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();
624 Change Paragraph::parEndChange() const
626 return d->changes_.lookup(size());
630 void Paragraph::setChange(Change const & change)
632 // beware of the imaginary end-of-par character!
633 d->changes_.set(change, 0, size() + 1);
636 * Propagate the change recursively - but not in case of DELETED!
638 * Imagine that your co-author makes changes in an existing inset. He
639 * sends your document to you and you come to the conclusion that the
640 * inset should go completely. If you erase it, LyX must not delete all
641 * text within the inset. Otherwise, the change tracked insertions of
642 * your co-author get lost and there is no way to restore them later.
644 * Conclusion: An inset's content should remain untouched if you delete it
647 if (!change.deleted()) {
648 for (pos_type pos = 0; pos < size(); ++pos) {
649 if (Inset * inset = getInset(pos))
650 inset->setChange(change);
656 void Paragraph::setChange(pos_type pos, Change const & change)
658 LASSERT(pos >= 0 && pos <= size(), return);
659 d->changes_.set(change, pos);
661 // see comment in setChange(Change const &) above
662 if (!change.deleted() && pos < size())
663 if (Inset * inset = getInset(pos))
664 inset->setChange(change);
668 Change const & Paragraph::lookupChange(pos_type pos) const
670 LBUFERR(pos >= 0 && pos <= size());
671 return d->changes_.lookup(pos);
675 void Paragraph::acceptChanges(pos_type start, pos_type end)
677 LASSERT(start >= 0 && start <= size(), return);
678 LASSERT(end > start && end <= size() + 1, return);
680 for (pos_type pos = start; pos < end; ++pos) {
681 switch (lookupChange(pos).type) {
682 case Change::UNCHANGED:
683 // accept changes in nested inset
684 if (Inset * inset = getInset(pos))
685 inset->acceptChanges();
688 case Change::INSERTED:
689 d->changes_.set(Change(Change::UNCHANGED), pos);
690 // also accept changes in nested inset
691 if (Inset * inset = getInset(pos))
692 inset->acceptChanges();
695 case Change::DELETED:
696 // Suppress access to non-existent
697 // "end-of-paragraph char"
699 eraseChar(pos, false);
710 void Paragraph::rejectChanges(pos_type start, pos_type end)
712 LASSERT(start >= 0 && start <= size(), return);
713 LASSERT(end > start && end <= size() + 1, return);
715 for (pos_type pos = start; pos < end; ++pos) {
716 switch (lookupChange(pos).type) {
717 case Change::UNCHANGED:
718 // reject changes in nested inset
719 if (Inset * inset = getInset(pos))
720 inset->rejectChanges();
723 case Change::INSERTED:
724 // Suppress access to non-existent
725 // "end-of-paragraph char"
727 eraseChar(pos, false);
733 case Change::DELETED:
734 d->changes_.set(Change(Change::UNCHANGED), pos);
736 // Do NOT reject changes within a deleted inset!
737 // There may be insertions of a co-author inside of it!
745 void Paragraph::Private::insertChar(pos_type pos, char_type c,
746 Change const & change)
748 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
751 changes_.insert(change, pos);
753 // This is actually very common when parsing buffers (and
754 // maybe inserting ascii text)
755 if (pos == pos_type(text_.size())) {
756 // when appending characters, no need to update tables
758 // but we want spell checking
759 requestSpellCheck(pos);
763 text_.insert(text_.begin() + pos, c);
765 // Update the font table.
766 fontlist_.increasePosAfterPos(pos);
769 insetlist_.increasePosAfterPos(pos);
771 // Update list of misspelled positions
772 speller_state_.increasePosAfterPos(pos);
776 bool Paragraph::insertInset(pos_type pos, Inset * inset,
777 Font const & font, Change const & change)
779 LASSERT(inset, return false);
780 LASSERT(pos >= 0 && pos <= size(), return false);
782 // Paragraph::insertInset() can be used in cut/copy/paste operation where
783 // d->inset_owner_ is not set yet.
784 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
787 d->insertChar(pos, META_INSET, change);
788 LASSERT(d->text_[pos] == META_INSET, return false);
790 // Add a new entry in the insetlist_.
791 d->insetlist_.insert(inset, pos);
793 // Some insets require run of spell checker
794 requestSpellCheck(pos);
800 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
802 LASSERT(pos >= 0 && pos <= size(), return false);
804 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
807 Change change = d->changes_.lookup(pos);
809 // set the character to DELETED if
810 // a) it was previously unchanged or
811 // b) it was inserted by a co-author
813 if (!change.changed() ||
814 (change.inserted() && !change.currentAuthor())) {
815 setChange(pos, Change(Change::DELETED));
816 // request run of spell checker
817 requestSpellCheck(pos);
821 if (change.deleted())
825 // Don't physically access the imaginary end-of-paragraph character.
826 // eraseChar() can only mark it as DELETED. A physical deletion of
827 // end-of-par must be handled externally.
833 d->changes_.erase(pos);
835 // if it is an inset, delete the inset entry
836 if (d->text_[pos] == META_INSET)
837 d->insetlist_.erase(pos);
839 d->text_.erase(d->text_.begin() + pos);
841 // Update the fontlist_
842 d->fontlist_.erase(pos);
844 // Update the insetlist_
845 d->insetlist_.decreasePosAfterPos(pos);
847 // Update list of misspelled positions
848 d->speller_state_.decreasePosAfterPos(pos);
849 d->speller_state_.refreshLast(size());
855 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
857 LASSERT(start >= 0 && start <= size(), return 0);
858 LASSERT(end >= start && end <= size() + 1, return 0);
861 for (pos_type count = end - start; count; --count) {
862 if (!eraseChar(i, trackChanges))
868 // Handle combining characters
869 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
870 otexstream & os, char_type c, char_type next,
871 OutputParams const & runparams)
873 // Writing next here may circumvent a possible font change between
874 // c and next. Since next is only output if it forms a surrogate pair
875 // with c we can ignore this:
876 // A font change inside a surrogate pair does not make sense and is
877 // hopefully impossible to input.
878 // FIXME: change tracking
879 // Is this correct WRT change tracking?
880 Encoding const & encoding = *(runparams.encoding);
881 docstring latex1 = encoding.latexChar(next).first;
882 if (runparams.inIPA) {
883 string const tipashortcut = Encodings::TIPAShortcut(next);
884 if (!tipashortcut.empty()) {
885 latex1 = from_ascii(tipashortcut);
888 docstring latex2 = encoding.latexChar(c).first;
890 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
891 // Encoding supports the combination:
892 // output as is (combining char after base char).
893 os << latex2 << latex1;
894 return latex1.length() + latex2.length();
897 os << latex1 << "{" << latex2 << "}";
898 return latex1.length() + latex2.length() + 2;
902 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
903 OutputParams const & runparams,
906 unsigned int & column,
908 Layout const & style)
910 if (style.pass_thru || runparams.pass_thru)
913 if (i + 1 < int(text_.size())) {
914 char_type next = text_[i + 1];
915 if (Encodings::isCombiningChar(next)) {
916 // This space has an accent, so we must always output it.
917 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
922 if (runparams.linelen > 0
923 && column > runparams.linelen
925 && text_[i - 1] != ' '
926 && (i + 1 < int(text_.size()))
927 // same in FreeSpacing mode
928 && !owner_->isFreeSpacing()
929 // In typewriter mode, we want to avoid
930 // ! . ? : at the end of a line
931 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
932 && (text_[i - 1] == '.'
933 || text_[i - 1] == '?'
934 || text_[i - 1] == ':'
935 || text_[i - 1] == '!'))) {
937 os.texrow().start(owner_->id(), i + 1);
939 } else if (style.free_spacing) {
948 void Paragraph::Private::latexInset(BufferParams const & bparams,
950 OutputParams & runparams,
953 Font const & outerfont,
955 Change & running_change,
956 Layout const & style,
958 unsigned int & column)
960 Inset * inset = owner_->getInset(i);
963 if (style.pass_thru) {
964 odocstringstream ods;
965 inset->plaintext(ods, runparams);
970 // FIXME: move this to InsetNewline::latex
971 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
972 // newlines are handled differently here than
973 // the default in simpleTeXSpecialChars().
974 if (!style.newline_allowed) {
978 bool needPar = false;
979 column += running_font.latexWriteEndChanges(
980 os, bparams, runparams,
981 basefont, basefont, needPar);
985 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
988 basefont = owner_->getLayoutFont(bparams, outerfont);
989 running_font = basefont;
991 if (runparams.moving_arg)
995 os.texrow().start(owner_->id(), i + 1);
999 if (owner_->isDeleted(i)) {
1000 if( ++runparams.inDeletedInset == 1)
1001 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1004 if (inset->canTrackChanges()) {
1005 column += Changes::latexMarkChange(os, bparams, running_change,
1006 Change(Change::UNCHANGED), runparams);
1007 running_change = Change(Change::UNCHANGED);
1011 odocstream::pos_type const len = os.os().tellp();
1013 if (inset->forceLTR(runparams)
1014 && running_font.isRightToLeft()
1015 // ERT is an exception, it should be output with no
1016 // decorations at all
1017 && inset->lyxCode() != ERT_CODE) {
1018 if (runparams.use_polyglossia) {
1020 } else if (running_font.language()->lang() == "farsi"
1021 || running_font.language()->lang() == "arabic_arabi")
1022 os << "\\textLR{" << termcmd;
1028 // FIXME: Bug: we can have an empty font change here!
1029 // if there has just been a font change, we are going to close it
1030 // right now, which means stupid latex code like \textsf{}. AFAIK,
1031 // this does not harm dvi output. A minor bug, thus (JMarc)
1033 // Some insets cannot be inside a font change command.
1034 // However, even such insets *can* be placed in \L or \R
1035 // or their equivalents (for RTL language switches), so we don't
1036 // close the language in those cases.
1037 // ArabTeX, though, cannot handle this special behavior, it seems.
1038 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1039 || running_font.language()->lang() == "arabic_arabtex";
1040 if (open_font && !inset->inheritFont()) {
1041 bool needPar = false;
1042 bool closeLanguage = arabtex
1043 || basefont.isRightToLeft() == running_font.isRightToLeft();
1044 unsigned int count = running_font.latexWriteEndChanges(os,
1045 bparams, runparams, basefont, basefont,
1046 needPar, closeLanguage);
1048 // if any font properties were closed, update the running_font,
1049 // making sure, however, to leave the language as it was
1051 // FIXME: probably a better way to keep track of the old
1052 // language, than copying the entire font?
