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 && running_font.isRightToLeft()
1147 // ERT is an exception, it should be output with no
1148 // decorations at all
1149 && inset->lyxCode() != ERT_CODE) {
1150 if (runparams.use_polyglossia) {
1152 } else if (running_font.language()->lang() == "farsi"
1153 || running_font.language()->lang() == "arabic_arabi")
1154 os << "\\textLR{" << termcmd;
1160 // FIXME: Bug: we can have an empty font change here!
1161 // if there has just been a font change, we are going to close it
1162 // right now, which means stupid latex code like \textsf{}. AFAIK,
1163 // this does not harm dvi output. A minor bug, thus (JMarc)
1165 // Some insets cannot be inside a font change command.
1166 // However, even such insets *can* be placed in \L or \R
1167 // or their equivalents (for RTL language switches), so we don't
1168 // close the language in those cases.
1169 // ArabTeX, though, cannot handle this special behavior, it seems.
1170 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1171 || running_font.language()->lang() == "arabic_arabtex";
1172 if (open_font && !inset->inheritFont()) {
1173 bool needPar = false;
1174 bool closeLanguage = arabtex
1175 || basefont.isRightToLeft() == running_font.isRightToLeft();
1176 unsigned int count = running_font.latexWriteEndChanges(os,
1177 bparams, runparams, basefont, basefont,
1178 needPar, closeLanguage);
1180 // if any font properties were closed, update the running_font,
1181 // making sure, however, to leave the language as it was
1183 // FIXME: probably a better way to keep track of the old
1184 // language, than copying the entire font?
1185 Font const copy_font(running_font);
1186 basefont = owner_->getLayoutFont(bparams, outerfont);
1187 running_font = basefont;
1189 running_font.setLanguage(copy_font.language());
1190 // leave font open if language is still open
1191 open_font = (running_font.language() == basefont.language());
1193 runparams.local_font = &basefont;
1197 size_t const previous_row_count = os.texrow().rows();
1200 runparams.lastid = id_;
1201 runparams.lastpos = i;
1202 inset->latex(os, runparams);
1203 } catch (EncodingException & e) {
1204 // add location information and throw again.
1213 if (os.texrow().rows() > previous_row_count) {
1214 os.texrow().start(owner_->id(), i + 1);
1217 column += (unsigned int)(os.os().tellp() - len);
1220 if (owner_->isDeleted(i))
1221 --runparams.inDeletedInset;
1225 void Paragraph::Private::latexSpecialChar(otexstream & os,
1226 BufferParams const & bparams,
1227 OutputParams const & runparams,
1228 Font const & running_font,
1229 Change const & running_change,
1230 Layout const & style,
1233 unsigned int & column)
1235 char_type const c = owner_->getUChar(bparams, runparams, i);
1237 if (style.pass_thru || runparams.pass_thru
1238 || contains(style.pass_thru_chars, c)
1239 || contains(runparams.pass_thru_chars, c)) {
1241 Encoding const * const enc = runparams.encoding;
1242 if (enc && !enc->encodable(c))
1243 throw EncodingException(c);
1249 // TIPA uses its own T3 encoding
1250 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1252 // If T1 font encoding is used, use the special
1253 // characters it provides.
1254 // NOTE: Some languages reset the font encoding internally to a
1255 // non-standard font encoding. If we are using such a language,
1256 // we do not output special T1 chars.
1257 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1258 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1259 && latexSpecialT1(c, os, i, column))
1261 // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
1262 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1263 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1266 // Otherwise, we use what LaTeX provides us.
1269 os << "\\textbackslash" << termcmd;
1273 os << "\\textless" << termcmd;
1277 os << "\\textgreater" << termcmd;
1281 os << "\\textbar" << termcmd;
1286 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1287 (end_pos == -1 || i + 1 < end_pos) &&
1288 text_[i+1] == '-') {
1289 // Prevent "--" becoming an en dash and "---" an em dash.
1290 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1296 os << "\\textquotedbl" << termcmd;
1301 case '%': case '#': case '{':
1309 os << "\\textasciitilde" << termcmd;
1314 os << "\\textasciicircum" << termcmd;
1321 // avoid being mistaken for optional arguments
1329 // Blanks are printed before font switching.
