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, true);
451 owner_->locateWord(last, endpos, WHOLE_WORD, true);
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 bool Paragraph::isDeleted(pos_type start, pos_type end) const
586 LASSERT(start >= 0 && start <= size(), return false);
587 LASSERT(end > start && end <= size() + 1, return false);
589 return d->changes_.isDeleted(start, end);
593 bool Paragraph::isChanged(pos_type start, pos_type end) const
595 LASSERT(start >= 0 && start <= size(), return false);
596 LASSERT(end > start && end <= size() + 1, return false);
598 return d->changes_.isChanged(start, end);
602 bool Paragraph::isChanged() const
604 return d->changes_.isChanged();
608 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
610 // keep the logic here in sync with the logic of eraseChars()
614 Change const & change = d->changes_.lookup(size());
615 return change.inserted() && change.currentAuthor();
618 Change Paragraph::parEndChange() const
620 return d->changes_.lookup(size());
624 void Paragraph::setChange(Change const & change)
626 // beware of the imaginary end-of-par character!
627 d->changes_.set(change, 0, size() + 1);
630 * Propagate the change recursively - but not in case of DELETED!
632 * Imagine that your co-author makes changes in an existing inset. He
633 * sends your document to you and you come to the conclusion that the
634 * inset should go completely. If you erase it, LyX must not delete all
635 * text within the inset. Otherwise, the change tracked insertions of
636 * your co-author get lost and there is no way to restore them later.
638 * Conclusion: An inset's content should remain untouched if you delete it
641 if (!change.deleted()) {
642 for (pos_type pos = 0; pos < size(); ++pos) {
643 if (Inset * inset = getInset(pos))
644 inset->setChange(change);
650 void Paragraph::setChange(pos_type pos, Change const & change)
652 LASSERT(pos >= 0 && pos <= size(), return);
653 d->changes_.set(change, pos);
655 // see comment in setChange(Change const &) above
656 if (!change.deleted() && pos < size())
657 if (Inset * inset = getInset(pos))
658 inset->setChange(change);
662 Change const & Paragraph::lookupChange(pos_type pos) const
664 LBUFERR(pos >= 0 && pos <= size());
665 return d->changes_.lookup(pos);
669 void Paragraph::acceptChanges(pos_type start, pos_type end)
671 LASSERT(start >= 0 && start <= size(), return);
672 LASSERT(end > start && end <= size() + 1, return);
674 for (pos_type pos = start; pos < end; ++pos) {
675 switch (lookupChange(pos).type) {
676 case Change::UNCHANGED:
677 // accept changes in nested inset
678 if (Inset * inset = getInset(pos))
679 inset->acceptChanges();
682 case Change::INSERTED:
683 d->changes_.set(Change(Change::UNCHANGED), pos);
684 // also accept changes in nested inset
685 if (Inset * inset = getInset(pos))
686 inset->acceptChanges();
689 case Change::DELETED:
690 // Suppress access to non-existent
691 // "end-of-paragraph char"
693 eraseChar(pos, false);
704 void Paragraph::rejectChanges(pos_type start, pos_type end)
706 LASSERT(start >= 0 && start <= size(), return);
707 LASSERT(end > start && end <= size() + 1, return);
709 for (pos_type pos = start; pos < end; ++pos) {
710 switch (lookupChange(pos).type) {
711 case Change::UNCHANGED:
712 // reject changes in nested inset
713 if (Inset * inset = getInset(pos))
714 inset->rejectChanges();
717 case Change::INSERTED:
718 // Suppress access to non-existent
719 // "end-of-paragraph char"
721 eraseChar(pos, false);
727 case Change::DELETED:
728 d->changes_.set(Change(Change::UNCHANGED), pos);
730 // Do NOT reject changes within a deleted inset!
731 // There may be insertions of a co-author inside of it!
739 void Paragraph::Private::insertChar(pos_type pos, char_type c,
740 Change const & change)
742 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
745 changes_.insert(change, pos);
747 // This is actually very common when parsing buffers (and
748 // maybe inserting ascii text)
749 if (pos == pos_type(text_.size())) {
750 // when appending characters, no need to update tables
752 // but we want spell checking
753 requestSpellCheck(pos);
757 text_.insert(text_.begin() + pos, c);
759 // Update the font table.
760 fontlist_.increasePosAfterPos(pos);
763 insetlist_.increasePosAfterPos(pos);
765 // Update list of misspelled positions
766 speller_state_.increasePosAfterPos(pos);
770 bool Paragraph::insertInset(pos_type pos, Inset * inset,
771 Font const & font, Change const & change)
773 LASSERT(inset, return false);
774 LASSERT(pos >= 0 && pos <= size(), return false);
776 // Paragraph::insertInset() can be used in cut/copy/paste operation where
777 // d->inset_owner_ is not set yet.
778 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
781 d->insertChar(pos, META_INSET, change);
782 LASSERT(d->text_[pos] == META_INSET, return false);
784 // Add a new entry in the insetlist_.
785 d->insetlist_.insert(inset, pos);
787 // Some insets require run of spell checker
788 requestSpellCheck(pos);
794 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
796 LASSERT(pos >= 0 && pos <= size(), return false);
798 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
801 Change change = d->changes_.lookup(pos);
803 // set the character to DELETED if
804 // a) it was previously unchanged or
805 // b) it was inserted by a co-author
807 if (!change.changed() ||
808 (change.inserted() && !change.currentAuthor())) {
809 setChange(pos, Change(Change::DELETED));
810 // request run of spell checker
811 requestSpellCheck(pos);
815 if (change.deleted())
819 // Don't physically access the imaginary end-of-paragraph character.
820 // eraseChar() can only mark it as DELETED. A physical deletion of
821 // end-of-par must be handled externally.
827 d->changes_.erase(pos);
829 // if it is an inset, delete the inset entry
830 if (d->text_[pos] == META_INSET)
831 d->insetlist_.erase(pos);
833 d->text_.erase(d->text_.begin() + pos);
835 // Update the fontlist_
836 d->fontlist_.erase(pos);
838 // Update the insetlist_
839 d->insetlist_.decreasePosAfterPos(pos);
841 // Update list of misspelled positions
842 d->speller_state_.decreasePosAfterPos(pos);
843 d->speller_state_.refreshLast(size());
849 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
851 LASSERT(start >= 0 && start <= size(), return 0);
852 LASSERT(end >= start && end <= size() + 1, return 0);
855 for (pos_type count = end - start; count; --count) {
856 if (!eraseChar(i, trackChanges))
862 // Handle combining characters
863 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
864 otexstream & os, char_type c, char_type next,
865 OutputParams const & runparams)
867 // Writing next here may circumvent a possible font change between
868 // c and next. Since next is only output if it forms a surrogate pair
869 // with c we can ignore this:
870 // A font change inside a surrogate pair does not make sense and is
871 // hopefully impossible to input.
872 // FIXME: change tracking
873 // Is this correct WRT change tracking?
874 Encoding const & encoding = *(runparams.encoding);
875 docstring latex1 = encoding.latexChar(next).first;
876 if (runparams.inIPA) {
877 string const tipashortcut = Encodings::TIPAShortcut(next);
878 if (!tipashortcut.empty()) {
879 latex1 = from_ascii(tipashortcut);
882 docstring latex2 = encoding.latexChar(c).first;
884 if (bparams.useNonTeXFonts || docstring(1, next) == latex1) {
885 // Encoding supports the combination:
886 // output as is (combining char after base char).
887 os << latex2 << latex1;
888 return latex1.length() + latex2.length();
891 os << latex1 << "{" << latex2 << "}";
892 return latex1.length() + latex2.length() + 2;
896 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
897 OutputParams const & runparams,
900 unsigned int & column,
902 Layout const & style)
904 if (style.pass_thru || runparams.pass_thru)
907 if (i + 1 < int(text_.size())) {
908 char_type next = text_[i + 1];
909 if (Encodings::isCombiningChar(next)) {
910 // This space has an accent, so we must always output it.
911 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
916 if (runparams.linelen > 0
917 && column > runparams.linelen
919 && text_[i - 1] != ' '
920 && (i + 1 < int(text_.size()))
921 // same in FreeSpacing mode
922 && !owner_->isFreeSpacing()
923 // In typewriter mode, we want to avoid
924 // ! . ? : at the end of a line
925 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
926 && (text_[i - 1] == '.'
927 || text_[i - 1] == '?'
928 || text_[i - 1] == ':'
929 || text_[i - 1] == '!'))) {
931 os.texrow().start(owner_->id(), i + 1);
933 } else if (style.free_spacing) {
942 void Paragraph::Private::latexInset(BufferParams const & bparams,
944 OutputParams & runparams,
947 Font const & outerfont,
949 Change & running_change,
950 Layout const & style,
952 unsigned int & column)
954 Inset * inset = owner_->getInset(i);
957 if (style.pass_thru) {
958 odocstringstream ods;
959 inset->plaintext(ods, runparams);
964 // FIXME: move this to InsetNewline::latex
965 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
966 // newlines are handled differently here than
967 // the default in simpleTeXSpecialChars().
968 if (!style.newline_allowed) {
972 bool needPar = false;
973 column += running_font.latexWriteEndChanges(
974 os, bparams, runparams,
975 basefont, basefont, needPar);
979 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
982 basefont = owner_->getLayoutFont(bparams, outerfont);
983 running_font = basefont;
985 if (runparams.moving_arg)
989 os.texrow().start(owner_->id(), i + 1);
993 if (owner_->isDeleted(i)) {
994 if( ++runparams.inDeletedInset == 1)
995 runparams.changeOfDeletedInset = owner_->lookupChange(i);
998 if (inset->canTrackChanges()) {
999 column += Changes::latexMarkChange(os, bparams, running_change,
1000 Change(Change::UNCHANGED), runparams);
1001 running_change = Change(Change::UNCHANGED);
1005 odocstream::pos_type const len = os.os().tellp();
1007 if (inset->forceLTR(runparams)
1008 && running_font.isRightToLeft()
1009 // ERT is an exception, it should be output with no
1010 // decorations at all
1011 && inset->lyxCode() != ERT_CODE) {
1012 if (runparams.use_polyglossia) {
1014 } else if (running_font.language()->lang() == "farsi"
1015 || running_font.language()->lang() == "arabic_arabi")
1016 os << "\\textLR{" << termcmd;
1022 // FIXME: Bug: we can have an empty font change here!
1023 // if there has just been a font change, we are going to close it
1024 // right now, which means stupid latex code like \textsf{}. AFAIK,
1025 // this does not harm dvi output. A minor bug, thus (JMarc)
1027 // Some insets cannot be inside a font change command.
1028 // However, even such insets *can* be placed in \L or \R
1029 // or their equivalents (for RTL language switches), so we don't
1030 // close the language in those cases.
1031 // ArabTeX, though, cannot handle this special behavior, it seems.