1053 Font const copy_font(running_font);
1054 basefont = owner_->getLayoutFont(bparams, outerfont);
1055 running_font = basefont;
1057 running_font.setLanguage(copy_font.language());
1058 // leave font open if language is still open
1059 open_font = (running_font.language() == basefont.language());
1061 runparams.local_font = &basefont;
1065 size_t const previous_row_count = os.texrow().rows();
1068 runparams.lastid = id_;
1069 runparams.lastpos = i;
1070 inset->latex(os, runparams);
1071 } catch (EncodingException & e) {
1072 // add location information and throw again.
1081 if (os.texrow().rows() > previous_row_count) {
1082 os.texrow().start(owner_->id(), i + 1);
1085 column += (unsigned int)(os.os().tellp() - len);
1088 if (owner_->isDeleted(i))
1089 --runparams.inDeletedInset;
1093 void Paragraph::Private::latexSpecialChar(otexstream & os,
1094 BufferParams const & bparams,
1095 OutputParams const & runparams,
1096 Font const & running_font,
1097 string & alien_script,
1098 Layout const & style,
1101 unsigned int & column)
1103 char_type const c = owner_->getUChar(bparams, runparams, i);
1105 if (style.pass_thru || runparams.pass_thru
1106 || contains(style.pass_thru_chars, c)
1107 || contains(runparams.pass_thru_chars, c)) {
1109 Encoding const * const enc = runparams.encoding;
1110 if (enc && !enc->encodable(c))
1111 throw EncodingException(c);
1117 // TIPA uses its own T3 encoding
1118 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1120 // If T1 font encoding is used, use the special
1121 // characters it provides.
1122 // NOTE: Some languages reset the font encoding internally to a
1123 // non-standard font encoding. If we are using such a language,
1124 // we do not output special T1 chars.
1125 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1126 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1127 && latexSpecialT1(c, os, i, column))
1129 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1130 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1131 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1134 // Otherwise, we use what LaTeX provides us.
1137 os << "\\textbackslash" << termcmd;
1141 os << "\\textless" << termcmd;
1145 os << "\\textgreater" << termcmd;
1149 os << "\\textbar" << termcmd;
1154 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1155 (end_pos == -1 || i + 1 < end_pos) &&
1156 text_[i+1] == '-') {
1157 // Prevent "--" becoming an en dash and "---" an em dash.
1158 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1164 os << "\\textquotedbl" << termcmd;
1169 case '%': case '#': case '{':
1177 os << "\\textasciitilde" << termcmd;
1182 os << "\\textasciicircum" << termcmd;
1189 // avoid being mistaken for optional arguments
1197 // Blanks are printed before font switching.
1198 // Sure? I am not! (try nice-latex)
1199 // I am sure it's correct. LyX might be smarter
1200 // in the future, but for now, nothing wrong is
1206 // XeTeX's dash behaviour is determined via a global setting
1207 if (bparams.use_dash_ligatures
1208 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1210 // TODO #10961: && not in inset Flex Code
1211 // TODO #10961: && not in layout LyXCode
1212 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1229 Encoding const & encoding = *(runparams.encoding);
1230 char_type next = '\0';
1231 if (i + 1 < int(text_.size())) {
1232 next = text_[i + 1];
1233 if (Encodings::isCombiningChar(next)) {
1234 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1239 pair<docstring, bool> latex = encoding.latexChar(c);
1240 docstring nextlatex;
1241 bool nexttipas = false;
1242 string nexttipashortcut;
1243 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1244 nextlatex = encoding.latexChar(next).first;
1245 if (runparams.inIPA) {
1246 nexttipashortcut = Encodings::TIPAShortcut(next);
1247 nexttipas = !nexttipashortcut.empty();
1251 if (runparams.inIPA) {
1252 string const tipashortcut = Encodings::TIPAShortcut(c);
1253 if (!tipashortcut.empty()) {
1254 latex.first = from_ascii(tipashortcut);
1255 latex.second = false;
1259 // eventually close "script wrapper" command (see `Paragraph::latex`)
1260 if (!alien_script.empty()
1261 && alien_script != Encodings::isKnownScriptChar(next)) {
1262 column += latex.first.length();
1263 alien_script.clear();
1264 os << latex.first << "}";
1268 && ((!prefixIs(nextlatex, '\\')
1269 && !prefixIs(nextlatex, '{')
1270 && !prefixIs(nextlatex, '}'))
1272 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1274 // Prevent eating of a following space or command corruption by
1275 // following characters
1276 if (next == ' ' || next == '\0') {
1277 column += latex.first.length() + 1;
1278 os << latex.first << "{}";
1280 column += latex.first.length();
1281 os << latex.first << " ";
1284 column += latex.first.length() - 1;
1292 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1293 pos_type i, unsigned int & column)
1299 // In T1 encoding, these characters exist
1300 // but we should avoid ligatures
1301 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1303 os << "\\textcompwordmark" << termcmd;
1310 // soul.sty breaks with \char`\"
1311 os << "\\textquotedbl" << termcmd;
1320 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1321 pos_type i, unsigned int & column)
1323 // TU encoding is currently on par with T1.
1324 return latexSpecialT1(c, os, i, column);
1328 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1329 pos_type /*i*/, unsigned int & column)
1339 os << "\\textvertline" << termcmd;
1348 void Paragraph::Private::validate(LaTeXFeatures & features) const
1350 if (layout_->inpreamble && inset_owner_) {
1351 // FIXME: Using a string stream here circumvents the encoding
1352 // switching machinery of odocstream. Therefore the
1353 // output is wrong if this paragraph contains content
1354 // that needs to switch encoding.
1355 Buffer const & buf = inset_owner_->buffer();
1356 otexstringstream os;
1357 os << layout_->preamble();
1358 size_t const length = os.length();
1359 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1360 buf.getParFromID(owner_->id()).pit(), os,
1361 features.runparams(), string(), 0, -1, true);
1362 if (os.length() > length)
1363 features.addPreambleSnippet(os.release(), true);
1366 if (features.runparams().flavor == OutputParams::HTML
1367 && layout_->htmltitle()) {
1368 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1371 // check the params.
1372 if (!params_.spacing().isDefault())
1373 features.require("setspace");
1376 features.useLayout(layout_->name());
1379 fontlist_.validate(features);
1381 // then the indentation
1382 if (!params_.leftIndent().zero())
1383 features.require("ParagraphLeftIndent");
1386 InsetList::const_iterator icit = insetlist_.begin();
1387 InsetList::const_iterator iend = insetlist_.end();
1388 for (; icit != iend; ++icit) {
1390 features.inDeletedInset(owner_->isDeleted(icit->pos));
1391 if (icit->inset->lyxCode() == FOOT_CODE) {
1392 // FIXME: an item inset would make things much easier.
1393 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1394 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1395 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1396 && (icit->pos < begin_of_body_
1397 || (icit->pos == begin_of_body_
1398 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1399 features.saveNoteEnv("description");
1401 icit->inset->validate(features);
1402 features.inDeletedInset(false);
1403 if (layout_->needprotect &&
1404 icit->inset->lyxCode() == FOOT_CODE)
1405 features.require("NeedLyXFootnoteCode");
1409 // then the contents
1410 BufferParams const bp = features.runparams().is_child
1411 ? features.buffer().masterParams() : features.buffer().params();
1412 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1413 char_type c = text_[i];
1414 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1416 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1417 features.require("textquotedblp");
1418 else if (bp.main_font_encoding() != "T1"
1419 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1420 features.require("textquotedbl");
1421 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1422 // features that depend on the font or input encoding
1423 string feats = ci.textpreamble();
1424 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1425 if (fontenc.empty())
1426 fontenc = features.runparams().main_fontenc;
1427 while (!feats.empty()) {
1429 feats = split(feats, feat, ',');
1430 if (contains(feat, "!=")) {
1431 // a feature that is required except for the spcified
1432 // font or input encodings
1434 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1435 // multiple encodings are separated by semicolon
1436 vector<string> context = getVectorFromString(contexts, ";");
1437 // require feature if the context matches neither current font
1438 // nor input encoding
1439 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1440 && std::find(context.begin(), context.end(),
1441 features.runparams().encoding->name()) == context.end())
1442 features.require(realfeature);
1443 } else if (contains(feat, '=')) {
1444 // a feature that is required only for the spcified
1445 // font or input encodings
1447 string const contexts = split(feat, realfeature, '=');
1448 // multiple encodings are separated by semicolon
1449 vector<string> context = getVectorFromString(contexts, ";");
1450 // require feature if the context matches either current font
1451 // or input encoding
1452 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1453 || std::find(context.begin(), context.end(),
1454 features.runparams().encoding->name()) != context.end())
1455 features.require(realfeature);
1458 } else if (!bp.use_dash_ligatures
1459 && (c == 0x2013 || c == 0x2014)
1460 && bp.useNonTeXFonts
1461 && features.runparams().flavor == OutputParams::XETEX)
1462 // XeTeX's dash behaviour is determined via a global setting
1463 features.require("xetexdashbreakstate");
1464 BufferEncodings::validate(c, features);
1468 /////////////////////////////////////////////////////////////////////
1472 /////////////////////////////////////////////////////////////////////
1475 Layout const emptyParagraphLayout;
1478 Paragraph::Paragraph()
1479 : d(new Paragraph::Private(this, emptyParagraphLayout))
1486 Paragraph::Paragraph(Paragraph const & par)
1487 : itemdepth(par.itemdepth),
1488 d(new Paragraph::Private(*par.d, this))
1494 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1495 : itemdepth(par.itemdepth),
1496 d(new Paragraph::Private(*par.d, this, beg, end))
1502 Paragraph & Paragraph::operator=(Paragraph const & par)
1504 // needed as we will destroy the private part before copying it
1506 itemdepth = par.itemdepth;
1510 d = new Private(*par.d, this);
1517 Paragraph::~Paragraph()
1526 // this shall be called just before every "os << ..." action.
1527 void flushString(ostream & os, docstring & s)
1536 void Paragraph::write(ostream & os, BufferParams const & bparams,
1537 depth_type & dth) const
1539 // The beginning or end of a deeper (i.e. nested) area?
