelled area after WORD_OK range
175 range.first = fp.last + 1;
176 result.push_back(SpellResultRange(range, it->result()));
181 if (state != SpellChecker::WORD_OK)
182 ranges_.push_back(SpellResultRange(fp, state));
185 void increasePosAfterPos(pos_type pos)
187 correctRangesAfterPos(pos, 1);
191 void decreasePosAfterPos(pos_type pos)
193 correctRangesAfterPos(pos, -1);
197 void refreshLast(pos_type pos)
199 if (pos < refresh_.last)
203 SpellChecker::Result getState(pos_type pos) const
205 SpellChecker::Result result = SpellChecker::WORD_OK;
206 RangesIterator et = ranges_.end();
207 RangesIterator it = ranges_.begin();
208 for (; it != et; ++it) {
209 if(it->contains(pos)) {
216 FontSpan const & getRange(pos_type pos) const
218 /// empty span to indicate mismatch
219 static FontSpan empty_;
220 RangesIterator et = ranges_.end();
221 RangesIterator it = ranges_.begin();
222 for (; it != et; ++it) {
223 if(it->contains(pos)) {
230 bool needsRefresh() const
232 return needs_refresh_;
235 SpellChecker::ChangeNumber currentChangeNumber() const
237 return current_change_number_;
240 void refreshRange(pos_type & first, pos_type & last) const
242 first = refresh_.first;
243 last = refresh_.last;
246 void needsRefresh(pos_type pos)
248 if (needs_refresh_ && pos != -1) {
249 if (pos < refresh_.first)
250 refresh_.first = pos;
251 if (pos > refresh_.last)
253 } else if (pos != -1) {
254 // init request check for neighbour positions too
255 refresh_.first = pos > 0 ? pos - 1 : 0;
256 // no need for special end of paragraph check
257 refresh_.last = pos + 1;
259 needs_refresh_ = pos != -1;
262 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
264 needs_refresh_ = true;
267 current_change_number_ = change_number;
270 typedef vector<SpellResultRange> Ranges;
271 typedef Ranges::const_iterator RangesIterator;
273 /// the area of the paragraph with pending spell check
276 /// spell state cache version number
277 SpellChecker::ChangeNumber current_change_number_;
280 void correctRangesAfterPos(pos_type pos, int offset)
282 RangesIterator et = ranges_.end();
283 Ranges::iterator it = ranges_.begin();
284 for (; it != et; ++it) {
285 it->shift(pos, offset);
291 /////////////////////////////////////////////////////////////////////
293 // Paragraph::Private
295 /////////////////////////////////////////////////////////////////////
297 class Paragraph::Private
299 // Enforce our own "copy" constructor
300 Private(Private const &) = delete;
301 Private & operator=(Private const &) = delete;
302 // Unique ID generator
303 static int make_id();
306 Private(Paragraph * owner, Layout const & layout);
307 /// "Copy constructor"
308 Private(Private const &, Paragraph * owner);
309 /// Copy constructor from \p beg to \p end
310 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
313 void insertChar(pos_type pos, char_type c, Change const & change);
315 /// Output the surrogate pair formed by \p c and \p next to \p os.
316 /// \return the number of characters written.
317 int latexSurrogatePair(BufferParams const &, otexstream & os,
318 char_type c, char_type next,
319 OutputParams const &);
321 /// Output a space in appropriate formatting (or a surrogate pair
322 /// if the next character is a combining character).
323 /// \return whether a surrogate pair was output.
324 bool simpleTeXBlanks(BufferParams const &,
325 OutputParams const &,
328 unsigned int & column,
330 Layout const & style);
332 /// This could go to ParagraphParameters if we want to.
333 int startTeXParParams(BufferParams const &, otexstream &,
334 OutputParams const &) const;
336 /// This could go to ParagraphParameters if we want to.
337 bool endTeXParParams(BufferParams const &, otexstream &,
338 OutputParams const &) const;
341 void latexInset(BufferParams const &,
346 Font const & outerfont,
348 Change & running_change,
349 Layout const & style,
351 unsigned int & column);
354 void latexSpecialChar(
356 BufferParams const & bparams,
357 OutputParams const & runparams,
358 Font const & running_font,
359 string & alien_script,
360 Layout const & style,
363 unsigned int & column);
370 unsigned int & column);
376 unsigned int & column);
382 unsigned int & column);
385 void validate(LaTeXFeatures & features) const;
387 /// Checks if the paragraph contains only text and no inset or font change.
388 bool onlyText(Buffer const & buf, Font const & outerfont,
389 pos_type initial) const;
391 /// a vector of speller skip positions
392 typedef vector<FontSpan> SkipPositions;
393 typedef SkipPositions::const_iterator SkipPositionsIterator;
395 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
397 Language * getSpellLanguage(pos_type const from) const;
399 Language * locateSpellRange(pos_type & from, pos_type & to,
400 SkipPositions & skips) const;
402 bool hasSpellerChange() const
404 SpellChecker::ChangeNumber speller_change_number = 0;
405 if (theSpellChecker())
406 speller_change_number = theSpellChecker()->changeNumber();
407 return speller_change_number > speller_state_.currentChangeNumber();
410 bool ignoreWord(docstring const & word) const ;
412 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
414 pos_type textsize = owner_->size();
415 // check for sane arguments
416 if (to <= from || from >= textsize)
418 FontSpan fp = FontSpan(from, to - 1);
419 speller_state_.setRange(fp, state);
422 void requestSpellCheck(pos_type pos)
425 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
427 speller_state_.needsRefresh(pos);
430 void readySpellCheck()
432 speller_state_.needsRefresh(-1);
435 bool needsSpellCheck() const
437 return speller_state_.needsRefresh();
440 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
442 speller_state_.refreshRange(first, last);
444 last = owner_->size();
447 pos_type endpos = last;
448 owner_->locateWord(first, endpos, WHOLE_WORD);
451 owner_->locateWord(last, endpos, WHOLE_WORD);
456 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
460 while (it != et && it->first < start) {
461 int skip = it->last - it->first + 1;
469 void markMisspelledWords(pos_type const & first, pos_type const & last,
470 SpellChecker::Result result,
471 docstring const & word,
472 SkipPositions const & skips);
474 InsetCode ownerCode() const
476 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
479 /// Which Paragraph owns us?
483 Inset const * inset_owner_;
492 ParagraphParameters params_;
494 /// for recording and looking up changes
498 InsetList insetlist_;
501 pos_type begin_of_body_;
503 typedef docstring TextContainer;
507 typedef set<docstring> Words;
508 typedef map<string, Words> LangWordsMap;
512 Layout const * layout_;
514 SpellCheckerState speller_state_;
518 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
519 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
526 int Paragraph::Private::make_id()
528 // The id is unique per session across buffers because it is used in
529 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
532 static atomic_uint next_id(0);
537 Paragraph::Private::Private(Private const & p, Paragraph * owner)
538 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
540 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
541 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
544 requestSpellCheck(p.text_.size());
548 Paragraph::Private::Private(Private const & p, Paragraph * owner,
549 pos_type beg, pos_type end)
550 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
551 params_(p.params_), changes_(p.changes_),
552 insetlist_(p.insetlist_, beg, end),
553 begin_of_body_(p.begin_of_body_), words_(p.words_),
556 if (beg >= pos_type(p.text_.size()))
558 text_ = p.text_.substr(beg, end - beg);
560 FontList::const_iterator fcit = fontlist_.begin();
561 FontList::const_iterator fend = fontlist_.end();
562 for (; fcit != fend; ++fcit) {
563 if (fcit->pos() < beg)
565 if (fcit->pos() >= end) {
566 // Add last entry in the fontlist_.
567 fontlist_.set(text_.size() - 1, fcit->font());
570 // Add a new entry in the fontlist_.
571 fontlist_.set(fcit->pos() - beg, fcit->font());
573 requestSpellCheck(p.text_.size());
577 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
578 bool output_active, TocBackend & backend) const
580 d->changes_.addToToc(cdit, buf, output_active, backend);
584 void Paragraph::addChangesToBuffer(Buffer const & buf) const
586 d->changes_.updateBuffer(buf);
590 bool Paragraph::isChangeUpdateRequired() const
592 return d->changes_.isUpdateRequired();
596 bool Paragraph::isDeleted(pos_type start, pos_type end) const
598 LASSERT(start >= 0 && start <= size(), return false);
599 LASSERT(end > start && end <= size() + 1, return false);
601 return d->changes_.isDeleted(start, end);
605 bool Paragraph::isChanged(pos_type start, pos_type end) const
607 LASSERT(start >= 0 && start <= size(), return false);
608 LASSERT(end > start && end <= size() + 1, return false);
610 return d->changes_.isChanged(start, end);
614 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
616 // keep the logic here in sync with the logic of eraseChars()
620 Change const & change = d->changes_.lookup(size());
621 return change.inserted() && change.currentAuthor();
625 void Paragraph::setChange(Change const & change)
627 // beware of the imaginary end-of-par character!
628 d->changes_.set(change, 0, size() + 1);
631 * Propagate the change recursively - but not in case of DELETED!
633 * Imagine that your co-author makes changes in an existing inset. He
634 * sends your document to you and you come to the conclusion that the
635 * inset should go completely. If you erase it, LyX must not delete all
636 * text within the inset. Otherwise, the change tracked insertions of
637 * your co-author get lost and there is no way to restore them later.
639 * Conclusion: An inset's content should remain untouched if you delete it
642 if (!change.deleted()) {
643 for (pos_type pos = 0; pos < size(); ++pos) {
644 if (Inset * inset = getInset(pos))
645 inset->setChange(change);
651 void Paragraph::setChange(pos_type pos, Change const & change)
653 LASSERT(pos >= 0 && pos <= size(), return);
654 d->changes_.set(change, pos);
656 // see comment in setChange(Change const &) above
657 if (!change.deleted() && pos < size())
658 if (Inset * inset = getInset(pos))
659 inset->setChange(change);
663 Change const & Paragraph::lookupChange(pos_type pos) const
665 LBUFERR(pos >= 0 && pos <= size());
666 return d->changes_.lookup(pos);
670 void Paragraph::acceptChanges(pos_type start, pos_type end)
672 LASSERT(start >= 0 && start <= size(), return);
673 LASSERT(end > start && end <= size() + 1, return);
675 for (pos_type pos = start; pos < end; ++pos) {
676 switch (lookupChange(pos).type) {
677 case Change::UNCHANGED:
678 // accept changes in nested inset
679 if (Inset * inset = getInset(pos))
680 inset->acceptChanges();
683 case Change::INSERTED:
684 d->changes_.set(Change(Change::UNCHANGED), pos);
685 // also accept changes in nested inset
686 if (Inset * inset = getInset(pos))
687 inset->acceptChanges();
690 case Change::DELETED:
691 // Suppress access to non-existent
692 // "end-of-paragraph char"
694 eraseChar(pos, false);
705 void Paragraph::rejectChanges(pos_type start, pos_type end)
707 LASSERT(start >= 0 && start <= size(), return);
708 LASSERT(end > start && end <= size() + 1, return);
710 for (pos_type pos = start; pos < end; ++pos) {
711 switch (lookupChange(pos).type) {
712 case Change::UNCHANGED:
713 // reject changes in nested inset
714 if (Inset * inset = getInset(pos))
715 inset->rejectChanges();
718 case Change::INSERTED:
719 // Suppress access to non-existent
720 // "end-of-paragraph char"
722 eraseChar(pos, false);
728 case Change::DELETED:
729 d->changes_.set(Change(Change::UNCHANGED), pos);
731 // Do NOT reject changes within a deleted inset!