1330 // Sure? I am not! (try nice-latex)
1331 // I am sure it's correct. LyX might be smarter
1332 // in the future, but for now, nothing wrong is
1338 // XeTeX's dash behaviour is determined via a global setting
1339 if (bparams.use_dash_ligatures
1340 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1342 // TODO #10961: && not in inset Flex Code
1343 // TODO #10961: && not in layout LyXCode
1344 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1361 Encoding const & encoding = *(runparams.encoding);
1362 char_type next = '\0';
1363 if (i + 1 < int(text_.size())) {
1364 next = text_[i + 1];
1365 if (Encodings::isCombiningChar(next)) {
1366 column += latexSurrogatePair(bparams, os, c, next, runparams) - 1;
1372 pair<docstring, bool> latex = encoding.latexChar(c);
1373 docstring nextlatex;
1374 bool nexttipas = false;
1375 string nexttipashortcut;
1376 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1377 nextlatex = encoding.latexChar(next).first;
1378 if (runparams.inIPA) {
1379 nexttipashortcut = Encodings::TIPAShortcut(next);
1380 nexttipas = !nexttipashortcut.empty();
1384 if (runparams.inIPA) {
1385 string const tipashortcut = Encodings::TIPAShortcut(c);
1386 if (!tipashortcut.empty()) {
1387 latex.first = from_ascii(tipashortcut);
1388 latex.second = false;
1393 fontenc = running_font.language()->fontenc(bparams);
1394 // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
1395 // whether they are encodable or not (it only depends on the font encoding)
1396 if (!runparams.isFullUnicode() && Encodings::isKnownScriptChar(c, script)) {
1397 docstring const wrapper = from_ascii("\\" + script + "{");
1398 docstring ltx = latex.first;
1399 if (!prefixIs(ltx, wrapper))
1400 ltx = wrapper + latex.first + from_ascii("}");
1401 column += writeScriptChars(bparams, runparams, os, ltx, running_change,
1402 encoding, fontenc, i) - 1;
1403 } else if (latex.second
1404 && ((!prefixIs(nextlatex, '\\')
1405 && !prefixIs(nextlatex, '{')
1406 && !prefixIs(nextlatex, '}'))
1408 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1410 // Prevent eating of a following
1411 // space or command corruption by
1412 // following characters
1413 if (next == ' ' || next == '\0') {
1414 column += latex.first.length() + 1;
1415 os << latex.first << "{}";
1417 column += latex.first.length();
1418 os << latex.first << " ";
1421 column += latex.first.length() - 1;
1429 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1430 pos_type i, unsigned int & column)
1436 // In T1 encoding, these characters exist
1437 // but we should avoid ligatures
1438 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1440 os << "\\textcompwordmark" << termcmd;
1447 // soul.sty breaks with \char`\"
1448 os << "\\textquotedbl" << termcmd;
1457 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1458 pos_type i, unsigned int & column)
1460 // TU encoding is currently on par with T1.
1461 return latexSpecialT1(c, os, i, column);
1465 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1466 pos_type /*i*/, unsigned int & column)
1476 os << "\\textvertline" << termcmd;
1485 void Paragraph::Private::validate(LaTeXFeatures & features) const
1487 if (layout_->inpreamble && inset_owner_) {
1488 // FIXME: Using a string stream here circumvents the encoding
1489 // switching machinery of odocstream. Therefore the
1490 // output is wrong if this paragraph contains content
1491 // that needs to switch encoding.
1492 Buffer const & buf = inset_owner_->buffer();
1493 otexstringstream os;
1494 os << layout_->preamble();
1495 size_t const length = os.length();
1496 TeXOnePar(buf, buf.text(), buf.getParFromID(owner_->id()).pit(), os,
1497 features.runparams(), string(), 0, -1, true);
1498 if (os.length() > length)
1499 features.addPreambleSnippet(os.release(), true);
1502 if (features.runparams().flavor == OutputParams::HTML
1503 && layout_->htmltitle()) {
1504 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1507 // check the params.
1508 if (!params_.spacing().isDefault())
1509 features.require("setspace");
1512 features.useLayout(layout_->name());
1515 fontlist_.validate(features);
1517 // then the indentation
1518 if (!params_.leftIndent().zero())
1519 features.require("ParagraphLeftIndent");
1522 InsetList::const_iterator icit = insetlist_.begin();
1523 InsetList::const_iterator iend = insetlist_.end();
1524 for (; icit != iend; ++icit) {
1526 features.inDeletedInset(owner_->isDeleted(icit->pos));
1527 if (icit->inset->lyxCode() == FOOT_CODE) {
1528 // FIXME: an item inset would make things much easier.
1529 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1530 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1531 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1532 && (icit->pos < begin_of_body_
1533 || (icit->pos == begin_of_body_
1534 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1535 features.saveNoteEnv("description");
1537 icit->inset->validate(features);
1538 features.inDeletedInset(false);
1539 if (layout_->needprotect &&
1540 icit->inset->lyxCode() == FOOT_CODE)
1541 features.require("NeedLyXFootnoteCode");
1545 // then the contents
1546 BufferParams const bp = features.runparams().is_child
1547 ? features.buffer().masterParams() : features.buffer().params();
1548 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1549 char_type c = text_[i];
1550 CharInfo const & ci = Encodings::unicodeCharInfo(c);
1552 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1553 features.require("textquotedblp");
1554 else if (bp.main_font_encoding() != "T1"
1555 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1556 features.require("textquotedbl");
1557 } else if (ci.textfeature() && contains(ci.textpreamble(), '=')) {
1558 // features that depend on the font or input encoding
1559 string feats = ci.textpreamble();
1560 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc(bp);
1561 if (fontenc.empty())
1562 fontenc = features.runparams().main_fontenc;
1563 while (!feats.empty()) {
1565 feats = split(feats, feat, ',');
1566 if (contains(feat, "!=")) {
1567 // a feature that is required except for the spcified
1568 // font or input encodings
1570 string const contexts = ltrim(split(feat, realfeature, '!'), "=");
1571 // multiple encodings are separated by semicolon
1572 vector<string> context = getVectorFromString(contexts, ";");
1573 // require feature if the context matches neither current font
1574 // nor input encoding
1575 if (std::find(context.begin(), context.end(), fontenc) == context.end()
1576 && std::find(context.begin(), context.end(),
1577 features.runparams().encoding->name()) == context.end())
1578 features.require(realfeature);
1579 } else if (contains(feat, '=')) {
1580 // a feature that is required only for the spcified
1581 // font or input encodings
1583 string const contexts = split(feat, realfeature, '=');
1584 // multiple encodings are separated by semicolon
1585 vector<string> context = getVectorFromString(contexts, ";");
1586 // require feature if the context matches either current font
1587 // or input encoding
1588 if (std::find(context.begin(), context.end(), fontenc) != context.end()
1589 || std::find(context.begin(), context.end(),
1590 features.runparams().encoding->name()) != context.end())
1591 features.require(realfeature);
1594 } else if (!bp.use_dash_ligatures
1595 && (c == 0x2013 || c == 0x2014)
1596 && bp.useNonTeXFonts
1597 && features.runparams().flavor == OutputParams::XETEX)
1598 // XeTeX's dash behaviour is determined via a global setting
1599 features.require("xetexdashbreakstate");
1600 BufferEncodings::validate(c, features);
1604 /////////////////////////////////////////////////////////////////////
1608 /////////////////////////////////////////////////////////////////////
1611 Layout const emptyParagraphLayout;
1614 Paragraph::Paragraph()
1615 : d(new Paragraph::Private(this, emptyParagraphLayout))
1622 Paragraph::Paragraph(Paragraph const & par)
1623 : itemdepth(par.itemdepth),
1624 d(new Paragraph::Private(*par.d, this))
1630 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1631 : itemdepth(par.itemdepth),
1632 d(new Paragraph::Private(*par.d, this, beg, end))
1638 Paragraph & Paragraph::operator=(Paragraph const & par)
1640 // needed as we will destroy the private part before copying it
1642 itemdepth = par.itemdepth;
1646 d = new Private(*par.d, this);
1653 Paragraph::~Paragraph()
1662 // this shall be called just before every "os << ..." action.
