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 /// Output consecutive unicode chars, belonging to the same script as
333 /// specified by the latex macro \p ltx, to \p os starting from \p i.
334 /// \return the number of characters written.
335 int writeScriptChars(BufferParams const &, OutputParams const &,
337 docstring const & ltx,
338 Change const &, Encoding const &,
339 std::string const, pos_type & i);
341 /// This could go to ParagraphParameters if we want to.
342 int startTeXParParams(BufferParams const &, otexstream &,
343 OutputParams const &) const;
345 /// This could go to ParagraphParameters if we want to.
346 bool endTeXParParams(BufferParams const &, otexstream &,
347 OutputParams const &) const;
350 void latexInset(BufferParams const &,
355 Font const & outerfont,
357 Change & running_change,
358 Layout const & style,
360 unsigned int & column);
363 void latexSpecialChar(
365 BufferParams const & bparams,
366 OutputParams const & runparams,
367 Font const & running_font,
368 Change const & running_change,
369 Layout const & style,
372 unsigned int & column);
379 unsigned int & column);
385 unsigned int & column);
391 unsigned int & column);
394 void validate(LaTeXFeatures & features) const;
396 /// Checks if the paragraph contains only text and no inset or font change.
397 bool onlyText(Buffer const & buf, Font const & outerfont,
398 pos_type initial) const;
400 /// a vector of speller skip positions
401 typedef vector<FontSpan> SkipPositions;
402 typedef SkipPositions::const_iterator SkipPositionsIterator;
404 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
406 Language * getSpellLanguage(pos_type const from) const;
408 Language * locateSpellRange(pos_type & from, pos_type & to,
409 SkipPositions & skips) const;
411 bool hasSpellerChange() const
413 SpellChecker::ChangeNumber speller_change_number = 0;
414 if (theSpellChecker())
415 speller_change_number = theSpellChecker()->changeNumber();
416 return speller_change_number > speller_state_.currentChangeNumber();
419 bool ignoreWord(docstring const & word) const ;
421 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
423 pos_type textsize = owner_->size();
424 // check for sane arguments
425 if (to <= from || from >= textsize)
427 FontSpan fp = FontSpan(from, to - 1);
428 speller_state_.setRange(fp, state);
431 void requestSpellCheck(pos_type pos)
434 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
436 speller_state_.needsRefresh(pos);
439 void readySpellCheck()
441 speller_state_.needsRefresh(-1);
444 bool needsSpellCheck() const
446 return speller_state_.needsRefresh();
449 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
451 speller_state_.refreshRange(first, last);
453 last = owner_->size();
456 pos_type endpos = last;
457 owner_->locateWord(first, endpos, WHOLE_WORD);
460 owner_->locateWord(last, endpos, WHOLE_WORD);
465 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
469 while (it != et && it->first < start) {
470 int skip = it->last - it->first + 1;
478 void markMisspelledWords(pos_type const & first, pos_type const & last,
479 SpellChecker::Result result,
480 docstring const & word,
481 SkipPositions const & skips);
483 InsetCode ownerCode() const
485 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
488 /// Which Paragraph owns us?
492 Inset const * inset_owner_;
501 ParagraphParameters params_;
503 /// for recording and looking up changes
507 InsetList insetlist_;
510 pos_type begin_of_body_;
512 typedef docstring TextContainer;
516 typedef set<docstring> Words;
517 typedef map<string, Words> LangWordsMap;
521 Layout const * layout_;
523 SpellCheckerState speller_state_;
527 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
528 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
535 int Paragraph::Private::make_id()
537 // The id is unique per session across buffers because it is used in
538 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
541 static atomic_uint next_id(0);
546 Paragraph::Private::Private(Private const & p, Paragraph * owner)
547 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
549 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
550 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
553 requestSpellCheck(p.text_.size());
557 Paragraph::Private::Private(Private const & p, Paragraph * owner,
558 pos_type beg, pos_type end)
559 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
560 params_(p.params_), changes_(p.changes_),
561 insetlist_(p.insetlist_, beg, end),
562 begin_of_body_(p.begin_of_body_), words_(p.words_),
565 if (beg >= pos_type(p.text_.size()))
567 text_ = p.text_.substr(beg, end - beg);
569 FontList::const_iterator fcit = fontlist_.begin();
570 FontList::const_iterator fend = fontlist_.end();
571 for (; fcit != fend; ++fcit) {
572 if (fcit->pos() < beg)
574 if (fcit->pos() >= end) {
575 // Add last entry in the fontlist_.
576 fontlist_.set(text_.size() - 1, fcit->font());
579 // Add a new entry in the fontlist_.
580 fontlist_.set(fcit->pos() - beg, fcit->font());
582 requestSpellCheck(p.text_.size());
586 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
587 bool output_active, TocBackend & backend) const
589 d->changes_.addToToc(cdit, buf, output_active, backend);
593 void Paragraph::addChangesToBuffer(Buffer const & buf) const
595 d->changes_.updateBuffer(buf);
599 bool Paragraph::isChangeUpdateRequired() const
601 return d->changes_.isUpdateRequired();
605 bool Paragraph::isDeleted(pos_type start, pos_type end) const
607 LASSERT(start >= 0 && start <= size(), return false);
608 LASSERT(end > start && end <= size() + 1, return false);
610 return d->changes_.isDeleted(start, end);
614 bool Paragraph::isChanged(pos_type start, pos_type end) const
616 LASSERT(start >= 0 && start <= size(), return false);
617 LASSERT(end > start && end <= size() + 1, return false);
619 return d->changes_.isChanged(start, end);
623 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
625 // keep the logic here in sync with the logic of eraseChars()
629 Change const & change = d->changes_.lookup(size());
630 return change.inserted() && change.currentAuthor();
634 void Paragraph::setChange(Change const & change)
636 // beware of the imaginary end-of-par character!
637 d->changes_.set(change, 0, size() + 1);
640 * Propagate the change recursively - but not in case of DELETED!
642 * Imagine that your co-author makes changes in an existing inset. He
643 * sends your document to you and you come to the conclusion that the
644 * inset should go completely. If you erase it, LyX must not delete all
645 * text within the inset. Otherwise, the change tracked insertions of
646 * your co-author get lost and there is no way to restore them later.
648 * Conclusion: An inset's content should remain untouched if you delete it
651 if (!change.deleted()) {
652 for (pos_type pos = 0; pos < size(); ++pos) {
653 if (Inset * inset = getInset(pos))
654 inset->setChange(change);
660 void Paragraph::setChange(pos_type pos, Change const & change)
662 LASSERT(pos >= 0 && pos <= size(), return);
663 d->changes_.set(change, pos);
665 // see comment in setChange(Change const &) above
666 if (!change.deleted() && pos < size())
667 if (Inset * inset = getInset(pos))
668 inset->setChange(change);
672 Change const & Paragraph::lookupChange(pos_type pos) const
674 LBUFERR(pos >= 0 && pos <= size());
675 return d->changes_.lookup(pos);
679 void Paragraph::acceptChanges(pos_type start, pos_type end)
681 LASSERT(start >= 0 && start <= size(), return);
682 LASSERT(end > start && end <= size() + 1, return);
684 for (pos_type pos = start; pos < end; ++pos) {
685 switch (lookupChange(pos).type) {
686 case Change::UNCHANGED:
687 // accept changes in nested inset
688 if (Inset * inset = getInset(pos))
689 inset->acceptChanges();
692 case Change::INSERTED:
693 d->changes_.set(Change(Change::UNCHANGED), pos);
694 // also accept changes in nested inset
695 if (Inset * inset = getInset(pos))
696 inset->acceptChanges();
699 case Change::DELETED:
700 // Suppress access to non-existent
701 // "end-of-paragraph char"
703 eraseChar(pos, false);
714 void Paragraph::rejectChanges(pos_type start, pos_type end)
716 LASSERT(start >= 0 && start <= size(), return);
717 LASSERT(end > start && end <= size() + 1, return);
719 for (pos_type pos = start; pos < end; ++pos) {
720 switch (lookupChange(pos).type) {
721 case Change::UNCHANGED:
722 // reject changes in nested inset
723 if (Inset * inset = getInset(pos))
724 inset->rejectChanges();
727 case Change::INSERTED:
728 // Suppress access to non-existent
729 // "end-of-paragraph char"
731 eraseChar(pos, false);
737 case Change::DELETED:
738 d->changes_.set(Change(Change::UNCHANGED), pos);
740 // Do NOT reject changes within a deleted inset!
741 // There may be insertions of a co-author inside of it!
749 void Paragraph::Private::insertChar(pos_type pos, char_type c,
750 Change const & change)
752 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
755 changes_.insert(change, pos);
757 // This is actually very common when parsing buffers (and
758 // maybe inserting ascii text)
759 if (pos == pos_type(text_.size())) {
760 // when appending characters, no need to update tables
762 // but we want spell checking
763 requestSpellCheck(pos);
767 text_.insert(text_.begin() + pos, c);
769 // Update the font table.
770 fontlist_.increasePosAfterPos(pos);
773 insetlist_.increasePosAfterPos(pos);
775 // Update list of misspelled positions
776 speller_state_.increasePosAfterPos(pos);
780 bool Paragraph::insertInset(pos_type pos, Inset * inset,
781 Font const & font, Change const & change)
783 LASSERT(inset, return false);
784 LASSERT(pos >= 0 && pos <= size(), return false);
786 // Paragraph::insertInset() can be used in cut/copy/paste operation where
787 // d->inset_owner_ is not set yet.
788 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
791 d->insertChar(pos, META_INSET, change);
792 LASSERT(d->text_[pos] == META_INSET, return false);
794 // Add a new entry in the insetlist_.
795 d->insetlist_.insert(inset, pos);
797 // Some insets require run of spell checker
798 requestSpellCheck(pos);
804 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
806 LASSERT(pos >= 0 && pos <= size(), return false);
808 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
811 Change change = d->changes_.lookup(pos);
813 // set the character to DELETED if
814 // a) it was previously unchanged or
815 // b) it was inserted by a co-author
817 if (!change.changed() ||
818 (change.inserted() && !change.currentAuthor())) {
819 setChange(pos, Change(Change::DELETED));
820 // request run of spell checker
821 requestSpellCheck(pos);
825 if (change.deleted())
829 // Don't physically access the imaginary end-of-paragraph character.
830 // eraseChar() can only mark it as DELETED. A physical deletion of
831 // end-of-par must be handled externally.
837 d->changes_.erase(pos);
839 // if it is an inset, delete the inset entry
840 if (d->text_[pos] == META_INSET)
841 d->insetlist_.erase(pos);
843 d->text_.erase(d->text_.begin() + pos);
845 // Update the fontlist_
846 d->fontlist_.erase(pos);
848 // Update the insetlist_
849 d->insetlist_.decreasePosAfterPos(pos);
851 // Update list of misspelled positions
852 d->speller_state_.decreasePosAfterPos(pos);
853 d->speller_state_.refreshLast(size());
859 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
861 LASSERT(start >= 0 && start <= size(), return 0);
862 LASSERT(end >= start && end <= size() + 1, return 0);
865 for (pos_type count = end - start; count; --count) {
866 if (!eraseChar(i, trackChanges))
873 int Paragraph::Private::latexSurrogatePair(BufferParams const & bparams,
874 otexstream & os, char_type c, char_type next,
875 OutputParams const & runparams)
877 // Writing next here may circumvent a possible font change between
878 // c and next. Since next is only output if it forms a surrogate pair
879 // with c we can ignore this:
880 // A font change inside a surrogate pair does not make sense and is
881 // hopefully impossible to input.