1032 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1033 || running_font.language()->lang() == "arabic_arabtex";
1034 if (open_font && !inset->inheritFont()) {
1035 bool needPar = false;
1036 bool closeLanguage = arabtex
1037 || basefont.isRightToLeft() == running_font.isRightToLeft();
1038 unsigned int count = running_font.latexWriteEndChanges(os,
1039 bparams, runparams, basefont, basefont,
1040 needPar, closeLanguage);
1042 // if any font properties were closed, update the running_font,
1043 // making sure, however, to leave the language as it was
1045 // FIXME: probably a better way to keep track of the old
1046 // language, than copying the entire font?
1047 Font const copy_font(running_font);
1048 basefont = owner_->getLayoutFont(bparams, outerfont);
1049 running_font = basefont;
1051 running_font.setLanguage(copy_font.language());
1052 // leave font open if language is still open
1053 open_font = (running_font.language() == basefont.language());
1055 runparams.local_font = &basefont;
1059 size_t const previous_row_count = os.texrow().rows();
1062 runparams.lastid = id_;
1063 runparams.lastpos = i;
1064 inset->latex(os, runparams);
1065 } catch (EncodingException & e) {
1066 // add location information and throw again.
1075 if (os.texrow().rows() > previous_row_count) {
1076 os.texrow().start(owner_->id(), i + 1);
1079 column += (unsigned int)(os.os().tellp() - len);
1082 if (owner_->isDeleted(i))
1083 --runparams.inDeletedInset;
1087 void Paragraph::Private::latexSpecialChar(otexstream & os,
1088 BufferParams const & bparams,
1089 OutputParams const & runparams,
1090 Font const & running_font,
1091 string & alien_script,
1092 Layout const & style,
1095 unsigned int & column)
1097 char_type const c = owner_->getUChar(bparams, runparams, i);
1099 if (style.pass_thru || runparams.pass_thru
1100 || contains(style.pass_thru_chars, c)
1101 || contains(runparams.pass_thru_chars, c)) {
1103 Encoding const * const enc = runparams.encoding;
1104 if (enc && !enc->encodable(c))
1105 throw EncodingException(c);
1111 // TIPA uses its own T3 encoding
1112 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1114 // If T1 font encoding is used, use the special
1115 // characters it provides.
1116 // NOTE: Some languages reset the font encoding internally to a
1117 // non-standard font encoding. If we are using such a language,
1118 // we do not output special T1 chars.
1119 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1120 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1121 && latexSpecialT1(c, os, i, column))
1123 // NOTE: "fontspec" (non-TeX fonts) sets the font encoding to "TU" (untill 2017 "EU1" or "EU2")
1124 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1125 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1128 // Otherwise, we use what LaTeX provides us.
1131 os << "\\textbackslash" << termcmd;
1135 os << "\\textless" << termcmd;
1139 os << "\\textgreater" << termcmd;
1143 os << "\\textbar" << termcmd;
1148 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1149 (end_pos == -1 || i + 1 < end_pos) &&
1150 text_[i+1] == '-') {
1151 // Prevent "--" becoming an en dash and "---" an em dash.
1152 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1158 os << "\\textquotedbl" << termcmd;
1163 case '%': case '#': case '{':
1171 os << "\\textasciitilde" << termcmd;
1176 os << "\\textasciicircum" << termcmd;
1183 // avoid being mistaken for optional arguments
1191 // Blanks are printed before font switching.
1192 // Sure? I am not! (try nice-latex)
1193 // I am sure it's correct. LyX might be smarter
1194 // in the future, but for now, nothing wrong is
1200 // XeTeX's dash behaviour is determined via a global setting
1201 if (bparams.use_dash_ligatures
1202 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1204 // TODO #10961: && not in inset Flex Code
1205 // TODO #10961: && not in layout LyXCode
1206 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1223 Encoding const & encoding = *(runparams.encoding);
1224 char_type next = '\0';
1225 if (i + 1 < int(text_.size())) {
1226 next = text_[i + 1];
1227 if (Encodings::isCombiningChar(next)) {
1228 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1233 pair<docstring, bool> latex = encoding.latexChar(c);
1234 docstring nextlatex;
1235 bool nexttipas = false;
1236 string nexttipashortcut;
1237 if (next != '\0' && next != META_INSET && !encoding.encodable(next)) {
1238 nextlatex = encoding.latexChar(next).first;
1239 if (runparams.inIPA) {
1240 nexttipashortcut = Encodings::TIPAShortcut(next);
1241 nexttipas = !nexttipashortcut.empty();
1245 if (runparams.inIPA) {
1246 string const tipashortcut = Encodings::TIPAShortcut(c);
1247 if (!tipashortcut.empty()) {
1248 latex.first = from_ascii(tipashortcut);
1249 latex.second = false;
1253 // eventually close "script wrapper" command (see `Paragraph::latex`)
1254 if (!alien_script.empty()
1255 && alien_script != Encodings::isKnownScriptChar(next)) {
1256 column += latex.first.length();
1257 alien_script.clear();
1258 os << latex.first << "}";
1262 && ((!prefixIs(nextlatex, '\\')
1263 && !prefixIs(nextlatex, '{')
1264 && !prefixIs(nextlatex, '}'))
1266 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1268 // Prevent eating of a following space or command corruption by
1269 // following characters
1270 if (next == ' ' || next == '\0') {
1271 column += latex.first.length() + 1;
1272 os << latex.first << "{}";
1274 column += latex.first.length();
1275 os << latex.first << " ";
1278 column += latex.first.length() - 1;
1286 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1287 pos_type i, unsigned int & column)
1293 // In T1 encoding, these characters exist
1294 // but we should avoid ligatures
1295 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1297 os << "\\textcompwordmark" << termcmd;
1304 // soul.sty breaks with \char`\"
1305 os << "\\textquotedbl" << termcmd;
1314 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1315 pos_type i, unsigned int & column)
1317 // TU encoding is currently on par with T1.
1318 return latexSpecialT1(c, os, i, column);
1322 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1323 pos_type /*i*/, unsigned int & column)
1333 os << "\\textvertline" << termcmd;
1342 void Paragraph::Private::validate(LaTeXFeatures & features) const
1344 if (layout_->inpreamble && inset_owner_) {
1345 // FIXME: Using a string stream here circumvents the encoding
1346 // switching machinery of odocstream. Therefore the
1347 // output is wrong if this paragraph contains content
1348 // that needs to switch encoding.
1349 Buffer const & buf = inset_owner_->buffer();
1350 otexstringstream os;
1351 os << layout_->preamble();
1352 size_t const length = os.length();
1353 TeXOnePar(buf, *inset_owner_->getText(int(buf.getParFromID(owner_->id()).idx())),
1354 buf.getParFromID(owner_->id()).pit(), os,
1355 features.runparams(), string(), 0, -1, true);
1356 if (os.length() > length)
1357 features.addPreambleSnippet(os.release(), true);
1360 if (features.runparams().flavor == OutputParams::HTML
1361 && layout_->htmltitle()) {
1362 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1365 // check the params.
1366 if (!params_.spacing().isDefault())
1367 features.require("setspace");
1370 features.useLayout(layout_->name());
1373 fontlist_.validate(features);
1375 // then the indentation
1376 if (!params_.leftIndent().zero())
1377 features.require("ParagraphLeftIndent");
1380 InsetList::const_iterator icit = insetlist_.begin();
1381 InsetList::const_iterator iend = insetlist_.end();
1382 for (; icit != iend; ++icit) {
1384 features.inDeletedInset(owner_->isDeleted(icit->pos));
1385 if (icit->inset->lyxCode() == FOOT_CODE) {
1386 // FIXME: an item inset would make things much easier.
1387 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1388 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1389 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1390 && (icit->pos < begin_of_body_
1391 || (icit->pos == begin_of_body_
1392 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1393 features.saveNoteEnv("description");
1395 icit->inset->validate(features);
1396 features.inDeletedInset(false);
1397 if (layout_->needprotect &&
1398 icit->inset->lyxCode() == FOOT_CODE)
1399 features.require("NeedLyXFootnoteCode");
1403 // then the contents
1404 BufferParams const bp = features.runparams().is_child
1405 ? features.buffer().masterParams() : features.buffer().params();
1406 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1407 char_type c = text_[i];
1408 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1410 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1411 features.require("textquotedblp");
1412 else if (bp.main_font_encoding() != "T1"
1413 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1414 features.require("textquotedbl");
1415 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1416 // features that depend on the font or input encoding
1417 string feats = ci.textpreamble();
1418 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1419 if (fontenc.empty())
1420 fontenc = features.runparams().main_fontenc;
1421 while (!feats.empty()) {
1423 feats = split(feats, feat, ',');
1424 if (contains(feat, "!=")) {
1425 // a feature that is required except for the spcified
1426 // font or input encodings
1428 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1429 // multiple encodings are separated by semicolon
1430 vector<string> context = getVectorFromString(contexts, ";");
1431 // require feature if the context matches neither current font
1432 // nor input encoding
1433 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1434 && std::find(context.begin(), context.end(),
1435 features.runparams().encoding->name()) == context.end())
1436 features.require(realfeature);
1437 } else if (contains(feat, '=')) {
1438 // a feature that is required only for the spcified
1439 // font or input encodings
1441 string const contexts = split(feat, realfeature, '=');
1442 // multiple encodings are separated by semicolon
1443 vector<string> context = getVectorFromString(contexts, ";");
1444 // require feature if the context matches either current font
1445 // or input encoding
1446 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1447 || std::find(context.begin(), context.end(),
1448 features.runparams().encoding->name()) != context.end())
1449 features.require(realfeature);
1452 } else if (!bp.use_dash_ligatures
1453 && (c == 0x2013 || c == 0x2014)
1454 && bp.useNonTeXFonts
1455 && features.runparams().flavor == OutputParams::XETEX)
1456 // XeTeX's dash behaviour is determined via a global setting
1457 features.require("xetexdashbreakstate");
1458 BufferEncodings::validate(c, features);
1462 /////////////////////////////////////////////////////////////////////
1466 /////////////////////////////////////////////////////////////////////
1469 Layout const emptyParagraphLayout;
1472 Paragraph::Paragraph()
1473 : d(new Paragraph::Private(this, emptyParagraphLayout))
1480 Paragraph::Paragraph(Paragraph const & par)
1481 : itemdepth(par.itemdepth),
1482 d(new Paragraph::Private(*par.d, this))
1488 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1489 : itemdepth(par.itemdepth),
1490 d(new Paragraph::Private(*par.d, this, beg, end))
1496 Paragraph & Paragraph::operator=(Paragraph const & par)
1498 // needed as we will destroy the private part before copying it
1500 itemdepth = par.itemdepth;
1504 d = new Private(*par.d, this);
1511 Paragraph::~Paragraph()
1520 // this shall be called just before every "os << ..." action.