1663 void flushString(ostream & os, docstring & s)
1672 void Paragraph::write(ostream & os, BufferParams const & bparams,
1673 depth_type & dth) const
1675 // The beginning or end of a deeper (i.e. nested) area?
1676 if (dth != d->params_.depth()) {
1677 if (d->params_.depth() > dth) {
1678 while (d->params_.depth() > dth) {
1679 os << "\n\\begin_deeper";
1683 while (d->params_.depth() < dth) {
1684 os << "\n\\end_deeper";
1690 // First write the layout
1691 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1693 d->params_.write(os);
1695 Font font1(inherit_font, bparams.language);
1697 Change running_change = Change(Change::UNCHANGED);
1699 // this string is used as a buffer to avoid repetitive calls
1700 // to to_utf8(), which turn out to be expensive (JMarc)
1701 docstring write_buffer;
1704 for (pos_type i = 0; i <= size(); ++i) {
1706 Change const & change = lookupChange(i);
1707 if (change != running_change)
1708 flushString(os, write_buffer);
1709 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1710 running_change = change;
1715 // Write font changes
1716 Font font2 = getFontSettings(bparams, i);
1717 if (font2 != font1) {
1718 flushString(os, write_buffer);
1719 font2.lyxWriteChanges(font1, os);
1724 char_type const c = d->text_[i];
1727 if (Inset const * inset = getInset(i)) {
1728 flushString(os, write_buffer);
1729 if (inset->directWrite()) {
1730 // international char, let it write
1731 // code directly so it's shorter in
1737 os << "\\begin_inset ";
1739 os << "\n\\end_inset\n\n";
1742 // FIXME This can be removed again once the mystery
1743 // crash has been resolved.
1748 flushString(os, write_buffer);
1749 os << "\n\\backslash\n";
1753 flushString(os, write_buffer);
1754 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1761 if ((column > 70 && c == ' ')
1763 flushString(os, write_buffer);
1767 // this check is to amend a bug. LyX sometimes
1768 // inserts '\0' this could cause problems.
1770 write_buffer.push_back(c);
1772 LYXERR0("NUL char in structure.");
1778 flushString(os, write_buffer);
1779 os << "\n\\end_layout\n";
1780 // FIXME This can be removed again once the mystery
1781 // crash has been resolved.
1786 void Paragraph::validate(LaTeXFeatures & features) const
1788 d->validate(features);
1789 bool fragile = features.runparams().moving_arg;
1790 fragile |= layout().needprotect;
1791 if (needsCProtection(fragile))
1792 features.require("cprotect");
1796 void Paragraph::insert(pos_type start, docstring const & str,
1797 Font const & font, Change const & change)
1799 for (size_t i = 0, n = str.size(); i != n ; ++i)
1800 insertChar(start + i, str[i], font, change);
1804 void Paragraph::appendChar(char_type c, Font const & font,
1805 Change const & change)
1808 d->changes_.insert(change, d->text_.size());
1809 // when appending characters, no need to update tables
1810 d->text_.push_back(c);
1811 setFont(d->text_.size() - 1, font);
1812 d->requestSpellCheck(d->text_.size() - 1);
1816 void Paragraph::appendString(docstring const & s, Font const & font,
1817 Change const & change)
1819 pos_type end = s.size();
1820 size_t oldsize = d->text_.size();
1821 size_t newsize = oldsize + end;
1822 size_t capacity = d->text_.capacity();
1823 if (newsize >= capacity)
1824 d->text_.reserve(max(capacity + 100, newsize));
1826 // when appending characters, no need to update tables
1829 // FIXME: Optimize this!
1830 for (size_t i = oldsize; i != newsize; ++i) {
1832 d->changes_.insert(change, i);
1833 d->requestSpellCheck(i);
1835 d->fontlist_.set(oldsize, font);
1836 d->fontlist_.set(newsize - 1, font);
1840 void Paragraph::insertChar(pos_type pos, char_type c,
1843 d->insertChar(pos, c, Change(trackChanges ?
1844 Change::INSERTED : Change::UNCHANGED));
1848 void Paragraph::insertChar(pos_type pos, char_type c,
1849 Font const & font, bool trackChanges)
1851 d->insertChar(pos, c, Change(trackChanges ?