882 // FIXME: change tracking
883 // Is this correct WRT change tracking?
884 Encoding const & encoding = *(runparams.encoding);
885 docstring latex1 = encoding.latexChar(next).first;
886 if (runparams.inIPA) {
887 string const tipashortcut = Encodings::TIPAShortcut(next);
888 if (!tipashortcut.empty()) {
889 latex1 = from_ascii(tipashortcut);
892 docstring latex2 = encoding.latexChar(c).first;
894 if (docstring(1, next) == latex1) {
895 // The encoding supports the combination:
896 // output as is (combining char after base char).
897 os << latex2 << latex1;
898 return latex1.length() + latex2.length();
901 // Handle combining characters in "script" context (i.e., \textgreek and \textcyrillic)
902 docstring::size_type const brace1 = latex2.find_first_of(from_ascii("{"));
903 docstring::size_type const brace2 = latex2.find_last_of(from_ascii("}"));
904 string script = to_ascii(latex2.substr(1, brace1 - 1));
905 // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
906 // whether they are encodable or not (it only depends on the font encoding)
907 if (!runparams.isFullUnicode())
908 // This will get us a script value to deal with below
909 Encodings::isKnownScriptChar(c, script);
913 if (runparams.local_font)
914 fontenc = runparams.local_font->language()->fontenc(bparams);
916 fontenc = bparams.language->fontenc(bparams);
917 docstring scriptmacro;
919 if (script == "textgreek" || script == "textcyrillic") {
920 // We separate the script macro (\text[greek|cyr]) from the rest,
921 // since we need to include the combining char in it (#6463).
922 // This is "the rest":
925 latex2 = latex2.substr(pos, length);
926 // We only need the script macro with non-native font encodings
927 if (Encodings::needsScriptWrapper(script, fontenc)) {
928 scriptmacro = from_ascii("\\" + script + "{");
929 cb = from_ascii("}");
935 // polutonikogreek does not play nice with brackets
936 if (!runparams.local_font
937 || runparams.local_font->language()->lang() != "polutonikogreek") {
938 lb = from_ascii("{");
939 rb = from_ascii("}");
942 os << scriptmacro << latex1 << lb << latex2 << rb << cb;
943 return latex1.length() + latex2.length() + lb.length() + rb.length() + cb.length();
947 bool Paragraph::Private::simpleTeXBlanks(BufferParams const & bparams,
948 OutputParams const & runparams,
951 unsigned int & column,
953 Layout const & style)
955 if (style.pass_thru || runparams.pass_thru)
958 if (i + 1 < int(text_.size())) {
959 char_type next = text_[i + 1];
960 if (Encodings::isCombiningChar(next)) {
961 // This space has an accent, so we must always output it.
962 column += latexSurrogatePair(bparams, os, ' ', next, runparams) - 1;
967 if (runparams.linelen > 0
968 && column > runparams.linelen
970 && text_[i - 1] != ' '
971 && (i + 1 < int(text_.size()))
972 // same in FreeSpacing mode
973 && !owner_->isFreeSpacing()
974 // In typewriter mode, we want to avoid
975 // ! . ? : at the end of a line
976 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
977 && (text_[i - 1] == '.'
978 || text_[i - 1] == '?'
979 || text_[i - 1] == ':'
980 || text_[i - 1] == '!'))) {
982 os.texrow().start(owner_->id(), i + 1);
984 } else if (style.free_spacing) {
993 int Paragraph::Private::writeScriptChars(BufferParams const & bparams,
994 OutputParams const & runparams,
996 docstring const & ltx,
997 Change const & runningChange,
998 Encoding const & encoding,
999 string const fontenc,
1002 // FIXME: modifying i here is not very nice...
1004 // We only arrive here when character text_[i] could not be translated
1005 // into the current latex encoding (or its latex translation has been forced,)
1006 // and it belongs to a known script.
1007 // Parameter ltx contains the latex translation of text_[i] as specified
1008 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
1009 // The latex macro name "textXXX" specifies the script to which text_[i]
1010 // belongs and we use it in order to check whether characters from the
1011 // same script immediately follow, such that we can collect them in a
1012 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
1013 // for the first char but only "<spec>" for all subsequent chars.
1014 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
1015 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
1016 string script = to_ascii(ltx.substr(1, brace1 - 1));
1018 int length = brace2;
1019 bool closing_brace = true;
1020 if (!Encodings::needsScriptWrapper(script, fontenc)) {
1021 // Correct font encoding is being used, so we can avoid \text[greek|cyr].
1024 closing_brace = false;
1026 os << ltx.substr(pos, length);
1027 int size = text_.size();
1028 while (i + 1 < size) {
1029 char_type const next = text_[i + 1];
1030 // Stop here if next character belongs to another script
1031 // or there is a change in change tracking status.
1032 if (!Encodings::isKnownScriptChar(next, script) ||
1033 runningChange != owner_->lookupChange(i + 1))
1037 FontList::const_iterator cit = fontlist_.begin();
1038 FontList::const_iterator end = fontlist_.end();
1039 for (; cit != end; ++cit) {
1040 if (cit->pos() >= i && !found) {
1041 prev_font = cit->font();
1044 if (cit->pos() >= i + 1)
1047 // Stop here if there is a font attribute or encoding change.
1048 if (found && cit != end && prev_font != cit->font())
1051 // Check whether we have a combining pair
1052 char_type next_next = '\0';
1054 next_next = text_[i + 2];
1055 if (Encodings::isCombiningChar(next_next)) {
1056 length += latexSurrogatePair(bparams, os, next, next_next, runparams) - 1;
1062 docstring const latex = encoding.latexChar(next).first;
1063 docstring::size_type const b1 =
1064 latex.find_first_of(from_ascii("{"));
1065 docstring::size_type const b2 =
1066 latex.find_last_of(from_ascii("}"));
1067 int const len = b2 - b1 - 1;
1068 os << latex.substr(b1 + 1, len);
1072 if (closing_brace) {
1080 void Paragraph::Private::latexInset(BufferParams const & bparams,
1082 OutputParams & runparams,
1083 Font & running_font,
1085 Font const & outerfont,
1087 Change & running_change,
1088 Layout const & style,
1090 unsigned int & column)
1092 Inset * inset = owner_->getInset(i);
1095 if (style.pass_thru) {
1096 odocstringstream ods;
1097 inset->plaintext(ods, runparams);
1102 // FIXME: move this to InsetNewline::latex
1103 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1104 // newlines are handled differently here than
1105 // the default in simpleTeXSpecialChars().
1106 if (!style.newline_allowed) {
1110 bool needPar = false;
1111 column += running_font.latexWriteEndChanges(
1112 os, bparams, runparams,
1113 basefont, basefont, needPar);
1117 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1120 basefont = owner_->getLayoutFont(bparams, outerfont);
1121 running_font = basefont;
1123 if (runparams.moving_arg)
1127 os.texrow().start(owner_->id(), i + 1);
1131 if (owner_->isDeleted(i)) {
1132 if( ++runparams.inDeletedInset == 1)
1133 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1136 if (inset->canTrackChanges()) {
1137 column += Changes::latexMarkChange(os, bparams, running_change,
1138 Change(Change::UNCHANGED), runparams);
1139 running_change = Change(Change::UNCHANGED);
1143 odocstream::pos_type const len = os.os().tellp();
1145 if (inset->forceLTR()
1146 && !runparams.use_polyglossia
1147 && running_font.isRightToLeft()
1148 // ERT is an exception, it should be output with no
1149 // decorations at all
1150 && inset->lyxCode() != ERT_CODE) {
1151 if (running_font.language()->lang() == "farsi")
1152 os << "\\beginL" << termcmd;
1158 // FIXME: Bug: we can have an empty font change here!
1159 // if there has just been a font change, we are going to close it
1160 // right now, which means stupid latex code like \textsf{}. AFAIK,
1161 // this does not harm dvi output. A minor bug, thus (JMarc)
1163 // Some insets cannot be inside a font change command.
1164 // However, even such insets *can* be placed in \L or \R
1165 // or their equivalents (for RTL language switches), so we don't
1166 // close the language in those cases.
1167 // ArabTeX, though, cannot handle this special behavior, it seems.
1168 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1169 || running_font.language()->lang() == "arabic_arabtex";
1170 if (open_font && !inset->inheritFont()) {
1171 bool needPar = false;
1172 bool closeLanguage = arabtex
1173 || basefont.isRightToLeft() == running_font.isRightToLeft();
1174 unsigned int count = running_font.latexWriteEndChanges(os,
1175 bparams, runparams, basefont, basefont,
1176 needPar, closeLanguage);
1178 // if any font properties were closed, update the running_font,
1179 // making sure, however, to leave the language as it was
1181 // FIXME: probably a better way to keep track of the old
1182 // language, than copying the entire font?
1183 Font const copy_font(running_font);
1184 basefont = owner_->getLayoutFont(bparams, outerfont);
1185 running_font = basefont;
1187 running_font.setLanguage(copy_font.language());
1188 // leave font open if language is still open
1189 open_font = (running_font.language() == basefont.language());
1191 runparams.local_font = &basefont;
1195 size_t const previous_row_count = os.texrow().rows();
1198 runparams.lastid = id_;
1199 runparams.lastpos = i;
1200 inset->latex(os, runparams);
1201 } catch (EncodingException & e) {
1202 // add location information and throw again.
1209 if (running_font.language()->lang() == "farsi")
1210 os << "\\endL" << termcmd;
1215 if (os.texrow().rows() > previous_row_count) {
1216 os.texrow().start(owner_->id(), i + 1);
1219 column += (unsigned int)(os.os().tellp() - len);
1222 if (owner_->isDeleted(i))
1223 --runparams.inDeletedInset;
1227 void Paragraph::Private::latexSpecialChar(otexstream & os,
1228 BufferParams const & bparams,
1229 OutputParams const & runparams,
1230 Font const & running_font,
1231 Change const & running_change,
1232 Layout const & style,
1235 unsigned int & column)
1237 char_type const c = owner_->getUChar(bparams, runparams, i);
1239 if (style.pass_thru || runparams.pass_thru
1240 || contains(style.pass_thru_chars, c)
1241 || contains(runparams.pass_thru_chars, c)) {
1243 Encoding const * const enc = runparams.encoding;
1244 if (enc && !enc->encodable(c))
1245 throw EncodingException(c);
1251 // TIPA uses its own T3 encoding
1252 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1254 // If T1 font encoding is used, use the special
1255 // characters it provides.
1256 // NOTE: Some languages reset the font encoding internally to a
1257 // non-standard font encoding. If we are using such a language,
1258 // we do not output special T1 chars.
1259 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1260 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1261 && latexSpecialT1(c, os, i, column))
1263 // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
1264 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1265 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1268 // Otherwise, we use what LaTeX provides us.
1271 os << "\\textbackslash" << termcmd;
1275 os << "\\textless" << termcmd;
1279 os << "\\textgreater" << termcmd;
1283 os << "\\textbar" << termcmd;
1288 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1289 (end_pos == -1 || i + 1 < end_pos) &&
1290 text_[i+1] == '-') {
1291 // Prevent "--" becoming an en dash and "---" an em dash.
1292 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1298 os << "\\textquotedbl" << termcmd;
1303 case '%': case '#': case '{':
1311 os << "\\textasciitilde" << termcmd;
1316 os << "\\textasciicircum" << termcmd;
1323 // avoid being mistaken for optional arguments
1331 // Blanks are printed before font switching.