1521 void flushString(ostream & os, docstring & s)
1530 void Paragraph::write(ostream & os, BufferParams const & bparams,
1531 depth_type & dth) const
1533 // The beginning or end of a deeper (i.e. nested) area?
1534 if (dth != d->params_.depth()) {
1535 if (d->params_.depth() > dth) {
1536 while (d->params_.depth() > dth) {
1537 os << "\n\\begin_deeper";
1541 while (d->params_.depth() < dth) {
1542 os << "\n\\end_deeper";
1548 // First write the layout
1549 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1551 d->params_.write(os);
1553 Font font1(inherit_font, bparams.language);
1555 Change running_change = Change(Change::UNCHANGED);
1557 // this string is used as a buffer to avoid repetitive calls
1558 // to to_utf8(), which turn out to be expensive (JMarc)
1559 docstring write_buffer;
1562 for (pos_type i = 0; i <= size(); ++i) {
1564 Change const & change = lookupChange(i);
1565 if (change != running_change)
1566 flushString(os, write_buffer);
1567 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1568 running_change = change;
1573 // Write font changes
1574 Font font2 = getFontSettings(bparams, i);
1575 if (font2 != font1) {
1576 flushString(os, write_buffer);
1577 font2.lyxWriteChanges(font1, os);
1582 char_type const c = d->text_[i];
1585 if (Inset const * inset = getInset(i)) {
1586 flushString(os, write_buffer);
1587 if (inset->directWrite()) {
1588 // international char, let it write
1589 // code directly so it's shorter in
1595 os << "\\begin_inset ";
1597 os << "\n\\end_inset\n\n";
1600 // FIXME This can be removed again once the mystery
1601 // crash has been resolved.
1606 flushString(os, write_buffer);
1607 os << "\n\\backslash\n";
1611 flushString(os, write_buffer);
1612 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1619 if ((column > 70 && c == ' ')
1621 flushString(os, write_buffer);
1625 // this check is to amend a bug. LyX sometimes
1626 // inserts '\0' this could cause problems.
1628 write_buffer.push_back(c);
1630 LYXERR0("NUL char in structure.");
1636 flushString(os, write_buffer);
1637 os << "\n\\end_layout\n";
1638 // FIXME This can be removed again once the mystery
1639 // crash has been resolved.
1644 void Paragraph::validate(LaTeXFeatures & features) const
1646 d->validate(features);
1647 bool fragile = features.runparams().moving_arg;
1648 fragile |= layout().needprotect;
1649 if (inInset().getLayout().isNeedProtect())
1651 if (needsCProtection(fragile))
1652 features.require("cprotect");
1656 void Paragraph::insert(pos_type start, docstring const & str,
1657 Font const & font, Change const & change)
1659 for (size_t i = 0, n = str.size(); i != n ; ++i)
1660 insertChar(start + i, str[i], font, change);
1664 void Paragraph::appendChar(char_type c, Font const & font,
1665 Change const & change)
1668 d->changes_.insert(change, d->text_.size());
1669 // when appending characters, no need to update tables
1670 d->text_.push_back(c);
1671 setFont(d->text_.size() - 1, font);
1672 d->requestSpellCheck(d->text_.size() - 1);
1676 void Paragraph::appendString(docstring const & s, Font const & font,
1677 Change const & change)
1679 pos_type end = s.size();
1680 size_t oldsize = d->text_.size();
1681 size_t newsize = oldsize + end;
1682 size_t capacity = d->text_.capacity();
1683 if (newsize >= capacity)
1684 d->text_.reserve(max(capacity + 100, newsize));
1686 // when appending characters, no need to update tables
1689 // FIXME: Optimize this!
1690 for (size_t i = oldsize; i != newsize; ++i) {
1692 d->changes_.insert(change, i);
1693 d->requestSpellCheck(i);
1695 d->fontlist_.set(oldsize, font);
1696 d->fontlist_.set(newsize - 1, font);
1700 void Paragraph::insertChar(pos_type pos, char_type c,
1703 d->insertChar(pos, c, Change(trackChanges ?
1704 Change::INSERTED : Change::UNCHANGED));
1708 void Paragraph::insertChar(pos_type pos, char_type c,
1709 Font const & font, bool trackChanges)
1711 d->insertChar(pos, c, Change(trackChanges ?
1712 Change::INSERTED : Change::UNCHANGED));
1717 void Paragraph::insertChar(pos_type pos, char_type c,
1718 Font const & font, Change const & change)
1720 d->insertChar(pos, c, change);
1725 void Paragraph::resetFonts(Font const & font)
1727 d->fontlist_.clear();
1728 d->fontlist_.set(0, font);
1729 d->fontlist_.set(d->text_.size() - 1, font);
1732 // Gets uninstantiated font setting at position.
1733 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1737 LYXERR0("pos: " << pos << " size: " << size());
1741 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1742 if (cit != d->fontlist_.end())
1745 if (pos == size() && !empty())
1746 return getFontSettings(bparams, pos - 1);
1748 // Optimisation: avoid a full font instantiation if there is no
1749 // language change from previous call.
1750 static Font previous_font;
1751 static Language const * previous_lang = 0;
1752 Language const * lang = getParLanguage(bparams);
1753 if (lang != previous_lang) {
1754 previous_lang = lang;
1755 previous_font = Font(inherit_font, lang);
1757 return previous_font;
1761 FontSpan Paragraph::fontSpan(pos_type pos) const
1763 LBUFERR(pos <= size());
1766 return FontSpan(pos, pos);
1769 FontList::const_iterator cit = d->fontlist_.begin();
1770 FontList::const_iterator end = d->fontlist_.end();
1771 for (; cit != end; ++cit) {
1772 if (cit->pos() >= pos) {
1773 if (pos >= beginOfBody())
1774 return FontSpan(max(start, beginOfBody()),
1777 return FontSpan(start,
1778 min(beginOfBody() - 1,
1781 start = cit->pos() + 1;
1784 // This should not happen, but if so, we take no chances.
1785 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1786 LASSERT(false, return FontSpan(pos, pos));
1790 // Gets uninstantiated font setting at position 0
1791 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1793 if (!empty() && !d->fontlist_.empty())
1794 return d->fontlist_.begin()->font();
1796 // Optimisation: avoid a full font instantiation if there is no
1797 // language change from previous call.
1798 static Font previous_font;
1799 static Language const * previous_lang = 0;
1800 if (bparams.language != previous_lang) {
1801 previous_lang = bparams.language;
1802 previous_font = Font(inherit_font, bparams.language);
1805 return previous_font;
1809 // Gets the fully instantiated font at a given position in a paragraph
1810 // This is basically the same function as Text::GetFont() in text2.cpp.
1811 // The difference is that this one is used for generating the LaTeX file,
1812 // and thus cosmetic "improvements" are disallowed: This has to deliver
1813 // the true picture of the buffer. (Asger)
1814 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1815 Font const & outerfont) const
1819 Font font = getFontSettings(bparams, pos);
1821 pos_type const body_pos = beginOfBody();
1822 FontInfo & fi = font.fontInfo();
1824 fi.realize(d->layout_->labelfont);
1826 fi.realize(d->layout_->font);
1828 fi.realize(outerfont.fontInfo());
1829 fi.realize(bparams.getFont().fontInfo());
1835 Font const Paragraph::getLabelFont
1836 (BufferParams const & bparams, Font const & outerfont) const
1838 FontInfo tmpfont = d->layout_->labelfont;
1839 tmpfont.realize(outerfont.fontInfo());
1840 tmpfont.realize(bparams.getFont().fontInfo());
1841 return Font(tmpfont, getParLanguage(bparams));
1845 Font const Paragraph::getLayoutFont
1846 (BufferParams const & bparams, Font const & outerfont) const
1848 FontInfo tmpfont = d->layout_->font;
1849 tmpfont.realize(outerfont.fontInfo());
1850 tmpfont.realize(bparams.getFont().fontInfo());
1851 return Font(tmpfont, getParLanguage(bparams));
1855 char_type Paragraph::getUChar(BufferParams const & bparams,
1856 OutputParams const & rp,
1859 char_type c = d->text_[pos];
1861 // Return unchanged character in LTR languages
1862 // or if we use poylglossia/bidi (XeTeX).
1863 if (rp.useBidiPackage()
1864 || !getFontSettings(bparams, pos).isRightToLeft())
1867 // Without polyglossia/bidi, we need to account for some special cases.
1868 // FIXME This needs to be audited!
1870 // * The input is as expected for all delimiters
1871 // => checked for Hebrew!
1872 // * The output matches the display in the LyX workarea
1873 // => checked for Hebrew!
1874 // * The special cases below are really necessary
1875 // => checked for Hebrew!
1876 // * In arabic_arabi, brackets are transformed to Arabic
1877 // Ornate Parentheses. Is this is really wanted?
1879 string const & lang = getFontSettings(bparams, pos).language()->lang();
1882 // 1. In the following languages, parentheses need to be reversed.
1883 // Also with polyglodia/luabidi
1884 bool const reverseparens = (lang == "hebrew" || rp.use_polyglossia);
1886 // 2. In the following languages, brackets don't need to be reversed.
1887 bool const reversebrackets = lang != "arabic_arabtex"
1888 && lang != "arabic_arabi"
1891 // Now swap delimiters if needed.