732 // There may be insertions of a co-author inside of it!
740 void Paragraph::Private::insertChar(pos_type pos, char_type c,
741 Change const & change)
743 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
746 changes_.insert(change, pos);
748 // This is actually very common when parsing buffers (and
749 // maybe inserting ascii text)
750 if (pos == pos_type(text_.size())) {
751 // when appending characters, no need to update tables
753 // but we want spell checking
754 requestSpellCheck(pos);
758 text_.insert(text_.begin() + pos, c);
760 // Update the font table.
761 fontlist_.increasePosAfterPos(pos);
764 insetlist_.increasePosAfterPos(pos);
766 // Update list of misspelled positions
767 speller_state_.increasePosAfterPos(pos);
771 bool Paragraph::insertInset(pos_type pos, Inset * inset,
772 Font const & font, Change const & change)
774 LASSERT(inset, return false);
775 LASSERT(pos >= 0 && pos <= size(), return false);
777 // Paragraph::insertInset() can be used in cut/copy/paste operation where
778 // d->inset_owner_ is not set yet.
779 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
782 d->insertChar(pos, META_INSET, change);
783 LASSERT(d->text_[pos] == META_INSET, return false);
785 // Add a new entry in the insetlist_.
786 d->insetlist_.insert(inset, pos);
788 // Some insets require run of spell checker
789 requestSpellCheck(pos);
795 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
797 LASSERT(pos >= 0 && pos <= size(), return false);
799 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
802 Change change = d->changes_.lookup(pos);
804 // set the character to DELETED if
805 // a) it was previously unchanged or
806 // b) it was inserted by a co-author
808 if (!change.changed() ||
809 (change.inserted() && !change.currentAuthor())) {
810 setChange(pos, Change(Change::DELETED));
811 // request run of spell checker
812 requestSpellCheck(pos);
816 if (change.deleted())
820 // Don't physically access the imaginary end-of-paragraph character.
821 // eraseChar() can only mark it as DELETED. A physical deletion of
822 // end-of-par must be handled externally.
828 d->changes_.erase(pos);
830 // if it is an inset, delete the inset entry
831 if (d->text_[pos] == META_INSET)
832 d->insetlist_.erase(pos);
834 d->text_.erase(d->text_.begin() + pos);
836 // Update the fontlist_
837 d->fontlist_.erase(pos);
839 // Update the insetlist_
840 d->insetlist_.decreasePosAfterPos(pos);
842 // Update list of misspelled positions
843 d->speller_state_.decreasePosAfterPos(pos);
844 d->speller_state_.refreshLast(size());
850 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
852 LASSERT(start >= 0 && start <= size(), return 0);
853 LASSERT(end >= start && end <= size() + 1, return 0);
856 for (pos_type count = end - start; count; --count) {
857 if (!eraseChar(i, trackChanges))
863 // Handle combining characters
864 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
865 otexstream & os, char_type c, char_type next,
866 OutputParams const & runparams)
868 // Writing next here may circumvent a possible font change between
869 // c and next. Since next is only output if it forms a surrogate pair
870 // with c we can ignore this:
871 // A font change inside a surrogate pair does not make sense and is
872 // hopefully impossible to input.
873 // FIXME: change tracking
874 // Is this correct WRT change tracking?
875 Encoding const & encoding = *(runparams.encoding);
876 docstring latex1 = encoding.latexChar(next).first;
877 if (runparams.inIPA) {
878 string const tipashortcut = Encodings::TIPAShortcut(next);
879 if (!tipashortcut.empty()) {
880 latex1 = from_ascii(tipashortcut);
883 docstring latex2 = encoding.latexChar(c).first;
885 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
886 // Encoding supports the combination:
887 // output as is (combining char after base char).
888 os << latex2 << latex1;
889 return latex1.length() + latex2.length();
892 os << latex1 << "{" << latex2 << "}";
893 return latex1.length() + latex2.length() + 2;
897 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
898 OutputParams const & runparams,
901 unsigned int & column,
903 Layout const & style)
905 if (style.pass_thru || runparams.pass_thru)
908 if (i + 1 < int(text_.size())) {
909 char_type next = text_[i + 1];
910 if (Encodings::isCombiningChar(next)) {
911 // This space has an accent, so we must always output it.
912 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
917 if (runparams.linelen > 0
918 && column > runparams.linelen
920 && text_[i - 1] != ' '
921 && (i + 1 < int(text_.size()))
922 // same in FreeSpacing mode
923 && !owner_->isFreeSpacing()
924 // In typewriter mode, we want to avoid
925 // ! . ? : at the end of a line
926 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
927 && (text_[i - 1] == '.'
928 || text_[i - 1] == '?'
929 || text_[i - 1] == ':'
930 || text_[i - 1] == '!'))) {
932 os.texrow().start(owner_->id(), i + 1);
934 } else if (style.free_spacing) {
943 void Paragraph::Private::latexInset(BufferParams const & bparams,
945 OutputParams & runparams,
948 Font const & outerfont,
950 Change & running_change,
951 Layout const & style,
953 unsigned int & column)
955 Inset * inset = owner_->getInset(i);
958 if (style.pass_thru) {
959 odocstringstream ods;
960 inset->plaintext(ods, runparams);
965 // FIXME: move this to InsetNewline::latex
966 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
967 // newlines are handled differently here than
968 // the default in simpleTeXSpecialChars().
969 if (!style.newline_allowed) {
973 bool needPar = false;
974 column += running_font.latexWriteEndChanges(
975 os, bparams, runparams,
976 basefont, basefont, needPar);
980 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
983 basefont = owner_->getLayoutFont(bparams, outerfont);
984 running_font = basefont;
986 if (runparams.moving_arg)
990 os.texrow().start(owner_->id(), i + 1);
994 if (owner_->isDeleted(i)) {
995 if( ++runparams.inDeletedInset == 1)
996 runparams.changeOfDeletedInset = owner_->lookupChange(i);
999 if (inset->canTrackChanges()) {
1000 column += Changes::latexMarkChange(os, bparams, running_change,
1001 Change(Change::UNCHANGED), runparams);
1002 running_change = Change(Change::UNCHANGED);
1006 odocstream::pos_type const len = os.os().tellp();
1008 if (inset->forceLTR()
1009 && running_font.isRightToLeft()
1010 // ERT is an exception, it should be output with no
1011 // decorations at all
1012 && inset->lyxCode() != ERT_CODE) {
1013 if (runparams.use_polyglossia) {
1015 } else if (running_font.language()->lang() == "farsi"
1016 || running_font.language()->lang() == "arabic_arabi")
1017 os << "\\textLR{" << termcmd;
1023 // FIXME: Bug: we can have an empty font change here!
1024 // if there has just been a font change, we are going to close it
1025 // right now, which means stupid latex code like \textsf{}. AFAIK,
1026 // this does not harm dvi output. A minor bug, thus (JMarc)
1028 // Some insets cannot be inside a font change command.
1029 // However, even such insets *can* be placed in \L or \R
1030 // or their equivalents (for RTL language switches), so we don't
1031 // close the language in those cases.
1032 // ArabTeX, though, cannot handle this special behavior, it seems.
1033 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1034 || running_font.language()->lang() == "arabic_arabtex";
1035 if (open_font && !inset->inheritFont()) {
1036 bool needPar = false;
1037 bool closeLanguage = arabtex
1038 || basefont.isRightToLeft() == running_font.isRightToLeft();
1039 unsigned int count = running_font.latexWriteEndChanges(os,
1040 bparams, runparams, basefont, basefont,
1041 needPar, closeLanguage);
1043 // if any font properties were closed, update the running_font,
1044 // making sure, however, to leave the language as it was
1046 // FIXME: probably a better way to keep track of the old
1047 // language, than copying the entire font?
1048 Font const copy_font(running_font);
1049 basefont = owner_->getLayoutFont(bparams, outerfont);
1050 running_font = basefont;
1052 running_font.setLanguage(copy_font.language());
1053 // leave font open if language is still open
1054 open_font = (running_font.language() == basefont.language());
1056 runparams.local_font = &basefont;
1060 size_t const previous_row_count = os.texrow().rows();
1063 runparams.lastid = id_;
1064 runparams.lastpos = i;
1065 inset->latex(os, runparams);
1066 } catch (EncodingException & e) {
1067 // add location information and throw again.
1076 if (os.texrow().rows() > previous_row_count) {
1077 os.texrow().start(owner_->id(), i + 1);
1080 column += (unsigned int)(os.os().tellp() - len);
1083 if (owner_->isDeleted(i))
1084 --runparams.inDeletedInset;
1088 void Paragraph::Private::latexSpecialChar(otexstream & os,
1089 BufferParams const & bparams,
1090 OutputParams const & runparams,
1091 Font const & running_font,
1092 string & alien_script,
1093 Layout const & style,
1096 unsigned int & column)
1098 char_type const c = owner_->getUChar(bparams, runparams, i);
1100 if (style.pass_thru || runparams.pass_thru
1101 || contains(style.pass_thru_chars, c)
1102 || contains(runparams.pass_thru_chars, c)) {
1104 Encoding const * const enc = runparams.encoding;
1105 if (enc && !enc->encodable(c))
1106 throw EncodingException(c);
1112 // TIPA uses its own T3 encoding
1113 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1115 // If T1 font encoding is used, use the special
1116 // characters it provides.
1117 // NOTE: Some languages reset the font encoding internally to a
1118 // non-standard font encoding. If we are using such a language,
1119 // we do not output special T1 chars.
1120 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1121 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1122 && latexSpecialT1(c, os, i, column))
1124 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1125 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1126 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1129 // Otherwise, we use what LaTeX provides us.
1132 os << "\\textbackslash" << termcmd;
1136 os << "\\textless" << termcmd;
1140 os << "\\textgreater" << termcmd;
1144 os << "\\textbar" << termcmd;
1149 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1150 (end_pos == -1 || i + 1 < end_pos) &&
1151 text_[i+1] == '-') {
1152 // Prevent "--" becoming an en dash and "---" an em dash.
1153 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1159 os << "\\textquotedbl" << termcmd;
1164 case '%': case '#': case '{':
1172 os << "\\textasciitilde" << termcmd;
1177 os << "\\textasciicircum" << termcmd;
1184 // avoid being mistaken for optional arguments
1192 // Blanks are printed before font switching.
1193 // Sure? I am not! (try nice-latex)
1194 // I am sure it's correct. LyX might be smarter
1195 // in the future, but for now, nothing wrong is
1201 // XeTeX's dash behaviour is determined via a global setting
1202 if (bparams.use_dash_ligatures
1203 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1205 // TODO #10961: && not in inset Flex Code
1206 // TODO #10961: && not in layout LyXCode
1207 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1224 Encoding const & encoding = *(runparams.encoding);
1225 char_type next = '\0';
1226 if (i + 1 < int(text_.size())) {
1227 next = text_[i + 1];
1228 if (Encodings::isCombiningChar(next)) {
1229 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1234 pair<docstring, bool> latex = encoding.latexChar(c);
1235 docstring nextlatex;
1236 bool nexttipas = false;
1237 string nexttipashortcut;
1238 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1239 nextlatex = encoding.latexChar(next).first;
1240 if (runparams.inIPA) {
1241 nexttipashortcut = Encodings::TIPAShortcut(next);
1242 nexttipas = !nexttipashortcut.empty();
1246 if (runparams.inIPA) {
1247 string const tipashortcut = Encodings::TIPAShortcut(c);
1248 if (!tipashortcut.empty()) {
1249 latex.first = from_ascii(tipashortcut);
1250 latex.second = false;
1254 // eventually close "script wrapper" command (see `Paragraph::latex`)
1255 if (!alien_script.empty()
1256 && alien_script != Encodings::isKnownScriptChar(next)) {
1257 column += latex.first.length();
1258 alien_script.clear();
1259 os << latex.first << "}";
1263 && ((!prefixIs(nextlatex, '\\')
1264 && !prefixIs(nextlatex, '{')
1265 && !prefixIs(nextlatex, '}'))
1267 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1269 // Prevent eating of a following space or command corruption by
1270 // following characters
1271 if (next == ' ' || next == '\0') {
1272 column += latex.first.length() + 1;
1273 os << latex.first << "{}";
1275 column += latex.first.length();
1276 os << latex.first << " ";
1279 column += latex.first.length() - 1;
1287 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1288 pos_type i, unsigned int & column)
1294 // In T1 encoding, these characters exist
1295 // but we should avoid ligatures
1296 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1298 os << "\\textcompwordmark" << termcmd;
1305 // soul.sty breaks with \char`\"
1306 os << "\\textquotedbl" << termcmd;
1315 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1316 pos_type i, unsigned int & column)
1318 // TU encoding is currently on par with T1.