1332 // Sure? I am not! (try nice-latex)
1333 // I am sure it's correct. LyX might be smarter
1334 // in the future, but for now, nothing wrong is
1340 // XeTeX's dash behaviour is determined via a global setting
1341 if (bparams.use_dash_ligatures
1342 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1344 // TODO #10961: && not in inset Flex Code
1345 // TODO #10961: && not in layout LyXCode
1346 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1363 Encoding const & encoding = *(runparams.encoding);
1364 char_type next = '\0';
1365 if (i + 1 < int(text_.size())) {
1366 next = text_[i + 1];
1367 if (Encodings::isCombiningChar(next)) {
1368 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1374 pair<docstring, bool> latex = encoding.latexChar(c);
1375 docstring nextlatex;
1376 bool nexttipas = false;
1377 string nexttipashortcut;
1378 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1379 nextlatex = encoding.latexChar(next).first;
1380 if (runparams.inIPA) {
1381 nexttipashortcut = Encodings::TIPAShortcut(next);
1382 nexttipas = !nexttipashortcut.empty();
1386 if (runparams.inIPA) {
1387 string const tipashortcut = Encodings::TIPAShortcut(c);
1388 if (!tipashortcut.empty()) {
1389 latex.first = from_ascii(tipashortcut);
1390 latex.second = false;
1395 fontenc = running_font.language()->fontenc(bparams);
1396 // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
1397 // whether they are encodable or not (it only depends on the font encoding)
1398 if (!runparams.isFullUnicode() && Encodings::isKnownScriptChar(c, script)) {
1399 docstring const wrapper = from_ascii("\\" + script + "{");
1400 docstring ltx = latex.first;
1401 if (!prefixIs(ltx, wrapper))
1402 ltx = wrapper + latex.first + from_ascii("}");
1403 column += writeScriptChars(bparams, runparams, os, ltx, running_change,
1404 encoding, fontenc, i) - 1;
1405 } else if (latex.second
1406 && ((!prefixIs(nextlatex, '\\')
1407 && !prefixIs(nextlatex, '{')
1408 && !prefixIs(nextlatex, '}'))
1410 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1412 // Prevent eating of a following
1413 // space or command corruption by
1414 // following characters
1415 if (next == ' ' || next == '\0') {
1416 column += latex.first.length() + 1;
1417 os << latex.first << "{}";
1419 column += latex.first.length();
1420 os << latex.first << " ";
1423 column += latex.first.length() - 1;
1431 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1432 pos_type i, unsigned int & column)
1438 // In T1 encoding, these characters exist
1439 // but we should avoid ligatures
1440 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1442 os << "\\textcompwordmark" << termcmd;
1449 // soul.sty breaks with \char`\"
1450 os << "\\textquotedbl" << termcmd;
1459 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1460 pos_type i, unsigned int & column)
1462 // TU encoding is currently on par with T1.
1463 return latexSpecialT1(c, os, i, column);
1467 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1468 pos_type /*i*/, unsigned int & column)
1478 os << "\\textvertline" << termcmd;
1487 void Paragraph::Private::validate(LaTeXFeatures & features) const
1489 if (layout_->inpreamble && inset_owner_) {
1490 // FIXME: Using a string stream here circumvents the encoding
1491 // switching machinery of odocstream. Therefore the
1492 // output is wrong if this paragraph contains content
1493 // that needs to switch encoding.
1494 Buffer const & buf = inset_owner_->buffer();
1495 otexstringstream os;
1496 os << layout_->preamble();
1497 size_t const length = os.length();
1498 TeXOnePar(buf, buf.text(), buf.getParFromID(owner_->id()).pit(), os,
1499 features.runparams(), string(), 0, -1, true);
1500 if (os.length() > length)
1501 features.addPreambleSnippet(os.release(), true);
1504 if (features.runparams().flavor == OutputParams::HTML
1505 && layout_->htmltitle()) {
1506 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1509 // check the params.
1510 if (!params_.spacing().isDefault())
1511 features.require("setspace");
1514 features.useLayout(layout_->name());
1517 fontlist_.validate(features);
1519 // then the indentation
1520 if (!params_.leftIndent().zero())
1521 features.require("ParagraphLeftIndent");
1524 InsetList::const_iterator icit = insetlist_.begin();
1525 InsetList::const_iterator iend = insetlist_.end();
1526 for (; icit != iend; ++icit) {
1528 features.inDeletedInset(owner_->isDeleted(icit->pos));
1529 if (icit->inset->lyxCode() == FOOT_CODE) {
1530 // FIXME: an item inset would make things much easier.
1531 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1532 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1533 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1534 && (icit->pos < begin_of_body_
1535 || (icit->pos == begin_of_body_
1536 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1537 features.saveNoteEnv("description");
1539 icit->inset->validate(features);
1540 features.inDeletedInset(false);
1541 if (layout_->needprotect &&
1542 icit->inset->lyxCode() == FOOT_CODE)
1543 features.require("NeedLyXFootnoteCode");
1547 // then the contents
1548 BufferParams const bp = features.runparams().is_child
1549 ? features.buffer().masterParams() : features.buffer().params();
1550 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1551 char_type c = text_[i];
1552 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1554 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1555 features.require("textquotedblp");
1556 else if (bp.main_font_encoding() != "T1"
1557 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1558 features.require("textquotedbl");
1559 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1560 // features that depend on the font or input encoding
1561 string feats = ci.textpreamble();
1562 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1563 if (fontenc.empty())
1564 fontenc = features.runparams().main_fontenc;
1565 while (!feats.empty()) {
1567 feats = split(feats, feat, ',');
1568 if (contains(feat, "!=")) {
1569 // a feature that is required except for the spcified
1570 // font or input encodings
1572 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1573 // multiple encodings are separated by semicolon
1574 vector<string> context = getVectorFromString(contexts, ";");
1575 // require feature if the context matches neither current font
1576 // nor input encoding
1577 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1578 && std::find(context.begin(), context.end(),
1579 features.runparams().encoding->name()) == context.end())
1580 features.require(realfeature);
1581 } else if (contains(feat, '=')) {
1582 // a feature that is required only for the spcified
1583 // font or input encodings
1585 string const contexts = split(feat, realfeature, '=');
1586 // multiple encodings are separated by semicolon
1587 vector<string> context = getVectorFromString(contexts, ";");
1588 // require feature if the context matches either current font
1589 // or input encoding
1590 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1591 || std::find(context.begin(), context.end(),
1592 features.runparams().encoding->name()) != context.end())
1593 features.require(realfeature);
1596 } else if (!bp.use_dash_ligatures
1597 && (c == 0x2013 || c == 0x2014)
1598 && bp.useNonTeXFonts
1599 && features.runparams().flavor == OutputParams::XETEX)
1600 // XeTeX's dash behaviour is determined via a global setting
1601 features.require("xetexdashbreakstate");
1602 BufferEncodings::validate(c, features);
1606 /////////////////////////////////////////////////////////////////////
1610 /////////////////////////////////////////////////////////////////////
1613 Layout const emptyParagraphLayout;
1616 Paragraph::Paragraph()
1617 : d(new Paragraph::Private(this, emptyParagraphLayout))
1624 Paragraph::Paragraph(Paragraph const & par)
1625 : itemdepth(par.itemdepth),
1626 d(new Paragraph::Private(*par.d, this))
1632 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1633 : itemdepth(par.itemdepth),
1634 d(new Paragraph::Private(*par.d, this, beg, end))
1640 Paragraph & Paragraph::operator=(Paragraph const & par)
1642 // needed as we will destroy the private part before copying it
1644 itemdepth = par.itemdepth;
1648 d = new Private(*par.d, this);
1655 Paragraph::~Paragraph()
1664 // this shall be called just before every "os << ..." action.
1665 void flushString(ostream & os, docstring & s)
1674 void Paragraph::write(ostream & os, BufferParams const & bparams,
1675 depth_type & dth) const
1677 // The beginning or end of a deeper (i.e. nested) area?
1678 if (dth != d->params_.depth()) {
1679 if (d->params_.depth() > dth) {
1680 while (d->params_.depth() > dth) {
1681 os << "\n\\begin_deeper";
1685 while (d->params_.depth() < dth) {
1686 os << "\n\\end_deeper";
1692 // First write the layout
1693 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1695 d->params_.write(os);
1697 Font font1(inherit_font, bparams.language);
1699 Change running_change = Change(Change::UNCHANGED);
1701 // this string is used as a buffer to avoid repetitive calls
1702 // to to_utf8(), which turn out to be expensive (JMarc)
1703 docstring write_buffer;
1706 for (pos_type i = 0; i <= size(); ++i) {
1708 Change const & change = lookupChange(i);
1709 if (change != running_change)
1710 flushString(os, write_buffer);
1711 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1712 running_change = change;
1717 // Write font changes
1718 Font font2 = getFontSettings(bparams, i);
1719 if (font2 != font1) {
1720 flushString(os, write_buffer);
1721 font2.lyxWriteChanges(font1, os);
1726 char_type const c = d->text_[i];
1729 if (Inset const * inset = getInset(i)) {
1730 flushString(os, write_buffer);
1731 if (inset->directWrite()) {
1732 // international char, let it write
1733 // code directly so it's shorter in
1739 os << "\\begin_inset ";
1741 os << "\n\\end_inset\n\n";
1744 // FIXME This can be removed again once the mystery
1745 // crash has been resolved.
1750 flushString(os, write_buffer);
1751 os << "\n\\backslash\n";
1755 flushString(os, write_buffer);
1756 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1763 if ((column > 70 && c == ' ')
1765 flushString(os, write_buffer);
1769 // this check is to amend a bug. LyX sometimes
1770 // inserts '\0' this could cause problems.
1772 write_buffer.push_back(c);
1774 LYXERR0("NUL char in structure.");
1780 flushString(os, write_buffer);
1781 os << "\n\\end_layout\n";
1782 // FIXME This can be removed again once the mystery
1783 // crash has been resolved.
1788 void Paragraph::validate(LaTeXFeatures & features) const
1790 d->validate(features);
1791 bool fragile = features.runparams().moving_arg;
1792 fragile |= layout().needprotect;
1793 if (needsCProtection(fragile))
1794 features.require("cprotect");
1798 void Paragraph::insert(pos_type start, docstring const & str,
1799 Font const & font, Change const & change)
1801 for (size_t i = 0, n = str.size(); i != n ; ++i)
1802 insertChar(start + i, str[i], font, change);
1806 void Paragraph::appendChar(char_type c, Font const & font,
1807 Change const & change)
1810 d->changes_.insert(change, d->text_.size());
1811 // when appending characters, no need to update tables
1812 d->text_.push_back(c);
1813 setFont(d->text_.size() - 1, font);
1814 d->requestSpellCheck(d->text_.size() - 1);
1818 void Paragraph::appendString(docstring const & s, Font const & font,
1819 Change const & change)
1821 pos_type end = s.size();
1822 size_t oldsize = d->text_.size();
1823 size_t newsize = oldsize + end;
1824 size_t capacity = d->text_.capacity();
1825 if (newsize >= capacity)
1826 d->text_.reserve(max(capacity + 100, newsize));
1828 // when appending characters, no need to update tables
1831 // FIXME: Optimize this!
1832 for (size_t i = oldsize; i != newsize; ++i) {
1834 d->changes_.insert(change, i);
1835 d->requestSpellCheck(i);
1837 d->fontlist_.set(oldsize, font);
1838 d->fontlist_.set(newsize - 1, font);
1842 void Paragraph::insertChar(pos_type pos, char_type c,
1845 d->insertChar(pos, c, Change(trackChanges ?
1846 Change::INSERTED : Change::UNCHANGED));
1850 void Paragraph::insertChar(pos_type pos, char_type c,
1851 Font const & font, bool trackChanges)
1853 d->insertChar(pos, c, Change(trackChanges ?