1540 if (dth != d->params_.depth()) {
1541 if (d->params_.depth() > dth) {
1542 while (d->params_.depth() > dth) {
1543 os << "\n\\begin_deeper";
1547 while (d->params_.depth() < dth) {
1548 os << "\n\\end_deeper";
1554 // First write the layout
1555 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1557 d->params_.write(os);
1559 Font font1(inherit_font, bparams.language);
1561 Change running_change = Change(Change::UNCHANGED);
1563 // this string is used as a buffer to avoid repetitive calls
1564 // to to_utf8(), which turn out to be expensive (JMarc)
1565 docstring write_buffer;
1568 for (pos_type i = 0; i <= size(); ++i) {
1570 Change const & change = lookupChange(i);
1571 if (change != running_change)
1572 flushString(os, write_buffer);
1573 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1574 running_change = change;
1579 // Write font changes
1580 Font font2 = getFontSettings(bparams, i);
1581 if (font2 != font1) {
1582 flushString(os, write_buffer);
1583 font2.lyxWriteChanges(font1, os);
1588 char_type const c = d->text_[i];
1591 if (Inset const * inset = getInset(i)) {
1592 flushString(os, write_buffer);
1593 if (inset->directWrite()) {
1594 // international char, let it write
1595 // code directly so it's shorter in
1601 os << "\\begin_inset ";
1603 os << "\n\\end_inset\n\n";
1606 // FIXME This can be removed again once the mystery
1607 // crash has been resolved.
1612 flushString(os, write_buffer);
1613 os << "\n\\backslash\n";
1617 flushString(os, write_buffer);
1618 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1625 if ((column > 70 && c == ' ')
1627 flushString(os, write_buffer);
1631 // this check is to amend a bug. LyX sometimes
1632 // inserts '\0' this could cause problems.
1634 write_buffer.push_back(c);
1636 LYXERR0("NUL char in structure.");
1642 flushString(os, write_buffer);
1643 os << "\n\\end_layout\n";
1644 // FIXME This can be removed again once the mystery
1645 // crash has been resolved.
1650 void Paragraph::validate(LaTeXFeatures & features) const
1652 d->validate(features);
1653 bool fragile = features.runparams().moving_arg;
1654 fragile |= layout().needprotect;
1655 if (inInset().getLayout().isNeedProtect())
1657 if (needsCProtection(fragile))
1658 features.require("cprotect");
1662 void Paragraph::insert(pos_type start, docstring const & str,
1663 Font const & font, Change const & change)
1665 for (size_t i = 0, n = str.size(); i != n ; ++i)
1666 insertChar(start + i, str[i], font, change);
1670 void Paragraph::appendChar(char_type c, Font const & font,
1671 Change const & change)
1674 d->changes_.insert(change, d->text_.size());
1675 // when appending characters, no need to update tables
1676 d->text_.push_back(c);
1677 setFont(d->text_.size() - 1, font);
1678 d->requestSpellCheck(d->text_.size() - 1);
1682 void Paragraph::appendString(docstring const & s, Font const & font,
1683 Change const & change)
1685 pos_type end = s.size();
1686 size_t oldsize = d->text_.size();
1687 size_t newsize = oldsize + end;
1688 size_t capacity = d->text_.capacity();
1689 if (newsize >= capacity)
1690 d->text_.reserve(max(capacity + 100, newsize));
1692 // when appending characters, no need to update tables
1695 // FIXME: Optimize this!
1696 for (size_t i = oldsize; i != newsize; ++i) {
1698 d->changes_.insert(change, i);
1699 d->requestSpellCheck(i);
1701 d->fontlist_.set(oldsize, font);
1702 d->fontlist_.set(newsize - 1, font);
1706 void Paragraph::insertChar(pos_type pos, char_type c,
1709 d->insertChar(pos, c, Change(trackChanges ?
1710 Change::INSERTED : Change::UNCHANGED));
1714 void Paragraph::insertChar(pos_type pos, char_type c,
1715 Font const & font, bool trackChanges)
1717 d->insertChar(pos, c, Change(trackChanges ?
1718 Change::INSERTED : Change::UNCHANGED));
1723 void Paragraph::insertChar(pos_type pos, char_type c,
1724 Font const & font, Change const & change)
1726 d->insertChar(pos, c, change);
1731 void Paragraph::resetFonts(Font const & font)
1733 d->fontlist_.clear();
1734 d->fontlist_.set(0, font);
1735 d->fontlist_.set(d->text_.size() - 1, font);
1738 // Gets uninstantiated font setting at position.
1739 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1743 LYXERR0("pos: " << pos << " size: " << size());
1747 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1748 if (cit != d->fontlist_.end())
1751 if (pos == size() && !empty())
1752 return getFontSettings(bparams, pos - 1);
1754 // Optimisation: avoid a full font instantiation if there is no
1755 // language change from previous call.
1756 static Font previous_font;
1757 static Language const * previous_lang = 0;
1758 Language const * lang = getParLanguage(bparams);
1759 if (lang != previous_lang) {
1760 previous_lang = lang;
1761 previous_font = Font(inherit_font, lang);
1763 return previous_font;
1767 FontSpan Paragraph::fontSpan(pos_type pos) const
1769 LBUFERR(pos <= size());
1772 return FontSpan(pos, pos);
1775 FontList::const_iterator cit = d->fontlist_.begin();
1776 FontList::const_iterator end = d->fontlist_.end();
1777 for (; cit != end; ++cit) {
1778 if (cit->pos() >= pos) {
1779 if (pos >= beginOfBody())
1780 return FontSpan(max(start, beginOfBody()),
1783 return FontSpan(start,
1784 min(beginOfBody() - 1,
1787 start = cit->pos() + 1;
1790 // This should not happen, but if so, we take no chances.
1791 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1792 LASSERT(false, return FontSpan(pos, pos));
1796 // Gets uninstantiated font setting at position 0
1797 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1799 if (!empty() && !d->fontlist_.empty())
1800 return d->fontlist_.begin()->font();
1802 // Optimisation: avoid a full font instantiation if there is no
1803 // language change from previous call.
1804 static Font previous_font;
1805 static Language const * previous_lang = 0;
1806 if (bparams.language != previous_lang) {
1807 previous_lang = bparams.language;
1808 previous_font = Font(inherit_font, bparams.language);
1811 return previous_font;
1815 // Gets the fully instantiated font at a given position in a paragraph
1816 // This is basically the same function as Text::GetFont() in text2.cpp.
1817 // The difference is that this one is used for generating the LaTeX file,
1818 // and thus cosmetic "improvements" are disallowed: This has to deliver
1819 // the true picture of the buffer. (Asger)
1820 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1821 Font const & outerfont) const
1825 Font font = getFontSettings(bparams, pos);
1827 pos_type const body_pos = beginOfBody();
1828 FontInfo & fi = font.fontInfo();
1830 fi.realize(d->layout_->labelfont);
1832 fi.realize(d->layout_->font);
1834 fi.realize(outerfont.fontInfo());
1835 fi.realize(bparams.getFont().fontInfo());
1841 Font const Paragraph::getLabelFont
1842 (BufferParams const & bparams, Font const & outerfont) const
1844 FontInfo tmpfont = d->layout_->labelfont;
1845 tmpfont.realize(outerfont.fontInfo());
1846 tmpfont.realize(bparams.getFont().fontInfo());
1847 return Font(tmpfont, getParLanguage(bparams));
1851 Font const Paragraph::getLayoutFont
1852 (BufferParams const & bparams, Font const & outerfont) const
1854 FontInfo tmpfont = d->layout_->font;
1855 tmpfont.realize(outerfont.fontInfo());
1856 tmpfont.realize(bparams.getFont().fontInfo());
1857 return Font(tmpfont, getParLanguage(bparams));
1861 char_type Paragraph::getUChar(BufferParams const & bparams,
1862 OutputParams const & rp,
1865 char_type c = d->text_[pos];
1867 // Return unchanged character in LTR languages
1868 // or if we use poylglossia/bidi (XeTeX).
1869 if (rp.useBidiPackage()
1870 || !getFontSettings(bparams, pos).isRightToLeft())
1873 // Without polyglossia/bidi, we need to account for some special cases.
1874 // FIXME This needs to be audited!
1876 // * The input is as expected for all delimiters
1877 // => checked for Hebrew!
1878 // * The output matches the display in the LyX workarea
1879 // => checked for Hebrew!
1880 // * The special cases below are really necessary
1881 // => checked for Hebrew!
1882 // * In arabic_arabi, brackets are transformed to Arabic
1883 // Ornate Parentheses. Is this is really wanted?
1885 string const & lang = getFontSettings(bparams, pos).language()->lang();
1888 // 1. In the following languages, parentheses need to be reversed.
1889 // Also with polyglodia/luabidi
1890 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1892 // 2. In the following languages, brackets don't need to be reversed.
1893 bool const reversebrackets = lang != "arabic_arabtex"
1894 && lang != "arabic_arabi"
1897 // Now swap delimiters if needed.