1902 if (reversebrackets)
1906 if (reversebrackets)
1927 void Paragraph::setFont(pos_type pos, Font const & font)
1929 LASSERT(pos <= size(), return);
1931 // First, reduce font against layout/label font
1932 // Update: The setCharFont() routine in text2.cpp already
1933 // reduces font, so we don't need to do that here. (Asger)
1935 d->fontlist_.set(pos, font);
1939 void Paragraph::makeSameLayout(Paragraph const & par)
1941 d->layout_ = par.d->layout_;
1942 d->params_ = par.d->params_;
1946 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1948 if (isFreeSpacing())
1954 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1955 if (eraseChar(pos, trackChanges))
1961 return count > 0 || pos > 0;
1965 bool Paragraph::hasSameLayout(Paragraph const & par) const
1967 return par.d->layout_ == d->layout_
1968 && d->params_.sameLayout(par.d->params_);
1972 depth_type Paragraph::getDepth() const
1974 return d->params_.depth();
1978 depth_type Paragraph::getMaxDepthAfter() const
1980 if (d->layout_->isEnvironment())
1981 return d->params_.depth() + 1;
1983 return d->params_.depth();
1987 LyXAlignment Paragraph::getAlign(BufferParams const & bparams) const
1989 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1990 return getDefaultAlign(bparams);
1992 return d->params_.align();
1996 LyXAlignment Paragraph::getDefaultAlign(BufferParams const & bparams) const
1998 LyXAlignment res = layout().align;
1999 if (isRTL(bparams)) {
2001 if (res == LYX_ALIGN_LEFT)
2002 res = LYX_ALIGN_RIGHT;
2003 else if (res == LYX_ALIGN_RIGHT)
2004 res = LYX_ALIGN_LEFT;
2010 docstring const & Paragraph::labelString() const
2012 return d->params_.labelString();
2016 // the next two functions are for the manual labels
2017 docstring const Paragraph::getLabelWidthString() const
2019 if (d->layout_->margintype == MARGIN_MANUAL
2020 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2021 return d->params_.labelWidthString();
2023 return _("Senseless with this layout!");
2027 void Paragraph::setLabelWidthString(docstring const & s)
2029 d->params_.labelWidthString(s);
2033 docstring Paragraph::expandLabel(Layout const & layout,
2034 BufferParams const & bparams) const
2036 return expandParagraphLabel(layout, bparams, true);
2040 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2041 BufferParams const & bparams) const
2043 return expandParagraphLabel(layout, bparams, false);
2047 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2048 BufferParams const & bparams, bool process_appendix) const
2050 DocumentClass const & tclass = bparams.documentClass();
2051 string const & lang = getParLanguage(bparams)->code();
2052 bool const in_appendix = process_appendix && d->params_.appendix();
2053 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2055 if (fmt.empty() && !layout.counter.empty())
2056 return tclass.counters().theCounter(layout.counter, lang);
2058 // handle 'inherited level parts' in 'fmt',
2059 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2060 size_t const i = fmt.find('@', 0);
2061 if (i != docstring::npos) {
2062 size_t const j = fmt.find('@', i + 1);
2063 if (j != docstring::npos) {
2064 docstring parent(fmt, i + 1, j - i - 1);
2065 docstring label = from_ascii("??");
2066 if (tclass.hasLayout(parent))
2067 label = expandParagraphLabel(tclass[parent], bparams,
2069 fmt = docstring(fmt, 0, i) + label
2070 + docstring(fmt, j + 1, docstring::npos);
2074 return tclass.counters().counterLabel(fmt, lang);
2078 void Paragraph::applyLayout(Layout const & new_layout)
2080 d->layout_ = &new_layout;
2081 LyXAlignment const oldAlign = d->params_.align();
2083 if (!(oldAlign & d->layout_->alignpossible)) {
2084 frontend::Alert::warning(_("Alignment not permitted"),
2085 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2086 d->params_.align(LYX_ALIGN_LAYOUT);
2091 pos_type Paragraph::beginOfBody() const
2093 return d->begin_of_body_;
2097 void Paragraph::setBeginOfBody()
2099 if (d->layout_->labeltype != LABEL_MANUAL) {
2100 d->begin_of_body_ = 0;
2104 // Unroll the first two cycles of the loop
2105 // and remember the previous character to
2106 // remove unnecessary getChar() calls
2108 pos_type end = size();
2109 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2112 char_type previous_char = d->text_[i];
2113 if (!(isNewline(i) || isEnvSeparator(i))) {
2115 while (i < end && previous_char != ' ') {
2116 char_type temp = d->text_[i];
2117 if (isNewline(i) || isEnvSeparator(i))
2120 previous_char = temp;
2126 d->begin_of_body_ = i;
2130 bool Paragraph::allowParagraphCustomization() const
2132 return inInset().allowParagraphCustomization();
2136 bool Paragraph::usePlainLayout() const
2138 return inInset().usePlainLayout();
2142 bool Paragraph::isPassThru() const
2144 return inInset().isPassThru() || d->layout_->pass_thru;
2149 // paragraphs inside floats need different alignment tags to avoid
2152 bool noTrivlistCentering(InsetCode code)
2154 return code == FLOAT_CODE
2155 || code == WRAP_CODE
2156 || code == CELL_CODE;
2160 string correction(string const & orig)
2162 if (orig == "flushleft")
2163 return "raggedright";
2164 if (orig == "flushright")
2165 return "raggedleft";
2166 if (orig == "center")
2172 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2173 InsetCode code, bool const lastpar, int & col)
2175 string macro = suffix + "{";
2176 if (noTrivlistCentering(code)) {
2178 // the last paragraph in non-trivlist-aligned
2179 // context is special (to avoid unwanted whitespace)
2180 if (suffix == "\\begin") {
2181 macro = "\\" + correction(env) + "{}";
2182 os << from_ascii(macro);
2183 col += macro.size();
2188 macro += correction(env);
2192 if (suffix == "\\par\\end") {
2196 os << from_ascii(macro);
2197 col += macro.size();
2198 if (suffix == "\\begin") {
2208 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2209 otexstream & os, OutputParams const & runparams) const
2214 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2215 (layout_->toggle_indent != ITOGGLE_NEVER) :
2216 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2218 if (canindent && params_.noindent() && !layout_->pass_thru) {
2219 os << "\\noindent ";
2223 LyXAlignment const curAlign = params_.align();
2225 if (curAlign == layout_->align)
2229 case LYX_ALIGN_NONE:
2230 case LYX_ALIGN_BLOCK:
2231 case LYX_ALIGN_LAYOUT:
2232 case LYX_ALIGN_SPECIAL:
2233 case LYX_ALIGN_DECIMAL:
2235 case LYX_ALIGN_LEFT:
2236 case LYX_ALIGN_RIGHT:
2237 case LYX_ALIGN_CENTER:
2238 if (runparams.moving_arg) {
2245 string const begin_tag = "\\begin";
2246 InsetCode code = ownerCode();
2247 bool const lastpar = runparams.isLastPar;
2248 // RTL in classic (PDF)LaTeX (without the Bidi package)
2249 // Luabibdi (used by LuaTeX) behaves like classic
2250 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2251 && !runparams.useBidiPackage();
2254 case LYX_ALIGN_NONE:
2255 case LYX_ALIGN_BLOCK:
2256 case LYX_ALIGN_LAYOUT:
2257 case LYX_ALIGN_SPECIAL:
2258 case LYX_ALIGN_DECIMAL:
2260 case LYX_ALIGN_LEFT: {
2262 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2263 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2265 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2267 } case LYX_ALIGN_RIGHT: {
2269 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2270 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2272 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2274 } case LYX_ALIGN_CENTER: {
2275 corrected_env(os, begin_tag, "center", code, lastpar, column);
2284 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2285 otexstream & os, OutputParams const & runparams) const
2287 LyXAlignment const curAlign = params_.align();
2289 if (curAlign == layout_->align)
2293 case LYX_ALIGN_NONE:
2294 case LYX_ALIGN_BLOCK:
2295 case LYX_ALIGN_LAYOUT:
2296 case LYX_ALIGN_SPECIAL:
2297 case LYX_ALIGN_DECIMAL:
2299 case LYX_ALIGN_LEFT:
2300 case LYX_ALIGN_RIGHT:
2301 case LYX_ALIGN_CENTER:
2302 if (runparams.moving_arg)
2307 bool output = false;
2309 string const end_tag = "\\par\\end";
2310 InsetCode code = ownerCode();
2311 bool const lastpar = runparams.isLastPar;
2312 // RTL in classic (PDF)LaTeX (without the Bidi package)
2313 // Luabibdi (used by LuaTeX) behaves like classic
2314 bool const rtl_classic = owner_->getParLanguage(bparams)->rightToLeft()
2315 && !runparams.useBidiPackage();
2318 case LYX_ALIGN_NONE:
2319 case LYX_ALIGN_BLOCK:
2320 case LYX_ALIGN_LAYOUT:
2321 case LYX_ALIGN_SPECIAL:
2322 case LYX_ALIGN_DECIMAL:
2324 case LYX_ALIGN_LEFT: {
2326 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2327 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2329 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2331 } case LYX_ALIGN_RIGHT: {
2333 // Classic (PDF)LaTeX switches the left/right logic in RTL mode
2334 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2336 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2338 } case LYX_ALIGN_CENTER: {
2339 corrected_env(os, end_tag, "center", code, lastpar, col);
2344 return output || lastpar;
2348 // This one spits out the text of the paragraph
2349 void Paragraph::latex(BufferParams const & bparams,
2350 Font const & outerfont,
2352 OutputParams const & runparams,
2353 int start_pos, int end_pos, bool force) const
2355 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2357 // FIXME This check should not be needed. Perhaps issue an
2358 // error if it triggers.
2359 Layout const & style = inInset().forcePlainLayout() ?
2360 bparams.documentClass().plainLayout() : *d->layout_;
2362 if (!force && style.inpreamble)
2365 bool const allowcust = allowParagraphCustomization();
2367 // Current base font for all inherited font changes, without any
2368 // change caused by an individual character, except for the language:
2369 // It is set to the language of the first character.
2370 // As long as we are in the label, this font is the base font of the
2371 // label. Before the first body character it is set to the base font
2375 // If there is an open font-encoding changing command (script wrapper),
2376 // alien_script is set to its name
2377 string alien_script;
2380 // Maybe we have to create a optional argument.
2381 pos_type body_pos = beginOfBody();
2382 unsigned int column = 0;
2385 // the optional argument is kept in curly brackets in
2386 // case it contains a ']'
2387 // This is not strictly needed, but if this is changed it
2388 // would be a file format change, and tex2lyx would need
2389 // to be adjusted, since it unconditionally removes the
2390 // braces when it parses \item.
2393 basefont = getLabelFont(bparams, outerfont);
2395 basefont = getLayoutFont(bparams, outerfont);
2398 // Which font is currently active?
2399 Font running_font(basefont);
2400 // Do we have an open font change?
2401 bool open_font = false;
2403 Change runningChange = Change(Change::UNCHANGED);
2405 Encoding const * const prev_encoding = runparams.encoding;
2407 os.texrow().start(id(), 0);
2409 // if the paragraph is empty, the loop will not be entered at all
2411 // For InTitle commands, we have already opened a group
2412 // in output_latex::TeXOnePar.
2413 if (style.isCommand() && !style.intitle) {
2417 if (!style.leftdelim().empty()) {
2418 os << style.leftdelim();
2419 column += style.leftdelim().size();
2422 column += d->startTeXParParams(bparams, os, runparams);
2425 // Whether a \par can be issued for insets typeset inline with text.
2426 // Yes if greater than 0. This has to be static.
2427 THREAD_LOCAL_STATIC int parInline = 0;
2429 for (pos_type i = 0; i < size(); ++i) {
2430 // First char in paragraph or after label?
2431 if (i == body_pos) {
2434 bool needPar = false;
2435 column += running_font.latexWriteEndChanges(
2436 os, bparams, runparams,
2437 basefont, basefont, needPar);
2440 basefont = getLayoutFont(bparams, outerfont);
2441 running_font = basefont;
2443 column += Changes::latexMarkChange(os, bparams,
2444 runningChange, Change(Change::UNCHANGED),
2446 runningChange = Change(Change::UNCHANGED);
2451 // For InTitle commands, we have already opened a group
2452 // in output_latex::TeXOnePar.