1854 Change::INSERTED : Change::UNCHANGED));
1859 void Paragraph::insertChar(pos_type pos, char_type c,
1860 Font const & font, Change const & change)
1862 d->insertChar(pos, c, change);
1867 void Paragraph::resetFonts(Font const & font)
1869 d->fontlist_.clear();
1870 d->fontlist_.set(0, font);
1871 d->fontlist_.set(d->text_.size() - 1, font);
1874 // Gets uninstantiated font setting at position.
1875 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1879 LYXERR0("pos: " << pos << " size: " << size());
1883 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1884 if (cit != d->fontlist_.end())
1887 if (pos == size() && !empty())
1888 return getFontSettings(bparams, pos - 1);
1890 // Optimisation: avoid a full font instantiation if there is no
1891 // language change from previous call.
1892 static Font previous_font;
1893 static Language const * previous_lang = 0;
1894 Language const * lang = getParLanguage(bparams);
1895 if (lang != previous_lang) {
1896 previous_lang = lang;
1897 previous_font = Font(inherit_font, lang);
1899 return previous_font;
1903 FontSpan Paragraph::fontSpan(pos_type pos) const
1905 LBUFERR(pos <= size());
1908 return FontSpan(pos, pos);
1911 FontList::const_iterator cit = d->fontlist_.begin();
1912 FontList::const_iterator end = d->fontlist_.end();
1913 for (; cit != end; ++cit) {
1914 if (cit->pos() >= pos) {
1915 if (pos >= beginOfBody())
1916 return FontSpan(max(start, beginOfBody()),
1919 return FontSpan(start,
1920 min(beginOfBody() - 1,
1923 start = cit->pos() + 1;
1926 // This should not happen, but if so, we take no chances.
1927 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1928 LASSERT(false, return FontSpan(pos, pos));
1932 // Gets uninstantiated font setting at position 0
1933 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1935 if (!empty() && !d->fontlist_.empty())
1936 return d->fontlist_.begin()->font();
1938 // Optimisation: avoid a full font instantiation if there is no
1939 // language change from previous call.
1940 static Font previous_font;
1941 static Language const * previous_lang = 0;
1942 if (bparams.language != previous_lang) {
1943 previous_lang = bparams.language;
1944 previous_font = Font(inherit_font, bparams.language);
1947 return previous_font;
1951 // Gets the fully instantiated font at a given position in a paragraph
1952 // This is basically the same function as Text::GetFont() in text2.cpp.
1953 // The difference is that this one is used for generating the LaTeX file,
1954 // and thus cosmetic "improvements" are disallowed: This has to deliver
1955 // the true picture of the buffer. (Asger)
1956 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1957 Font const & outerfont) const
1961 Font font = getFontSettings(bparams, pos);
1963 pos_type const body_pos = beginOfBody();
1964 FontInfo & fi = font.fontInfo();
1966 fi.realize(d->layout_->labelfont);
1968 fi.realize(d->layout_->font);
1970 fi.realize(outerfont.fontInfo());
1971 fi.realize(bparams.getFont().fontInfo());
1977 Font const Paragraph::getLabelFont
1978 (BufferParams const & bparams, Font const & outerfont) const
1980 FontInfo tmpfont = d->layout_->labelfont;
1981 tmpfont.realize(outerfont.fontInfo());
1982 tmpfont.realize(bparams.getFont().fontInfo());
1983 return Font(tmpfont, getParLanguage(bparams));
1987 Font const Paragraph::getLayoutFont
1988 (BufferParams const & bparams, Font const & outerfont) const
1990 FontInfo tmpfont = d->layout_->font;
1991 tmpfont.realize(outerfont.fontInfo());
1992 tmpfont.realize(bparams.getFont().fontInfo());
1993 return Font(tmpfont, getParLanguage(bparams));
1997 char_type Paragraph::getUChar(BufferParams const & bparams,
1998 OutputParams const & rp,
2001 char_type c = d->text_[pos];
2003 // Return unchanged character in LTR languages.
2004 if (!getFontSettings(bparams, pos).isRightToLeft())
2007 // FIXME This is a complete mess due to all the language-specific
2008 // special cases. We need to unify this eventually, but this
2009 // requires a file format change and some thought.
2010 // We also need to unify the input of parentheses in different RTL
2011 // languages. Currently, some have their own methods (Arabic:
2012 // 18599/lyxsvn, Hebrew: e5f42f67d/lyxgit), some don't (Urdu, Syriac).
2013 // Also note that the representation in the LyX file is probably wrong
2014 // (see FIXME in TextMetrics::breakRow).
2015 // Most likely, we should simply rely on Qt's unicode handling here.
2016 string const & lang = getFontSettings(bparams, pos).language()->lang();
2018 // With polyglossia, brackets and stuff need not be reversed in RTL scripts
2019 // FIXME: The special casing for Hebrew parens is due to the special
2020 // handling on input (for Hebrew in e5f42f67d/lyxgit); see #8251.
2022 if (rp.use_polyglossia) {
2025 if (lang == "hebrew")
2029 if (lang == "hebrew")
2036 // In the following languages, brackets don't need to be reversed.
2037 // Furthermore, in arabic_arabi, they are transformed to Arabic
2038 // Ornate Parentheses (dunno if this is really wanted)
2039 bool const reversebrackets = lang != "arabic_arabtex"
2040 && lang != "arabic_arabi"
2045 if (reversebrackets)
2049 if (reversebrackets)
2070 void Paragraph::setFont(pos_type pos, Font const & font)
2072 LASSERT(pos <= size(), return);
2074 // First, reduce font against layout/label font
2075 // Update: The setCharFont() routine in text2.cpp already
2076 // reduces font, so we don't need to do that here. (Asger)
2078 d->fontlist_.set(pos, font);
2082 void Paragraph::makeSameLayout(Paragraph const & par)
2084 d->layout_ = par.d->layout_;
2085 d->params_ = par.d->params_;
2089 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2091 if (isFreeSpacing())
2097 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2098 if (eraseChar(pos, trackChanges))
2104 return count > 0 || pos > 0;
2108 bool Paragraph::hasSameLayout(Paragraph const & par) const
2110 return par.d->layout_ == d->layout_
2111 && d->params_.sameLayout(par.d->params_);
2115 depth_type Paragraph::getDepth() const
2117 return d->params_.depth();
2121 depth_type Paragraph::getMaxDepthAfter() const
2123 if (d->layout_->isEnvironment())
2124 return d->params_.depth() + 1;
2126 return d->params_.depth();
2130 LyXAlignment Paragraph::getAlign() const
2132 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2133 return d->layout_->align;
2135 return d->params_.align();
2139 docstring const & Paragraph::labelString() const
2141 return d->params_.labelString();
2145 // the next two functions are for the manual labels
2146 docstring const Paragraph::getLabelWidthString() const
2148 if (d->layout_->margintype == MARGIN_MANUAL
2149 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2150 return d->params_.labelWidthString();
2152 return _("Senseless with this layout!");
2156 void Paragraph::setLabelWidthString(docstring const & s)
2158 d->params_.labelWidthString(s);
2162 docstring Paragraph::expandLabel(Layout const & layout,
2163 BufferParams const & bparams) const
2165 return expandParagraphLabel(layout, bparams, true);
2169 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2170 BufferParams const & bparams) const
2172 return expandParagraphLabel(layout, bparams, false);
2176 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2177 BufferParams const & bparams, bool process_appendix) const
2179 DocumentClass const & tclass = bparams.documentClass();
2180 string const & lang = getParLanguage(bparams)->code();
2181 bool const in_appendix = process_appendix && d->params_.appendix();
2182 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2184 if (fmt.empty() && !layout.counter.empty())
2185 return tclass.counters().theCounter(layout.counter, lang);
2187 // handle 'inherited level parts' in 'fmt',
2188 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2189 size_t const i = fmt.find('@', 0);
2190 if (i != docstring::npos) {
2191 size_t const j = fmt.find('@', i + 1);
2192 if (j != docstring::npos) {
2193 docstring parent(fmt, i + 1, j - i - 1);
2194 docstring label = from_ascii("??");
2195 if (tclass.hasLayout(parent))
2196 label = expandParagraphLabel(tclass[parent], bparams,
2198 fmt = docstring(fmt, 0, i) + label
2199 + docstring(fmt, j + 1, docstring::npos);
2203 return tclass.counters().counterLabel(fmt, lang);
2207 void Paragraph::applyLayout(Layout const & new_layout)
2209 d->layout_ = &new_layout;
2210 LyXAlignment const oldAlign = d->params_.align();
2212 if (!(oldAlign & d->layout_->alignpossible)) {
2213 frontend::Alert::warning(_("Alignment not permitted"),
2214 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2215 d->params_.align(LYX_ALIGN_LAYOUT);
2220 pos_type Paragraph::beginOfBody() const
2222 return d->begin_of_body_;
2226 void Paragraph::setBeginOfBody()
2228 if (d->layout_->labeltype != LABEL_MANUAL) {
2229 d->begin_of_body_ = 0;
2233 // Unroll the first two cycles of the loop
2234 // and remember the previous character to
2235 // remove unnecessary getChar() calls
2237 pos_type end = size();
2238 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2241 char_type previous_char = d->text_[i];
2242 if (!(isNewline(i) || isEnvSeparator(i))) {
2244 while (i < end && previous_char != ' ') {
2245 char_type temp = d->text_[i];
2246 if (isNewline(i) || isEnvSeparator(i))
2249 previous_char = temp;
2255 d->begin_of_body_ = i;
2259 bool Paragraph::allowParagraphCustomization() const
2261 return inInset().allowParagraphCustomization();
2265 bool Paragraph::usePlainLayout() const
2267 return inInset().usePlainLayout();
2271 bool Paragraph::isPassThru() const
2273 return inInset().isPassThru() || d->layout_->pass_thru;
2278 // paragraphs inside floats need different alignment tags to avoid
2281 bool noTrivlistCentering(InsetCode code)
2283 return code == FLOAT_CODE
2284 || code == WRAP_CODE
2285 || code == CELL_CODE;
2289 string correction(string const & orig)
2291 if (orig == "flushleft")
2292 return "raggedright";
2293 if (orig == "flushright")
2294 return "raggedleft";
2295 if (orig == "center")
2301 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2302 InsetCode code, bool const lastpar, int & col)
2304 string macro = suffix + "{";
2305 if (noTrivlistCentering(code)) {
2307 // the last paragraph in non-trivlist-aligned
2308 // context is special (to avoid unwanted whitespace)
2309 if (suffix == "\\begin") {
2310 macro = "\\" + correction(env) + "{}";
2311 os << from_ascii(macro);
2312 col += macro.size();
2317 macro += correction(env);
2321 if (suffix == "\\par\\end") {
2325 os << from_ascii(macro);
2326 col += macro.size();
2327 if (suffix == "\\begin") {
2337 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2338 otexstream & os, OutputParams const & runparams) const
2343 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2344 (layout_->toggle_indent != ITOGGLE_NEVER) :
2345 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2347 if (canindent && params_.noindent() && !layout_->pass_thru) {
2348 os << "\\noindent ";
2352 LyXAlignment const curAlign = params_.align();
2354 if (curAlign == layout_->align)
2358 case LYX_ALIGN_NONE:
2359 case LYX_ALIGN_BLOCK:
2360 case LYX_ALIGN_LAYOUT:
2361 case LYX_ALIGN_SPECIAL:
2362 case LYX_ALIGN_DECIMAL:
2364 case LYX_ALIGN_LEFT:
2365 case LYX_ALIGN_RIGHT:
2366 case LYX_ALIGN_CENTER:
2367 if (runparams.moving_arg) {
2374 string const begin_tag = "\\begin";
2375 InsetCode code = ownerCode();
2376 bool const lastpar = runparams.isLastPar;
2379 case LYX_ALIGN_NONE:
2380 case LYX_ALIGN_BLOCK:
2381 case LYX_ALIGN_LAYOUT:
2382 case LYX_ALIGN_SPECIAL:
2383 case LYX_ALIGN_DECIMAL:
2385 case LYX_ALIGN_LEFT: {
2386 if (!owner_->getParLanguage(bparams)->rightToLeft())
2387 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2389 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2391 } case LYX_ALIGN_RIGHT: {
2392 if (!owner_->getParLanguage(bparams)->rightToLeft())
2393 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2395 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2397 } case LYX_ALIGN_CENTER: {
2398 corrected_env(os, begin_tag, "center", code, lastpar, column);
2407 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2408 otexstream & os, OutputParams const & runparams) const
2410 LyXAlignment const curAlign = params_.align();
2412 if (curAlign == layout_->align)
2416 case LYX_ALIGN_NONE:
2417 case LYX_ALIGN_BLOCK:
2418 case LYX_ALIGN_LAYOUT:
2419 case LYX_ALIGN_SPECIAL:
2420 case LYX_ALIGN_DECIMAL:
2422 case LYX_ALIGN_LEFT:
2423 case LYX_ALIGN_RIGHT:
2424 case LYX_ALIGN_CENTER:
2425 if (runparams.moving_arg)
2430 bool output = false;
2432 string const end_tag = "\\par\\end";
2433 InsetCode code = ownerCode();
2434 bool const lastpar = runparams.isLastPar;
2437 case LYX_ALIGN_NONE:
2438 case LYX_ALIGN_BLOCK:
2439 case LYX_ALIGN_LAYOUT:
2440 case LYX_ALIGN_SPECIAL:
2441 case LYX_ALIGN_DECIMAL:
2443 case LYX_ALIGN_LEFT: {
2444 if (!owner_->getParLanguage(bparams)->rightToLeft())
2445 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2447 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2449 } case LYX_ALIGN_RIGHT: {
2450 if (!owner_->getParLanguage(bparams)->rightToLeft())
2451 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2453 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2455 } case LYX_ALIGN_CENTER: {
2456 corrected_env(os, end_tag, "center", code, lastpar, col);
2461 return output || lastpar;
2465 // This one spits out the text of the paragraph
2466 void Paragraph::latex(BufferParams const & bparams,
2467 Font const & outerfont,
2469 OutputParams const & runparams,
2470 int start_pos, int end_pos, bool force) const
2472 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2474 // FIXME This check should not be needed. Perhaps issue an
2475 // error if it triggers.