1852 Change::INSERTED : Change::UNCHANGED));
1857 void Paragraph::insertChar(pos_type pos, char_type c,
1858 Font const & font, Change const & change)
1860 d->insertChar(pos, c, change);
1865 void Paragraph::resetFonts(Font const & font)
1867 d->fontlist_.clear();
1868 d->fontlist_.set(0, font);
1869 d->fontlist_.set(d->text_.size() - 1, font);
1872 // Gets uninstantiated font setting at position.
1873 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1877 LYXERR0("pos: " << pos << " size: " << size());
1881 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1882 if (cit != d->fontlist_.end())
1885 if (pos == size() && !empty())
1886 return getFontSettings(bparams, pos - 1);
1888 // Optimisation: avoid a full font instantiation if there is no
1889 // language change from previous call.
1890 static Font previous_font;
1891 static Language const * previous_lang = 0;
1892 Language const * lang = getParLanguage(bparams);
1893 if (lang != previous_lang) {
1894 previous_lang = lang;
1895 previous_font = Font(inherit_font, lang);
1897 return previous_font;
1901 FontSpan Paragraph::fontSpan(pos_type pos) const
1903 LBUFERR(pos <= size());
1906 return FontSpan(pos, pos);
1909 FontList::const_iterator cit = d->fontlist_.begin();
1910 FontList::const_iterator end = d->fontlist_.end();
1911 for (; cit != end; ++cit) {
1912 if (cit->pos() >= pos) {
1913 if (pos >= beginOfBody())
1914 return FontSpan(max(start, beginOfBody()),
1917 return FontSpan(start,
1918 min(beginOfBody() - 1,
1921 start = cit->pos() + 1;
1924 // This should not happen, but if so, we take no chances.
1925 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1926 LASSERT(false, return FontSpan(pos, pos));
1930 // Gets uninstantiated font setting at position 0
1931 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1933 if (!empty() && !d->fontlist_.empty())
1934 return d->fontlist_.begin()->font();
1936 // Optimisation: avoid a full font instantiation if there is no
1937 // language change from previous call.
1938 static Font previous_font;
1939 static Language const * previous_lang = 0;
1940 if (bparams.language != previous_lang) {
1941 previous_lang = bparams.language;
1942 previous_font = Font(inherit_font, bparams.language);
1945 return previous_font;
1949 // Gets the fully instantiated font at a given position in a paragraph
1950 // This is basically the same function as Text::GetFont() in text2.cpp.
1951 // The difference is that this one is used for generating the LaTeX file,
1952 // and thus cosmetic "improvements" are disallowed: This has to deliver
1953 // the true picture of the buffer. (Asger)
1954 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1955 Font const & outerfont) const
1959 Font font = getFontSettings(bparams, pos);
1961 pos_type const body_pos = beginOfBody();
1962 FontInfo & fi = font.fontInfo();
1964 fi.realize(d->layout_->labelfont);
1966 fi.realize(d->layout_->font);
1968 fi.realize(outerfont.fontInfo());
1969 fi.realize(bparams.getFont().fontInfo());
1975 Font const Paragraph::getLabelFont
1976 (BufferParams const & bparams, Font const & outerfont) const
1978 FontInfo tmpfont = d->layout_->labelfont;
1979 tmpfont.realize(outerfont.fontInfo());
1980 tmpfont.realize(bparams.getFont().fontInfo());
1981 return Font(tmpfont, getParLanguage(bparams));
1985 Font const Paragraph::getLayoutFont
1986 (BufferParams const & bparams, Font const & outerfont) const
1988 FontInfo tmpfont = d->layout_->font;
1989 tmpfont.realize(outerfont.fontInfo());
1990 tmpfont.realize(bparams.getFont().fontInfo());
1991 return Font(tmpfont, getParLanguage(bparams));
1995 char_type Paragraph::getUChar(BufferParams const & bparams,
1996 OutputParams const & rp,
1999 char_type c = d->text_[pos];
2001 // Return unchanged character in LTR languages.
2002 if (!getFontSettings(bparams, pos).isRightToLeft())
2005 // FIXME This is a complete mess due to all the language-specific
2006 // special cases. We need to unify this eventually, but this
2007 // requires a file format change and some thought.
2008 // We also need to unify the input of parentheses in different RTL
2009 // languages. Currently, some have their own methods (Arabic:
2010 // 18599/lyxsvn, Hebrew: e5f42f67d/lyxgit), some don't (Urdu, Syriac).
2011 // Also note that the representation in the LyX file is probably wrong
2012 // (see FIXME in TextMetrics::breakRow).
2013 // Most likely, we should simply rely on Qt's unicode handling here.
2014 string const & lang = getFontSettings(bparams, pos).language()->lang();
2016 // With polyglossia, brackets and stuff need not be reversed in RTL scripts
2017 // FIXME: The special casing for Hebrew parens is due to the special
2018 // handling on input (for Hebrew in e5f42f67d/lyxgit); see #8251.
2020 if (rp.use_polyglossia) {
2023 if (lang == "hebrew")
2027 if (lang == "hebrew")
2034 // In the following languages, brackets don't need to be reversed.