2453 if (style.isCommand() && !style.intitle) {
2458 if (!style.leftdelim().empty()) {
2459 os << style.leftdelim();
2460 column += style.leftdelim().size();
2464 column += d->startTeXParParams(bparams, os,
2468 runparams.wasDisplayMath = runparams.inDisplayMath;
2469 runparams.inDisplayMath = false;
2470 bool deleted_display_math = false;
2471 Change const & change = runparams.inDeletedInset
2472 ? runparams.changeOfDeletedInset : lookupChange(i);
2474 char_type const c = d->text_[i];
2476 // Check whether a display math inset follows
2478 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2480 runparams.ctObject = getInset(i)->CtObject(runparams);
2482 InsetMath const * im = getInset(i)->asInsetMath();
2483 if (im && im->asHullInset()
2484 && im->asHullInset()->outerDisplay()) {
2485 runparams.inDisplayMath = true;
2486 // runparams.inDeletedInset will be set by
2487 // latexInset later, but we need this info
2488 // before it is called. On the other hand, we
2489 // cannot set it here because it is a counter.
2490 deleted_display_math = isDeleted(i);
2492 if (bparams.output_changes && deleted_display_math
2493 && runningChange == change
2494 && change.type == Change::DELETED
2495 && !os.afterParbreak()) {
2496 // A display math in the same paragraph follows.
2497 // We have to close and then reopen \lyxdeleted,
2498 // otherwise the math will be shifted up.
2499 OutputParams rp = runparams;
2501 bool needPar = false;
2502 column += running_font.latexWriteEndChanges(
2503 os, bparams, rp, basefont,
2507 basefont = (body_pos > i) ? getLabelFont(bparams, outerfont)
2508 : 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 = (body_pos > i) ? getLabelFont(bparams, outerfont)
2529 : getLayoutFont(bparams, outerfont);
2530 running_font = basefont;
2531 column += Changes::latexMarkChange(os, bparams, runningChange,
2533 runningChange = change;
2536 // do not output text which is marked deleted
2537 // if change tracking output is disabled
2538 if (!bparams.output_changes && change.deleted()) {
2544 // Fully instantiated font
2545 Font const current_font = getFont(bparams, i, outerfont);
2547 Font const last_font = running_font;
2549 // Do we need to close the previous font?
2551 (current_font != running_font ||
2552 current_font.language() != running_font.language()))
2554 // ensure there is no open script-wrapper
2555 if (!alien_script.empty()) {
2558 alien_script.clear();
2560 bool needPar = false;
2561 column += running_font.latexWriteEndChanges(
2562 os, bparams, runparams, basefont,
2563 (i == body_pos-1) ? basefont : current_font,
2565 running_font = basefont;
2569 // if necessary, close language environment before opening CJK
2570 string const running_lang = running_font.language()->babel();
2571 string const lang_end_command = lyxrc.language_command_end;
2572 if (!lang_end_command.empty() && !bparams.useNonTeXFonts
2573 && !running_lang.empty()
2574 && running_lang == openLanguageName()
2575 && current_font.language()->encoding()->package() == Encoding::CJK) {
2576 string end_tag = subst(lang_end_command, "$$lang", running_lang);
2577 os << from_ascii(end_tag);
2578 column += end_tag.length();
2582 // Switch file encoding if necessary (and allowed)
2583 if (!runparams.pass_thru && !style.pass_thru &&
2584 runparams.encoding->package() != Encoding::none &&
2585 current_font.language()->encoding()->package() != Encoding::none) {
2586 pair<bool, int> const enc_switch =
2587 switchEncoding(os.os(), bparams, runparams,
2588 *(current_font.language()->encoding()));
2589 if (enc_switch.first) {
2590 column += enc_switch.second;
2591 runparams.encoding = current_font.language()->encoding();
2595 // A display math inset inside an ulem command will be output
2596 // as a box of width \linewidth, so we have to either disable
2597 // indentation if the inset starts a paragraph, or start a new
2598 // line to accommodate such box. This has to be done before
2599 // writing any font changing commands.
2600 if (runparams.inDisplayMath && !deleted_display_math
2601 && runparams.inulemcmd) {
2602 if (os.afterParbreak())
2608 // Do we need to change font?
2609 if ((current_font != running_font ||
2610 current_font.language() != running_font.language()) &&
2613 bool const in_ct_deletion = (bparams.output_changes
2614 && runningChange == change
2615 && change.type == Change::DELETED
2616 && !os.afterParbreak());
2617 if (in_ct_deletion) {
2618 // We have to close and then reopen \lyxdeleted,
2619 // as strikeout needs to be on lowest level.
2620 bool needPar = false;
2621 OutputParams rp = runparams;
2622 column += running_font.latexWriteEndChanges(
2623 os, bparams, rp, basefont,
2628 odocstringstream ods;
2629 column += current_font.latexWriteStartChanges(ods, bparams,
2630 runparams, basefont,
2632 // Check again for display math in ulem commands as a
2633 // font change may also occur just before a math inset.
2634 if (runparams.inDisplayMath && !deleted_display_math
2635 && runparams.inulemcmd) {
2636 if (os.afterParbreak())
2641 running_font = current_font;
2643 docstring fontchange = ods.str();
2644 // check whether the fontchange ends with a \\textcolor
2645 // modifier and the text starts with a space (bug 4473)
2646 docstring const last_modifier = rsplit(fontchange, '\\');
2647 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2648 os << fontchange << from_ascii("{}");
2649 // check if the fontchange ends with a trailing blank
2650 // (like "\small " (see bug 3382)
2651 else if (suffixIs(fontchange, ' ') && c == ' ')
2652 os << fontchange.substr(0, fontchange.size() - 1)
2653 << from_ascii("{}");
2656 if (in_ct_deletion) {
2657 // We have to close and then reopen \lyxdeleted,
2658 // as strikeout needs to be on lowest level.
2659 OutputParams rp = runparams;
2660 column += Changes::latexMarkChange(os, bparams,
2661 Change(Change::UNCHANGED), change, rp);
2665 // FIXME: think about end_pos implementation...
2666 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2667 // FIXME: integrate this case in latexSpecialChar
2668 // Do not print the separation of the optional argument
2669 // if style.pass_thru is false. This works because
2670 // latexSpecialChar ignores spaces if
2671 // style.pass_thru is false.
2672 if (i != body_pos - 1) {
2673 if (d->simpleTeXBlanks(bparams, runparams, os,
2674 i, column, current_font, style)) {
2675 // A surrogate pair was output. We
2676 // must not call latexSpecialChar
2677 // in this iteration, since it would output
2678 // the combining character again.
2685 OutputParams rp = runparams;
2686 rp.free_spacing = style.free_spacing;
2687 rp.local_font = ¤t_font;
2688 rp.intitle = style.intitle;
2690 // Two major modes: LaTeX or plain
2691 // Handle here those cases common to both modes
2692 // and then split to handle the two modes separately.
2693 if (c == META_INSET) {
2694 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2695 // Greyedout notes and, in general, all insets
2696 // with InsetLayout::isDisplay() == false,
2697 // are typeset inline with the text. So, we
2698 // can add a \par to the last paragraph of
2699 // such insets only if nothing else follows.
2700 bool incremented = false;
2701 Inset const * inset = getInset(i);
2702 InsetText const * textinset = inset
2703 ? inset->asInsetText()
2705 if (i + 1 == size() && textinset
2706 && !inset->getLayout().isDisplay()) {
2707 ParagraphList const & pars =
2708 textinset->text().paragraphs();
2709 pit_type const pit = pars.size() - 1;
2710 Font const lastfont =
2711 pit < 0 || pars[pit].empty()
2712 ? pars[pit].getLayoutFont(
2715 : pars[pit].getFont(bparams,
2716 pars[pit].size() - 1,
2718 if (lastfont.fontInfo().size() !=
2719 basefont.fontInfo().size()) {
2724 d->latexInset(bparams, os, rp, running_font,
2725 basefont, outerfont, open_font,
2726 runningChange, style, i, column);
2730 if (runparams.ctObject == OutputParams::CT_DISPLAYOBJECT
2731 || runparams.ctObject == OutputParams::CT_UDISPLAYOBJECT) {
2732 // Close \lyx*deleted and force its
2733 // reopening (if needed)
2736 runningChange = Change(Change::UNCHANGED);
2737 runparams.ctObject = OutputParams::CT_NORMAL;
2740 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2741 if (!bparams.useNonTeXFonts)
2742 script = Encodings::isKnownScriptChar(c);
2743 if (script != alien_script) {
2744 if (!alien_script.empty()) {
2746 alien_script.clear();
2748 string fontenc = running_font.language()->fontenc(bparams);
2750 && !Encodings::fontencSupportsScript(fontenc, script)) {
2751 column += script.length() + 2;
2752 os << "\\" << script << "{";
2753 alien_script = script;
2757 d->latexSpecialChar(os, bparams, rp, running_font,
2758 alien_script, style, i, end_pos, column);
2759 } catch (EncodingException & e) {
2760 if (runparams.dryrun) {
2761 os << "<" << _("LyX Warning: ")
2762 << _("uncodable character") << " '";
2766 // add location information and throw again.
2774 // Set the encoding to that returned from latexSpecialChar (see
2775 // comment for encoding member in OutputParams.h)
2776 runparams.encoding = rp.encoding;
2778 // Also carry on the info on a closed ulem command for insets
2779 // such as Note that do not produce any output, so that no
2780 // command is ever executed but its opening was recorded.
2781 runparams.inulemcmd = rp.inulemcmd;
2783 // And finally, pass the post_macros upstream
2784 runparams.post_macro = rp.post_macro;
2787 // Close wrapper for alien script
2788 if (!alien_script.empty()) {
2790 alien_script.clear();
2793 // If we have an open font definition, we have to close it
2795 // Make sure that \\par is done with the font of the last
2796 // character if this has another size as the default.
2797 // This is necessary because LaTeX (and LyX on the screen)
2798 // calculates the space between the baselines according
2799 // to this font. (Matthias)
2801 // We must not change the font for the last paragraph
2802 // of non-multipar insets, tabular cells or commands,
2803 // since this produces unwanted whitespace.