2476 Layout const & style = inInset().forcePlainLayout() ?
2477 bparams.documentClass().plainLayout() : *d->layout_;
2479 if (!force && style.inpreamble)
2482 bool const allowcust = allowParagraphCustomization();
2484 // Current base font for all inherited font changes, without any
2485 // change caused by an individual character, except for the language:
2486 // It is set to the language of the first character.
2487 // As long as we are in the label, this font is the base font of the
2488 // label. Before the first body character it is set to the base font
2492 // Maybe we have to create a optional argument.
2493 pos_type body_pos = beginOfBody();
2494 unsigned int column = 0;
2497 // the optional argument is kept in curly brackets in
2498 // case it contains a ']'
2499 // This is not strictly needed, but if this is changed it
2500 // would be a file format change, and tex2lyx would need
2501 // to be adjusted, since it unconditionally removes the
2502 // braces when it parses \item.
2505 basefont = getLabelFont(bparams, outerfont);
2507 basefont = getLayoutFont(bparams, outerfont);
2510 // Which font is currently active?
2511 Font running_font(basefont);
2512 // Do we have an open font change?
2513 bool open_font = false;
2515 Change runningChange = Change(Change::UNCHANGED);
2517 Encoding const * const prev_encoding = runparams.encoding;
2519 os.texrow().start(id(), 0);
2521 // if the paragraph is empty, the loop will not be entered at all
2523 // For InTitle commands, we have already opened a group
2524 // in output_latex::TeXOnePar.
2525 if (style.isCommand() && !style.intitle) {
2529 if (!style.leftdelim().empty()) {
2530 os << style.leftdelim();
2531 column += style.leftdelim().size();
2534 column += d->startTeXParParams(bparams, os, runparams);
2537 // Whether a \par can be issued for insets typeset inline with text.
2538 // Yes if greater than 0. This has to be static.
2539 THREAD_LOCAL_STATIC int parInline = 0;
2541 for (pos_type i = 0; i < size(); ++i) {
2542 // First char in paragraph or after label?
2543 if (i == body_pos) {
2546 bool needPar = false;
2547 column += running_font.latexWriteEndChanges(
2548 os, bparams, runparams,
2549 basefont, basefont, needPar);
2552 basefont = getLayoutFont(bparams, outerfont);
2553 running_font = basefont;
2555 column += Changes::latexMarkChange(os, bparams,
2556 runningChange, Change(Change::UNCHANGED),
2558 runningChange = Change(Change::UNCHANGED);
2563 // For InTitle commands, we have already opened a group
2564 // in output_latex::TeXOnePar.
2565 if (style.isCommand() && !style.intitle) {
2570 if (!style.leftdelim().empty()) {
2571 os << style.leftdelim();
2572 column += style.leftdelim().size();
2576 column += d->startTeXParParams(bparams, os,
2580 runparams.wasDisplayMath = runparams.inDisplayMath;
2581 runparams.inDisplayMath = false;
2582 bool deleted_display_math = false;
2583 Change const & change = runparams.inDeletedInset
2584 ? runparams.changeOfDeletedInset : lookupChange(i);
2586 // Check whether a display math inset follows
2587 if (d->text_[i] == META_INSET
2588 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2589 InsetMath const * im = getInset(i)->asInsetMath();
2590 if (im && im->asHullInset()
2591 && im->asHullInset()->outerDisplay()) {
2592 runparams.inDisplayMath = true;
2593 // runparams.inDeletedInset will be set by
2594 // latexInset later, but we need this info
2595 // before it is called. On the other hand, we
2596 // cannot set it here because it is a counter.
2597 deleted_display_math = isDeleted(i);
2599 if (bparams.output_changes && deleted_display_math
2600 && runningChange == change
2601 && change.type == Change::DELETED
2602 && !os.afterParbreak()) {
2603 // A display math in the same paragraph follows.
2604 // We have to close and then reopen \lyxdeleted,
2605 // otherwise the math will be shifted up.
2606 OutputParams rp = runparams;
2608 bool needPar = false;
2609 column += running_font.latexWriteEndChanges(
2610 os, bparams, rp, basefont,
2614 basefont = getLayoutFont(bparams, outerfont);
2615 running_font = basefont;
2616 column += Changes::latexMarkChange(os, bparams,
2617 Change(Change::INSERTED), change, rp);
2621 if (bparams.output_changes && runningChange != change) {
2623 bool needPar = false;
2624 column += running_font.latexWriteEndChanges(
2625 os, bparams, runparams,
2626 basefont, basefont, needPar);
2629 basefont = getLayoutFont(bparams, outerfont);
2630 running_font = basefont;
2631 column += Changes::latexMarkChange(os, bparams, runningChange,
2633 runningChange = change;
2636 // do not output text which is marked deleted
2637 // if change tracking output is disabled
2638 if (!bparams.output_changes && change.deleted()) {
2644 // Fully instantiated font
2645 Font const current_font = getFont(bparams, i, outerfont);
2647 Font const last_font = running_font;
2649 // Do we need to close the previous font?
2651 (current_font != running_font ||
2652 current_font.language() != running_font.language()))
2654 bool needPar = false;
2655 column += running_font.latexWriteEndChanges(
2656 os, bparams, runparams, basefont,
2657 (i == body_pos-1) ? basefont : current_font,
2659 running_font = basefont;
2663 string const running_lang = runparams.use_polyglossia ?
2664 running_font.language()->polyglossia() : running_font.language()->babel();
2665 // close babel's font environment before opening CJK.
2666 string const lang_end_command = runparams.use_polyglossia ?
2667 "\\end{$$lang}" : lyxrc.language_command_end;
2668 bool const using_begin_end = runparams.use_polyglossia ||
2669 !lang_end_command.empty();
2670 if (!running_lang.empty() &&
2671 (!using_begin_end || running_lang == openLanguageName()) &&
2672 current_font.language()->encoding()->package() == Encoding::CJK) {
2673 string end_tag = subst(lang_end_command,
2676 os << from_ascii(end_tag);
2677 column += end_tag.length();
2678 if (using_begin_end)
2682 // Switch file encoding if necessary (and allowed)
2683 if (!runparams.pass_thru && !style.pass_thru &&
2684 runparams.encoding->package() != Encoding::none &&
2685 current_font.language()->encoding()->package() != Encoding::none) {
2686 pair<bool, int> const enc_switch =
2687 switchEncoding(os.os(), bparams, runparams,
2688 *(current_font.language()->encoding()));
2689 if (enc_switch.first) {
2690 column += enc_switch.second;
2691 runparams.encoding = current_font.language()->encoding();
2695 char_type const c = d->text_[i];
2697 // A display math inset inside an ulem command will be output
2698 // as a box of width \linewidth, so we have to either disable
2699 // indentation if the inset starts a paragraph, or start a new
2700 // line to accommodate such box. This has to be done before
2701 // writing any font changing commands.
2702 if (runparams.inDisplayMath && !deleted_display_math
2703 && runparams.inulemcmd) {
2704 if (os.afterParbreak())
2710 // Do we need to change font?
2711 if ((current_font != running_font ||
2712 current_font.language() != running_font.language()) &&
2715 odocstringstream ods;
2716 column += current_font.latexWriteStartChanges(ods, bparams,
2717 runparams, basefont,
2719 // Check again for display math in ulem commands as a
2720 // font change may also occur just before a math inset.
2721 if (runparams.inDisplayMath && !deleted_display_math
2722 && runparams.inulemcmd) {
2723 if (os.afterParbreak())
2728 running_font = current_font;
2730 docstring fontchange = ods.str();
2731 // check whether the fontchange ends with a \\textcolor
2732 // modifier and the text starts with a space (bug 4473)
2733 docstring const last_modifier = rsplit(fontchange, '\\');
2734 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2735 os << fontchange << from_ascii("{}");
2736 // check if the fontchange ends with a trailing blank
2737 // (like "\small " (see bug 3382)
2738 else if (suffixIs(fontchange, ' ') && c == ' ')
2739 os << fontchange.substr(0, fontchange.size() - 1)
2740 << from_ascii("{}");
2745 // FIXME: think about end_pos implementation...
2746 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2747 // FIXME: integrate this case in latexSpecialChar
2748 // Do not print the separation of the optional argument
2749 // if style.pass_thru is false. This works because
2750 // latexSpecialChar ignores spaces if
2751 // style.pass_thru is false.
2752 if (i != body_pos - 1) {
2753 if (d->simpleTeXBlanks(bparams, runparams, os,
2754 i, column, current_font, style)) {
2755 // A surrogate pair was output. We
2756 // must not call latexSpecialChar
2757 // in this iteration, since it would output
2758 // the combining character again.
2765 OutputParams rp = runparams;
2766 rp.free_spacing = style.free_spacing;
2767 rp.local_font = ¤t_font;
2768 rp.intitle = style.intitle;
2770 // Two major modes: LaTeX or plain
2771 // Handle here those cases common to both modes
2772 // and then split to handle the two modes separately.