2035 // Furthermore, in arabic_arabi, they are transformed to Arabic
2036 // Ornate Parentheses (dunno if this is really wanted)
2037 bool const reversebrackets = lang != "arabic_arabtex"
2038 && lang != "arabic_arabi"
2043 if (reversebrackets)
2047 if (reversebrackets)
2068 void Paragraph::setFont(pos_type pos, Font const & font)
2070 LASSERT(pos <= size(), return);
2072 // First, reduce font against layout/label font
2073 // Update: The setCharFont() routine in text2.cpp already
2074 // reduces font, so we don't need to do that here. (Asger)
2076 d->fontlist_.set(pos, font);
2080 void Paragraph::makeSameLayout(Paragraph const & par)
2082 d->layout_ = par.d->layout_;
2083 d->params_ = par.d->params_;
2087 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2089 if (isFreeSpacing())
2095 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2096 if (eraseChar(pos, trackChanges))
2102 return count > 0 || pos > 0;
2106 bool Paragraph::hasSameLayout(Paragraph const & par) const
2108 return par.d->layout_ == d->layout_
2109 && d->params_.sameLayout(par.d->params_);
2113 depth_type Paragraph::getDepth() const
2115 return d->params_.depth();
2119 depth_type Paragraph::getMaxDepthAfter() const
2121 if (d->layout_->isEnvironment())
2122 return d->params_.depth() + 1;
2124 return d->params_.depth();
2128 LyXAlignment Paragraph::getAlign() const
2130 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2131 return d->layout_->align;
2133 return d->params_.align();
2137 docstring const & Paragraph::labelString() const
2139 return d->params_.labelString();
2143 // the next two functions are for the manual labels
2144 docstring const Paragraph::getLabelWidthString() const
2146 if (d->layout_->margintype == MARGIN_MANUAL
2147 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2148 return d->params_.labelWidthString();
2150 return _("Senseless with this layout!");
2154 void Paragraph::setLabelWidthString(docstring const & s)
2156 d->params_.labelWidthString(s);
2160 docstring Paragraph::expandLabel(Layout const & layout,
2161 BufferParams const & bparams) const
2163 return expandParagraphLabel(layout, bparams, true);
2167 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2168 BufferParams const & bparams) const
2170 return expandParagraphLabel(layout, bparams, false);
2174 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2175 BufferParams const & bparams, bool process_appendix) const
2177 DocumentClass const & tclass = bparams.documentClass();
2178 string const & lang = getParLanguage(bparams)->code();
2179 bool const in_appendix = process_appendix && d->params_.appendix();
2180 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2182 if (fmt.empty() && !layout.counter.empty())
2183 return tclass.counters().theCounter(layout.counter, lang);
2185 // handle 'inherited level parts' in 'fmt',
2186 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2187 size_t const i = fmt.find('@', 0);
2188 if (i != docstring::npos) {
2189 size_t const j = fmt.find('@', i + 1);
2190 if (j != docstring::npos) {
2191 docstring parent(fmt, i + 1, j - i - 1);
2192 docstring label = from_ascii("??");
2193 if (tclass.hasLayout(parent))
2194 label = expandParagraphLabel(tclass[parent], bparams,
2196 fmt = docstring(fmt, 0, i) + label
2197 + docstring(fmt, j + 1, docstring::npos);
2201 return tclass.counters().counterLabel(fmt, lang);
2205 void Paragraph::applyLayout(Layout const & new_layout)
2207 d->layout_ = &new_layout;
2208 LyXAlignment const oldAlign = d->params_.align();
2210 if (!(oldAlign & d->layout_->alignpossible)) {
2211 frontend::Alert::warning(_("Alignment not permitted"),
2212 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2213 d->params_.align(LYX_ALIGN_LAYOUT);
2218 pos_type Paragraph::beginOfBody() const
2220 return d->begin_of_body_;
2224 void Paragraph::setBeginOfBody()
2226 if (d->layout_->labeltype != LABEL_MANUAL) {
2227 d->begin_of_body_ = 0;
2231 // Unroll the first two cycles of the loop
2232 // and remember the previous character to
2233 // remove unnecessary getChar() calls
2235 pos_type end = size();
2236 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2239 char_type previous_char = d->text_[i];
2240 if (!(isNewline(i) || isEnvSeparator(i))) {
2242 while (i < end && previous_char != ' ') {
2243 char_type temp = d->text_[i];
2244 if (isNewline(i) || isEnvSeparator(i))
2247 previous_char = temp;
2253 d->begin_of_body_ = i;
2257 bool Paragraph::allowParagraphCustomization() const
2259 return inInset().allowParagraphCustomization();
2263 bool Paragraph::usePlainLayout() const
2265 return inInset().usePlainLayout();
2269 bool Paragraph::isPassThru() const
2271 return inInset().isPassThru() || d->layout_->pass_thru;
2276 // paragraphs inside floats need different alignment tags to avoid
2279 bool noTrivlistCentering(InsetCode code)
2281 return code == FLOAT_CODE
2282 || code == WRAP_CODE
2283 || code == CELL_CODE;
2287 string correction(string const & orig)
2289 if (orig == "flushleft")
2290 return "raggedright";
2291 if (orig == "flushright")
2292 return "raggedleft";
2293 if (orig == "center")
2299 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2300 InsetCode code, bool const lastpar, int & col)
2302 string macro = suffix + "{";
2303 if (noTrivlistCentering(code)) {
2305 // the last paragraph in non-trivlist-aligned
2306 // context is special (to avoid unwanted whitespace)
2307 if (suffix == "\\begin") {