1908 if (reversebrackets)
1912 if (reversebrackets)
1933 void Paragraph::setFont(pos_type pos, Font const & font)
1935 LASSERT(pos <= size(), return);
1937 // First, reduce font against layout/label font
1938 // Update: The setCharFont() routine in text2.cpp already
1939 // reduces font, so we don't need to do that here. (Asger)
1941 d->fontlist_.set(pos, font);
1945 void Paragraph::makeSameLayout(Paragraph const & par)
1947 d->layout_ = par.d->layout_;
1948 d->params_ = par.d->params_;
1952 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1954 if (isFreeSpacing())
1960 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1961 if (eraseChar(pos, trackChanges))
1967 return count > 0 || pos > 0;
1971 bool Paragraph::hasSameLayout(Paragraph const & par) const
1973 return par.d->layout_ == d->layout_
1974 && d->params_.sameLayout(par.d->params_);
1978 depth_type Paragraph::getDepth() const
1980 return d->params_.depth();
1984 depth_type Paragraph::getMaxDepthAfter() const
1986 if (d->layout_->isEnvironment())
1987 return d->params_.depth() + 1;
1989 return d->params_.depth();
1993 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
1995 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1996 return getDefaultAlign(bparams);
1998 return d->params_.align();
2002 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
2004 LyXAlignment res = layout().align;
2005 if (isRTL(bparams)) {
2007 if (res == LYX_ALIGN_LEFT)
2008 res = LYX_ALIGN_RIGHT;
2009 else if (res == LYX_ALIGN_RIGHT)
2010 res = LYX_ALIGN_LEFT;
2016 docstring const & Paragraph::labelString() const
2018 return d->params_.labelString();
2022 // the next two functions are for the manual labels
2023 docstring const Paragraph::getLabelWidthString() const
2025 if (d->layout_->margintype == MARGIN_MANUAL
2026 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2027 return d->params_.labelWidthString();
2029 return _("Senseless with this layout!");
2033 void Paragraph::setLabelWidthString(docstring const & s)
2035 d->params_.labelWidthString(s);
2039 docstring Paragraph::expandLabel(Layout const & layout,
2040 BufferParams const & bparams) const
2042 return expandParagraphLabel(layout, bparams, true);
2046 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2047 BufferParams const & bparams) const
2049 return expandParagraphLabel(layout, bparams, false);
2053 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2054 BufferParams const & bparams, bool process_appendix) const
2056 DocumentClass const & tclass = bparams.documentClass();
2057 string const & lang = getParLanguage(bparams)->code();
2058 bool const in_appendix = process_appendix && d->params_.appendix();
2059 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2061 if (fmt.empty() && !layout.counter.empty())
2062 return tclass.counters().theCounter(layout.counter, lang);
2064 // handle 'inherited level parts' in 'fmt',
2065 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2066 size_t const i = fmt.find('@', 0);
2067 if (i != docstring::npos) {
2068 size_t const j = fmt.find('@', i + 1);
2069 if (j != docstring::npos) {
2070 docstring parent(fmt, i + 1, j - i - 1);
2071 docstring label = from_ascii("??");
2072 if (tclass.hasLayout(parent))
2073 label = expandParagraphLabel(tclass[parent], bparams,
2075 fmt = docstring(fmt, 0, i) + label
2076 + docstring(fmt, j + 1, docstring::npos);
2080 return tclass.counters().counterLabel(fmt, lang);
2084 void Paragraph::applyLayout(Layout const & new_layout)
2086 d->layout_ = &new_layout;
2087 LyXAlignment const oldAlign = d->params_.align();
2089 if (!(oldAlign & d->layout_->alignpossible)) {
2090 frontend::Alert::warning(_("Alignment not permitted"),
2091 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2092 d->params_.align(LYX_ALIGN_LAYOUT);
2097 pos_type Paragraph::beginOfBody() const
2099 return d->begin_of_body_;
2103 void Paragraph::setBeginOfBody()
2105 if (d->layout_->labeltype != LABEL_MANUAL) {
2106 d->begin_of_body_ = 0;
2110 // Unroll the first two cycles of the loop
2111 // and remember the previous character to
2112 // remove unnecessary getChar() calls
2114 pos_type end = size();
2115 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2118 char_type previous_char = d->text_[i];
2119 if (!(isNewline(i) || isEnvSeparator(i))) {
2121 while (i < end && previous_char != ' ') {
2122 char_type temp = d->text_[i];
2123 if (isNewline(i) || isEnvSeparator(i))
2126 previous_char = temp;
2132 d->begin_of_body_ = i;
2136 bool Paragraph::allowParagraphCustomization() const
2138 return inInset().allowParagraphCustomization();
2142 bool Paragraph::usePlainLayout() const
2144 return inInset().usePlainLayout();
2148 bool Paragraph::isPassThru() const
2150 return inInset().isPassThru() || d->layout_->pass_thru;
2155 // paragraphs inside floats need different alignment tags to avoid
2158 bool noTrivlistCentering(InsetCode code)
2160 return code == FLOAT_CODE
2161 || code == WRAP_CODE
2162 || code == CELL_CODE;
2166 string correction(string const & orig)
2168 if (orig == "flushleft")
2169 return "raggedright";
2170 if (orig == "flushright")
2171 return "raggedleft";
2172 if (orig == "center")
2178 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2179 InsetCode code, bool const lastpar, int & col)
2181 string macro = suffix + "{";
2182 if (noTrivlistCentering(code)) {
2184 // the last paragraph in non-trivlist-aligned
2185 // context is special (to avoid unwanted whitespace)
2186 if (suffix == "\\begin") {
2187 macro = "\\" + correction(env) + "{}";
2188 os << from_ascii(macro);
2189 col += macro.size();
2194 macro += correction(env);
2198 if (suffix == "\\par\\end") {
2202 os << from_ascii(macro);
2203 col += macro.size();
2204 if (suffix == "\\begin") {
2214 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2215 otexstream & os, OutputParams const & runparams) const
2220 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2221 (layout_->toggle_indent != ITOGGLE_NEVER) :
2222 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2224 if (canindent && params_.noindent() && !layout_->pass_thru) {
2225 os << "\\noindent ";
2229 LyXAlignment const curAlign = params_.align();
2231 if (curAlign == layout_->align)
2235 case LYX_ALIGN_NONE:
2236 case LYX_ALIGN_BLOCK:
2237 case LYX_ALIGN_LAYOUT:
2238 case LYX_ALIGN_SPECIAL:
2239 case LYX_ALIGN_DECIMAL:
2241 case LYX_ALIGN_LEFT:
2242 case LYX_ALIGN_RIGHT:
2243 case LYX_ALIGN_CENTER:
2244 if (runparams.moving_arg) {
2251 string const begin_tag = "\\begin";
2252 InsetCode code = ownerCode();
2253 bool const lastpar = runparams.isLastPar;
2254 // RTL in classic (PDF)LaTeX (without the Bidi package)
2255 // Luabibdi (used by LuaTeX) behaves like classic
2256 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2257 && !runparams.useBidiPackage();
2260 case LYX_ALIGN_NONE:
2261 case LYX_ALIGN_BLOCK:
2262 case LYX_ALIGN_LAYOUT:
2263 case LYX_ALIGN_SPECIAL:
2264 case LYX_ALIGN_DECIMAL:
2266 case LYX_ALIGN_LEFT: {
2268 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2269 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2271 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2273 } case LYX_ALIGN_RIGHT: {
2275 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2276 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2278 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2280 } case LYX_ALIGN_CENTER: {
2281 corrected_env(os, begin_tag, "center", code, lastpar, column);
2290 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2291 otexstream & os, OutputParams const & runparams) const
2293 LyXAlignment const curAlign = params_.align();
2295 if (curAlign == layout_->align)
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:
2306 case LYX_ALIGN_RIGHT:
2307 case LYX_ALIGN_CENTER:
2308 if (runparams.moving_arg)
2313 bool output = false;
2315 string const end_tag = "\\par\\end";
2316 InsetCode code = ownerCode();
2317 bool const lastpar = runparams.isLastPar;
2318 // RTL in classic (PDF)LaTeX (without the Bidi package)
2319 // Luabibdi (used by LuaTeX) behaves like classic
2320 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2321 && !runparams.useBidiPackage();
2324 case LYX_ALIGN_NONE:
2325 case LYX_ALIGN_BLOCK:
2326 case LYX_ALIGN_LAYOUT:
2327 case LYX_ALIGN_SPECIAL:
2328 case LYX_ALIGN_DECIMAL:
2330 case LYX_ALIGN_LEFT: {
2332 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2333 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2335 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2337 } case LYX_ALIGN_RIGHT: {
2339 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2340 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2342 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2344 } case LYX_ALIGN_CENTER: {
2345 corrected_env(os, end_tag, "center", code, lastpar, col);
2350 return output || lastpar;
2354 // This one spits out the text of the paragraph
2355 void Paragraph::latex(BufferParams const & bparams,
2356 Font const & outerfont,
2358 OutputParams const & runparams,
2359 int start_pos, int end_pos, bool force) const
2361 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2363 // FIXME This check should not be needed. Perhaps issue an
2364 // error if it triggers.
2365 Layout const & style = inInset().forcePlainLayout() ?
2366 bparams.documentClass().plainLayout() : *d->layout_;
2368 if (!force && style.inpreamble)
2371 bool const allowcust = allowParagraphCustomization();
2373 // Current base font for all inherited font changes, without any
2374 // change caused by an individual character, except for the language:
2375 // It is set to the language of the first character.
2376 // As long as we are in the label, this font is the base font of the
2377 // label. Before the first body character it is set to the base font
2381 // If there is an open font-encoding changing command (script wrapper),
2382 // alien_script is set to its name
2383 string alien_script;
2386 // Maybe we have to create a optional argument.
2387 pos_type body_pos = beginOfBody();
2388 unsigned int column = 0;
2391 // the optional argument is kept in curly brackets in
2392 // case it contains a ']'
2393 // This is not strictly needed, but if this is changed it
2394 // would be a file format change, and tex2lyx would need
2395 // to be adjusted, since it unconditionally removes the
2396 // braces when it parses \item.
2399 basefont = getLabelFont(bparams, outerfont);
2401 basefont = getLayoutFont(bparams, outerfont);
2404 // Which font is currently active?
2405 Font running_font(basefont);
2406 // Do we have an open font change?
2407 bool open_font = false;
2409 Change runningChange = Change(Change::UNCHANGED);
2411 Encoding const * const prev_encoding = runparams.encoding;
2413 os.texrow().start(id(), 0);
2415 // if the paragraph is empty, the loop will not be entered at all
2417 // For InTitle commands, we have already opened a group
2418 // in output_latex::TeXOnePar.
2419 if (style.isCommand() && !style.intitle) {
2423 if (!style.leftdelim().empty()) {
2424 os << style.leftdelim();
2425 column += style.leftdelim().size();
2428 column += d->startTeXParParams(bparams, os, runparams);
2431 // Whether a \par can be issued for insets typeset inline with text.
2432 // Yes if greater than 0. This has to be static.
2433 THREAD_LOCAL_STATIC int parInline = 0;
2435 for (pos_type i = 0; i < size(); ++i) {
2436 // First char in paragraph or after label?
2437 if (i == body_pos) {
2440 bool needPar = false;
2441 column += running_font.latexWriteEndChanges(
2442 os, bparams, runparams,
2443 basefont, basefont, needPar);
2446 basefont = getLayoutFont(bparams, outerfont);
2447 running_font = basefont;
2449 column += Changes::latexMarkChange(os, bparams,
2450 runningChange, Change(Change::UNCHANGED),
2452 runningChange = Change(Change::UNCHANGED);
2457 // For InTitle commands, we have already opened a group
2458 // in output_latex::TeXOnePar.