2773 if (c == META_INSET) {
2774 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2775 // Greyedout notes and, in general, all insets
2776 // with InsetLayout::isDisplay() == false,
2777 // are typeset inline with the text. So, we
2778 // can add a \par to the last paragraph of
2779 // such insets only if nothing else follows.
2780 bool incremented = false;
2781 Inset const * inset = getInset(i);
2782 InsetText const * textinset = inset
2783 ? inset->asInsetText()
2785 if (i + 1 == size() && textinset
2786 && !inset->getLayout().isDisplay()) {
2787 ParagraphList const & pars =
2788 textinset->text().paragraphs();
2789 pit_type const pit = pars.size() - 1;
2790 Font const lastfont =
2791 pit < 0 || pars[pit].empty()
2792 ? pars[pit].getLayoutFont(
2795 : pars[pit].getFont(bparams,
2796 pars[pit].size() - 1,
2798 if (lastfont.fontInfo().size() !=
2799 basefont.fontInfo().size()) {
2804 d->latexInset(bparams, os, rp, running_font,
2805 basefont, outerfont, open_font,
2806 runningChange, style, i, column);
2810 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2812 d->latexSpecialChar(os, bparams, rp,
2813 running_font, runningChange,
2814 style, i, end_pos, column);
2815 } catch (EncodingException & e) {
2816 if (runparams.dryrun) {
2817 os << "<" << _("LyX Warning: ")
2818 << _("uncodable character") << " '";
2822 // add location information and throw again.
2830 // Set the encoding to that returned from latexSpecialChar (see
2831 // comment for encoding member in OutputParams.h)
2832 runparams.encoding = rp.encoding;
2834 // Also carry on the info on a closed ulem command for insets
2835 // such as Note that do not produce any output, so that no
2836 // command is ever executed but its opening was recorded.
2837 runparams.inulemcmd = rp.inulemcmd;
2840 // If we have an open font definition, we have to close it
2842 // Make sure that \\par is done with the font of the last
2843 // character if this has another size as the default.
2844 // This is necessary because LaTeX (and LyX on the screen)
2845 // calculates the space between the baselines according
2846 // to this font. (Matthias)
2848 // We must not change the font for the last paragraph
2849 // of non-multipar insets, tabular cells or commands,
2850 // since this produces unwanted whitespace.
2852 Font const font = empty()
2853 ? getLayoutFont(bparams, outerfont)
2854 : getFont(bparams, size() - 1, outerfont);
2856 InsetText const * textinset = inInset().asInsetText();
2858 bool const maintext = textinset
2859 ? textinset->text().isMainText()
2862 size_t const numpars = textinset
2863 ? textinset->text().paragraphs().size()
2866 bool needPar = false;
2868 if (style.resfont.size() != font.fontInfo().size()
2869 && (!runparams.isLastPar || maintext
2870 || (numpars > 1 && d->ownerCode() != CELL_CODE
2871 && (inInset().getLayout().isDisplay()
2873 && !style.isCommand()) {
2876 #ifdef FIXED_LANGUAGE_END_DETECTION
2878 running_font.latexWriteEndChanges(os, bparams,
2879 runparams, basefont,
2880 next_->getFont(bparams, 0, outerfont),
2883 running_font.latexWriteEndChanges(os, bparams,
2884 runparams, basefont, basefont, needPar);
2887 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2888 //FIXME: there as we start another \selectlanguage with the next paragraph if
2889 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2890 running_font.latexWriteEndChanges(os, bparams, runparams,
2891 basefont, basefont, needPar);
2894 // The \par could not be inserted at the same nesting
2895 // level of the font size change, so do it now.
2896 os << "{\\" << font.latexSize() << "\\par}";
2900 column += Changes::latexMarkChange(os, bparams, runningChange,
2901 Change(Change::UNCHANGED), runparams);
2903 // Needed if there is an optional argument but no contents.
2904 if (body_pos > 0 && body_pos == size()) {
2908 if (!style.rightdelim().empty()) {
2909 os << style.rightdelim();
2910 column += style.rightdelim().size();
2913 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2914 && runparams.encoding != prev_encoding) {
2915 runparams.encoding = prev_encoding;
2916 os << setEncoding(prev_encoding->iconvName());
2919 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2923 bool Paragraph::emptyTag() const
2925 for (pos_type i = 0; i < size(); ++i) {
2926 if (Inset const * inset = getInset(i)) {
2927 InsetCode lyx_code = inset->lyxCode();
2928 // FIXME testing like that is wrong. What is
2930 if (lyx_code != TOC_CODE &&
2931 lyx_code != INCLUDE_CODE &&
2932 lyx_code != GRAPHICS_CODE &&
2933 lyx_code != ERT_CODE &&
2934 lyx_code != LISTINGS_CODE &&
2935 lyx_code != FLOAT_CODE &&
2936 lyx_code != TABULAR_CODE) {
2940 char_type c = d->text_[i];
2941 if (c != ' ' && c != '\t')
2949 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2952 for (pos_type i = 0; i < size(); ++i) {
2953 if (Inset const * inset = getInset(i)) {
2954 InsetCode lyx_code = inset->lyxCode();
2955 if (lyx_code == LABEL_CODE) {
2956 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2957 docstring const & id = il->getParam("name");
2958 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2966 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2970 for (i = 0; i < size(); ++i) {
2971 if (Inset const * inset = getInset(i)) {
2972 inset->docbook(os, runparams);
2974 char_type c = d->text_[i];
2977 os << sgml::escapeChar(c);
2984 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2988 for (i = 0; i < size(); ++i) {
2989 if (Inset const * inset = getInset(i)) {
2990 inset->xhtml(xs, runparams);
2992 char_type c = d->text_[i];
3002 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3005 pos_type size = text_.size();
3006 for (pos_type i = initial; i < size; ++i) {
3007 Font font = owner_->getFont(buf.params(), i, outerfont);
3008 if (text_[i] == META_INSET)
3010 if (i != initial && font != font_old)
3019 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3021 OutputParams const & runparams,
3022 Font const & outerfont,
3023 pos_type initial) const
3025 bool emph_flag = false;
3027 Layout const & style = *d->layout_;
3029 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3031 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3034 // parsing main loop
3035 for (pos_type i = initial; i < size(); ++i) {
3036 Font font = getFont(buf.params(), i, outerfont);
3038 // handle <emphasis> tag
3039 if (font_old.emph() != font.fontInfo().emph()) {
3040 if (font.fontInfo().emph() == FONT_ON) {
3043 } else if (i != initial) {
3044 os << "</emphasis>";
3049 if (Inset const * inset = getInset(i)) {
3050 inset->docbook(os, runparams);
3052 char_type c = d->text_[i];
3054 if (style.pass_thru)
3057 os << sgml::escapeChar(c);
3059 font_old = font.fontInfo();
3063 os << "</emphasis>";
3066 if (style.free_spacing)
3068 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3074 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
3075 vector<html::EndFontTag> & tagsToClose,
3076 bool & flag, FontState curstate, html::FontTypes type)
3078 if (curstate == FONT_ON) {
3079 tagsToOpen.push_back(html::FontTag(type));
3082 tagsToClose.push_back(html::EndFontTag(type));
3089 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3091 OutputParams const & runparams,
3092 Font const & outerfont,
3093 bool start_paragraph, bool close_paragraph,
3094 pos_type initial) const
3098 // track whether we have opened these tags
3099 bool emph_flag = false;
3100 bool bold_flag = false;
3101 bool noun_flag = false;
3102 bool ubar_flag = false;
3103 bool dbar_flag = false;
3104 bool sout_flag = false;
3105 bool xout_flag = false;
3106 bool wave_flag = false;
3108 bool shap_flag = false;
3110 bool faml_flag = false;
3112 bool size_flag = false;
3114 Layout const & style = *d->layout_;
3116 if (start_paragraph)
3117 xs.startDivision(allowEmpty());
3120 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3122 FontShape curr_fs = INHERIT_SHAPE;
3123 FontFamily curr_fam = INHERIT_FAMILY;
3124 FontSize curr_size = FONT_SIZE_INHERIT;
3126 string const default_family =
3127 buf.masterBuffer()->params().fonts_default_family;
3129 vector<html::FontTag> tagsToOpen;
3130 vector<html::EndFontTag> tagsToClose;
3132 // parsing main loop
3133 for (pos_type i = initial; i < size(); ++i) {
3134 // let's not show deleted material in the output
3138 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3141 FontState curstate = font.fontInfo().emph();
3142 if (font_old.emph() != curstate)
3143 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3146 curstate = font.fontInfo().noun();
3147 if (font_old.noun() != curstate)
3148 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3151 curstate = font.fontInfo().underbar();
3152 if (font_old.underbar() != curstate)
3153 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3156 curstate = font.fontInfo().strikeout();
3157 if (font_old.strikeout() != curstate)
3158 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3161 curstate = font.fontInfo().xout();
3162 if (font_old.xout() != curstate)
3163 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3166 curstate = font.fontInfo().uuline();
3167 if (font_old.uuline() != curstate)
3168 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3171 curstate = font.fontInfo().uwave();
3172 if (font_old.uwave() != curstate)
3173 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3176 // a little hackish, but allows us to reuse what we have.
3177 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3178 if (font_old.series() != font.fontInfo().series())
3179 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3182 curr_fs = font.fontInfo().shape();
3183 FontShape old_fs = font_old.shape();
3184 if (old_fs != curr_fs) {
3188 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3191 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3193 case SMALLCAPS_SHAPE:
3194 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3200 // the other tags are for internal use
3208 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3212 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3215 case SMALLCAPS_SHAPE:
3216 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3223 // the other tags are for internal use
3230 curr_fam = font.fontInfo().family();
3231 FontFamily old_fam = font_old.family();
3232 if (old_fam != curr_fam) {
3236 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3239 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3241 case TYPEWRITER_FAMILY:
3242 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3244 case INHERIT_FAMILY:
3247 // the other tags are for internal use
3255 // we will treat a "default" font family as roman, since we have
3256 // no other idea what to do.
3257 if (default_family != "rmdefault" && default_family != "default") {
3258 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3263 if (default_family != "sfdefault") {
3264 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3268 case TYPEWRITER_FAMILY:
3269 if (default_family != "ttdefault") {
3270 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3274 case INHERIT_FAMILY:
3277 // the other tags are for internal use
3284 curr_size = font.fontInfo().size();
3285 FontSize old_size = font_old.size();
3286 if (old_size != curr_size) {
3289 case FONT_SIZE_TINY:
3290 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3292 case FONT_SIZE_SCRIPT:
3293 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3295 case FONT_SIZE_FOOTNOTE:
3296 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3298 case FONT_SIZE_SMALL:
3299 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3301 case FONT_SIZE_LARGE:
3302 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3304 case FONT_SIZE_LARGER:
3305 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3307 case FONT_SIZE_LARGEST:
3308 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3310 case FONT_SIZE_HUGE:
3311 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3313 case FONT_SIZE_HUGER:
3314 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3316 case FONT_SIZE_INCREASE:
3317 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3319 case FONT_SIZE_DECREASE:
3320 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3322 case FONT_SIZE_INHERIT:
3323 case FONT_SIZE_NORMAL:
3326 // the other tags are for internal use
3332 switch (curr_size) {
3333 case FONT_SIZE_TINY:
3334 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3337 case FONT_SIZE_SCRIPT:
3338 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3341 case FONT_SIZE_FOOTNOTE:
3342 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3345 case FONT_SIZE_SMALL:
3346 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3349 case FONT_SIZE_LARGE:
3350 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3353 case FONT_SIZE_LARGER:
3354 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3357 case FONT_SIZE_LARGEST:
3358 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3361 case FONT_SIZE_HUGE:
3362 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3365 case FONT_SIZE_HUGER:
3366 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3369 case FONT_SIZE_INCREASE:
3370 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3373 case FONT_SIZE_DECREASE:
3374 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3377 case FONT_SIZE_NORMAL:
3378 case FONT_SIZE_INHERIT:
3381 // the other tags are for internal use
3388 // Other such tags? What about the other text ranges?