1319 return latexSpecialT1(c, os, i, column);
1323 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1324 pos_type /*i*/, unsigned int & column)
1334 os << "\\textvertline" << termcmd;
1343 void Paragraph::Private::validate(LaTeXFeatures & features) const
1345 if (layout_->inpreamble && inset_owner_) {
1346 // FIXME: Using a string stream here circumvents the encoding
1347 // switching machinery of odocstream. Therefore the
1348 // output is wrong if this paragraph contains content
1349 // that needs to switch encoding.
1350 Buffer const & buf = inset_owner_->buffer();
1351 otexstringstream os;
1352 os << layout_->preamble();
1353 size_t const length = os.length();
1354 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1355 buf.getParFromID(owner_->id()).pit(), os,
1356 features.runparams(), string(), 0, -1, true);
1357 if (os.length() > length)
1358 features.addPreambleSnippet(os.release(), true);
1361 if (features.runparams().flavor == OutputParams::HTML
1362 && layout_->htmltitle()) {
1363 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1366 // check the params.
1367 if (!params_.spacing().isDefault())
1368 features.require("setspace");
1371 features.useLayout(layout_->name());
1374 fontlist_.validate(features);
1376 // then the indentation
1377 if (!params_.leftIndent().zero())
1378 features.require("ParagraphLeftIndent");
1381 InsetList::const_iterator icit = insetlist_.begin();
1382 InsetList::const_iterator iend = insetlist_.end();
1383 for (; icit != iend; ++icit) {
1385 features.inDeletedInset(owner_->isDeleted(icit->pos));
1386 if (icit->inset->lyxCode() == FOOT_CODE) {
1387 // FIXME: an item inset would make things much easier.
1388 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1389 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1390 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1391 && (icit->pos < begin_of_body_
1392 || (icit->pos == begin_of_body_
1393 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1394 features.saveNoteEnv("description");
1396 icit->inset->validate(features);
1397 features.inDeletedInset(false);
1398 if (layout_->needprotect &&
1399 icit->inset->lyxCode() == FOOT_CODE)
1400 features.require("NeedLyXFootnoteCode");
1404 // then the contents
1405 BufferParams const bp = features.runparams().is_child
1406 ? features.buffer().masterParams() : features.buffer().params();
1407 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1408 char_type c = text_[i];
1409 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1411 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1412 features.require("textquotedblp");
1413 else if (bp.main_font_encoding() != "T1"
1414 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1415 features.require("textquotedbl");
1416 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1417 // features that depend on the font or input encoding
1418 string feats = ci.textpreamble();
1419 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1420 if (fontenc.empty())
1421 fontenc = features.runparams().main_fontenc;
1422 while (!feats.empty()) {
1424 feats = split(feats, feat, ',');
1425 if (contains(feat, "!=")) {
1426 // a feature that is required except for the spcified
1427 // font or input encodings
1429 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1430 // multiple encodings are separated by semicolon
1431 vector<string> context = getVectorFromString(contexts, ";");
1432 // require feature if the context matches neither current font
1433 // nor input encoding
1434 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1435 && std::find(context.begin(), context.end(),
1436 features.runparams().encoding->name()) == context.end())
1437 features.require(realfeature);
1438 } else if (contains(feat, '=')) {
1439 // a feature that is required only for the spcified
1440 // font or input encodings
1442 string const contexts = split(feat, realfeature, '=');
1443 // multiple encodings are separated by semicolon
1444 vector<string> context = getVectorFromString(contexts, ";");
1445 // require feature if the context matches either current font
1446 // or input encoding
1447 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1448 || std::find(context.begin(), context.end(),
1449 features.runparams().encoding->name()) != context.end())
1450 features.require(realfeature);
1453 } else if (!bp.use_dash_ligatures
1454 && (c == 0x2013 || c == 0x2014)
1455 && bp.useNonTeXFonts
1456 && features.runparams().flavor == OutputParams::XETEX)
1457 // XeTeX's dash behaviour is determined via a global setting
1458 features.require("xetexdashbreakstate");
1459 BufferEncodings::validate(c, features);
1463 /////////////////////////////////////////////////////////////////////
1467 /////////////////////////////////////////////////////////////////////
1470 Layout const emptyParagraphLayout;
1473 Paragraph::Paragraph()
1474 : d(new Paragraph::Private(this, emptyParagraphLayout))
1481 Paragraph::Paragraph(Paragraph const & par)
1482 : itemdepth(par.itemdepth),
1483 d(new Paragraph::Private(*par.d, this))
1489 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1490 : itemdepth(par.itemdepth),
1491 d(new Paragraph::Private(*par.d, this, beg, end))
1497 Paragraph & Paragraph::operator=(Paragraph const & par)
1499 // needed as we will destroy the private part before copying it
1501 itemdepth = par.itemdepth;
1505 d = new Private(*par.d, this);
1512 Paragraph::~Paragraph()
1521 // this shall be called just before every "os << ..." action.
1522 void flushString(ostream & os, docstring & s)
1531 void Paragraph::write(ostream & os, BufferParams const & bparams,
1532 depth_type & dth) const
1534 // The beginning or end of a deeper (i.e. nested) area?
1535 if (dth != d->params_.depth()) {
1536 if (d->params_.depth() > dth) {
1537 while (d->params_.depth() > dth) {
1538 os << "\n\\begin_deeper";
1542 while (d->params_.depth() < dth) {
1543 os << "\n\\end_deeper";
1549 // First write the layout
1550 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1552 d->params_.write(os);
1554 Font font1(inherit_font, bparams.language);
1556 Change running_change = Change(Change::UNCHANGED);
1558 // this string is used as a buffer to avoid repetitive calls
1559 // to to_utf8(), which turn out to be expensive (JMarc)
1560 docstring write_buffer;
1563 for (pos_type i = 0; i <= size(); ++i) {
1565 Change const & change = lookupChange(i);
1566 if (change != running_change)
1567 flushString(os, write_buffer);
1568 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1569 running_change = change;
1574 // Write font changes
1575 Font font2 = getFontSettings(bparams, i);
1576 if (font2 != font1) {
1577 flushString(os, write_buffer);
1578 font2.lyxWriteChanges(font1, os);
1583 char_type const c = d->text_[i];
1586 if (Inset const * inset = getInset(i)) {
1587 flushString(os, write_buffer);
1588 if (inset->directWrite()) {
1589 // international char, let it write
1590 // code directly so it's shorter in
1596 os << "\\begin_inset ";
1598 os << "\n\\end_inset\n\n";
1601 // FIXME This can be removed again once the mystery
1602 // crash has been resolved.
1607 flushString(os, write_buffer);
1608 os << "\n\\backslash\n";
1612 flushString(os, write_buffer);
1613 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1620 if ((column > 70 && c == ' ')
1622 flushString(os, write_buffer);
1626 // this check is to amend a bug. LyX sometimes
1627 // inserts '\0' this could cause problems.
1629 write_buffer.push_back(c);
1631 LYXERR0("NUL char in structure.");
1637 flushString(os, write_buffer);
1638 os << "\n\\end_layout\n";
1639 // FIXME This can be removed again once the mystery
1640 // crash has been resolved.
1645 void Paragraph::validate(LaTeXFeatures & features) const
1647 d->validate(features);
1648 bool fragile = features.runparams().moving_arg;
1649 fragile |= layout().needprotect;
1650 if (needsCProtection(fragile))
1651 features.require("cprotect");
1655 void Paragraph::insert(pos_type start, docstring const & str,
1656 Font const & font, Change const & change)
1658 for (size_t i = 0, n = str.size(); i != n ; ++i)
1659 insertChar(start + i, str[i], font, change);
1663 void Paragraph::appendChar(char_type c, Font const & font,
1664 Change const & change)
1667 d->changes_.insert(change, d->text_.size());
1668 // when appending characters, no need to update tables
1669 d->text_.push_back(c);
1670 setFont(d->text_.size() - 1, font);
1671 d->requestSpellCheck(d->text_.size() - 1);
1675 void Paragraph::appendString(docstring const & s, Font const & font,
1676 Change const & change)
1678 pos_type end = s.size();
1679 size_t oldsize = d->text_.size();
1680 size_t newsize = oldsize + end;
1681 size_t capacity = d->text_.capacity();
1682 if (newsize >= capacity)
1683 d->text_.reserve(max(capacity + 100, newsize));
1685 // when appending characters, no need to update tables
1688 // FIXME: Optimize this!
1689 for (size_t i = oldsize; i != newsize; ++i) {
1691 d->changes_.insert(change, i);
1692 d->requestSpellCheck(i);
1694 d->fontlist_.set(oldsize, font);
1695 d->fontlist_.set(newsize - 1, font);
1699 void Paragraph::insertChar(pos_type pos, char_type c,
1702 d->insertChar(pos, c, Change(trackChanges ?
1703 Change::INSERTED : Change::UNCHANGED));
1707 void Paragraph::insertChar(pos_type pos, char_type c,
1708 Font const & font, bool trackChanges)
1710 d->insertChar(pos, c, Change(trackChanges ?
1711 Change::INSERTED : Change::UNCHANGED));
1716 void Paragraph::insertChar(pos_type pos, char_type c,
1717 Font const & font, Change const & change)
1719 d->insertChar(pos, c, change);
1724 void Paragraph::resetFonts(Font const & font)
1726 d->fontlist_.clear();
1727 d->fontlist_.set(0, font);
1728 d->fontlist_.set(d->text_.size() - 1, font);
1731 // Gets uninstantiated font setting at position.
1732 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1736 LYXERR0("pos: " << pos << " size: " << size());
1740 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1741 if (cit != d->fontlist_.end())
1744 if (pos == size() && !empty())
1745 return getFontSettings(bparams, pos - 1);
1747 // Optimisation: avoid a full font instantiation if there is no
1748 // language change from previous call.
1749 static Font previous_font;
1750 static Language const * previous_lang = 0;
1751 Language const * lang = getParLanguage(bparams);
1752 if (lang != previous_lang) {
1753 previous_lang = lang;
1754 previous_font = Font(inherit_font, lang);
1756 return previous_font;
1760 FontSpan Paragraph::fontSpan(pos_type pos) const
1762 LBUFERR(pos <= size());
1765 return FontSpan(pos, pos);
1768 FontList::const_iterator cit = d->fontlist_.begin();
1769 FontList::const_iterator end = d->fontlist_.end();
1770 for (; cit != end; ++cit) {
1771 if (cit->pos() >= pos) {
1772 if (pos >= beginOfBody())
1773 return FontSpan(max(start, beginOfBody()),
1776 return FontSpan(start,
1777 min(beginOfBody() - 1,
1780 start = cit->pos() + 1;
1783 // This should not happen, but if so, we take no chances.