2459 if (style.isCommand() && !style.intitle) {
2464 if (!style.leftdelim().empty()) {
2465 os << style.leftdelim();
2466 column += style.leftdelim().size();
2470 column += d->startTeXParParams(bparams, os,
2474 runparams.wasDisplayMath = runparams.inDisplayMath;
2475 runparams.inDisplayMath = false;
2476 bool deleted_display_math = false;
2477 Change const & change = runparams.inDeletedInset
2478 ? runparams.changeOfDeletedInset : lookupChange(i);
2480 // Check whether a display math inset follows
2481 if (d->text_[i] == META_INSET
2482 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2483 InsetMath const * im = getInset(i)->asInsetMath();
2484 if (im && im->asHullInset()
2485 && im->asHullInset()->outerDisplay()) {
2486 runparams.inDisplayMath = true;
2487 // runparams.inDeletedInset will be set by
2488 // latexInset later, but we need this info
2489 // before it is called. On the other hand, we
2490 // cannot set it here because it is a counter.
2491 deleted_display_math = isDeleted(i);
2493 if (bparams.output_changes && deleted_display_math
2494 && runningChange == change
2495 && change.type == Change::DELETED
2496 && !os.afterParbreak()) {
2497 // A display math in the same paragraph follows.
2498 // We have to close and then reopen \lyxdeleted,
2499 // otherwise the math will be shifted up.
2500 OutputParams rp = runparams;
2502 bool needPar = false;
2503 column += running_font.latexWriteEndChanges(
2504 os, bparams, rp, basefont,
2508 basefont = getLayoutFont(bparams, outerfont);
2509 running_font = basefont;
2510 column += Changes::latexMarkChange(os, bparams,
2511 Change(Change::INSERTED), change, rp);
2515 if (bparams.output_changes && runningChange != change) {
2516 if (!alien_script.empty()) {
2519 alien_script.clear();
2522 bool needPar = false;
2523 column += running_font.latexWriteEndChanges(
2524 os, bparams, runparams,
2525 basefont, basefont, needPar);
2528 basefont = getLayoutFont(bparams, outerfont);
2529 running_font = basefont;
2530 column += Changes::latexMarkChange(os, bparams, runningChange,
2532 runningChange = change;
2535 // do not output text which is marked deleted
2536 // if change tracking output is disabled
2537 if (!bparams.output_changes && change.deleted()) {
2543 // Fully instantiated font
2544 Font const current_font = getFont(bparams, i, outerfont);
2546 Font const last_font = running_font;
2548 // Do we need to close the previous font?
2550 (current_font != running_font ||
2551 current_font.language() != running_font.language()))
2553 // ensure there is no open script-wrapper
2554 if (!alien_script.empty()) {
2557 alien_script.clear();
2559 bool needPar = false;
2560 column += running_font.latexWriteEndChanges(
2561 os, bparams, runparams, basefont,
2562 (i == body_pos-1) ? basefont : current_font,
2564 running_font = basefont;
2568 // if necessary, close language environment before opening CJK
2569 string const running_lang = running_font.language()->babel();
2570 string const lang_end_command = lyxrc.language_command_end;
2571 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2572 && !running_lang.empty()
2573 && running_lang == openLanguageName()
2574 && current_font.language()->encoding()->package() == Encoding::CJK) {
2575 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2576 os << from_ascii(end_tag);
2577 column += end_tag.length();
2581 // Switch file encoding if necessary (and allowed)
2582 if (!runparams.pass_thru && !style.pass_thru &&
2583 runparams.encoding->package() != Encoding::none &&
2584 current_font.language()->encoding()->package() != Encoding::none) {
2585 pair<bool, int> const enc_switch =
2586 switchEncoding(os.os(), bparams, runparams,
2587 *(current_font.language()->encoding()));
2588 if (enc_switch.first) {
2589 column += enc_switch.second;
2590 runparams.encoding = current_font.language()->encoding();
2594 char_type const c = d->text_[i];
2596 // A display math inset inside an ulem command will be output
2597 // as a box of width \linewidth, so we have to either disable
2598 // indentation if the inset starts a paragraph, or start a new
2599 // line to accommodate such box. This has to be done before
2600 // writing any font changing commands.
2601 if (runparams.inDisplayMath && !deleted_display_math
2602 && runparams.inulemcmd) {
2603 if (os.afterParbreak())
2609 // Do we need to change font?
2610 if ((current_font != running_font ||
2611 current_font.language() != running_font.language()) &&
2614 odocstringstream ods;
2615 column += current_font.latexWriteStartChanges(ods, bparams,
2616 runparams, basefont,
2618 // Check again for display math in ulem commands as a
2619 // font change may also occur just before a math inset.
2620 if (runparams.inDisplayMath && !deleted_display_math
2621 && runparams.inulemcmd) {
2622 if (os.afterParbreak())
2627 running_font = current_font;
2629 docstring fontchange = ods.str();
2630 // check whether the fontchange ends with a \\textcolor
2631 // modifier and the text starts with a space (bug 4473)
2632 docstring const last_modifier = rsplit(fontchange, '\\');
2633 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2634 os << fontchange << from_ascii("{}");
2635 // check if the fontchange ends with a trailing blank
2636 // (like "\small " (see bug 3382)
2637 else if (suffixIs(fontchange, ' ') && c == ' ')
2638 os << fontchange.substr(0, fontchange.size() - 1)
2639 << from_ascii("{}");
2644 // FIXME: think about end_pos implementation...
2645 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2646 // FIXME: integrate this case in latexSpecialChar
2647 // Do not print the separation of the optional argument
2648 // if style.pass_thru is false. This works because
2649 // latexSpecialChar ignores spaces if
2650 // style.pass_thru is false.
2651 if (i != body_pos - 1) {
2652 if (d->simpleTeXBlanks(bparams, runparams, os,
2653 i, column, current_font, style)) {
2654 // A surrogate pair was output. We
2655 // must not call latexSpecialChar
2656 // in this iteration, since it would output
2657 // the combining character again.
2664 OutputParams rp = runparams;
2665 rp.free_spacing = style.free_spacing;
2666 rp.local_font = ¤t_font;
2667 rp.intitle = style.intitle;
2669 // Two major modes: LaTeX or plain
2670 // Handle here those cases common to both modes
2671 // and then split to handle the two modes separately.
2672 if (c == META_INSET) {
2673 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2674 // Greyedout notes and, in general, all insets
2675 // with InsetLayout::isDisplay() == false,
2676 // are typeset inline with the text. So, we
2677 // can add a \par to the last paragraph of
2678 // such insets only if nothing else follows.
2679 bool incremented = false;
2680 Inset const * inset = getInset(i);
2681 InsetText const * textinset = inset
2682 ? inset->asInsetText()
2684 if (i + 1 == size() && textinset
2685 && !inset->getLayout().isDisplay()) {
2686 ParagraphList const & pars =
2687 textinset->text().paragraphs();
2688 pit_type const pit = pars.size() - 1;
2689 Font const lastfont =
2690 pit < 0 || pars[pit].empty()
2691 ? pars[pit].getLayoutFont(
2694 : pars[pit].getFont(bparams,
2695 pars[pit].size() - 1,
2697 if (lastfont.fontInfo().size() !=
2698 basefont.fontInfo().size()) {
2703 d->latexInset(bparams, os, rp, running_font,
2704 basefont, outerfont, open_font,
2705 runningChange, style, i, column);
2709 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2710 if (!bparams.useNonTeXFonts)
2711 script = Encodings::isKnownScriptChar(c);
2712 if (script != alien_script) {
2713 if (!alien_script.empty()) {
2715 alien_script.clear();
2717 string fontenc = running_font.language()->fontenc(bparams);
2719 && !Encodings::fontencSupportsScript(fontenc, script)) {
2720 column += script.length() + 2;
2721 os << "\\" << script << "{";
2722 alien_script = script;
2726 d->latexSpecialChar(os, bparams, rp, running_font,
2727 alien_script, style, i, end_pos, column);
2728 } catch (EncodingException & e) {
2729 if (runparams.dryrun) {
2730 os << "<" << _("LyX Warning: ")
2731 << _("uncodable character") << " '";
2735 // add location information and throw again.
2743 // Set the encoding to that returned from latexSpecialChar (see
2744 // comment for encoding member in OutputParams.h)
2745 runparams.encoding = rp.encoding;
2747 // Also carry on the info on a closed ulem command for insets
2748 // such as Note that do not produce any output, so that no
2749 // command is ever executed but its opening was recorded.
2750 runparams.inulemcmd = rp.inulemcmd;
2752 // And finally, pass the post_macros upstream
2753 runparams.post_macro = rp.post_macro;
2756 // Close wrapper for alien script
2757 if (!alien_script.empty()) {
2759 alien_script.clear();
2762 // If we have an open font definition, we have to close it
2764 // Make sure that \\par is done with the font of the last
2765 // character if this has another size as the default.
2766 // This is necessary because LaTeX (and LyX on the screen)
2767 // calculates the space between the baselines according
2768 // to this font. (Matthias)
2770 // We must not change the font for the last paragraph
2771 // of non-multipar insets, tabular cells or commands,
2772 // since this produces unwanted whitespace.
2774 Font const font = empty()
2775 ? getLayoutFont(bparams, outerfont)
2776 : getFont(bparams, size() - 1, outerfont);
2778 InsetText const * textinset = inInset().asInsetText();
2780 bool const maintext = textinset
2781 ? textinset->text().isMainText()
2784 size_t const numpars = textinset
2785 ? textinset->text().paragraphs().size()
2788 bool needPar = false;
2790 if (style.resfont.size() != font.fontInfo().size()
2791 && (!runparams.isLastPar || maintext
2792 || (numpars > 1 && d->ownerCode() != CELL_CODE
2793 && (inInset().getLayout().isDisplay()
2795 && !style.isCommand()) {
2798 #ifdef FIXED_LANGUAGE_END_DETECTION
2800 running_font.latexWriteEndChanges(os, bparams,
2801 runparams, basefont,
2802 next_->getFont(bparams, 0, outerfont),
2805 running_font.latexWriteEndChanges(os, bparams,
2806 runparams, basefont, basefont, needPar);
2809 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2810 //FIXME: there as we start another \selectlanguage with the next paragraph if
2811 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2812 running_font.latexWriteEndChanges(os, bparams, runparams,
2813 basefont, basefont, needPar);
2816 // The \par could not be inserted at the same nesting
2817 // level of the font size change, so do it now.