2805 Font const font = empty()
2806 ? getLayoutFont(bparams, outerfont)
2807 : getFont(bparams, size() - 1, outerfont);
2809 InsetText const * textinset = inInset().asInsetText();
2811 bool const maintext = textinset
2812 ? textinset->text().isMainText()
2815 size_t const numpars = textinset
2816 ? textinset->text().paragraphs().size()
2819 bool needPar = false;
2821 if (style.resfont.size() != font.fontInfo().size()
2822 && (!runparams.isLastPar || maintext
2823 || (numpars > 1 && d->ownerCode() != CELL_CODE
2824 && (inInset().getLayout().isDisplay()
2826 && !style.isCommand()) {
2829 #ifdef FIXED_LANGUAGE_END_DETECTION
2831 running_font.latexWriteEndChanges(os, bparams,
2832 runparams, basefont,
2833 next_->getFont(bparams, 0, outerfont),
2836 running_font.latexWriteEndChanges(os, bparams,
2837 runparams, basefont, basefont, needPar);
2840 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2841 //FIXME: there as we start another \selectlanguage with the next paragraph if
2842 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2843 running_font.latexWriteEndChanges(os, bparams, runparams,
2844 basefont, basefont, needPar);
2847 // The \par could not be inserted at the same nesting
2848 // level of the font size change, so do it now.
2849 os << "{\\" << font.latexSize() << "\\par}";
2853 column += Changes::latexMarkChange(os, bparams, runningChange,
2854 Change(Change::UNCHANGED), runparams);
2856 // Needed if there is an optional argument but no contents.
2857 if (body_pos > 0 && body_pos == size()) {
2861 if (!style.rightdelim().empty()) {
2862 os << style.rightdelim();
2863 column += style.rightdelim().size();
2866 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2867 && runparams.encoding != prev_encoding) {
2868 runparams.encoding = prev_encoding;
2869 os << setEncoding(prev_encoding->iconvName());
2872 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2876 bool Paragraph::emptyTag() const
2878 for (pos_type i = 0; i < size(); ++i) {
2879 if (Inset const * inset = getInset(i)) {
2880 InsetCode lyx_code = inset->lyxCode();
2881 // FIXME testing like that is wrong. What is
2883 if (lyx_code != TOC_CODE &&
2884 lyx_code != INCLUDE_CODE &&
2885 lyx_code != GRAPHICS_CODE &&
2886 lyx_code != ERT_CODE &&
2887 lyx_code != LISTINGS_CODE &&
2888 lyx_code != FLOAT_CODE &&
2889 lyx_code != TABULAR_CODE) {
2893 char_type c = d->text_[i];
2894 if (c != ' ' && c != '\t')
2902 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2905 for (pos_type i = 0; i < size(); ++i) {
2906 if (Inset const * inset = getInset(i)) {
2907 InsetCode lyx_code = inset->lyxCode();
2908 if (lyx_code == LABEL_CODE) {
2909 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2910 docstring const & id = il->getParam("name");
2911 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2919 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2923 for (i = 0; i < size(); ++i) {
2924 if (Inset const * inset = getInset(i)) {
2925 inset->docbook(os, runparams);
2927 char_type c = d->text_[i];
2930 os << sgml::escapeChar(c);
2937 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2941 for (i = 0; i < size(); ++i) {
2942 if (Inset const * inset = getInset(i)) {
2943 inset->xhtml(xs, runparams);
2945 char_type c = d->text_[i];
2955 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2958 pos_type size = text_.size();
2959 for (pos_type i = initial; i < size; ++i) {
2960 Font font = owner_->getFont(buf.params(), i, outerfont);
2961 if (text_[i] == META_INSET)
2963 if (i != initial && font != font_old)
2972 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2974 OutputParams const & runparams,
2975 Font const & outerfont,
2976 pos_type initial) const
2978 bool emph_flag = false;
2980 Layout const & style = *d->layout_;
2982 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2984 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2987 // parsing main loop
2988 for (pos_type i = initial; i < size(); ++i) {
2989 Font font = getFont(buf.params(), i, outerfont);
2991 // handle <emphasis> tag
2992 if (font_old.emph() != font.fontInfo().emph()) {
2993 if (font.fontInfo().emph() == FONT_ON) {
2996 } else if (i != initial) {
2997 os << "</emphasis>";
3002 if (Inset const * inset = getInset(i)) {
3003 inset->docbook(os, runparams);
3005 char_type c = d->text_[i];
3007 if (style.pass_thru)
3010 os << sgml::escapeChar(c);
3012 font_old = font.fontInfo();
3016 os << "</emphasis>";
3019 if (style.free_spacing)
3021 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3027 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
3028 vector<html::EndFontTag> & tagsToClose,
3029 bool & flag, FontState curstate, html::FontTypes type)
3031 if (curstate == FONT_ON) {
3032 tagsToOpen.push_back(html::FontTag(type));
3035 tagsToClose.push_back(html::EndFontTag(type));
3042 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3044 OutputParams const & runparams,
3045 Font const & outerfont,
3046 bool start_paragraph, bool close_paragraph,
3047 pos_type initial) const
3051 // track whether we have opened these tags
3052 bool emph_flag = false;
3053 bool bold_flag = false;
3054 bool noun_flag = false;
3055 bool ubar_flag = false;
3056 bool dbar_flag = false;
3057 bool sout_flag = false;
3058 bool xout_flag = false;
3059 bool wave_flag = false;
3061 bool shap_flag = false;
3063 bool faml_flag = false;
3065 bool size_flag = false;
3067 Layout const & style = *d->layout_;
3069 if (start_paragraph)
3070 xs.startDivision(allowEmpty());
3073 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3075 FontShape curr_fs = INHERIT_SHAPE;
3076 FontFamily curr_fam = INHERIT_FAMILY;
3077 FontSize curr_size = INHERIT_SIZE;
3079 string const default_family =
3080 buf.masterBuffer()->params().fonts_default_family;
3082 vector<html::FontTag> tagsToOpen;
3083 vector<html::EndFontTag> tagsToClose;
3085 // parsing main loop
3086 for (pos_type i = initial; i < size(); ++i) {
3087 // let's not show deleted material in the output
3091 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3094 FontState curstate = font.fontInfo().emph();
3095 if (font_old.emph() != curstate)
3096 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3099 curstate = font.fontInfo().noun();
3100 if (font_old.noun() != curstate)
3101 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3104 curstate = font.fontInfo().underbar();
3105 if (font_old.underbar() != curstate)
3106 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3109 curstate = font.fontInfo().strikeout();
3110 if (font_old.strikeout() != curstate)
3111 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3114 curstate = font.fontInfo().xout();
3115 if (font_old.xout() != curstate)
3116 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3119 curstate = font.fontInfo().uuline();
3120 if (font_old.uuline() != curstate)
3121 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3124 curstate = font.fontInfo().uwave();
3125 if (font_old.uwave() != curstate)
3126 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3129 // a little hackish, but allows us to reuse what we have.
3130 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3131 if (font_old.series() != font.fontInfo().series())
3132 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3135 curr_fs = font.fontInfo().shape();
3136 FontShape old_fs = font_old.shape();
3137 if (old_fs != curr_fs) {
3141 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3144 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3146 case SMALLCAPS_SHAPE:
3147 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3153 // the other tags are for internal use
3161 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3165 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3168 case SMALLCAPS_SHAPE:
3169 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3176 // the other tags are for internal use
3183 curr_fam = font.fontInfo().family();
3184 FontFamily old_fam = font_old.family();
3185 if (old_fam != curr_fam) {
3189 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3192 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3194 case TYPEWRITER_FAMILY:
3195 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3197 case INHERIT_FAMILY:
3200 // the other tags are for internal use
3208 // we will treat a "default" font family as roman, since we have
3209 // no other idea what to do.
3210 if (default_family != "rmdefault" && default_family != "default") {
3211 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3216 if (default_family != "sfdefault") {
3217 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3221 case TYPEWRITER_FAMILY:
3222 if (default_family != "ttdefault") {
3223 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3227 case INHERIT_FAMILY:
3230 // the other tags are for internal use
3237 curr_size = font.fontInfo().size();
3238 FontSize old_size = font_old.size();
3239 if (old_size != curr_size) {
3243 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3246 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3249 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3252 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3255 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3258 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3261 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3264 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3267 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3270 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3273 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3279 // the other tags are for internal use
3285 switch (curr_size) {
3287 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3291 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3295 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3299 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3303 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3307 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3311 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3315 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3319 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3323 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3327 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3334 // the other tags are for internal use
3341 // Other such tags? What about the other text ranges?
3343 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3344 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3345 for (; cit != cen; ++cit)
3348 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3349 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3350 for (; sit != sen; ++sit)
3353 tagsToClose.clear();
3356 Inset const * inset = getInset(i);
3358 if (!runparams.for_toc || inset->isInToc()) {
3359 OutputParams np = runparams;
3360 np.local_font = &font;
3361 // If the paragraph has size 1, then we are in the "special
3362 // case" where we do not output the containing paragraph info
3363 if (!inset->getLayout().htmlisblock() && size() != 1)
3364 np.html_in_par = true;
3365 retval += inset->xhtml(xs, np);
3368 char_type c = getUChar(buf.masterBuffer()->params(),
3370 if (c == ' ' && (style.free_spacing || runparams.free_spacing))
3371 xs << XHTMLStream::ESCAPE_NONE << " ";
3375 font_old = font.fontInfo();
3379 // I'm worried about what happens if a branch, say, is itself
3380 // wrapped in some font stuff. I think that will not work.
3382 if (close_paragraph)
3389 bool Paragraph::isHfill(pos_type pos) const
3391 Inset const * inset = getInset(pos);
3392 return inset && inset->isHfill();
3396 bool Paragraph::isNewline(pos_type pos) const
3398 // U+2028 LINE SEPARATOR
3399 // U+2029 PARAGRAPH SEPARATOR
3400 char_type const c = d->text_[pos];
3401 if (c == 0x2028 || c == 0x2029)
3403 Inset const * inset = getInset(pos);
3404 return inset && inset->lyxCode() == NEWLINE_CODE;
3408 bool Paragraph::isEnvSeparator(pos_type pos) const
3410 Inset const * inset = getInset(pos);
3411 return inset && inset->lyxCode() == SEPARATOR_CODE;
3415 bool Paragraph::isLineSeparator(pos_type pos) const
3417 char_type const c = d->text_[pos];
3418 if (isLineSeparatorChar(c))
3420 Inset const * inset = getInset(pos);
3421 return inset && inset->isLineSeparator();
3425 bool Paragraph::isWordSeparator(pos_type pos, bool const ignore_deleted) const
3429 if (ignore_deleted && isDeleted(pos))
3431 if (Inset const * inset = getInset(pos))
3432 return !inset->isLetter();
3433 // if we have a hard hyphen (no en- or emdash) or apostrophe
3434 // we pass this to the spell checker
3435 // FIXME: this method is subject to change, visit
3436 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3437 // to get an impression how complex this is.