1784 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1785 LASSERT(false, return FontSpan(pos, pos));
1789 // Gets uninstantiated font setting at position 0
1790 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1792 if (!empty() && !d->fontlist_.empty())
1793 return d->fontlist_.begin()->font();
1795 // Optimisation: avoid a full font instantiation if there is no
1796 // language change from previous call.
1797 static Font previous_font;
1798 static Language const * previous_lang = 0;
1799 if (bparams.language != previous_lang) {
1800 previous_lang = bparams.language;
1801 previous_font = Font(inherit_font, bparams.language);
1804 return previous_font;
1808 // Gets the fully instantiated font at a given position in a paragraph
1809 // This is basically the same function as Text::GetFont() in text2.cpp.
1810 // The difference is that this one is used for generating the LaTeX file,
1811 // and thus cosmetic "improvements" are disallowed: This has to deliver
1812 // the true picture of the buffer. (Asger)
1813 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1814 Font const & outerfont) const
1818 Font font = getFontSettings(bparams, pos);
1820 pos_type const body_pos = beginOfBody();
1821 FontInfo & fi = font.fontInfo();
1823 fi.realize(d->layout_->labelfont);
1825 fi.realize(d->layout_->font);
1827 fi.realize(outerfont.fontInfo());
1828 fi.realize(bparams.getFont().fontInfo());
1834 Font const Paragraph::getLabelFont
1835 (BufferParams const & bparams, Font const & outerfont) const
1837 FontInfo tmpfont = d->layout_->labelfont;
1838 tmpfont.realize(outerfont.fontInfo());
1839 tmpfont.realize(bparams.getFont().fontInfo());
1840 return Font(tmpfont, getParLanguage(bparams));
1844 Font const Paragraph::getLayoutFont
1845 (BufferParams const & bparams, Font const & outerfont) const
1847 FontInfo tmpfont = d->layout_->font;
1848 tmpfont.realize(outerfont.fontInfo());
1849 tmpfont.realize(bparams.getFont().fontInfo());
1850 return Font(tmpfont, getParLanguage(bparams));
1854 char_type Paragraph::getUChar(BufferParams const & bparams,
1855 OutputParams const & rp,
1858 char_type c = d->text_[pos];
1860 // Return unchanged character in LTR languages
1861 // or if we use poylglossia/bidi.
1862 if (rp.use_polyglossia || !getFontSettings(bparams, pos).isRightToLeft())
1865 // Without polyglossia/bidi, we need to account for some special cases.
1866 // FIXME This needs to be audited!
1868 // * The input is as expected for all delimiters
1869 // => checked for Hebrew!
1870 // * The output matches the display in the LyX workarea
1871 // => checked for Hebrew!
1872 // * The special cases below are really necessary
1873 // => checked for Hebrew!
1874 // * In arabic_arabi, brackets are transformed to Arabic
1875 // Ornate Parentheses. Is this is really wanted?
1877 string const & lang = getFontSettings(bparams, pos).language()->lang();
1880 // 1. In the following languages, parentheses need to be reversed.
1881 bool const reverseparens = lang == "hebrew";
1883 // 2. In the following languages, brackets don't need to be reversed.
1884 bool const reversebrackets = lang != "arabic_arabtex"
1885 && lang != "arabic_arabi"
1888 // Now swap delimiters if needed.
1899 if (reversebrackets)
1903 if (reversebrackets)
1924 void Paragraph::setFont(pos_type pos, Font const & font)
1926 LASSERT(pos <= size(), return);
1928 // First, reduce font against layout/label font
1929 // Update: The setCharFont() routine in text2.cpp already
1930 // reduces font, so we don't need to do that here. (Asger)
1932 d->fontlist_.set(pos, font);
1936 void Paragraph::makeSameLayout(Paragraph const & par)
1938 d->layout_ = par.d->layout_;
1939 d->params_ = par.d->params_;
1943 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1945 if (isFreeSpacing())
1951 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1952 if (eraseChar(pos, trackChanges))
1958 return count > 0 || pos > 0;
1962 bool Paragraph::hasSameLayout(Paragraph const & par) const
1964 return par.d->layout_ == d->layout_
1965 && d->params_.sameLayout(par.d->params_);
1969 depth_type Paragraph::getDepth() const
1971 return d->params_.depth();
1975 depth_type Paragraph::getMaxDepthAfter() const
1977 if (d->layout_->isEnvironment())
1978 return d->params_.depth() + 1;
1980 return d->params_.depth();
1984 LyXAlignment Paragraph::getAlign() const
1986 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1987 return d->layout_->align;
1989 return d->params_.align();
1993 docstring const & Paragraph::labelString() const
1995 return d->params_.labelString();
1999 // the next two functions are for the manual labels
2000 docstring const Paragraph::getLabelWidthString() const
2002 if (d->layout_->margintype == MARGIN_MANUAL
2003 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2004 return d->params_.labelWidthString();
2006 return _("Senseless with this layout!");
2010 void Paragraph::setLabelWidthString(docstring const & s)
2012 d->params_.labelWidthString(s);
2016 docstring Paragraph::expandLabel(Layout const & layout,
2017 BufferParams const & bparams) const
2019 return expandParagraphLabel(layout, bparams, true);
2023 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2024 BufferParams const & bparams) const
2026 return expandParagraphLabel(layout, bparams, false);
2030 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2031 BufferParams const & bparams, bool process_appendix) const
2033 DocumentClass const & tclass = bparams.documentClass();
2034 string const & lang = getParLanguage(bparams)->code();
2035 bool const in_appendix = process_appendix && d->params_.appendix();
2036 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2038 if (fmt.empty() && !layout.counter.empty())
2039 return tclass.counters().theCounter(layout.counter, lang);
2041 // handle 'inherited level parts' in 'fmt',
2042 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2043 size_t const i = fmt.find('@', 0);
2044 if (i != docstring::npos) {
2045 size_t const j = fmt.find('@', i + 1);
2046 if (j != docstring::npos) {
2047 docstring parent(fmt, i + 1, j - i - 1);
2048 docstring label = from_ascii("??");
2049 if (tclass.hasLayout(parent))
2050 label = expandParagraphLabel(tclass[parent], bparams,
2052 fmt = docstring(fmt, 0, i) + label
2053 + docstring(fmt, j + 1, docstring::npos);
2057 return tclass.counters().counterLabel(fmt, lang);
2061 void Paragraph::applyLayout(Layout const & new_layout)
2063 d->layout_ = &new_layout;
2064 LyXAlignment const oldAlign = d->params_.align();
2066 if (!(oldAlign & d->layout_->alignpossible)) {
2067 frontend::Alert::warning(_("Alignment not permitted"),
2068 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2069 d->params_.align(LYX_ALIGN_LAYOUT);
2074 pos_type Paragraph::beginOfBody() const
2076 return d->begin_of_body_;
2080 void Paragraph::setBeginOfBody()
2082 if (d->layout_->labeltype != LABEL_MANUAL) {
2083 d->begin_of_body_ = 0;
2087 // Unroll the first two cycles of the loop
2088 // and remember the previous character to
2089 // remove unnecessary getChar() calls
2091 pos_type end = size();
2092 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2095 char_type previous_char = d->text_[i];
2096 if (!(isNewline(i) || isEnvSeparator(i))) {
2098 while (i < end && previous_char != ' ') {
2099 char_type temp = d->text_[i];
2100 if (isNewline(i) || isEnvSeparator(i))
2103 previous_char = temp;
2109 d->begin_of_body_ = i;
2113 bool Paragraph::allowParagraphCustomization() const
2115 return inInset().allowParagraphCustomization();
2119 bool Paragraph::usePlainLayout() const
2121 return inInset().usePlainLayout();
2125 bool Paragraph::isPassThru() const
2127 return inInset().isPassThru() || d->layout_->pass_thru;
2132 // paragraphs inside floats need different alignment tags to avoid
2135 bool noTrivlistCentering(InsetCode code)
2137 return code == FLOAT_CODE
2138 || code == WRAP_CODE
2139 || code == CELL_CODE;
2143 string correction(string const & orig)
2145 if (orig == "flushleft")
2146 return "raggedright";
2147 if (orig == "flushright")
2148 return "raggedleft";
2149 if (orig == "center")
2155 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2156 InsetCode code, bool const lastpar, int & col)
2158 string macro = suffix + "{";
2159 if (noTrivlistCentering(code)) {
2161 // the last paragraph in non-trivlist-aligned
2162 // context is special (to avoid unwanted whitespace)
2163 if (suffix == "\\begin") {
2164 macro = "\\" + correction(env) + "{}";
2165 os << from_ascii(macro);
2166 col += macro.size();
2171 macro += correction(env);
2175 if (suffix == "\\par\\end") {
2179 os << from_ascii(macro);
2180 col += macro.size();
2181 if (suffix == "\\begin") {
2191 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2192 otexstream & os, OutputParams const & runparams) const
2197 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2198 (layout_->toggle_indent != ITOGGLE_NEVER) :
2199 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2201 if (canindent && params_.noindent() && !layout_->pass_thru) {
2202 os << "\\noindent ";
2206 LyXAlignment const curAlign = params_.align();
2208 if (curAlign == layout_->align)
2212 case LYX_ALIGN_NONE:
2213 case LYX_ALIGN_BLOCK:
2214 case LYX_ALIGN_LAYOUT:
2215 case LYX_ALIGN_SPECIAL:
2216 case LYX_ALIGN_DECIMAL:
2218 case LYX_ALIGN_LEFT:
2219 case LYX_ALIGN_RIGHT:
2220 case LYX_ALIGN_CENTER:
2221 if (runparams.moving_arg) {
2228 string const begin_tag = "\\begin";
2229 InsetCode code = ownerCode();
2230 bool const lastpar = runparams.isLastPar;
2231 // RTL in classic (PDF)LaTeX (without the Bidi package)
2232 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2233 && !runparams.use_polyglossia;
2236 case LYX_ALIGN_NONE:
2237 case LYX_ALIGN_BLOCK:
2238 case LYX_ALIGN_LAYOUT:
2239 case LYX_ALIGN_SPECIAL:
2240 case LYX_ALIGN_DECIMAL:
2242 case LYX_ALIGN_LEFT: {
2244 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2245 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2247 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2249 } case LYX_ALIGN_RIGHT: {
2251 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2252 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2254 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2256 } case LYX_ALIGN_CENTER: {
2257 corrected_env(os, begin_tag, "center", code, lastpar, column);
2266 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2267 otexstream & os, OutputParams const & runparams) const
2269 LyXAlignment const curAlign = params_.align();
2271 if (curAlign == layout_->align)
2275 case LYX_ALIGN_NONE:
2276 case LYX_ALIGN_BLOCK:
2277 case LYX_ALIGN_LAYOUT:
2278 case LYX_ALIGN_SPECIAL:
2279 case LYX_ALIGN_DECIMAL:
2281 case LYX_ALIGN_LEFT:
2282 case LYX_ALIGN_RIGHT:
2283 case LYX_ALIGN_CENTER:
2284 if (runparams.moving_arg)
2289 bool output = false;
2291 string const end_tag = "\\par\\end";
2292 InsetCode code = ownerCode();
2293 bool const lastpar = runparams.isLastPar;
2294 // RTL in classic (PDF)LaTeX (without the Bidi package)
2295 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2296 && !runparams.use_polyglossia;
2299 case LYX_ALIGN_NONE:
2300 case LYX_ALIGN_BLOCK:
2301 case LYX_ALIGN_LAYOUT:
2302 case LYX_ALIGN_SPECIAL:
2303 case LYX_ALIGN_DECIMAL:
2305 case LYX_ALIGN_LEFT: {
2307 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2308 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2310 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2312 } case LYX_ALIGN_RIGHT: {
2314 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2315 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2317 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2319 } case LYX_ALIGN_CENTER: {
2320 corrected_env(os, end_tag, "center", code, lastpar, col);
2325 return output || lastpar;
2329 // This one spits out the text of the paragraph
2330 void Paragraph::latex(BufferParams const & bparams,
2331 Font const & outerfont,
2333 OutputParams const & runparams,
2334 int start_pos, int end_pos, bool force) const
2336 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2338 // FIXME This check should not be needed. Perhaps issue an
2339 // error if it triggers.