2818 os << "{\\" << font.latexSize() << "\\par}";
2822 column += Changes::latexMarkChange(os, bparams, runningChange,
2823 Change(Change::UNCHANGED), runparams);
2825 // Needed if there is an optional argument but no contents.
2826 if (body_pos > 0 && body_pos == size()) {
2830 if (!style.rightdelim().empty()) {
2831 os << style.rightdelim();
2832 column += style.rightdelim().size();
2835 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2836 && runparams.encoding != prev_encoding) {
2837 runparams.encoding = prev_encoding;
2838 os << setEncoding(prev_encoding->iconvName());
2841 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2845 bool Paragraph::emptyTag() const
2847 for (pos_type i = 0; i < size(); ++i) {
2848 if (Inset const * inset = getInset(i)) {
2849 InsetCode lyx_code = inset->lyxCode();
2850 // FIXME testing like that is wrong. What is
2852 if (lyx_code != TOC_CODE &&
2853 lyx_code != INCLUDE_CODE &&
2854 lyx_code != GRAPHICS_CODE &&
2855 lyx_code != ERT_CODE &&
2856 lyx_code != LISTINGS_CODE &&
2857 lyx_code != FLOAT_CODE &&
2858 lyx_code != TABULAR_CODE) {
2862 char_type c = d->text_[i];
2863 if (c != ' ' && c != '\t')
2871 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2874 for (pos_type i = 0; i < size(); ++i) {
2875 if (Inset const * inset = getInset(i)) {
2876 InsetCode lyx_code = inset->lyxCode();
2877 if (lyx_code == LABEL_CODE) {
2878 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2879 docstring const & id = il->getParam("name");
2880 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2888 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2892 for (i = 0; i < size(); ++i) {
2893 if (Inset const * inset = getInset(i)) {
2894 inset->docbook(os, runparams);
2896 char_type c = d->text_[i];
2899 os << sgml::escapeChar(c);
2906 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2910 for (i = 0; i < size(); ++i) {
2911 if (Inset const * inset = getInset(i)) {
2912 inset->xhtml(xs, runparams);
2914 char_type c = d->text_[i];
2924 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2927 pos_type size = text_.size();
2928 for (pos_type i = initial; i < size; ++i) {
2929 Font font = owner_->getFont(buf.params(), i, outerfont);
2930 if (text_[i] == META_INSET)
2932 if (i != initial && font != font_old)
2941 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2943 OutputParams const & runparams,
2944 Font const & outerfont,
2945 pos_type initial) const
2947 bool emph_flag = false;
2949 Layout const & style = *d->layout_;
2951 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2953 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2956 // parsing main loop
2957 for (pos_type i = initial; i < size(); ++i) {
2958 Font font = getFont(buf.params(), i, outerfont);
2960 // handle <emphasis> tag
2961 if (font_old.emph() != font.fontInfo().emph()) {
2962 if (font.fontInfo().emph() == FONT_ON) {
2965 } else if (i != initial) {
2966 os << "</emphasis>";
2971 if (Inset const * inset = getInset(i)) {
2972 inset->docbook(os, runparams);
2974 char_type c = d->text_[i];
2976 if (style.pass_thru)
2979 os << sgml::escapeChar(c);
2981 font_old = font.fontInfo();
2985 os << "</emphasis>";
2988 if (style.free_spacing)
2990 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2996 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2997 vector<html::EndFontTag> & tagsToClose,
2998 bool & flag, FontState curstate, html::FontTypes type)
3000 if (curstate == FONT_ON) {
3001 tagsToOpen.push_back(html::FontTag(type));
3004 tagsToClose.push_back(html::EndFontTag(type));
3011 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3013 OutputParams const & runparams,
3014 Font const & outerfont,
3015 bool start_paragraph, bool close_paragraph,
3016 pos_type initial) const
3020 // track whether we have opened these tags
3021 bool emph_flag = false;
3022 bool bold_flag = false;
3023 bool noun_flag = false;
3024 bool ubar_flag = false;
3025 bool dbar_flag = false;
3026 bool sout_flag = false;
3027 bool xout_flag = false;
3028 bool wave_flag = false;
3030 bool shap_flag = false;
3032 bool faml_flag = false;
3034 bool size_flag = false;
3036 Layout const & style = *d->layout_;
3038 if (start_paragraph)
3039 xs.startDivision(allowEmpty());
3042 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3044 FontShape curr_fs = INHERIT_SHAPE;
3045 FontFamily curr_fam = INHERIT_FAMILY;
3046 FontSize curr_size = INHERIT_SIZE;
3048 string const default_family =
3049 buf.masterBuffer()->params().fonts_default_family;
3051 vector<html::FontTag> tagsToOpen;
3052 vector<html::EndFontTag> tagsToClose;
3054 // parsing main loop
3055 for (pos_type i = initial; i < size(); ++i) {
3056 // let's not show deleted material in the output
3060 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3063 FontState curstate = font.fontInfo().emph();
3064 if (font_old.emph() != curstate)
3065 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3068 curstate = font.fontInfo().noun();
3069 if (font_old.noun() != curstate)
3070 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3073 curstate = font.fontInfo().underbar();
3074 if (font_old.underbar() != curstate)
3075 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3078 curstate = font.fontInfo().strikeout();
3079 if (font_old.strikeout() != curstate)
3080 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3083 curstate = font.fontInfo().xout();
3084 if (font_old.xout() != curstate)
3085 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3088 curstate = font.fontInfo().uuline();
3089 if (font_old.uuline() != curstate)
3090 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3093 curstate = font.fontInfo().uwave();
3094 if (font_old.uwave() != curstate)
3095 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3098 // a little hackish, but allows us to reuse what we have.
3099 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3100 if (font_old.series() != font.fontInfo().series())
3101 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3104 curr_fs = font.fontInfo().shape();
3105 FontShape old_fs = font_old.shape();
3106 if (old_fs != curr_fs) {
3110 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3113 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3115 case SMALLCAPS_SHAPE:
3116 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3122 // the other tags are for internal use
3130 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3134 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3137 case SMALLCAPS_SHAPE:
3138 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3145 // the other tags are for internal use
3152 curr_fam = font.fontInfo().family();
3153 FontFamily old_fam = font_old.family();
3154 if (old_fam != curr_fam) {
3158 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3161 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3163 case TYPEWRITER_FAMILY:
3164 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3166 case INHERIT_FAMILY:
3169 // the other tags are for internal use
3177 // we will treat a "default" font family as roman, since we have
3178 // no other idea what to do.
3179 if (default_family != "rmdefault" && default_family != "default") {
3180 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3185 if (default_family != "sfdefault") {
3186 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3190 case TYPEWRITER_FAMILY:
3191 if (default_family != "ttdefault") {
3192 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3196 case INHERIT_FAMILY:
3199 // the other tags are for internal use
3206 curr_size = font.fontInfo().size();
3207 FontSize old_size = font_old.size();
3208 if (old_size != curr_size) {
3212 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3215 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3218 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3221 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3224 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3227 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3230 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3233 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3236 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3239 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3242 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3248 // the other tags are for internal use
3254 switch (curr_size) {
3256 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3260 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3264 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3268 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3272 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3276 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3280 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3284 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3288 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3292 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3296 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3303 // the other tags are for internal use
3310 // Other such tags? What about the other text ranges?
3312 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3313 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3314 for (; cit != cen; ++cit)
3317 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3318 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3319 for (; sit != sen; ++sit)
3322 tagsToClose.clear();
3325 Inset const * inset = getInset(i);
3327 if (!runparams.for_toc || inset->isInToc()) {
3328 OutputParams np = runparams;
3329 np.local_font = &font;
3330 // If the paragraph has size 1, then we are in the "special
3331 // case" where we do not output the containing paragraph info
3332 if (!inset->getLayout().htmlisblock() && size() != 1)
3333 np.html_in_par = true;
3334 retval += inset->xhtml(xs, np);
3337 char_type c = getUChar(buf.masterBuffer()->params(),
3339 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3340 xs << XHTMLStream::ESCAPE_NONE << " ";
3344 font_old = font.fontInfo();
3348 // I'm worried about what happens if a branch, say, is itself
3349 // wrapped in some font stuff. I think that will not work.
3351 if (close_paragraph)
3358 bool Paragraph::isHfill(pos_type pos) const
3360 Inset const * inset = getInset(pos);
3361 return inset && inset->isHfill();
3365 bool Paragraph::isNewline(pos_type pos) const
3367 // U+2028 LINE SEPARATOR
3368 // U+2029 PARAGRAPH SEPARATOR
3369 char_type const c = d->text_[pos];
3370 if (c == 0x2028 || c == 0x2029)
3372 Inset const * inset = getInset(pos);
3373 return inset && inset->lyxCode() == NEWLINE_CODE;
3377 bool Paragraph::isEnvSeparator(pos_type pos) const
3379 Inset const * inset = getInset(pos);
3380 return inset && inset->lyxCode() == SEPARATOR_CODE;
3384 bool Paragraph::isLineSeparator(pos_type pos) const
3386 char_type const c = d->text_[pos];
3387 if (isLineSeparatorChar(c))
3389 Inset const * inset = getInset(pos);
3390 return inset && inset->isLineSeparator();
3394 bool Paragraph::isWordSeparator(pos_type pos) const
3398 if (Inset const * inset = getInset(pos))
3399 return !inset->isLetter();
3400 // if we have a hard hyphen (no en- or emdash) or apostrophe
3401 // we pass this to the spell checker
3402 // FIXME: this method is subject to change, visit
3403 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3404 // to get an impression how complex this is.