3390 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3391 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3392 for (; cit != cen; ++cit)
3395 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3396 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3397 for (; sit != sen; ++sit)
3400 tagsToClose.clear();
3403 Inset const * inset = getInset(i);
3405 if (!runparams.for_toc || inset->isInToc()) {
3406 OutputParams np = runparams;
3407 np.local_font = &font;
3408 // If the paragraph has size 1, then we are in the "special
3409 // case" where we do not output the containing paragraph info
3410 if (!inset->getLayout().htmlisblock() && size() != 1)
3411 np.html_in_par = true;
3412 retval += inset->xhtml(xs, np);
3415 char_type c = getUChar(buf.masterBuffer()->params(),
3419 font_old = font.fontInfo();
3423 // I'm worried about what happens if a branch, say, is itself
3424 // wrapped in some font stuff. I think that will not work.
3426 if (close_paragraph)
3433 bool Paragraph::isHfill(pos_type pos) const
3435 Inset const * inset = getInset(pos);
3436 return inset && inset->isHfill();
3440 bool Paragraph::isNewline(pos_type pos) const
3442 // U+2028 LINE SEPARATOR
3443 // U+2029 PARAGRAPH SEPARATOR
3444 char_type const c = d->text_[pos];
3445 if (c == 0x2028 || c == 0x2029)
3447 Inset const * inset = getInset(pos);
3448 return inset && inset->lyxCode() == NEWLINE_CODE;
3452 bool Paragraph::isEnvSeparator(pos_type pos) const
3454 Inset const * inset = getInset(pos);
3455 return inset && inset->lyxCode() == SEPARATOR_CODE;
3459 bool Paragraph::isLineSeparator(pos_type pos) const
3461 char_type const c = d->text_[pos];
3462 if (isLineSeparatorChar(c))
3464 Inset const * inset = getInset(pos);
3465 return inset && inset->isLineSeparator();
3469 bool Paragraph::isWordSeparator(pos_type pos) const
3473 if (Inset const * inset = getInset(pos))
3474 return !inset->isLetter();
3475 // if we have a hard hyphen (no en- or emdash) or apostrophe
3476 // we pass this to the spell checker
3477 // FIXME: this method is subject to change, visit
3478 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3479 // to get an impression how complex this is.
3480 if (isHardHyphenOrApostrophe(pos))
3482 char_type const c = d->text_[pos];
3483 // We want to pass the escape chars to the spellchecker
3484 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3485 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3489 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3491 pos_type const psize = size();
3494 char_type const c = d->text_[pos];
3495 if (c != '-' && c != '\'')
3497 int nextpos = pos + 1;
3498 int prevpos = pos > 0 ? pos - 1 : 0;
3499 if ((nextpos == psize || isSpace(nextpos))
3500 && (pos == 0 || isSpace(prevpos)))
3506 bool Paragraph::needsCProtection(bool const fragile) const
3508 // first check the layout of the paragraph, but only in insets
3509 InsetText const * textinset = inInset().asInsetText();
3510 bool const maintext = textinset
3511 ? textinset->text().isMainText()
3514 if (!maintext && layout().needcprotect) {
3515 // Environments need cprotection regardless the content
3516 if (layout().latextype == LATEX_ENVIRONMENT)
3519 // Commands need cprotection if they contain specific chars
3520 int const nchars_escape = 9;
3521 static char_type const chars_escape[nchars_escape] = {
3522 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3524 docstring const pars = asString();
3525 for (int k = 0; k < nchars_escape; k++) {
3526 if (contains(pars, chars_escape[k]))
3531 // now check whether we have insets that need cprotection
3532 pos_type size = d->text_.size();
3533 for (pos_type i = 0; i < size; ++i)
3534 if (isInset(i) && getInset(i)->needsCProtection(maintext, fragile))
3541 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3543 return d->speller_state_.getRange(pos);
3547 bool Paragraph::isChar(pos_type pos) const
3549 if (Inset const * inset = getInset(pos))
3550 return inset->isChar();
3551 char_type const c = d->text_[pos];
3552 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3556 bool Paragraph::isSpace(pos_type pos) const
3558 if (Inset const * inset = getInset(pos))
3559 return inset->isSpace();
3560 char_type const c = d->text_[pos];
3561 return lyx::isSpace(c);
3566 Paragraph::getParLanguage(BufferParams const & bparams) const
3569 return getFirstFontSettings(bparams).language();
3570 // FIXME: we should check the prev par as well (Lgb)
3571 return bparams.language;
3575 bool Paragraph::isRTL(BufferParams const & bparams) const
3577 return getParLanguage(bparams)->rightToLeft()
3578 && !inInset().getLayout().forceLTR();
3582 void Paragraph::changeLanguage(BufferParams const & bparams,
3583 Language const * from, Language const * to)
3585 // change language including dummy font change at the end
3586 for (pos_type i = 0; i <= size(); ++i) {
3587 Font font = getFontSettings(bparams, i);
3588 if (font.language() == from) {
3589 font.setLanguage(to);
3591 d->requestSpellCheck(i);
3597 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3599 Language const * doc_language = bparams.language;
3600 for (auto const & f : d->fontlist_)
3601 if (f.font().language() != ignore_language &&
3602 f.font().language() != latex_language &&
3603 f.font().language() != doc_language)
3609 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3611 for (auto const & f : d->fontlist_) {
3612 Language const * lang = f.font().language();
3613 if (lang != ignore_language &&
3614 lang != latex_language)
3620 docstring Paragraph::asString(int options) const
3622 return asString(0, size(), options);
3626 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3628 odocstringstream os;
3631 && options & AS_STR_LABEL
3632 && !d->params_.labelString().empty())
3633 os << d->params_.labelString() << ' ';
3635 for (pos_type i = beg; i < end; ++i) {
3636 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3638 char_type const c = d->text_[i];
3639 if (isPrintable(c) || c == '\t'
3640 || (c == '\n' && (options & AS_STR_NEWLINES)))
3642 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3643 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3644 LASSERT(runparams != 0, return docstring());
3645 getInset(i)->plaintext(os, *runparams);
3647 getInset(i)->toString(os);
3656 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3657 bool const shorten, bool const label) const
3659 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3660 if (label && !labelString().empty())
3661 os += labelString() + ' ';
3662 if (!layout().isTocCaption())
3664 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3667 char_type const c = d->text_[i];
3670 else if (c == META_INSET)
3671 getInset(i)->forOutliner(os, tmplen, false);
3674 Text::shortenForOutliner(os, maxlen);
3678 void Paragraph::setInsetOwner(Inset const * inset)
3680 d->inset_owner_ = inset;
3684 int Paragraph::id() const
3690 void Paragraph::setId(int id)
3696 Layout const & Paragraph::layout() const
3702 void Paragraph::setLayout(Layout const & layout)
3704 d->layout_ = &layout;
3708 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3710 setLayout(tc.defaultLayout());
3714 void Paragraph::setPlainLayout(DocumentClass const & tc)
3716 setLayout(tc.plainLayout());
3720 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3722 if (usePlainLayout())
3723 setPlainLayout(tclass);
3725 setDefaultLayout(tclass);
3729 Inset const & Paragraph::inInset() const
3731 LBUFERR(d->inset_owner_);
3732 return *d->inset_owner_;
3736 ParagraphParameters & Paragraph::params()
3742 ParagraphParameters const & Paragraph::params() const
3748 bool Paragraph::isFreeSpacing() const
3750 if (d->layout_->free_spacing)
3752 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3756 bool Paragraph::allowEmpty() const
3758 if (d->layout_->keepempty)
3760 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3764 bool Paragraph::brokenBiblio() const
3766 // there is a problem if there is no bibitem at position 0 or
3767 // if there is another bibitem in the paragraph.
3768 return d->layout_->labeltype == LABEL_BIBLIO
3769 && (d->insetlist_.find(BIBITEM_CODE) != 0
3770 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3774 int Paragraph::fixBiblio(Buffer const & buffer)
3776 // FIXME: What about the case where paragraph is not BIBLIO
3777 // but there is an InsetBibitem?
3778 // FIXME: when there was already an inset at 0, the return value is 1,
3779 // which does not tell whether another inset has been remove; the
3780 // cursor cannot be correctly updated.
3782 if (d->layout_->labeltype != LABEL_BIBLIO)
3785 bool const track_changes = buffer.params().track_changes;
3786 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3787 bool const hasbibitem0 = bibitem_pos == 0;
3790 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3791 // There was an InsetBibitem at pos 0, and no other one => OK
3792 if (bibitem_pos == -1)
3794 // there is a bibitem at the 0 position, but since
3795 // there is a second one, we copy the second on the
3796 // first. We're assuming there are at most two of
3797 // these, which there should be.
3798 // FIXME: why does it make sense to do that rather
3799 // than keep the first? (JMarc)
3800 Inset * inset = releaseInset(bibitem_pos);
3801 d->insetlist_.begin()->inset = inset;
3802 return -bibitem_pos;
3805 // We need to create an inset at the beginning
3807 if (bibitem_pos > 0) {
3808 // there was one somewhere in the paragraph, let's move it
3809 inset = d->insetlist_.release(bibitem_pos);
3810 eraseChar(bibitem_pos, track_changes);
3813 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3814 InsetCommandParams(BIBITEM_CODE));
3816 Font font(inherit_font, buffer.params().language);
3817 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3818 : Change::UNCHANGED));
3824 void Paragraph::checkAuthors(AuthorList const & authorList)
3826 d->changes_.checkAuthors(authorList);
3830 bool Paragraph::isChanged(pos_type pos) const
3832 return lookupChange(pos).changed();
3836 bool Paragraph::isInserted(pos_type pos) const
3838 return lookupChange(pos).inserted();
3842 bool Paragraph::isDeleted(pos_type pos) const
3844 return lookupChange(pos).deleted();
3848 InsetList const & Paragraph::insetList() const
3850 return d->insetlist_;
3854 void Paragraph::setInsetBuffers(Buffer & b)
3856 d->insetlist_.setBuffer(b);
3860 void Paragraph::resetBuffer()
3862 d->insetlist_.resetBuffer();
3866 Inset * Paragraph::releaseInset(pos_type pos)
3868 Inset * inset = d->insetlist_.release(pos);
3869 /// does not honour change tracking!
3870 eraseChar(pos, false);
3875 Inset * Paragraph::getInset(pos_type pos)
3877 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3878 ? d->insetlist_.get(pos) : 0;
3882 Inset const * Paragraph::getInset(pos_type pos) const
3884 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3885 ? d->insetlist_.get(pos) : 0;
3889 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3890 pos_type & right, TextCase action)
3892 // process sequences of modified characters; in change
3893 // tracking mode, this approach results in much better
3894 // usability than changing case on a char-by-char basis
3895 // We also need to track the current font, since font
3896 // changes within sequences can occur.
3897 vector<pair<char_type, Font> > changes;
3899 bool const trackChanges = bparams.track_changes;
3901 bool capitalize = true;
3903 for (; pos < right; ++pos) {
3904 char_type oldChar = d->text_[pos];
3905 char_type newChar = oldChar;
3907 // ignore insets and don't play with deleted text!