2340 Layout const & style = inInset().forcePlainLayout() ?
2341 bparams.documentClass().plainLayout() : *d->layout_;
2343 if (!force && style.inpreamble)
2346 bool const allowcust = allowParagraphCustomization();
2348 // Current base font for all inherited font changes, without any
2349 // change caused by an individual character, except for the language:
2350 // It is set to the language of the first character.
2351 // As long as we are in the label, this font is the base font of the
2352 // label. Before the first body character it is set to the base font
2356 // If there is an open font-encoding changing command (script wrapper),
2357 // alien_script is set to its name
2358 string alien_script;
2361 // Maybe we have to create a optional argument.
2362 pos_type body_pos = beginOfBody();
2363 unsigned int column = 0;
2366 // the optional argument is kept in curly brackets in
2367 // case it contains a ']'
2368 // This is not strictly needed, but if this is changed it
2369 // would be a file format change, and tex2lyx would need
2370 // to be adjusted, since it unconditionally removes the
2371 // braces when it parses \item.
2374 basefont = getLabelFont(bparams, outerfont);
2376 basefont = getLayoutFont(bparams, outerfont);
2379 // Which font is currently active?
2380 Font running_font(basefont);
2381 // Do we have an open font change?
2382 bool open_font = false;
2384 Change runningChange = Change(Change::UNCHANGED);
2386 Encoding const * const prev_encoding = runparams.encoding;
2388 os.texrow().start(id(), 0);
2390 // if the paragraph is empty, the loop will not be entered at all
2392 // For InTitle commands, we have already opened a group
2393 // in output_latex::TeXOnePar.
2394 if (style.isCommand() && !style.intitle) {
2398 if (!style.leftdelim().empty()) {
2399 os << style.leftdelim();
2400 column += style.leftdelim().size();
2403 column += d->startTeXParParams(bparams, os, runparams);
2406 // Whether a \par can be issued for insets typeset inline with text.
2407 // Yes if greater than 0. This has to be static.
2408 THREAD_LOCAL_STATIC int parInline = 0;
2410 for (pos_type i = 0; i < size(); ++i) {
2411 // First char in paragraph or after label?
2412 if (i == body_pos) {
2415 bool needPar = false;
2416 column += running_font.latexWriteEndChanges(
2417 os, bparams, runparams,
2418 basefont, basefont, needPar);
2421 basefont = getLayoutFont(bparams, outerfont);
2422 running_font = basefont;
2424 column += Changes::latexMarkChange(os, bparams,
2425 runningChange, Change(Change::UNCHANGED),
2427 runningChange = Change(Change::UNCHANGED);
2432 // For InTitle commands, we have already opened a group
2433 // in output_latex::TeXOnePar.
2434 if (style.isCommand() && !style.intitle) {
2439 if (!style.leftdelim().empty()) {
2440 os << style.leftdelim();
2441 column += style.leftdelim().size();
2445 column += d->startTeXParParams(bparams, os,
2449 runparams.wasDisplayMath = runparams.inDisplayMath;
2450 runparams.inDisplayMath = false;
2451 bool deleted_display_math = false;
2452 Change const & change = runparams.inDeletedInset
2453 ? runparams.changeOfDeletedInset : lookupChange(i);
2455 // Check whether a display math inset follows
2456 if (d->text_[i] == META_INSET
2457 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2458 InsetMath const * im = getInset(i)->asInsetMath();
2459 if (im && im->asHullInset()
2460 && im->asHullInset()->outerDisplay()) {
2461 runparams.inDisplayMath = true;
2462 // runparams.inDeletedInset will be set by
2463 // latexInset later, but we need this info
2464 // before it is called. On the other hand, we
2465 // cannot set it here because it is a counter.
2466 deleted_display_math = isDeleted(i);
2468 if (bparams.output_changes && deleted_display_math
2469 && runningChange == change
2470 && change.type == Change::DELETED
2471 && !os.afterParbreak()) {
2472 // A display math in the same paragraph follows.
2473 // We have to close and then reopen \lyxdeleted,
2474 // otherwise the math will be shifted up.
2475 OutputParams rp = runparams;
2477 bool needPar = false;
2478 column += running_font.latexWriteEndChanges(
2479 os, bparams, rp, basefont,
2483 basefont = getLayoutFont(bparams, outerfont);
2484 running_font = basefont;
2485 column += Changes::latexMarkChange(os, bparams,
2486 Change(Change::INSERTED), change, rp);
2490 if (bparams.output_changes && runningChange != change) {
2491 if (!alien_script.empty()) {
2494 alien_script.clear();
2497 bool needPar = false;
2498 column += running_font.latexWriteEndChanges(
2499 os, bparams, runparams,
2500 basefont, basefont, needPar);
2503 basefont = getLayoutFont(bparams, outerfont);
2504 running_font = basefont;
2505 column += Changes::latexMarkChange(os, bparams, runningChange,
2507 runningChange = change;
2510 // do not output text which is marked deleted
2511 // if change tracking output is disabled
2512 if (!bparams.output_changes && change.deleted()) {
2518 // Fully instantiated font
2519 Font const current_font = getFont(bparams, i, outerfont);
2521 Font const last_font = running_font;
2523 // Do we need to close the previous font?
2525 (current_font != running_font ||
2526 current_font.language() != running_font.language()))
2528 // ensure there is no open script-wrapper
2529 if (!alien_script.empty()) {
2532 alien_script.clear();
2534 bool needPar = false;
2535 column += running_font.latexWriteEndChanges(
2536 os, bparams, runparams, basefont,
2537 (i == body_pos-1) ? basefont : current_font,
2539 running_font = basefont;
2543 string const running_lang = runparams.use_polyglossia ?
2544 running_font.language()->polyglossia() : running_font.language()->babel();
2545 // close babel's font environment before opening CJK.
2546 string const lang_end_command = runparams.use_polyglossia ?
2547 "\\end{$$lang}" : lyxrc.language_command_end;
2548 bool const using_begin_end = runparams.use_polyglossia ||
2549 !lang_end_command.empty();
2550 if (!running_lang.empty() &&
2551 (!using_begin_end || running_lang == openLanguageName()) &&
2552 current_font.language()->encoding()->package() == Encoding::CJK) {
2553 string end_tag = subst(lang_end_command,
2556 os << from_ascii(end_tag);
2557 column += end_tag.length();
2558 if (using_begin_end)
2562 // Switch file encoding if necessary (and allowed)
2563 if (!runparams.pass_thru && !style.pass_thru &&
2564 runparams.encoding->package() != Encoding::none &&
2565 current_font.language()->encoding()->package() != Encoding::none) {
2566 pair<bool, int> const enc_switch =
2567 switchEncoding(os.os(), bparams, runparams,
2568 *(current_font.language()->encoding()));
2569 if (enc_switch.first) {
2570 column += enc_switch.second;
2571 runparams.encoding = current_font.language()->encoding();
2575 char_type const c = d->text_[i];
2577 // A display math inset inside an ulem command will be output
2578 // as a box of width \linewidth, so we have to either disable
2579 // indentation if the inset starts a paragraph, or start a new
2580 // line to accommodate such box. This has to be done before
2581 // writing any font changing commands.
2582 if (runparams.inDisplayMath && !deleted_display_math
2583 && runparams.inulemcmd) {
2584 if (os.afterParbreak())
2590 // Do we need to change font?
2591 if ((current_font != running_font ||
2592 current_font.language() != running_font.language()) &&
2595 odocstringstream ods;
2596 column += current_font.latexWriteStartChanges(ods, bparams,
2597 runparams, basefont,
2599 // Check again for display math in ulem commands as a
2600 // font change may also occur just before a math inset.
2601 if (runparams.inDisplayMath && !deleted_display_math
2602 && runparams.inulemcmd) {
2603 if (os.afterParbreak())
2608 running_font = current_font;
2610 docstring fontchange = ods.str();
2611 // check whether the fontchange ends with a \\textcolor
2612 // modifier and the text starts with a space (bug 4473)
2613 docstring const last_modifier = rsplit(fontchange, '\\');
2614 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2615 os << fontchange << from_ascii("{}");
2616 // check if the fontchange ends with a trailing blank
2617 // (like "\small " (see bug 3382)
2618 else if (suffixIs(fontchange, ' ') && c == ' ')
2619 os << fontchange.substr(0, fontchange.size() - 1)
2620 << from_ascii("{}");
2625 // FIXME: think about end_pos implementation...
2626 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2627 // FIXME: integrate this case in latexSpecialChar
2628 // Do not print the separation of the optional argument
2629 // if style.pass_thru is false. This works because
2630 // latexSpecialChar ignores spaces if
2631 // style.pass_thru is false.
2632 if (i != body_pos - 1) {
2633 if (d->simpleTeXBlanks(bparams, runparams, os,
2634 i, column, current_font, style)) {
2635 // A surrogate pair was output. We
2636 // must not call latexSpecialChar
2637 // in this iteration, since it would output
2638 // the combining character again.
2645 OutputParams rp = runparams;
2646 rp.free_spacing = style.free_spacing;
2647 rp.local_font = ¤t_font;
2648 rp.intitle = style.intitle;
2650 // Two major modes: LaTeX or plain
2651 // Handle here those cases common to both modes
2652 // and then split to handle the two modes separately.
2653 if (c == META_INSET) {
2654 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2655 // Greyedout notes and, in general, all insets
2656 // with InsetLayout::isDisplay() == false,
2657 // are typeset inline with the text. So, we
2658 // can add a \par to the last paragraph of
2659 // such insets only if nothing else follows.
2660 bool incremented = false;
2661 Inset const * inset = getInset(i);
2662 InsetText const * textinset = inset
2663 ? inset->asInsetText()
2665 if (i + 1 == size() && textinset
2666 && !inset->getLayout().isDisplay()) {
2667 ParagraphList const & pars =
2668 textinset->text().paragraphs();
2669 pit_type const pit = pars.size() - 1;
2670 Font const lastfont =
2671 pit < 0 || pars[pit].empty()
2672 ? pars[pit].getLayoutFont(
2675 : pars[pit].getFont(bparams,
2676 pars[pit].size() - 1,
2678 if (lastfont.fontInfo().size() !=
2679 basefont.fontInfo().size()) {
2684 d->latexInset(bparams, os, rp, running_font,
2685 basefont, outerfont, open_font,
2686 runningChange, style, i, column);
2690 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2691 if (!bparams.useNonTeXFonts)
2692 script = Encodings::isKnownScriptChar(c);
2693 if (script != alien_script) {
2694 if (!alien_script.empty()) {
2696 alien_script.clear();
2698 string fontenc = running_font.language()->fontenc(bparams);
2700 && !Encodings::fontencSupportsScript(fontenc, script)) {
2701 column += script.length() + 2;
2702 os << "\\" << script << "{";
2703 alien_script = script;
2707 d->latexSpecialChar(os, bparams, rp, running_font,
2708 alien_script, style, i, end_pos, column);
2709 } catch (EncodingException & e) {
2710 if (runparams.dryrun) {
2711 os << "<" << _("LyX Warning: ")
2712 << _("uncodable character") << " '";
2716 // add location information and throw again.