2308 macro = "\\" + correction(env) + "{}";
2309 os << from_ascii(macro);
2310 col += macro.size();
2315 macro += correction(env);
2319 if (suffix == "\\par\\end") {
2323 os << from_ascii(macro);
2324 col += macro.size();
2325 if (suffix == "\\begin") {
2335 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2336 otexstream & os, OutputParams const & runparams) const
2341 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2342 (layout_->toggle_indent != ITOGGLE_NEVER) :
2343 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2345 if (canindent && params_.noindent() && !layout_->pass_thru) {
2346 os << "\\noindent ";
2350 LyXAlignment const curAlign = params_.align();
2352 if (curAlign == layout_->align)
2356 case LYX_ALIGN_NONE:
2357 case LYX_ALIGN_BLOCK:
2358 case LYX_ALIGN_LAYOUT:
2359 case LYX_ALIGN_SPECIAL:
2360 case LYX_ALIGN_DECIMAL:
2362 case LYX_ALIGN_LEFT:
2363 case LYX_ALIGN_RIGHT:
2364 case LYX_ALIGN_CENTER:
2365 if (runparams.moving_arg) {
2372 string const begin_tag = "\\begin";
2373 InsetCode code = ownerCode();
2374 bool const lastpar = runparams.isLastPar;
2377 case LYX_ALIGN_NONE:
2378 case LYX_ALIGN_BLOCK:
2379 case LYX_ALIGN_LAYOUT:
2380 case LYX_ALIGN_SPECIAL:
2381 case LYX_ALIGN_DECIMAL:
2383 case LYX_ALIGN_LEFT: {
2384 if (!owner_->getParLanguage(bparams)->rightToLeft())
2385 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2387 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2389 } case LYX_ALIGN_RIGHT: {
2390 if (!owner_->getParLanguage(bparams)->rightToLeft())
2391 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2393 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2395 } case LYX_ALIGN_CENTER: {
2396 corrected_env(os, begin_tag, "center", code, lastpar, column);
2405 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2406 otexstream & os, OutputParams const & runparams) const
2408 LyXAlignment const curAlign = params_.align();
2410 if (curAlign == layout_->align)
2414 case LYX_ALIGN_NONE:
2415 case LYX_ALIGN_BLOCK:
2416 case LYX_ALIGN_LAYOUT:
2417 case LYX_ALIGN_SPECIAL:
2418 case LYX_ALIGN_DECIMAL:
2420 case LYX_ALIGN_LEFT:
2421 case LYX_ALIGN_RIGHT:
2422 case LYX_ALIGN_CENTER:
2423 if (runparams.moving_arg)
2428 bool output = false;
2430 string const end_tag = "\\par\\end";
2431 InsetCode code = ownerCode();
2432 bool const lastpar = runparams.isLastPar;
2435 case LYX_ALIGN_NONE:
2436 case LYX_ALIGN_BLOCK:
2437 case LYX_ALIGN_LAYOUT:
2438 case LYX_ALIGN_SPECIAL:
2439 case LYX_ALIGN_DECIMAL:
2441 case LYX_ALIGN_LEFT: {
2442 if (!owner_->getParLanguage(bparams)->rightToLeft())
2443 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2445 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2447 } case LYX_ALIGN_RIGHT: {
2448 if (!owner_->getParLanguage(bparams)->rightToLeft())
2449 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2451 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2453 } case LYX_ALIGN_CENTER: {
2454 corrected_env(os, end_tag, "center", code, lastpar, col);
2459 return output || lastpar;
2463 // This one spits out the text of the paragraph
2464 void Paragraph::latex(BufferParams const & bparams,
2465 Font const & outerfont,
2467 OutputParams const & runparams,
2468 int start_pos, int end_pos, bool force) const
2470 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2472 // FIXME This check should not be needed. Perhaps issue an
2473 // error if it triggers.
2474 Layout const & style = inInset().forcePlainLayout() ?
2475 bparams.documentClass().plainLayout() : *d->layout_;
2477 if (!force && style.inpreamble)
2480 bool const allowcust = allowParagraphCustomization();
2482 // Current base font for all inherited font changes, without any
2483 // change caused by an individual character, except for the language:
2484 // It is set to the language of the first character.
2485 // As long as we are in the label, this font is the base font of the
2486 // label. Before the first body character it is set to the base font
2490 // Maybe we have to create a optional argument.
2491 pos_type body_pos = beginOfBody();
2492 unsigned int column = 0;
2495 // the optional argument is kept in curly brackets in
2496 // case it contains a ']'
2497 // This is not strictly needed, but if this is changed it
2498 // would be a file format change, and tex2lyx would need
2499 // to be adjusted, since it unconditionally removes the
2500 // braces when it parses \item.
2503 basefont = getLabelFont(bparams, outerfont);
2505 basefont = getLayoutFont(bparams, outerfont);
2508 // Which font is currently active?
2509 Font running_font(basefont);
2510 // Do we have an open font change?
2511 bool open_font = false;
2513 Change runningChange = Change(Change::UNCHANGED);
2515 Encoding const * const prev_encoding = runparams.encoding;
2517 os.texrow().start(id(), 0);
2519 // if the paragraph is empty, the loop will not be entered at all
2521 // For InTitle commands, we have already opened a group
2522 // in output_latex::TeXOnePar.
2523 if (style.isCommand() && !style.intitle) {
2527 if (!style.leftdelim().empty()) {
2528 os << style.leftdelim();
2529 column += style.leftdelim().size();
2532 column += d->startTeXParParams(bparams, os, runparams);
2535 // Whether a \par can be issued for insets typeset inline with text.
2536 // Yes if greater than 0. This has to be static.
2537 THREAD_LOCAL_STATIC int parInline = 0;
2539 for (pos_type i = 0; i < size(); ++i) {
2540 // First char in paragraph or after label?