3908 if (oldChar != META_INSET && !isDeleted(pos)) {
3910 case text_lowercase:
3911 newChar = lowercase(oldChar);
3913 case text_capitalization:
3915 newChar = uppercase(oldChar);
3919 case text_uppercase:
3920 newChar = uppercase(oldChar);
3925 if (isWordSeparator(pos) || isDeleted(pos)) {
3926 // permit capitalization again
3930 if (oldChar != newChar) {
3931 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3932 if (pos != right - 1)
3934 // step behind the changing area
3938 int erasePos = pos - changes.size();
3939 for (size_t i = 0; i < changes.size(); i++) {
3940 insertChar(pos, changes[i].first,
3943 if (!eraseChar(erasePos, trackChanges)) {
3946 ++right; // expand selection
3954 int Paragraph::find(docstring const & str, bool cs, bool mw,
3955 pos_type start_pos, bool del) const
3957 pos_type pos = start_pos;
3958 int const strsize = str.length();
3960 pos_type const parsize = d->text_.size();
3961 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3962 // Ignore "invisible" letters such as ligature breaks
3963 // and hyphenation chars while searching
3964 while (pos < parsize - 1 && isInset(pos)) {
3965 odocstringstream os;
3966 getInset(pos)->toString(os);
3967 if (!getInset(pos)->isLetter() || !os.str().empty())
3971 if (cs && str[i] != d->text_[pos])
3973 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3975 if (!del && isDeleted(pos))
3982 // if necessary, check whether string matches word
3984 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3987 && !isWordSeparator(pos))
3991 return pos - start_pos;
3995 char_type Paragraph::getChar(pos_type pos) const
3997 return d->text_[pos];
4001 pos_type Paragraph::size() const
4003 return d->text_.size();
4007 bool Paragraph::empty() const
4009 return d->text_.empty();
4013 bool Paragraph::isInset(pos_type pos) const
4015 return d->text_[pos] == META_INSET;
4019 bool Paragraph::isSeparator(pos_type pos) const
4021 //FIXME: Are we sure this can be the only separator?
4022 return d->text_[pos] == ' ';
4026 void Paragraph::deregisterWords()
4028 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4029 Private::LangWordsMap::const_iterator ite = d->words_.end();
4030 for (; itl != ite; ++itl) {
4031 WordList & wl = theWordList(itl->first);
4032 Private::Words::const_iterator it = (itl->second).begin();
4033 Private::Words::const_iterator et = (itl->second).end();
4034 for (; it != et; ++it)
4041 void Paragraph::locateWord(pos_type & from, pos_type & to,
4042 word_location const loc) const
4045 case WHOLE_WORD_STRICT:
4046 if (from == 0 || from == size()
4047 || isWordSeparator(from)
4048 || isWordSeparator(from - 1)) {
4055 // If we are already at the beginning of a word, do nothing
4056 if (!from || isWordSeparator(from - 1))
4061 // always move the cursor to the beginning of previous word
4062 while (from && !isWordSeparator(from - 1))
4066 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4069 // no need to move the 'from' cursor
4073 while (to < size() && !isWordSeparator(to))
4078 void Paragraph::collectWords()
4080 for (pos_type pos = 0; pos < size(); ++pos) {
4081 if (isWordSeparator(pos))
4083 pos_type from = pos;
4084 locateWord(from, pos, WHOLE_WORD);
4085 // Work around MSVC warning: The statement
4086 // if (pos < from + lyxrc.completion_minlength)
4087 // triggers a signed vs. unsigned warning.
4088 // I don't know why this happens, it could be a MSVC bug, or
4089 // related to LLP64 (windows) vs. LP64 (unix) programming
4090 // model, or the C++ standard might be ambigous in the section
4091 // defining the "usual arithmetic conversions". However, using
4092 // a temporary variable is safe and works on all compilers.
4093 pos_type const endpos = from + lyxrc.completion_minlength;
4096 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4097 if (cit == d->fontlist_.end())
4099 Language const * lang = cit->font().language();
4100 docstring const word = asString(from, pos, AS_STR_NONE);
4101 d->words_[lang->lang()].insert(word);
4106 void Paragraph::registerWords()
4108 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4109 Private::LangWordsMap::const_iterator ite = d->words_.end();
4110 for (; itl != ite; ++itl) {
4111 WordList & wl = theWordList(itl->first);
4112 Private::Words::const_iterator it = (itl->second).begin();
4113 Private::Words::const_iterator et = (itl->second).end();
4114 for (; it != et; ++it)
4120 void Paragraph::updateWords()
4128 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4130 SkipPositionsIterator begin = skips.begin();
4131 SkipPositions::iterator end = skips.end();
4132 if (pos > 0 && begin < end) {
4134 if (end->last == pos - 1) {
4139 skips.insert(end, FontSpan(pos, pos));
4143 Language * Paragraph::Private::locateSpellRange(
4144 pos_type & from, pos_type & to,
4145 SkipPositions & skips) const
4147 // skip leading white space
4148 while (from < to && owner_->isWordSeparator(from))
4150 // don't check empty range
4153 // get current language
4154 Language * lang = getSpellLanguage(from);
4155 pos_type last = from;
4156 bool samelang = true;
4157 bool sameinset = true;
4158 while (last < to && samelang && sameinset) {
4159 // hop to end of word
4160 while (last < to && !owner_->isWordSeparator(last)) {
4161 if (owner_->getInset(last)) {
4162 appendSkipPosition(skips, last);
4163 } else if (owner_->isDeleted(last)) {
4164 appendSkipPosition(skips, last);
4168 // hop to next word while checking for insets
4169 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4170 if (Inset const * inset = owner_->getInset(last))
4171 sameinset = inset->isChar() && inset->isLetter();
4172 if (sameinset && owner_->isDeleted(last)) {
4173 appendSkipPosition(skips, last);
4178 if (sameinset && last < to) {
4179 // now check for language change
4180 samelang = lang == getSpellLanguage(last);
4183 // if language change detected backstep is needed
4191 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4194 const_cast<Language *>(owner_->getFontSettings(
4195 inset_owner_->buffer().params(), from).language());
4196 if (lang == inset_owner_->buffer().params().language
4197 && !lyxrc.spellchecker_alt_lang.empty()) {
4199 string const lang_variety =
4200 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4201 lang->setCode(lang_code);
4202 lang->setVariety(lang_variety);
4208 void Paragraph::requestSpellCheck(pos_type pos)
4210 d->requestSpellCheck(pos);
4214 bool Paragraph::needsSpellCheck() const
4216 SpellChecker::ChangeNumber speller_change_number = 0;
4217 if (theSpellChecker())
4218 speller_change_number = theSpellChecker()->changeNumber();
4219 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4220 d->speller_state_.needsCompleteRefresh(speller_change_number);
4222 return d->needsSpellCheck();
4226 bool Paragraph::Private::ignoreWord(docstring const & word) const
4228 // Ignore words with digits
4229 // FIXME: make this customizable
4230 // (note that some checkers ignore words with digits by default)
4231 docstring::const_iterator cit = word.begin();
4232 docstring::const_iterator const end = word.end();
4233 for (; cit != end; ++cit) {
4234 if (isNumber((*cit)))
4241 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4242 WordLangTuple & wl, docstring_list & suggestions,
4243 bool do_suggestion, bool check_learned) const
4245 SpellChecker::Result result = SpellChecker::WORD_OK;
4246 SpellChecker * speller = theSpellChecker();
4250 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4253 locateWord(from, to, WHOLE_WORD);
4254 if (from == to || from >= size())
4257 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4258 Language * lang = d->getSpellLanguage(from);
4260 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4263 wl = WordLangTuple(word, lang);
4268 if (needsSpellCheck() || check_learned) {
4270 if (!d->ignoreWord(word)) {
4271 bool const trailing_dot = to < size() && d->text_[to] == '.';
4272 result = speller->check(wl);
4273 if (SpellChecker::misspelled(result) && trailing_dot) {
4274 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4275 result = speller->check(wl);
4276 if (!SpellChecker::misspelled(result)) {
4277 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4279 from << ".." << to << "]");
4281 // spell check with dot appended failed too
4282 // restore original word/lang value
4283 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4284 wl = WordLangTuple(word, lang);
4288 if (!SpellChecker::misspelled(result)) {
4289 // area up to the begin of the next word is not misspelled
4290 while (end < size() && isWordSeparator(end))
4293 d->setMisspelled(from, end, result);
4295 result = d->speller_state_.getState(from);
4299 suggestions.clear();
4301 if (SpellChecker::misspelled(result)) {
4302 LYXERR(Debug::GUI, "misspelled word: \"" <<
4304 from << ".." << to << "]");
4306 speller->suggest(wl, suggestions);
4312 void Paragraph::anonymize()
4314 // This is a very crude anonymization for now
4315 for (char_type & c : d->text_)
4316 if (isLetterChar(c) || isNumber(c))
4321 void Paragraph::Private::markMisspelledWords(
4322 pos_type const & first, pos_type const & last,
4323 SpellChecker::Result result,
4324 docstring const & word,
4325 SkipPositions const & skips)
4327 if (!SpellChecker::misspelled(result)) {
4328 setMisspelled(first, last, SpellChecker::WORD_OK);
4332 SpellChecker * speller = theSpellChecker();
4333 // locate and enumerate the error positions
4334 int nerrors = speller->numMisspelledWords();
4336 SkipPositionsIterator it = skips.begin();
4337 SkipPositionsIterator et = skips.end();
4338 for (int index = 0; index < nerrors; ++index) {
4341 speller->misspelledWord(index, wstart, wlen);
4342 /// should not happen if speller supports range checks
4343 if (!wlen) continue;
4344 docstring const misspelled = word.substr(wstart, wlen);
4345 wstart += first + numskipped;
4346 if (snext < wstart) {
4347 /// mark the range of correct spelling
4348 numskipped += countSkips(it, et, wstart);
4349 setMisspelled(snext,
4350 wstart - 1, SpellChecker::WORD_OK);
4352 snext = wstart + wlen;
4353 numskipped += countSkips(it, et, snext);
4354 /// mark the range of misspelling
4355 setMisspelled(wstart, snext, result);
4356 LYXERR(Debug::GUI, "misspelled word: \"" <<
4357 misspelled << "\" [" <<
4358 wstart << ".." << (snext-1) << "]");
4361 if (snext <= last) {
4362 /// mark the range of correct spelling at end
4363 setMisspelled(snext, last, SpellChecker::WORD_OK);
4368 void Paragraph::spellCheck() const
4370 SpellChecker * speller = theSpellChecker();
4371 if (!speller || empty() ||!needsSpellCheck())
4375 d->rangeOfSpellCheck(start, endpos);
4376 if (speller->canCheckParagraph()) {
4377 // loop until we leave the range
4378 for (pos_type first = start; first < endpos; ) {
4379 pos_type last = endpos;
4380 Private::SkipPositions skips;
4381 Language * lang = d->locateSpellRange(first, last, skips);
4382 if (first >= endpos)
4384 // start the spell checker on the unit of meaning
4385 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4386 WordLangTuple wl = WordLangTuple(word, lang);
4387 SpellChecker::Result result = word.size() ?
4388 speller->check(wl) : SpellChecker::WORD_OK;
4389 d->markMisspelledWords(first, last, result, word, skips);
4393 static docstring_list suggestions;
4394 pos_type to = endpos;
4395 while (start < endpos) {
4397 spellCheck(start, to, wl, suggestions, false);
4401 d->readySpellCheck();
4405 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4407 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4408 if (result || pos <= 0 || pos > size())
4410 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4411 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4416 string Paragraph::magicLabel() const
4419 ss << "magicparlabel-" << id();