2724 // Set the encoding to that returned from latexSpecialChar (see
2725 // comment for encoding member in OutputParams.h)
2726 runparams.encoding = rp.encoding;
2728 // Also carry on the info on a closed ulem command for insets
2729 // such as Note that do not produce any output, so that no
2730 // command is ever executed but its opening was recorded.
2731 runparams.inulemcmd = rp.inulemcmd;
2733 // And finally, pass the post_macros upstream
2734 runparams.post_macro = rp.post_macro;
2737 // Close wrapper for alien script
2738 if (!alien_script.empty()) {
2740 alien_script.clear();
2743 // If we have an open font definition, we have to close it
2745 // Make sure that \\par is done with the font of the last
2746 // character if this has another size as the default.
2747 // This is necessary because LaTeX (and LyX on the screen)
2748 // calculates the space between the baselines according
2749 // to this font. (Matthias)
2751 // We must not change the font for the last paragraph
2752 // of non-multipar insets, tabular cells or commands,
2753 // since this produces unwanted whitespace.
2755 Font const font = empty()
2756 ? getLayoutFont(bparams, outerfont)
2757 : getFont(bparams, size() - 1, outerfont);
2759 InsetText const * textinset = inInset().asInsetText();
2761 bool const maintext = textinset
2762 ? textinset->text().isMainText()
2765 size_t const numpars = textinset
2766 ? textinset->text().paragraphs().size()
2769 bool needPar = false;
2771 if (style.resfont.size() != font.fontInfo().size()
2772 && (!runparams.isLastPar || maintext
2773 || (numpars > 1 && d->ownerCode() != CELL_CODE
2774 && (inInset().getLayout().isDisplay()
2776 && !style.isCommand()) {
2779 #ifdef FIXED_LANGUAGE_END_DETECTION
2781 running_font.latexWriteEndChanges(os, bparams,
2782 runparams, basefont,
2783 next_->getFont(bparams, 0, outerfont),
2786 running_font.latexWriteEndChanges(os, bparams,
2787 runparams, basefont, basefont, needPar);
2790 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2791 //FIXME: there as we start another \selectlanguage with the next paragraph if
2792 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2793 running_font.latexWriteEndChanges(os, bparams, runparams,
2794 basefont, basefont, needPar);
2797 // The \par could not be inserted at the same nesting
2798 // level of the font size change, so do it now.
2799 os << "{\\" << font.latexSize() << "\\par}";
2803 column += Changes::latexMarkChange(os, bparams, runningChange,
2804 Change(Change::UNCHANGED), runparams);
2806 // Needed if there is an optional argument but no contents.
2807 if (body_pos > 0 && body_pos == size()) {
2811 if (!style.rightdelim().empty()) {
2812 os << style.rightdelim();
2813 column += style.rightdelim().size();
2816 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2817 && runparams.encoding != prev_encoding) {
2818 runparams.encoding = prev_encoding;
2819 os << setEncoding(prev_encoding->iconvName());
2822 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2826 bool Paragraph::emptyTag() const
2828 for (pos_type i = 0; i < size(); ++i) {
2829 if (Inset const * inset = getInset(i)) {
2830 InsetCode lyx_code = inset->lyxCode();
2831 // FIXME testing like that is wrong. What is
2833 if (lyx_code != TOC_CODE &&
2834 lyx_code != INCLUDE_CODE &&
2835 lyx_code != GRAPHICS_CODE &&
2836 lyx_code != ERT_CODE &&
2837 lyx_code != LISTINGS_CODE &&
2838 lyx_code != FLOAT_CODE &&
2839 lyx_code != TABULAR_CODE) {
2843 char_type c = d->text_[i];
2844 if (c != ' ' && c != '\t')
2852 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2855 for (pos_type i = 0; i < size(); ++i) {
2856 if (Inset const * inset = getInset(i)) {
2857 InsetCode lyx_code = inset->lyxCode();
2858 if (lyx_code == LABEL_CODE) {
2859 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2860 docstring const & id = il->getParam("name");
2861 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2869 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2873 for (i = 0; i < size(); ++i) {
2874 if (Inset const * inset = getInset(i)) {
2875 inset->docbook(os, runparams);
2877 char_type c = d->text_[i];
2880 os << sgml::escapeChar(c);
2887 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2891 for (i = 0; i < size(); ++i) {
2892 if (Inset const * inset = getInset(i)) {
2893 inset->xhtml(xs, runparams);
2895 char_type c = d->text_[i];
2905 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2908 pos_type size = text_.size();
2909 for (pos_type i = initial; i < size; ++i) {
2910 Font font = owner_->getFont(buf.params(), i, outerfont);
2911 if (text_[i] == META_INSET)
2913 if (i != initial && font != font_old)
2922 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2924 OutputParams const & runparams,
2925 Font const & outerfont,
2926 pos_type initial) const
2928 bool emph_flag = false;
2930 Layout const & style = *d->layout_;
2932 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2934 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2937 // parsing main loop
2938 for (pos_type i = initial; i < size(); ++i) {
2939 Font font = getFont(buf.params(), i, outerfont);
2941 // handle <emphasis> tag
2942 if (font_old.emph() != font.fontInfo().emph()) {
2943 if (font.fontInfo().emph() == FONT_ON) {
2946 } else if (i != initial) {
2947 os << "</emphasis>";
2952 if (Inset const * inset = getInset(i)) {
2953 inset->docbook(os, runparams);
2955 char_type c = d->text_[i];
2957 if (style.pass_thru)
2960 os << sgml::escapeChar(c);
2962 font_old = font.fontInfo();
2966 os << "</emphasis>";
2969 if (style.free_spacing)
2971 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2977 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2978 vector<html::EndFontTag> & tagsToClose,
2979 bool & flag, FontState curstate, html::FontTypes type)
2981 if (curstate == FONT_ON) {
2982 tagsToOpen.push_back(html::FontTag(type));
2985 tagsToClose.push_back(html::EndFontTag(type));
2992 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2994 OutputParams const & runparams,
2995 Font const & outerfont,
2996 bool start_paragraph, bool close_paragraph,
2997 pos_type initial) const
3001 // track whether we have opened these tags
3002 bool emph_flag = false;
3003 bool bold_flag = false;
3004 bool noun_flag = false;
3005 bool ubar_flag = false;
3006 bool dbar_flag = false;
3007 bool sout_flag = false;
3008 bool xout_flag = false;
3009 bool wave_flag = false;
3011 bool shap_flag = false;
3013 bool faml_flag = false;
3015 bool size_flag = false;
3017 Layout const & style = *d->layout_;
3019 if (start_paragraph)
3020 xs.startDivision(allowEmpty());
3023 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3025 FontShape curr_fs = INHERIT_SHAPE;
3026 FontFamily curr_fam = INHERIT_FAMILY;
3027 FontSize curr_size = FONT_SIZE_INHERIT;
3029 string const default_family =
3030 buf.masterBuffer()->params().fonts_default_family;
3032 vector<html::FontTag> tagsToOpen;
3033 vector<html::EndFontTag> tagsToClose;
3035 // parsing main loop
3036 for (pos_type i = initial; i < size(); ++i) {
3037 // let's not show deleted material in the output
3041 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3044 FontState curstate = font.fontInfo().emph();
3045 if (font_old.emph() != curstate)
3046 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3049 curstate = font.fontInfo().noun();
3050 if (font_old.noun() != curstate)
3051 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3054 curstate = font.fontInfo().underbar();
3055 if (font_old.underbar() != curstate)
3056 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3059 curstate = font.fontInfo().strikeout();
3060 if (font_old.strikeout() != curstate)
3061 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3064 curstate = font.fontInfo().xout();
3065 if (font_old.xout() != curstate)
3066 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3069 curstate = font.fontInfo().uuline();
3070 if (font_old.uuline() != curstate)
3071 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3074 curstate = font.fontInfo().uwave();
3075 if (font_old.uwave() != curstate)
3076 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3079 // a little hackish, but allows us to reuse what we have.
3080 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3081 if (font_old.series() != font.fontInfo().series())
3082 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3085 curr_fs = font.fontInfo().shape();
3086 FontShape old_fs = font_old.shape();
3087 if (old_fs != curr_fs) {
3091 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3094 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3096 case SMALLCAPS_SHAPE:
3097 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3103 // the other tags are for internal use
3111 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3115 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3118 case SMALLCAPS_SHAPE:
3119 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3126 // the other tags are for internal use
3133 curr_fam = font.fontInfo().family();
3134 FontFamily old_fam = font_old.family();
3135 if (old_fam != curr_fam) {
3139 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3142 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3144 case TYPEWRITER_FAMILY:
3145 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3147 case INHERIT_FAMILY:
3150 // the other tags are for internal use
3158 // we will treat a "default" font family as roman, since we have
3159 // no other idea what to do.
3160 if (default_family != "rmdefault" && default_family != "default") {
3161 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3166 if (default_family != "sfdefault") {
3167 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3171 case TYPEWRITER_FAMILY:
3172 if (default_family != "ttdefault") {
3173 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3177 case INHERIT_FAMILY:
3180 // the other tags are for internal use
3187 curr_size = font.fontInfo().size();
3188 FontSize old_size = font_old.size();
3189 if (old_size != curr_size) {
3192 case FONT_SIZE_TINY:
3193 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3195 case FONT_SIZE_SCRIPT:
3196 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3198 case FONT_SIZE_FOOTNOTE:
3199 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3201 case FONT_SIZE_SMALL:
3202 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3204 case FONT_SIZE_LARGE:
3205 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3207 case FONT_SIZE_LARGER:
3208 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3210 case FONT_SIZE_LARGEST:
3211 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3213 case FONT_SIZE_HUGE:
3214 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3216 case FONT_SIZE_HUGER:
3217 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3219 case FONT_SIZE_INCREASE:
3220 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3222 case FONT_SIZE_DECREASE:
3223 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3225 case FONT_SIZE_INHERIT:
3226 case FONT_SIZE_NORMAL:
3229 // the other tags are for internal use
3235 switch (curr_size) {
3236 case FONT_SIZE_TINY:
3237 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3240 case FONT_SIZE_SCRIPT:
3241 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3244 case FONT_SIZE_FOOTNOTE:
3245 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3248 case FONT_SIZE_SMALL:
3249 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3252 case FONT_SIZE_LARGE:
3253 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3256 case FONT_SIZE_LARGER:
3257 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3260 case FONT_SIZE_LARGEST:
3261 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3264 case FONT_SIZE_HUGE:
3265 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3268 case FONT_SIZE_HUGER:
3269 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3272 case FONT_SIZE_INCREASE:
3273 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3276 case FONT_SIZE_DECREASE:
3277 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3280 case FONT_SIZE_NORMAL:
3281 case FONT_SIZE_INHERIT:
3284 // the other tags are for internal use
3291 // Other such tags? What about the other text ranges?