3405 if (isHardHyphenOrApostrophe(pos))
3407 char_type const c = d->text_[pos];
3408 // We want to pass the escape chars to the spellchecker
3409 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3410 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3414 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3416 pos_type const psize = size();
3419 char_type const c = d->text_[pos];
3420 if (c != '-' && c != '\'')
3422 int nextpos = pos + 1;
3423 int prevpos = pos > 0 ? pos - 1 : 0;
3424 if ((nextpos == psize || isSpace(nextpos))
3425 && (pos == 0 || isSpace(prevpos)))
3431 bool Paragraph::needsCProtection(bool const fragile) const
3433 // first check the layout of the paragraph, but only in insets
3434 InsetText const * textinset = inInset().asInsetText();
3435 bool const maintext = textinset
3436 ? textinset->text().isMainText()
3439 if (!maintext && layout().needcprotect) {
3440 // Environments need cprotection regardless the content
3441 if (layout().latextype == LATEX_ENVIRONMENT)
3444 // Commands need cprotection if they contain specific chars
3445 int const nchars_escape = 9;
3446 static char_type const chars_escape[nchars_escape] = {
3447 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3449 docstring const pars = asString();
3450 for (int k = 0; k < nchars_escape; k++) {
3451 if (contains(pars, chars_escape[k]))
3456 // now check whether we have insets that need cprotection
3457 pos_type size = pos_type(d->text_.size());
3458 for (pos_type i = 0; i < size; ++i) {
3461 Inset const * ins = getInset(i);
3462 if (ins->needsCProtection(maintext, fragile))
3464 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3465 // Environments need cprotection regardless the content
3467 // Now check math environments
3468 InsetMath const * im = getInset(i)->asInsetMath();
3469 if (!im || im->cell(0).empty())
3471 switch(im->cell(0)[0]->lyxCode()) {
3472 case MATH_AMSARRAY_CODE:
3473 case MATH_SUBSTACK_CODE:
3475 case MATH_XYMATRIX_CODE:
3476 // these need cprotection
3487 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3489 return d->speller_state_.getRange(pos);
3493 bool Paragraph::isChar(pos_type pos) const
3495 if (Inset const * inset = getInset(pos))
3496 return inset->isChar();
3497 char_type const c = d->text_[pos];
3498 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3502 bool Paragraph::isSpace(pos_type pos) const
3504 if (Inset const * inset = getInset(pos))
3505 return inset->isSpace();
3506 char_type const c = d->text_[pos];
3507 return lyx::isSpace(c);
3512 Paragraph::getParLanguage(BufferParams const & bparams) const
3515 return getFirstFontSettings(bparams).language();
3516 // FIXME: we should check the prev par as well (Lgb)
3517 return bparams.language;
3521 bool Paragraph::isRTL(BufferParams const & bparams) const
3523 return getParLanguage(bparams)->rightToLeft()
3524 && !inInset().getLayout().forceLTR();
3528 void Paragraph::changeLanguage(BufferParams const & bparams,
3529 Language const * from, Language const * to)
3531 // change language including dummy font change at the end
3532 for (pos_type i = 0; i <= size(); ++i) {
3533 Font font = getFontSettings(bparams, i);
3534 if (font.language() == from) {
3535 font.setLanguage(to);
3537 d->requestSpellCheck(i);
3543 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3545 Language const * doc_language = bparams.language;
3546 for (auto const & f : d->fontlist_)
3547 if (f.font().language() != ignore_language &&
3548 f.font().language() != latex_language &&
3549 f.font().language() != doc_language)
3555 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3557 for (auto const & f : d->fontlist_) {
3558 Language const * lang = f.font().language();
3559 if (lang != ignore_language &&
3560 lang != latex_language)
3566 docstring Paragraph::asString(int options) const
3568 return asString(0, size(), options);
3572 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3574 odocstringstream os;
3577 && options & AS_STR_LABEL
3578 && !d->params_.labelString().empty())
3579 os << d->params_.labelString() << ' ';
3581 for (pos_type i = beg; i < end; ++i) {
3582 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3584 char_type const c = d->text_[i];
3585 if (isPrintable(c) || c == '\t'
3586 || (c == '\n' && (options & AS_STR_NEWLINES)))
3588 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3589 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3590 LASSERT(runparams != 0, return docstring());
3591 getInset(i)->plaintext(os, *runparams);
3593 getInset(i)->toString(os);
3602 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3603 bool const shorten, bool const label) const
3605 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3606 if (label && !labelString().empty())
3607 os += labelString() + ' ';
3608 if (!layout().isTocCaption())
3610 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3613 char_type const c = d->text_[i];
3616 else if (c == META_INSET)
3617 getInset(i)->forOutliner(os, tmplen, false);
3620 Text::shortenForOutliner(os, maxlen);
3624 void Paragraph::setInsetOwner(Inset const * inset)
3626 d->inset_owner_ = inset;
3630 int Paragraph::id() const
3636 void Paragraph::setId(int id)
3642 Layout const & Paragraph::layout() const
3648 void Paragraph::setLayout(Layout const & layout)
3650 d->layout_ = &layout;
3654 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3656 setLayout(tc.defaultLayout());
3660 void Paragraph::setPlainLayout(DocumentClass const & tc)
3662 setLayout(tc.plainLayout());
3666 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3668 if (usePlainLayout())
3669 setPlainLayout(tclass);
3671 setDefaultLayout(tclass);
3675 Inset const & Paragraph::inInset() const
3677 LBUFERR(d->inset_owner_);
3678 return *d->inset_owner_;
3682 ParagraphParameters & Paragraph::params()
3688 ParagraphParameters const & Paragraph::params() const
3694 bool Paragraph::isFreeSpacing() const
3696 if (d->layout_->free_spacing)
3698 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3702 bool Paragraph::allowEmpty() const
3704 if (d->layout_->keepempty)
3706 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3710 bool Paragraph::brokenBiblio() const
3712 // There is a problem if there is no bibitem at position 0 in
3713 // paragraphs that need one, if there is another bibitem in the
3714 // paragraph or if this paragraph is not supposed to have
3715 // a bibitem inset at all.
3716 return ((d->layout_->labeltype == LABEL_BIBLIO
3717 && (d->insetlist_.find(BIBITEM_CODE) != 0
3718 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
3719 || (d->layout_->labeltype != LABEL_BIBLIO
3720 && d->insetlist_.find(BIBITEM_CODE) != -1));
3724 int Paragraph::fixBiblio(Buffer const & buffer)
3726 // FIXME: when there was already an inset at 0, the return value is 1,
3727 // which does not tell whether another inset has been remove; the
3728 // cursor cannot be correctly updated.
3730 bool const track_changes = buffer.params().track_changes;
3731 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3733 // The case where paragraph is not BIBLIO
3734 if (d->layout_->labeltype != LABEL_BIBLIO) {
3735 if (bibitem_pos == -1)
3736 // No InsetBibitem => OK
3738 // There is an InsetBibitem: remove it!
3739 d->insetlist_.release(bibitem_pos);
3740 eraseChar(bibitem_pos, track_changes);
3741 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
3744 bool const hasbibitem0 = bibitem_pos == 0;
3746 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3747 // There was an InsetBibitem at pos 0,
3748 // and no other one => OK
3749 if (bibitem_pos == -1)
3751 // there is a bibitem at the 0 position, but since
3752 // there is a second one, we copy the second on the
3753 // first. We're assuming there are at most two of
3754 // these, which there should be.
3755 // FIXME: why does it make sense to do that rather
3756 // than keep the first? (JMarc)
3757 Inset * inset = releaseInset(bibitem_pos);
3758 d->insetlist_.begin()->inset = inset;
3759 return -bibitem_pos;
3762 // We need to create an inset at the beginning
3763 Inset * inset = nullptr;
3764 if (bibitem_pos > 0) {
3765 // there was one somewhere in the paragraph, let's move it
3766 inset = d->insetlist_.release(bibitem_pos);
3767 eraseChar(bibitem_pos, track_changes);
3770 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3771 InsetCommandParams(BIBITEM_CODE));
3773 Font font(inherit_font, buffer.params().language);
3774 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3775 : Change::UNCHANGED));
3777 // This is needed to get the counters right
3778 buffer.updateBuffer();
3783 void Paragraph::checkAuthors(AuthorList const & authorList)
3785 d->changes_.checkAuthors(authorList);
3789 bool Paragraph::isChanged(pos_type pos) const
3791 return lookupChange(pos).changed();
3795 bool Paragraph::isInserted(pos_type pos) const
3797 return lookupChange(pos).inserted();
3801 bool Paragraph::isDeleted(pos_type pos) const
3803 return lookupChange(pos).deleted();
3807 InsetList const & Paragraph::insetList() const
3809 return d->insetlist_;
3813 void Paragraph::setInsetBuffers(Buffer & b)
3815 d->insetlist_.setBuffer(b);
3819 void Paragraph::resetBuffer()
3821 d->insetlist_.resetBuffer();
3825 Inset * Paragraph::releaseInset(pos_type pos)
3827 Inset * inset = d->insetlist_.release(pos);
3828 /// does not honour change tracking!
3829 eraseChar(pos, false);
3834 Inset * Paragraph::getInset(pos_type pos)
3836 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3837 ? d->insetlist_.get(pos) : 0;
3841 Inset const * Paragraph::getInset(pos_type pos) const
3843 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3844 ? d->insetlist_.get(pos) : 0;
3848 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3849 pos_type & right, TextCase action)
3851 // process sequences of modified characters; in change
3852 // tracking mode, this approach results in much better
3853 // usability than changing case on a char-by-char basis
3854 // We also need to track the current font, since font
3855 // changes within sequences can occur.
3856 vector<pair<char_type, Font> > changes;
3858 bool const trackChanges = bparams.track_changes;
3860 bool capitalize = true;
3862 for (; pos < right; ++pos) {
3863 char_type oldChar = d->text_[pos];
3864 char_type newChar = oldChar;
3866 // ignore insets and don't play with deleted text!
3867 if (oldChar != META_INSET && !isDeleted(pos)) {
3869 case text_lowercase:
3870 newChar = lowercase(oldChar);
3872 case text_capitalization:
3874 newChar = uppercase(oldChar);
3878 case text_uppercase:
3879 newChar = uppercase(oldChar);
3884 if (isWordSeparator(pos) || isDeleted(pos)) {
3885 // permit capitalization again
3889 if (oldChar != newChar) {
3890 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3891 if (pos != right - 1)
3893 // step behind the changing area
3897 int erasePos = pos - changes.size();
3898 for (size_t i = 0; i < changes.size(); i++) {
3899 insertChar(pos, changes[i].first,
3902 if (!eraseChar(erasePos, trackChanges)) {
3905 ++right; // expand selection
3913 int Paragraph::find(docstring const & str, bool cs, bool mw,
3914 pos_type start_pos, bool del) const
3916 pos_type pos = start_pos;
3917 int const strsize = str.length();
3919 pos_type const parsize = d->text_.size();
3920 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3921 // Ignore "invisible" letters such as ligature breaks
3922 // and hyphenation chars while searching
3923 while (pos < parsize - 1 && isInset(pos)) {
3924 odocstringstream os;
3925 getInset(pos)->toString(os);
3926 if (!getInset(pos)->isLetter() || !os.str().empty())
3930 if (cs && str[i] != d->text_[pos])
3932 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3934 if (!del && isDeleted(pos))
3941 // if necessary, check whether string matches word
3943 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3946 && !isWordSeparator(pos))
3950 return pos - start_pos;
3954 char_type Paragraph::getChar(pos_type pos) const
3956 return d->text_[pos];
3960 pos_type Paragraph::size() const
3962 return d->text_.size();
3966 bool Paragraph::empty() const
3968 return d->text_.empty();
3972 bool Paragraph::isInset(pos_type pos) const
3974 return d->text_[pos] == META_INSET;
3978 bool Paragraph::isSeparator(pos_type pos) const
3980 //FIXME: Are we sure this can be the only separator?