3438 if (isHardHyphenOrApostrophe(pos))
3440 char_type const c = d->text_[pos];
3441 // We want to pass the escape chars to the spellchecker
3442 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3443 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3447 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3449 pos_type const psize = size();
3452 char_type const c = d->text_[pos];
3453 if (c != '-' && c != '\'')
3455 int nextpos = pos + 1;
3456 int prevpos = pos > 0 ? pos - 1 : 0;
3457 if ((nextpos == psize || isSpace(nextpos))
3458 && (pos == 0 || isSpace(prevpos)))
3464 bool Paragraph::needsCProtection(bool const fragile) const
3466 // first check the layout of the paragraph, but only in insets
3467 InsetText const * textinset = inInset().asInsetText();
3468 bool const maintext = textinset
3469 ? textinset->text().isMainText()
3472 if (!maintext && layout().needcprotect) {
3473 // Environments need cprotection regardless the content
3474 if (layout().latextype == LATEX_ENVIRONMENT)
3477 // Commands need cprotection if they contain specific chars
3478 int const nchars_escape = 9;
3479 static char_type const chars_escape[nchars_escape] = {
3480 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3482 docstring const pars = asString();
3483 for (int k = 0; k < nchars_escape; k++) {
3484 if (contains(pars, chars_escape[k]))
3489 // now check whether we have insets that need cprotection
3490 pos_type size = pos_type(d->text_.size());
3491 for (pos_type i = 0; i < size; ++i) {
3494 Inset const * ins = getInset(i);
3495 if (ins->needsCProtection(maintext, fragile))
3497 if (ins->getLayout().latextype() == InsetLayout::ENVIRONMENT)
3498 // Environments need cprotection regardless the content
3500 // Now check math environments
3501 InsetMath const * im = getInset(i)->asInsetMath();
3502 if (!im || im->cell(0).empty())
3504 switch(im->cell(0)[0]->lyxCode()) {
3505 case MATH_AMSARRAY_CODE:
3506 case MATH_SUBSTACK_CODE:
3508 case MATH_XYMATRIX_CODE:
3509 // these need cprotection
3520 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3522 return d->speller_state_.getRange(pos);
3526 bool Paragraph::isChar(pos_type pos) const
3528 if (Inset const * inset = getInset(pos))
3529 return inset->isChar();
3530 char_type const c = d->text_[pos];
3531 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3535 bool Paragraph::isSpace(pos_type pos) const
3537 if (Inset const * inset = getInset(pos))
3538 return inset->isSpace();
3539 char_type const c = d->text_[pos];
3540 return lyx::isSpace(c);
3545 Paragraph::getParLanguage(BufferParams const & bparams) const
3548 return getFirstFontSettings(bparams).language();
3549 // FIXME: we should check the prev par as well (Lgb)
3550 return bparams.language;
3554 bool Paragraph::isRTL(BufferParams const & bparams) const
3556 return getParLanguage(bparams)->rightToLeft()
3557 && !inInset().getLayout().forceLTR();
3561 void Paragraph::changeLanguage(BufferParams const & bparams,
3562 Language const * from, Language const * to)
3564 // change language including dummy font change at the end
3565 for (pos_type i = 0; i <= size(); ++i) {
3566 Font font = getFontSettings(bparams, i);
3567 if (font.language() == from) {
3568 font.setLanguage(to);
3570 d->requestSpellCheck(i);
3576 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3578 Language const * doc_language = bparams.language;
3579 for (auto const & f : d->fontlist_)
3580 if (f.font().language() != ignore_language &&
3581 f.font().language() != latex_language &&
3582 f.font().language() != doc_language)
3588 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3590 for (auto const & f : d->fontlist_) {
3591 Language const * lang = f.font().language();
3592 if (lang != ignore_language &&
3593 lang != latex_language)
3599 docstring Paragraph::asString(int options) const
3601 return asString(0, size(), options);
3605 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3607 odocstringstream os;
3610 && options & AS_STR_LABEL
3611 && !d->params_.labelString().empty())
3612 os << d->params_.labelString() << ' ';
3614 for (pos_type i = beg; i < end; ++i) {
3615 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3617 char_type const c = d->text_[i];
3618 if (isPrintable(c) || c == '\t'
3619 || (c == '\n' && (options & AS_STR_NEWLINES)))
3621 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3622 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3623 LASSERT(runparams != 0, return docstring());
3624 getInset(i)->plaintext(os, *runparams);
3626 getInset(i)->toString(os);
3635 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3636 bool const shorten, bool const label) const
3638 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3639 if (label && !labelString().empty())
3640 os += labelString() + ' ';
3641 if (!layout().isTocCaption())
3643 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3646 char_type const c = d->text_[i];
3649 else if (c == META_INSET)
3650 getInset(i)->forOutliner(os, tmplen, false);
3653 Text::shortenForOutliner(os, maxlen);
3657 void Paragraph::setInsetOwner(Inset const * inset)
3659 d->inset_owner_ = inset;
3663 int Paragraph::id() const
3669 void Paragraph::setId(int id)
3675 Layout const & Paragraph::layout() const
3681 void Paragraph::setLayout(Layout const & layout)
3683 d->layout_ = &layout;
3687 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3689 setLayout(tc.defaultLayout());
3693 void Paragraph::setPlainLayout(DocumentClass const & tc)
3695 setLayout(tc.plainLayout());
3699 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3701 if (usePlainLayout())
3702 setPlainLayout(tclass);
3704 setDefaultLayout(tclass);
3708 Inset const & Paragraph::inInset() const
3710 LBUFERR(d->inset_owner_);
3711 return *d->inset_owner_;
3715 ParagraphParameters & Paragraph::params()
3721 ParagraphParameters const & Paragraph::params() const
3727 bool Paragraph::isFreeSpacing() const
3729 if (d->layout_->free_spacing)
3731 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3735 bool Paragraph::allowEmpty() const
3737 if (d->layout_->keepempty)
3739 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3743 bool Paragraph::brokenBiblio() const
3745 // There is a problem if there is no bibitem at position 0 in
3746 // paragraphs that need one, if there is another bibitem in the
3747 // paragraph or if this paragraph is not supposed to have
3748 // a bibitem inset at all.
3749 return ((d->layout_->labeltype == LABEL_BIBLIO
3750 && (d->insetlist_.find(BIBITEM_CODE) != 0
3751 || d->insetlist_.find(BIBITEM_CODE, 1) > 0))
3752 || (d->layout_->labeltype != LABEL_BIBLIO
3753 && d->insetlist_.find(BIBITEM_CODE) != -1));
3757 int Paragraph::fixBiblio(Buffer const & buffer)
3759 // FIXME: when there was already an inset at 0, the return value is 1,
3760 // which does not tell whether another inset has been remove; the
3761 // cursor cannot be correctly updated.
3763 bool const track_changes = buffer.params().track_changes;
3764 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3766 // The case where paragraph is not BIBLIO
3767 if (d->layout_->labeltype != LABEL_BIBLIO) {
3768 if (bibitem_pos == -1)
3769 // No InsetBibitem => OK
3771 // There is an InsetBibitem: remove it!
3772 d->insetlist_.release(bibitem_pos);
3773 eraseChar(bibitem_pos, track_changes);
3774 return (bibitem_pos == 0) ? -1 : -bibitem_pos;
3777 bool const hasbibitem0 = bibitem_pos == 0;
3779 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3780 // There was an InsetBibitem at pos 0,
3781 // and no other one => OK
3782 if (bibitem_pos == -1)
3784 // there is a bibitem at the 0 position, but since
3785 // there is a second one, we copy the second on the
3786 // first. We're assuming there are at most two of
3787 // these, which there should be.
3788 // FIXME: why does it make sense to do that rather
3789 // than keep the first? (JMarc)
3790 Inset * inset = releaseInset(bibitem_pos);
3791 d->insetlist_.begin()->inset = inset;
3792 return -bibitem_pos;
3795 // We need to create an inset at the beginning
3796 Inset * inset = nullptr;
3797 if (bibitem_pos > 0) {
3798 // there was one somewhere in the paragraph, let's move it
3799 inset = d->insetlist_.release(bibitem_pos);
3800 eraseChar(bibitem_pos, track_changes);
3803 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3804 InsetCommandParams(BIBITEM_CODE));
3806 Font font(inherit_font, buffer.params().language);
3807 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3808 : Change::UNCHANGED));
3810 // This is needed to get the counters right
3811 buffer.updateBuffer();
3816 void Paragraph::checkAuthors(AuthorList const & authorList)
3818 d->changes_.checkAuthors(authorList);
3822 bool Paragraph::isChanged(pos_type pos) const
3824 return lookupChange(pos).changed();
3828 bool Paragraph::isInserted(pos_type pos) const
3830 return lookupChange(pos).inserted();
3834 bool Paragraph::isDeleted(pos_type pos) const
3836 return lookupChange(pos).deleted();
3840 InsetList const & Paragraph::insetList() const
3842 return d->insetlist_;
3846 void Paragraph::setInsetBuffers(Buffer & b)
3848 d->insetlist_.setBuffer(b);
3852 void Paragraph::resetBuffer()
3854 d->insetlist_.resetBuffer();
3858 Inset * Paragraph::releaseInset(pos_type pos)
3860 Inset * inset = d->insetlist_.release(pos);
3861 /// does not honour change tracking!
3862 eraseChar(pos, false);
3867 Inset * Paragraph::getInset(pos_type pos)
3869 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3870 ? d->insetlist_.get(pos) : 0;
3874 Inset const * Paragraph::getInset(pos_type pos) const
3876 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3877 ? d->insetlist_.get(pos) : 0;
3881 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3882 pos_type & right, TextCase action)
3884 // process sequences of modified characters; in change
3885 // tracking mode, this approach results in much better
3886 // usability than changing case on a char-by-char basis
3887 // We also need to track the current font, since font
3888 // changes within sequences can occur.
3889 vector<pair<char_type, Font> > changes;
3891 bool const trackChanges = bparams.track_changes;
3893 bool capitalize = true;
3895 for (; pos < right; ++pos) {
3896 char_type oldChar = d->text_[pos];
3897 char_type newChar = oldChar;
3899 // ignore insets and don't play with deleted text!