3293 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3294 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3295 for (; cit != cen; ++cit)
3298 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3299 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3300 for (; sit != sen; ++sit)
3303 tagsToClose.clear();
3306 Inset const * inset = getInset(i);
3308 if (!runparams.for_toc || inset->isInToc()) {
3309 OutputParams np = runparams;
3310 np.local_font = &font;
3311 // If the paragraph has size 1, then we are in the "special
3312 // case" where we do not output the containing paragraph info
3313 if (!inset->getLayout().htmlisblock() && size() != 1)
3314 np.html_in_par = true;
3315 retval += inset->xhtml(xs, np);
3318 char_type c = getUChar(buf.masterBuffer()->params(),
3320 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3321 xs << XHTMLStream::ESCAPE_NONE << " ";
3325 font_old = font.fontInfo();
3329 // I'm worried about what happens if a branch, say, is itself
3330 // wrapped in some font stuff. I think that will not work.
3332 if (close_paragraph)
3339 bool Paragraph::isHfill(pos_type pos) const
3341 Inset const * inset = getInset(pos);
3342 return inset && inset->isHfill();
3346 bool Paragraph::isNewline(pos_type pos) const
3348 // U+2028 LINE SEPARATOR
3349 // U+2029 PARAGRAPH SEPARATOR
3350 char_type const c = d->text_[pos];
3351 if (c == 0x2028 || c == 0x2029)
3353 Inset const * inset = getInset(pos);
3354 return inset && inset->lyxCode() == NEWLINE_CODE;
3358 bool Paragraph::isEnvSeparator(pos_type pos) const
3360 Inset const * inset = getInset(pos);
3361 return inset && inset->lyxCode() == SEPARATOR_CODE;
3365 bool Paragraph::isLineSeparator(pos_type pos) const
3367 char_type const c = d->text_[pos];
3368 if (isLineSeparatorChar(c))
3370 Inset const * inset = getInset(pos);
3371 return inset && inset->isLineSeparator();
3375 bool Paragraph::isWordSeparator(pos_type pos) const
3379 if (Inset const * inset = getInset(pos))
3380 return !inset->isLetter();
3381 // if we have a hard hyphen (no en- or emdash) or apostrophe
3382 // we pass this to the spell checker
3383 // FIXME: this method is subject to change, visit
3384 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3385 // to get an impression how complex this is.
3386 if (isHardHyphenOrApostrophe(pos))
3388 char_type const c = d->text_[pos];
3389 // We want to pass the escape chars to the spellchecker
3390 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3391 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3395 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3397 pos_type const psize = size();
3400 char_type const c = d->text_[pos];
3401 if (c != '-' && c != '\'')
3403 int nextpos = pos + 1;
3404 int prevpos = pos > 0 ? pos - 1 : 0;
3405 if ((nextpos == psize || isSpace(nextpos))
3406 && (pos == 0 || isSpace(prevpos)))
3412 bool Paragraph::needsCProtection(bool const fragile) const
3414 // first check the layout of the paragraph, but only in insets
3415 InsetText const * textinset = inInset().asInsetText();
3416 bool const maintext = textinset
3417 ? textinset->text().isMainText()
3420 if (!maintext && layout().needcprotect) {
3421 // Environments need cprotection regardless the content
3422 if (layout().latextype == LATEX_ENVIRONMENT)
3425 // Commands need cprotection if they contain specific chars
3426 int const nchars_escape = 9;
3427 static char_type const chars_escape[nchars_escape] = {
3428 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3430 docstring const pars = asString();
3431 for (int k = 0; k < nchars_escape; k++) {
3432 if (contains(pars, chars_escape[k]))
3437 // now check whether we have insets that need cprotection
3438 pos_type size = d->text_.size();
3439 for (pos_type i = 0; i < size; ++i)
3440 if (isInset(i) && getInset(i)->needsCProtection(maintext, fragile))
3447 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3449 return d->speller_state_.getRange(pos);
3453 bool Paragraph::isChar(pos_type pos) const
3455 if (Inset const * inset = getInset(pos))
3456 return inset->isChar();
3457 char_type const c = d->text_[pos];
3458 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3462 bool Paragraph::isSpace(pos_type pos) const
3464 if (Inset const * inset = getInset(pos))
3465 return inset->isSpace();
3466 char_type const c = d->text_[pos];
3467 return lyx::isSpace(c);
3472 Paragraph::getParLanguage(BufferParams const & bparams) const
3475 return getFirstFontSettings(bparams).language();
3476 // FIXME: we should check the prev par as well (Lgb)
3477 return bparams.language;
3481 bool Paragraph::isRTL(BufferParams const & bparams) const
3483 return getParLanguage(bparams)->rightToLeft()
3484 && !inInset().getLayout().forceLTR();
3488 void Paragraph::changeLanguage(BufferParams const & bparams,
3489 Language const * from, Language const * to)
3491 // change language including dummy font change at the end
3492 for (pos_type i = 0; i <= size(); ++i) {
3493 Font font = getFontSettings(bparams, i);
3494 if (font.language() == from) {
3495 font.setLanguage(to);
3497 d->requestSpellCheck(i);
3503 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3505 Language const * doc_language = bparams.language;
3506 for (auto const & f : d->fontlist_)
3507 if (f.font().language() != ignore_language &&
3508 f.font().language() != latex_language &&
3509 f.font().language() != doc_language)
3515 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3517 for (auto const & f : d->fontlist_) {
3518 Language const * lang = f.font().language();
3519 if (lang != ignore_language &&
3520 lang != latex_language)
3526 docstring Paragraph::asString(int options) const
3528 return asString(0, size(), options);
3532 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3534 odocstringstream os;
3537 && options & AS_STR_LABEL
3538 && !d->params_.labelString().empty())
3539 os << d->params_.labelString() << ' ';
3541 for (pos_type i = beg; i < end; ++i) {
3542 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3544 char_type const c = d->text_[i];
3545 if (isPrintable(c) || c == '\t'
3546 || (c == '\n' && (options & AS_STR_NEWLINES)))
3548 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3549 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3550 LASSERT(runparams != 0, return docstring());
3551 getInset(i)->plaintext(os, *runparams);
3553 getInset(i)->toString(os);
3562 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3563 bool const shorten, bool const label) const
3565 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3566 if (label && !labelString().empty())
3567 os += labelString() + ' ';
3568 if (!layout().isTocCaption())
3570 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3573 char_type const c = d->text_[i];
3576 else if (c == META_INSET)
3577 getInset(i)->forOutliner(os, tmplen, false);
3580 Text::shortenForOutliner(os, maxlen);
3584 void Paragraph::setInsetOwner(Inset const * inset)
3586 d->inset_owner_ = inset;
3590 int Paragraph::id() const
3596 void Paragraph::setId(int id)
3602 Layout const & Paragraph::layout() const
3608 void Paragraph::setLayout(Layout const & layout)
3610 d->layout_ = &layout;
3614 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3616 setLayout(tc.defaultLayout());
3620 void Paragraph::setPlainLayout(DocumentClass const & tc)
3622 setLayout(tc.plainLayout());
3626 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3628 if (usePlainLayout())
3629 setPlainLayout(tclass);
3631 setDefaultLayout(tclass);
3635 Inset const & Paragraph::inInset() const
3637 LBUFERR(d->inset_owner_);
3638 return *d->inset_owner_;
3642 ParagraphParameters & Paragraph::params()
3648 ParagraphParameters const & Paragraph::params() const
3654 bool Paragraph::isFreeSpacing() const
3656 if (d->layout_->free_spacing)
3658 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3662 bool Paragraph::allowEmpty() const
3664 if (d->layout_->keepempty)
3666 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3670 bool Paragraph::brokenBiblio() const
3672 // There is a problem if there is no bibitem at position 0 in
3673 // paragraphs that need one, if there is another bibitem in the
3674 // paragraph or if this paragraph is not supposed to have
3675 // a bibitem inset at all.
3676 return ((d->layout_->labeltype == LABEL_BIBLIO
3677 && (d->insetlist_.find(BIBITEM_CODE) != 0
3678 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
3679 || (d->layout_->labeltype != LABEL_BIBLIO
3680 && d->insetlist_.find(BIBITEM_CODE) != -1));
3684 int Paragraph::fixBiblio(Buffer const & buffer)
3686 // FIXME: when there was already an inset at 0, the return value is 1,
3687 // which does not tell whether another inset has been remove; the
3688 // cursor cannot be correctly updated.
3690 bool const track_changes = buffer.params().track_changes;
3691 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3693 // The case where paragraph is not BIBLIO
3694 if (d->layout_->labeltype != LABEL_BIBLIO) {
3695 if (bibitem_pos == -1)
3696 // No InsetBibitem => OK
3698 // There is an InsetBibitem: remove it!
3699 d->insetlist_.release(bibitem_pos);
3700 eraseChar(bibitem_pos, track_changes);
3701 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
3704 bool const hasbibitem0 = bibitem_pos == 0;
3706 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3707 // There was an InsetBibitem at pos 0,
3708 // and no other one => OK
3709 if (bibitem_pos == -1)
3711 // there is a bibitem at the 0 position, but since
3712 // there is a second one, we copy the second on the
3713 // first. We're assuming there are at most two of
3714 // these, which there should be.
3715 // FIXME: why does it make sense to do that rather
3716 // than keep the first? (JMarc)
3717 Inset * inset = releaseInset(bibitem_pos);
3718 d->insetlist_.begin()->inset = inset;
3719 return -bibitem_pos;
3722 // We need to create an inset at the beginning
3723 Inset * inset = nullptr;
3724 if (bibitem_pos > 0) {
3725 // there was one somewhere in the paragraph, let's move it
3726 inset = d->insetlist_.release(bibitem_pos);
3727 eraseChar(bibitem_pos, track_changes);
3730 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3731 InsetCommandParams(BIBITEM_CODE));
3733 Font font(inherit_font, buffer.params().language);
3734 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3735 : Change::UNCHANGED));
3737 // This is needed to get the counters right
3738 buffer.updateBuffer();
3743 void Paragraph::checkAuthors(AuthorList const & authorList)
3745 d->changes_.checkAuthors(authorList);
3749 bool Paragraph::isChanged(pos_type pos) const
3751 return lookupChange(pos).changed();
3755 bool Paragraph::isInserted(pos_type pos) const
3757 return lookupChange(pos).inserted();
3761 bool Paragraph::isDeleted(pos_type pos) const
3763 return lookupChange(pos).deleted();
3767 InsetList const & Paragraph::insetList() const
3769 return d->insetlist_;
3773 void Paragraph::setInsetBuffers(Buffer & b)
3775 d->insetlist_.setBuffer(b);
3779 void Paragraph::resetBuffer()
3781 d->insetlist_.resetBuffer();
3785 Inset * Paragraph::releaseInset(pos_type pos)
3787 Inset * inset = d->insetlist_.release(pos);
3788 /// does not honour change tracking!
3789 eraseChar(pos, false);
3794 Inset * Paragraph::getInset(pos_type pos)
3796 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3797 ? d->insetlist_.get(pos) : 0;
3801 Inset const * Paragraph::getInset(pos_type pos) const
3803 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3804 ? d->insetlist_.get(pos) : 0;
3808 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3809 pos_type & right, TextCase action)
3811 // process sequences of modified characters; in change
3812 // tracking mode, this approach results in much better
3813 // usability than changing case on a char-by-char basis
3814 // We also need to track the current font, since font
3815 // changes within sequences can occur.
3816 vector<pair<char_type, Font> > changes;
3818 bool const trackChanges = bparams.track_changes;
3820 bool capitalize = true;
3822 for (; pos < right; ++pos) {
3823 char_type oldChar = d->text_[pos];
3824 char_type newChar = oldChar;
3826 // ignore insets and don't play with deleted text!