3981 return d->text_[pos] == ' ';
3985 void Paragraph::deregisterWords()
3987 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3988 Private::LangWordsMap::const_iterator ite = d->words_.end();
3989 for (; itl != ite; ++itl) {
3990 WordList & wl = theWordList(itl->first);
3991 Private::Words::const_iterator it = (itl->second).begin();
3992 Private::Words::const_iterator et = (itl->second).end();
3993 for (; it != et; ++it)
4000 void Paragraph::locateWord(pos_type & from, pos_type & to,
4001 word_location const loc) const
4004 case WHOLE_WORD_STRICT:
4005 if (from == 0 || from == size()
4006 || isWordSeparator(from)
4007 || isWordSeparator(from - 1)) {
4014 // If we are already at the beginning of a word, do nothing
4015 if (!from || isWordSeparator(from - 1))
4020 // always move the cursor to the beginning of previous word
4021 while (from && !isWordSeparator(from - 1))
4025 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4028 // no need to move the 'from' cursor
4032 while (to < size() && !isWordSeparator(to))
4037 void Paragraph::collectWords()
4039 for (pos_type pos = 0; pos < size(); ++pos) {
4040 if (isWordSeparator(pos))
4042 pos_type from = pos;
4043 locateWord(from, pos, WHOLE_WORD);
4044 // Work around MSVC warning: The statement
4045 // if (pos < from + lyxrc.completion_minlength)
4046 // triggers a signed vs. unsigned warning.
4047 // I don't know why this happens, it could be a MSVC bug, or
4048 // related to LLP64 (windows) vs. LP64 (unix) programming
4049 // model, or the C++ standard might be ambigous in the section
4050 // defining the "usual arithmetic conversions". However, using
4051 // a temporary variable is safe and works on all compilers.
4052 pos_type const endpos = from + lyxrc.completion_minlength;
4055 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4056 if (cit == d->fontlist_.end())
4058 Language const * lang = cit->font().language();
4059 docstring const word = asString(from, pos, AS_STR_NONE);
4060 d->words_[lang->lang()].insert(word);
4065 void Paragraph::registerWords()
4067 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4068 Private::LangWordsMap::const_iterator ite = d->words_.end();
4069 for (; itl != ite; ++itl) {
4070 WordList & wl = theWordList(itl->first);
4071 Private::Words::const_iterator it = (itl->second).begin();
4072 Private::Words::const_iterator et = (itl->second).end();
4073 for (; it != et; ++it)
4079 void Paragraph::updateWords()
4087 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4089 SkipPositionsIterator begin = skips.begin();
4090 SkipPositions::iterator end = skips.end();
4091 if (pos > 0 && begin < end) {
4093 if (end->last == pos - 1) {
4098 skips.insert(end, FontSpan(pos, pos));
4102 Language * Paragraph::Private::locateSpellRange(
4103 pos_type & from, pos_type & to,
4104 SkipPositions & skips) const
4106 // skip leading white space
4107 while (from < to && owner_->isWordSeparator(from))
4109 // don't check empty range
4112 // get current language
4113 Language * lang = getSpellLanguage(from);
4114 pos_type last = from;
4115 bool samelang = true;
4116 bool sameinset = true;
4117 while (last < to && samelang && sameinset) {
4118 // hop to end of word
4119 while (last < to && !owner_->isWordSeparator(last)) {
4120 if (owner_->getInset(last)) {
4121 appendSkipPosition(skips, last);
4122 } else if (owner_->isDeleted(last)) {
4123 appendSkipPosition(skips, last);
4127 // hop to next word while checking for insets
4128 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4129 if (Inset const * inset = owner_->getInset(last))
4130 sameinset = inset->isChar() && inset->isLetter();
4131 if (sameinset && owner_->isDeleted(last)) {
4132 appendSkipPosition(skips, last);
4137 if (sameinset && last < to) {
4138 // now check for language change
4139 samelang = lang == getSpellLanguage(last);
4142 // if language change detected backstep is needed
4150 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4153 const_cast<Language *>(owner_->getFontSettings(
4154 inset_owner_->buffer().params(), from).language());
4155 if (lang == inset_owner_->buffer().params().language
4156 && !lyxrc.spellchecker_alt_lang.empty()) {
4158 string const lang_variety =
4159 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4160 lang->setCode(lang_code);
4161 lang->setVariety(lang_variety);
4167 void Paragraph::requestSpellCheck(pos_type pos)
4169 d->requestSpellCheck(pos);
4173 bool Paragraph::needsSpellCheck() const
4175 SpellChecker::ChangeNumber speller_change_number = 0;
4176 if (theSpellChecker())
4177 speller_change_number = theSpellChecker()->changeNumber();
4178 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4179 d->speller_state_.needsCompleteRefresh(speller_change_number);
4181 return d->needsSpellCheck();
4185 bool Paragraph::Private::ignoreWord(docstring const & word) const
4187 // Ignore words with digits
4188 // FIXME: make this customizable
4189 // (note that some checkers ignore words with digits by default)
4190 docstring::const_iterator cit = word.begin();
4191 docstring::const_iterator const end = word.end();
4192 for (; cit != end; ++cit) {
4193 if (isNumber((*cit)))
4200 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4201 WordLangTuple & wl, docstring_list & suggestions,
4202 bool do_suggestion, bool check_learned) const
4204 SpellChecker::Result result = SpellChecker::WORD_OK;
4205 SpellChecker * speller = theSpellChecker();
4209 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4212 locateWord(from, to, WHOLE_WORD);
4213 if (from == to || from >= size())
4216 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4217 Language * lang = d->getSpellLanguage(from);
4219 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4222 wl = WordLangTuple(word, lang);
4227 if (needsSpellCheck() || check_learned) {
4229 if (!d->ignoreWord(word)) {
4230 bool const trailing_dot = to < size() && d->text_[to] == '.';
4231 result = speller->check(wl);
4232 if (SpellChecker::misspelled(result) && trailing_dot) {
4233 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4234 result = speller->check(wl);
4235 if (!SpellChecker::misspelled(result)) {
4236 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4238 from << ".." << to << "]");
4240 // spell check with dot appended failed too
4241 // restore original word/lang value
4242 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4243 wl = WordLangTuple(word, lang);
4247 if (!SpellChecker::misspelled(result)) {
4248 // area up to the begin of the next word is not misspelled
4249 while (end < size() && isWordSeparator(end))
4252 d->setMisspelled(from, end, result);
4254 result = d->speller_state_.getState(from);
4258 suggestions.clear();
4260 if (SpellChecker::misspelled(result)) {
4261 LYXERR(Debug::GUI, "misspelled word: \"" <<
4263 from << ".." << to << "]");
4265 speller->suggest(wl, suggestions);
4271 void Paragraph::anonymize()
4273 // This is a very crude anonymization for now
4274 for (char_type & c : d->text_)
4275 if (isLetterChar(c) || isNumber(c))
4280 void Paragraph::Private::markMisspelledWords(
4281 pos_type const & first, pos_type const & last,
4282 SpellChecker::Result result,
4283 docstring const & word,
4284 SkipPositions const & skips)
4286 if (!SpellChecker::misspelled(result)) {
4287 setMisspelled(first, last, SpellChecker::WORD_OK);
4291 SpellChecker * speller = theSpellChecker();
4292 // locate and enumerate the error positions
4293 int nerrors = speller->numMisspelledWords();
4295 SkipPositionsIterator it = skips.begin();
4296 SkipPositionsIterator et = skips.end();
4297 for (int index = 0; index < nerrors; ++index) {
4300 speller->misspelledWord(index, wstart, wlen);
4301 /// should not happen if speller supports range checks
4302 if (!wlen) continue;
4303 docstring const misspelled = word.substr(wstart, wlen);
4304 wstart += first + numskipped;
4305 if (snext < wstart) {
4306 /// mark the range of correct spelling
4307 numskipped += countSkips(it, et, wstart);
4308 setMisspelled(snext,
4309 wstart - 1, SpellChecker::WORD_OK);
4311 snext = wstart + wlen;
4312 numskipped += countSkips(it, et, snext);
4313 /// mark the range of misspelling
4314 setMisspelled(wstart, snext, result);
4315 LYXERR(Debug::GUI, "misspelled word: \"" <<
4316 misspelled << "\" [" <<
4317 wstart << ".." << (snext-1) << "]");
4320 if (snext <= last) {
4321 /// mark the range of correct spelling at end
4322 setMisspelled(snext, last, SpellChecker::WORD_OK);
4327 void Paragraph::spellCheck() const
4329 SpellChecker * speller = theSpellChecker();
4330 if (!speller || empty() ||!needsSpellCheck())
4334 d->rangeOfSpellCheck(start, endpos);
4335 if (speller->canCheckParagraph()) {
4336 // loop until we leave the range
4337 for (pos_type first = start; first < endpos; ) {
4338 pos_type last = endpos;
4339 Private::SkipPositions skips;
4340 Language * lang = d->locateSpellRange(first, last, skips);
4341 if (first >= endpos)
4343 // start the spell checker on the unit of meaning
4344 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4345 WordLangTuple wl = WordLangTuple(word, lang);
4346 SpellChecker::Result result = word.size() ?
4347 speller->check(wl) : SpellChecker::WORD_OK;
4348 d->markMisspelledWords(first, last, result, word, skips);
4352 static docstring_list suggestions;
4353 pos_type to = endpos;
4354 while (start < endpos) {
4356 spellCheck(start, to, wl, suggestions, false);
4360 d->readySpellCheck();
4364 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4366 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4367 if (result || pos <= 0 || pos > size())
4369 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4370 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4375 string Paragraph::magicLabel() const
4378 ss << "magicparlabel-" << id();