3900 if (oldChar != META_INSET && !isDeleted(pos)) {
3902 case text_lowercase:
3903 newChar = lowercase(oldChar);
3905 case text_capitalization:
3907 newChar = uppercase(oldChar);
3911 case text_uppercase:
3912 newChar = uppercase(oldChar);
3917 if (isWordSeparator(pos) || isDeleted(pos)) {
3918 // permit capitalization again
3922 if (oldChar != newChar) {
3923 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3924 if (pos != right - 1)
3926 // step behind the changing area
3930 int erasePos = pos - changes.size();
3931 for (size_t i = 0; i < changes.size(); i++) {
3932 insertChar(pos, changes[i].first,
3935 if (!eraseChar(erasePos, trackChanges)) {
3938 ++right; // expand selection
3946 int Paragraph::find(docstring const & str, bool cs, bool mw,
3947 pos_type start_pos, bool del) const
3949 pos_type pos = start_pos;
3950 int const strsize = str.length();
3952 pos_type const parsize = d->text_.size();
3953 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3954 // Ignore "invisible" letters such as ligature breaks
3955 // and hyphenation chars while searching
3956 while (pos < parsize - 1 && isInset(pos)) {
3957 odocstringstream os;
3958 getInset(pos)->toString(os);
3959 if (!getInset(pos)->isLetter() || !os.str().empty())
3963 if (cs && str[i] != d->text_[pos])
3965 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3967 if (!del && isDeleted(pos))
3974 // if necessary, check whether string matches word
3976 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3979 && !isWordSeparator(pos))
3983 return pos - start_pos;
3987 char_type Paragraph::getChar(pos_type pos) const
3989 return d->text_[pos];
3993 pos_type Paragraph::size() const
3995 return d->text_.size();
3999 bool Paragraph::empty() const
4001 return d->text_.empty();
4005 bool Paragraph::isInset(pos_type pos) const
4007 return d->text_[pos] == META_INSET;
4011 bool Paragraph::isSeparator(pos_type pos) const
4013 //FIXME: Are we sure this can be the only separator?
4014 return d->text_[pos] == ' ';
4018 void Paragraph::deregisterWords()
4020 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4021 Private::LangWordsMap::const_iterator ite = d->words_.end();
4022 for (; itl != ite; ++itl) {
4023 WordList & wl = theWordList(itl->first);
4024 Private::Words::const_iterator it = (itl->second).begin();
4025 Private::Words::const_iterator et = (itl->second).end();
4026 for (; it != et; ++it)
4033 void Paragraph::locateWord(pos_type & from, pos_type & to,
4034 word_location const loc, bool const ignore_deleted) const
4037 case WHOLE_WORD_STRICT:
4038 if (from == 0 || from == size()
4039 || isWordSeparator(from, ignore_deleted)
4040 || isWordSeparator(from - 1, ignore_deleted)) {
4047 // If we are already at the beginning of a word, do nothing
4048 if (!from || isWordSeparator(from - 1, ignore_deleted))
4053 // always move the cursor to the beginning of previous word
4054 while (from && !isWordSeparator(from - 1, ignore_deleted))
4058 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4061 // no need to move the 'from' cursor
4065 while (to < size() && !isWordSeparator(to, ignore_deleted))
4070 void Paragraph::collectWords()
4072 for (pos_type pos = 0; pos < size(); ++pos) {
4073 if (isWordSeparator(pos))
4075 pos_type from = pos;
4076 locateWord(from, pos, WHOLE_WORD);
4077 // Work around MSVC warning: The statement
4078 // if (pos < from + lyxrc.completion_minlength)
4079 // triggers a signed vs. unsigned warning.
4080 // I don't know why this happens, it could be a MSVC bug, or
4081 // related to LLP64 (windows) vs. LP64 (unix) programming
4082 // model, or the C++ standard might be ambigous in the section
4083 // defining the "usual arithmetic conversions". However, using
4084 // a temporary variable is safe and works on all compilers.
4085 pos_type const endpos = from + lyxrc.completion_minlength;
4088 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4089 if (cit == d->fontlist_.end())
4091 Language const * lang = cit->font().language();
4092 docstring const word = asString(from, pos, AS_STR_NONE);
4093 d->words_[lang->lang()].insert(word);
4098 void Paragraph::registerWords()
4100 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4101 Private::LangWordsMap::const_iterator ite = d->words_.end();
4102 for (; itl != ite; ++itl) {
4103 WordList & wl = theWordList(itl->first);
4104 Private::Words::const_iterator it = (itl->second).begin();
4105 Private::Words::const_iterator et = (itl->second).end();
4106 for (; it != et; ++it)
4112 void Paragraph::updateWords()
4120 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4122 SkipPositionsIterator begin = skips.begin();
4123 SkipPositions::iterator end = skips.end();
4124 if (pos > 0 && begin < end) {
4126 if (end->last == pos - 1) {
4131 skips.insert(end, FontSpan(pos, pos));
4135 Language * Paragraph::Private::locateSpellRange(
4136 pos_type & from, pos_type & to,
4137 SkipPositions & skips) const
4139 // skip leading white space
4140 while (from < to && owner_->isWordSeparator(from))
4142 // don't check empty range
4145 // get current language
4146 Language * lang = getSpellLanguage(from);
4147 pos_type last = from;
4148 bool samelang = true;
4149 bool sameinset = true;
4150 while (last < to && samelang && sameinset) {
4151 // hop to end of word
4152 while (last < to && !owner_->isWordSeparator(last)) {
4153 if (owner_->getInset(last)) {
4154 appendSkipPosition(skips, last);
4155 } else if (owner_->isDeleted(last)) {
4156 appendSkipPosition(skips, last);
4160 // hop to next word while checking for insets
4161 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4162 if (Inset const * inset = owner_->getInset(last))
4163 sameinset = inset->isChar() && inset->isLetter();
4164 if (sameinset && owner_->isDeleted(last)) {
4165 appendSkipPosition(skips, last);
4170 if (sameinset && last < to) {
4171 // now check for language change
4172 samelang = lang == getSpellLanguage(last);
4175 // if language change detected backstep is needed
4183 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4186 const_cast<Language *>(owner_->getFontSettings(
4187 inset_owner_->buffer().params(), from).language());
4188 if (lang == inset_owner_->buffer().params().language
4189 && !lyxrc.spellchecker_alt_lang.empty()) {
4191 string const lang_variety =
4192 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4193 lang->setCode(lang_code);
4194 lang->setVariety(lang_variety);
4200 void Paragraph::requestSpellCheck(pos_type pos)
4202 d->requestSpellCheck(pos);
4206 bool Paragraph::needsSpellCheck() const
4208 SpellChecker::ChangeNumber speller_change_number = 0;
4209 if (theSpellChecker())
4210 speller_change_number = theSpellChecker()->changeNumber();
4211 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4212 d->speller_state_.needsCompleteRefresh(speller_change_number);
4214 return d->needsSpellCheck();
4218 bool Paragraph::Private::ignoreWord(docstring const & word) const
4220 // Ignore words with digits
4221 // FIXME: make this customizable
4222 // (note that some checkers ignore words with digits by default)
4223 docstring::const_iterator cit = word.begin();
4224 docstring::const_iterator const end = word.end();
4225 for (; cit != end; ++cit) {
4226 if (isNumber((*cit)))
4233 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4234 WordLangTuple & wl, docstring_list & suggestions,
4235 bool do_suggestion, bool check_learned) const
4237 SpellChecker::Result result = SpellChecker::WORD_OK;
4238 SpellChecker * speller = theSpellChecker();
4242 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4245 locateWord(from, to, WHOLE_WORD, true);
4246 if (from == to || from >= size())
4249 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4250 Language * lang = d->getSpellLanguage(from);
4252 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4255 wl = WordLangTuple(word, lang);
4260 if (needsSpellCheck() || check_learned) {
4262 if (!d->ignoreWord(word)) {
4263 bool const trailing_dot = to < size() && d->text_[to] == '.';
4264 result = speller->check(wl);
4265 if (SpellChecker::misspelled(result) && trailing_dot) {
4266 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4267 result = speller->check(wl);
4268 if (!SpellChecker::misspelled(result)) {
4269 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4271 from << ".." << to << "]");
4273 // spell check with dot appended failed too
4274 // restore original word/lang value
4275 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4276 wl = WordLangTuple(word, lang);
4280 if (!SpellChecker::misspelled(result)) {
4281 // area up to the begin of the next word is not misspelled
4282 while (end < size() && isWordSeparator(end))
4285 d->setMisspelled(from, end, result);
4287 result = d->speller_state_.getState(from);
4291 suggestions.clear();
4293 if (SpellChecker::misspelled(result)) {
4294 LYXERR(Debug::GUI, "misspelled word: \"" <<
4296 from << ".." << to << "]");
4298 speller->suggest(wl, suggestions);
4304 void Paragraph::anonymize()
4306 // This is a very crude anonymization for now
4307 for (char_type & c : d->text_)
4308 if (isLetterChar(c) || isNumber(c))
4313 void Paragraph::Private::markMisspelledWords(
4314 pos_type const & first, pos_type const & last,
4315 SpellChecker::Result result,
4316 docstring const & word,
4317 SkipPositions const & skips)
4319 if (!SpellChecker::misspelled(result)) {
4320 setMisspelled(first, last, SpellChecker::WORD_OK);
4324 SpellChecker * speller = theSpellChecker();
4325 // locate and enumerate the error positions
4326 int nerrors = speller->numMisspelledWords();
4328 SkipPositionsIterator it = skips.begin();
4329 SkipPositionsIterator et = skips.end();
4330 for (int index = 0; index < nerrors; ++index) {
4333 speller->misspelledWord(index, wstart, wlen);
4334 /// should not happen if speller supports range checks
4335 if (!wlen) continue;
4336 docstring const misspelled = word.substr(wstart, wlen);
4337 wstart += first + numskipped;
4338 if (snext < wstart) {
4339 /// mark the range of correct spelling
4340 numskipped += countSkips(it, et, wstart);
4341 setMisspelled(snext,
4342 wstart - 1, SpellChecker::WORD_OK);
4344 snext = wstart + wlen;
4345 numskipped += countSkips(it, et, snext);
4346 /// mark the range of misspelling
4347 setMisspelled(wstart, snext, result);
4348 LYXERR(Debug::GUI, "misspelled word: \"" <<
4349 misspelled << "\" [" <<
4350 wstart << ".." << (snext-1) << "]");
4353 if (snext <= last) {
4354 /// mark the range of correct spelling at end
4355 setMisspelled(snext, last, SpellChecker::WORD_OK);
4360 void Paragraph::spellCheck() const
4362 SpellChecker * speller = theSpellChecker();
4363 if (!speller || empty() ||!needsSpellCheck())
4367 d->rangeOfSpellCheck(start, endpos);
4368 if (speller->canCheckParagraph()) {
4369 // loop until we leave the range
4370 for (pos_type first = start; first < endpos; ) {
4371 pos_type last = endpos;
4372 Private::SkipPositions skips;
4373 Language * lang = d->locateSpellRange(first, last, skips);
4374 if (first >= endpos)
4376 // start the spell checker on the unit of meaning
4377 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4378 WordLangTuple wl = WordLangTuple(word, lang);
4379 SpellChecker::Result result = word.size() ?
4380 speller->check(wl) : SpellChecker::WORD_OK;
4381 d->markMisspelledWords(first, last, result, word, skips);
4385 static docstring_list suggestions;
4386 pos_type to = endpos;
4387 while (start < endpos) {
4389 spellCheck(start, to, wl, suggestions, false);
4393 d->readySpellCheck();
4397 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4399 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4400 if (result || pos <= 0 || pos > size())
4402 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4403 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4408 string Paragraph::magicLabel() const
4411 ss << "magicparlabel-" << id();