3827 if (oldChar != META_INSET && !isDeleted(pos)) {
3829 case text_lowercase:
3830 newChar = lowercase(oldChar);
3832 case text_capitalization:
3834 newChar = uppercase(oldChar);
3838 case text_uppercase:
3839 newChar = uppercase(oldChar);
3844 if (isWordSeparator(pos) || isDeleted(pos)) {
3845 // permit capitalization again
3849 if (oldChar != newChar) {
3850 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3851 if (pos != right - 1)
3853 // step behind the changing area
3857 int erasePos = pos - changes.size();
3858 for (size_t i = 0; i < changes.size(); i++) {
3859 insertChar(pos, changes[i].first,
3862 if (!eraseChar(erasePos, trackChanges)) {
3865 ++right; // expand selection
3873 int Paragraph::find(docstring const & str, bool cs, bool mw,
3874 pos_type start_pos, bool del) const
3876 pos_type pos = start_pos;
3877 int const strsize = str.length();
3879 pos_type const parsize = d->text_.size();
3880 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3881 // Ignore "invisible" letters such as ligature breaks
3882 // and hyphenation chars while searching
3883 while (pos < parsize - 1 && isInset(pos)) {
3884 odocstringstream os;
3885 getInset(pos)->toString(os);
3886 if (!getInset(pos)->isLetter() || !os.str().empty())
3890 if (cs && str[i] != d->text_[pos])
3892 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3894 if (!del && isDeleted(pos))
3901 // if necessary, check whether string matches word
3903 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3906 && !isWordSeparator(pos))
3910 return pos - start_pos;
3914 char_type Paragraph::getChar(pos_type pos) const
3916 return d->text_[pos];
3920 pos_type Paragraph::size() const
3922 return d->text_.size();
3926 bool Paragraph::empty() const
3928 return d->text_.empty();
3932 bool Paragraph::isInset(pos_type pos) const
3934 return d->text_[pos] == META_INSET;
3938 bool Paragraph::isSeparator(pos_type pos) const
3940 //FIXME: Are we sure this can be the only separator?
3941 return d->text_[pos] == ' ';
3945 void Paragraph::deregisterWords()
3947 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3948 Private::LangWordsMap::const_iterator ite = d->words_.end();
3949 for (; itl != ite; ++itl) {
3950 WordList & wl = theWordList(itl->first);
3951 Private::Words::const_iterator it = (itl->second).begin();
3952 Private::Words::const_iterator et = (itl->second).end();
3953 for (; it != et; ++it)
3960 void Paragraph::locateWord(pos_type & from, pos_type & to,
3961 word_location const loc) const
3964 case WHOLE_WORD_STRICT:
3965 if (from == 0 || from == size()
3966 || isWordSeparator(from)
3967 || isWordSeparator(from - 1)) {
3974 // If we are already at the beginning of a word, do nothing
3975 if (!from || isWordSeparator(from - 1))
3980 // always move the cursor to the beginning of previous word
3981 while (from && !isWordSeparator(from - 1))
3985 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3988 // no need to move the 'from' cursor
3992 while (to < size() && !isWordSeparator(to))
3997 void Paragraph::collectWords()
3999 for (pos_type pos = 0; pos < size(); ++pos) {
4000 if (isWordSeparator(pos))
4002 pos_type from = pos;
4003 locateWord(from, pos, WHOLE_WORD);
4004 // Work around MSVC warning: The statement
4005 // if (pos < from + lyxrc.completion_minlength)
4006 // triggers a signed vs. unsigned warning.
4007 // I don't know why this happens, it could be a MSVC bug, or
4008 // related to LLP64 (windows) vs. LP64 (unix) programming
4009 // model, or the C++ standard might be ambigous in the section
4010 // defining the "usual arithmetic conversions". However, using
4011 // a temporary variable is safe and works on all compilers.
4012 pos_type const endpos = from + lyxrc.completion_minlength;
4015 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4016 if (cit == d->fontlist_.end())
4018 Language const * lang = cit->font().language();
4019 docstring const word = asString(from, pos, AS_STR_NONE);
4020 d->words_[lang->lang()].insert(word);
4025 void Paragraph::registerWords()
4027 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4028 Private::LangWordsMap::const_iterator ite = d->words_.end();
4029 for (; itl != ite; ++itl) {
4030 WordList & wl = theWordList(itl->first);
4031 Private::Words::const_iterator it = (itl->second).begin();
4032 Private::Words::const_iterator et = (itl->second).end();
4033 for (; it != et; ++it)
4039 void Paragraph::updateWords()
4047 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4049 SkipPositionsIterator begin = skips.begin();
4050 SkipPositions::iterator end = skips.end();
4051 if (pos > 0 && begin < end) {
4053 if (end->last == pos - 1) {
4058 skips.insert(end, FontSpan(pos, pos));
4062 Language * Paragraph::Private::locateSpellRange(
4063 pos_type & from, pos_type & to,
4064 SkipPositions & skips) const
4066 // skip leading white space
4067 while (from < to && owner_->isWordSeparator(from))
4069 // don't check empty range
4072 // get current language
4073 Language * lang = getSpellLanguage(from);
4074 pos_type last = from;
4075 bool samelang = true;
4076 bool sameinset = true;
4077 while (last < to && samelang && sameinset) {
4078 // hop to end of word
4079 while (last < to && !owner_->isWordSeparator(last)) {
4080 if (owner_->getInset(last)) {
4081 appendSkipPosition(skips, last);
4082 } else if (owner_->isDeleted(last)) {
4083 appendSkipPosition(skips, last);
4087 // hop to next word while checking for insets
4088 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4089 if (Inset const * inset = owner_->getInset(last))
4090 sameinset = inset->isChar() && inset->isLetter();
4091 if (sameinset && owner_->isDeleted(last)) {
4092 appendSkipPosition(skips, last);
4097 if (sameinset && last < to) {
4098 // now check for language change
4099 samelang = lang == getSpellLanguage(last);
4102 // if language change detected backstep is needed
4110 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4113 const_cast<Language *>(owner_->getFontSettings(
4114 inset_owner_->buffer().params(), from).language());
4115 if (lang == inset_owner_->buffer().params().language
4116 && !lyxrc.spellchecker_alt_lang.empty()) {
4118 string const lang_variety =
4119 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4120 lang->setCode(lang_code);
4121 lang->setVariety(lang_variety);
4127 void Paragraph::requestSpellCheck(pos_type pos)
4129 d->requestSpellCheck(pos);
4133 bool Paragraph::needsSpellCheck() const
4135 SpellChecker::ChangeNumber speller_change_number = 0;
4136 if (theSpellChecker())
4137 speller_change_number = theSpellChecker()->changeNumber();
4138 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4139 d->speller_state_.needsCompleteRefresh(speller_change_number);
4141 return d->needsSpellCheck();
4145 bool Paragraph::Private::ignoreWord(docstring const & word) const
4147 // Ignore words with digits
4148 // FIXME: make this customizable
4149 // (note that some checkers ignore words with digits by default)
4150 docstring::const_iterator cit = word.begin();
4151 docstring::const_iterator const end = word.end();
4152 for (; cit != end; ++cit) {
4153 if (isNumber((*cit)))
4160 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4161 WordLangTuple & wl, docstring_list & suggestions,
4162 bool do_suggestion, bool check_learned) const
4164 SpellChecker::Result result = SpellChecker::WORD_OK;
4165 SpellChecker * speller = theSpellChecker();
4169 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4172 locateWord(from, to, WHOLE_WORD);
4173 if (from == to || from >= size())
4176 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4177 Language * lang = d->getSpellLanguage(from);
4179 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4182 wl = WordLangTuple(word, lang);
4187 if (needsSpellCheck() || check_learned) {
4189 if (!d->ignoreWord(word)) {
4190 bool const trailing_dot = to < size() && d->text_[to] == '.';
4191 result = speller->check(wl);
4192 if (SpellChecker::misspelled(result) && trailing_dot) {
4193 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4194 result = speller->check(wl);
4195 if (!SpellChecker::misspelled(result)) {
4196 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4198 from << ".." << to << "]");
4200 // spell check with dot appended failed too
4201 // restore original word/lang value
4202 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4203 wl = WordLangTuple(word, lang);
4207 if (!SpellChecker::misspelled(result)) {
4208 // area up to the begin of the next word is not misspelled
4209 while (end < size() && isWordSeparator(end))
4212 d->setMisspelled(from, end, result);
4214 result = d->speller_state_.getState(from);
4218 suggestions.clear();
4220 if (SpellChecker::misspelled(result)) {
4221 LYXERR(Debug::GUI, "misspelled word: \"" <<
4223 from << ".." << to << "]");
4225 speller->suggest(wl, suggestions);
4231 void Paragraph::anonymize()
4233 // This is a very crude anonymization for now
4234 for (char_type & c : d->text_)
4235 if (isLetterChar(c) || isNumber(c))
4240 void Paragraph::Private::markMisspelledWords(
4241 pos_type const & first, pos_type const & last,
4242 SpellChecker::Result result,
4243 docstring const & word,
4244 SkipPositions const & skips)
4246 if (!SpellChecker::misspelled(result)) {
4247 setMisspelled(first, last, SpellChecker::WORD_OK);
4251 SpellChecker * speller = theSpellChecker();
4252 // locate and enumerate the error positions
4253 int nerrors = speller->numMisspelledWords();
4255 SkipPositionsIterator it = skips.begin();
4256 SkipPositionsIterator et = skips.end();
4257 for (int index = 0; index < nerrors; ++index) {
4260 speller->misspelledWord(index, wstart, wlen);
4261 /// should not happen if speller supports range checks
4262 if (!wlen) continue;
4263 docstring const misspelled = word.substr(wstart, wlen);
4264 wstart += first + numskipped;
4265 if (snext < wstart) {
4266 /// mark the range of correct spelling
4267 numskipped += countSkips(it, et, wstart);
4268 setMisspelled(snext,
4269 wstart - 1, SpellChecker::WORD_OK);
4271 snext = wstart + wlen;
4272 numskipped += countSkips(it, et, snext);
4273 /// mark the range of misspelling
4274 setMisspelled(wstart, snext, result);
4275 LYXERR(Debug::GUI, "misspelled word: \"" <<
4276 misspelled << "\" [" <<
4277 wstart << ".." << (snext-1) << "]");
4280 if (snext <= last) {
4281 /// mark the range of correct spelling at end
4282 setMisspelled(snext, last, SpellChecker::WORD_OK);
4287 void Paragraph::spellCheck() const
4289 SpellChecker * speller = theSpellChecker();
4290 if (!speller || empty() ||!needsSpellCheck())
4294 d->rangeOfSpellCheck(start, endpos);
4295 if (speller->canCheckParagraph()) {
4296 // loop until we leave the range
4297 for (pos_type first = start; first < endpos; ) {
4298 pos_type last = endpos;
4299 Private::SkipPositions skips;
4300 Language * lang = d->locateSpellRange(first, last, skips);
4301 if (first >= endpos)
4303 // start the spell checker on the unit of meaning
4304 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4305 WordLangTuple wl = WordLangTuple(word, lang);
4306 SpellChecker::Result result = word.size() ?
4307 speller->check(wl) : SpellChecker::WORD_OK;
4308 d->markMisspelledWords(first, last, result, word, skips);
4312 static docstring_list suggestions;
4313 pos_type to = endpos;
4314 while (start < endpos) {
4316 spellCheck(start, to, wl, suggestions, false);
4320 d->readySpellCheck();
4324 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4326 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4327 if (result || pos <= 0 || pos > size())
4329 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4330 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4335 string Paragraph::magicLabel() const
4338 ss << "magicparlabel-" << id();