2541 if (i == body_pos) {
2544 bool needPar = false;
2545 column += running_font.latexWriteEndChanges(
2546 os, bparams, runparams,
2547 basefont, basefont, needPar);
2550 basefont = getLayoutFont(bparams, outerfont);
2551 running_font = basefont;
2553 column += Changes::latexMarkChange(os, bparams,
2554 runningChange, Change(Change::UNCHANGED),
2556 runningChange = Change(Change::UNCHANGED);
2561 // For InTitle commands, we have already opened a group
2562 // in output_latex::TeXOnePar.
2563 if (style.isCommand() && !style.intitle) {
2568 if (!style.leftdelim().empty()) {
2569 os << style.leftdelim();
2570 column += style.leftdelim().size();
2574 column += d->startTeXParParams(bparams, os,
2578 runparams.wasDisplayMath = runparams.inDisplayMath;
2579 runparams.inDisplayMath = false;
2580 bool deleted_display_math = false;
2581 Change const & change = runparams.inDeletedInset
2582 ? runparams.changeOfDeletedInset : lookupChange(i);
2584 // Check whether a display math inset follows
2585 if (d->text_[i] == META_INSET
2586 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2587 InsetMath const * im = getInset(i)->asInsetMath();
2588 if (im && im->asHullInset()
2589 && im->asHullInset()->outerDisplay()) {
2590 runparams.inDisplayMath = true;
2591 // runparams.inDeletedInset will be set by
2592 // latexInset later, but we need this info
2593 // before it is called. On the other hand, we
2594 // cannot set it here because it is a counter.
2595 deleted_display_math = isDeleted(i);
2597 if (bparams.output_changes && deleted_display_math
2598 && runningChange == change
2599 && change.type == Change::DELETED
2600 && !os.afterParbreak()) {
2601 // A display math in the same paragraph follows.
2602 // We have to close and then reopen \lyxdeleted,
2603 // otherwise the math will be shifted up.
2604 OutputParams rp = runparams;
2606 bool needPar = false;
2607 column += running_font.latexWriteEndChanges(
2608 os, bparams, rp, basefont,
2612 basefont = getLayoutFont(bparams, outerfont);
2613 running_font = basefont;
2614 column += Changes::latexMarkChange(os, bparams,
2615 Change(Change::INSERTED), change, rp);
2619 if (bparams.output_changes && runningChange != change) {
2621 bool needPar = false;
2622 column += running_font.latexWriteEndChanges(
2623 os, bparams, runparams,
2624 basefont, basefont, needPar);
2627 basefont = getLayoutFont(bparams, outerfont);
2628 running_font = basefont;
2629 column += Changes::latexMarkChange(os, bparams, runningChange,
2631 runningChange = change;
2634 // do not output text which is marked deleted
2635 // if change tracking output is disabled
2636 if (!bparams.output_changes && change.deleted()) {
2642 // Fully instantiated font
2643 Font const current_font = getFont(bparams, i, outerfont);
2645 Font const last_font = running_font;
2647 // Do we need to close the previous font?
2649 (current_font != running_font ||
2650 current_font.language() != running_font.language()))
2652 bool needPar = false;
2653 column += running_font.latexWriteEndChanges(
2654 os, bparams, runparams, basefont,
2655 (i == body_pos-1) ? basefont : current_font,
2657 running_font = basefont;
2661 string const running_lang = runparams.use_polyglossia ?
2662 running_font.language()->polyglossia() : running_font.language()->babel();
2663 // close babel's font environment before opening CJK.
2664 string const lang_end_command = runparams.use_polyglossia ?
2665 "\\end{$$lang}" : lyxrc.language_command_end;
2666 bool const using_begin_end = runparams.use_polyglossia ||
2667 !lang_end_command.empty();
2668 if (!running_lang.empty() &&
2669 (!using_begin_end || running_lang == openLanguageName()) &&
2670 current_font.language()->encoding()->package() == Encoding::CJK) {
2671 string end_tag = subst(lang_end_command,
2674 os << from_ascii(end_tag);
2675 column += end_tag.length();
2676 if (using_begin_end)
2680 // Switch file encoding if necessary (and allowed)
2681 if (!runparams.pass_thru && !style.pass_thru &&
2682 runparams.encoding->package() != Encoding::none &&
2683 current_font.language()->encoding()->package() != Encoding::none) {
2684 pair<bool, int> const enc_switch =
2685 switchEncoding(os.os(), bparams, runparams,
2686 *(current_font.language()->encoding()));
2687 if (enc_switch.first) {
2688 column += enc_switch.second;
2689 runparams.encoding = current_font.language()->encoding();
2693 char_type const c = d->text_[i];
2695 // A display math inset inside an ulem command will be output
2696 // as a box of width \linewidth, so we have to either disable
2697 // indentation if the inset starts a paragraph, or start a new
2698 // line to accommodate such box. This has to be done before
2699 // writing any font changing commands.
2700 if (runparams.inDisplayMath && !deleted_display_math
2701 && runparams.inulemcmd) {
2702 if (os.afterParbreak())
2708 // Do we need to change font?
2709 if ((current_font != running_font ||
2710 current_font.language() != running_font.language()) &&
2713 odocstringstream ods;
2714 column += current_font.latexWriteStartChanges(ods, bparams,
2715 runparams, basefont,
2717 // Check again for display math in ulem commands as a
2718 // font change may also occur just before a math inset.
2719 if (runparams.inDisplayMath && !deleted_display_math
2720 && runparams.inulemcmd) {
2721 if (os.afterParbreak())
2726 running_font = current_font;
2728 docstring fontchange = ods.str();
2729 // check whether the fontchange ends with a \\textcolor
2730 // modifier and the text starts with a space (bug 4473)
2731 docstring const last_modifier = rsplit(fontchange, '\\');
2732 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2733 os << fontchange << from_ascii("{}");
2734 // check if the fontchange ends with a trailing blank
2735 // (like "\small " (see bug 3382)
2736 else if (suffixIs(fontchange, ' ') && c == ' ')
2737 os << fontchange.substr(0, fontchange.size() - 1)
2738 << from_ascii("{}");
2743 // FIXME: think about end_pos implementation...
2744 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2745 // FIXME: integrate this case in latexSpecialChar
2746 // Do not print the separation of the optional argument
2747 // if style.pass_thru is false. This works because
2748 // latexSpecialChar ignores spaces if
2749 // style.pass_thru is false.
2750 if (i != body_pos - 1) {
2751 if (d->simpleTeXBlanks(bparams, runparams, os,
2752 i, column, current_font, style)) {
2753 // A surrogate pair was output. We
2754 // must not call latexSpecialChar
2755 // in this iteration, since it would output
2756 // the combining character again.
2763 OutputParams rp = runparams;
2764 rp.free_spacing = style.free_spacing;
2765 rp.local_font = ¤t_font;
2766 rp.intitle = style.intitle;
2768 // Two major modes: LaTeX or plain
2769 // Handle here those cases common to both modes
2770 // and then split to handle the two modes separately.
2771 if (c == META_INSET) {
2772 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2773 // Greyedout notes and, in general, all insets
2774 // with InsetLayout::isDisplay() == false,
2775 // are typeset inline with the text. So, we
2776 // can add a \par to the last paragraph of
2777 // such insets only if nothing else follows.
2778 bool incremented = false;
2779 Inset const * inset = getInset(i);
2780 InsetText const * textinset = inset
2781 ? inset->asInsetText()
2783 if (i + 1 == size() && textinset
2784 && !inset->getLayout().isDisplay()) {
2785 ParagraphList const & pars =
2786 textinset->text().paragraphs();
2787 pit_type const pit = pars.size() - 1;
2788 Font const lastfont =
2789 pit < 0 || pars[pit].empty()
2790 ? pars[pit].getLayoutFont(
2793 : pars[pit].getFont(bparams,
2794 pars[pit].size() - 1,
2796 if (lastfont.fontInfo().size() !=
2797 basefont.fontInfo().size()) {
2802 d->latexInset(bparams, os, rp, running_font,
2803 basefont, outerfont, open_font,
2804 runningChange, style, i, column);
2808 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2810 d->latexSpecialChar(os, bparams, rp,
2811 running_font, runningChange,
2812 style, i, end_pos, column);
2813 } catch (EncodingException & e) {
2814 if (runparams.dryrun) {
2815 os << "<" << _("LyX Warning: ")
2816 << _("uncodable character") << " '";
2820 // add location information and throw again.
2828 // Set the encoding to that returned from latexSpecialChar (see
2829 // comment for encoding member in OutputParams.h)
2830 runparams.encoding = rp.encoding;
2832 // Also carry on the info on a closed ulem command for insets
2833 // such as Note that do not produce any output, so that no
2834 // command is ever executed but its opening was recorded.
2835 runparams.inulemcmd = rp.inulemcmd;
2838 // If we have an open font definition, we have to close it
2840 // Make sure that \\par is done with the font of the last
2841 // character if this has another size as the default.
2842 // This is necessary because LaTeX (and LyX on the screen)
2843 // calculates the space between the baselines according
2844 // to this font. (Matthias)
2846 // We must not change the font for the last paragraph
2847 // of non-multipar insets, tabular cells or commands,
2848 // since this produces unwanted whitespace.
2850 Font const font = empty()
2851 ? getLayoutFont(bparams, outerfont)
2852 : getFont(bparams, size() - 1, outerfont);
2854 InsetText const * textinset = inInset().asInsetText();
2856 bool const maintext = textinset
2857 ? textinset->text().isMainText()
2860 size_t const numpars = textinset
2861 ? textinset->text().paragraphs().size()
2864 bool needPar = false;
2866 if (style.resfont.size() != font.fontInfo().size()
2867 && (!runparams.isLastPar || maintext
2868 || (numpars > 1 && d->ownerCode() != CELL_CODE
2869 && (inInset().getLayout().isDisplay()
2871 && !style.isCommand()) {
2874 #ifdef FIXED_LANGUAGE_END_DETECTION
2876 running_font.latexWriteEndChanges(os, bparams,
2877 runparams, basefont,
2878 next_->getFont(bparams, 0, outerfont),
2881 running_font.latexWriteEndChanges(os, bparams,
2882 runparams, basefont, basefont, needPar);
2885 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2886 //FIXME: there as we start another \selectlanguage with the next paragraph if
2887 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2888 running_font.latexWriteEndChanges(os, bparams, runparams,
2889 basefont, basefont, needPar);
2892 // The \par could not be inserted at the same nesting
2893 // level of the font size change, so do it now.
2894 os << "{\\" << font.latexSize() << "\\par}";
2898 column += Changes::latexMarkChange(os, bparams, runningChange,
2899 Change(Change::UNCHANGED), runparams);
2901 // Needed if there is an optional argument but no contents.
2902 if (body_pos > 0 && body_pos == size()) {
2906 if (!style.rightdelim().empty()) {
2907 os << style.rightdelim();
2908 column += style.rightdelim().size();
2911 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2912 && runparams.encoding != prev_encoding) {
2913 runparams.encoding = prev_encoding;
2914 os << setEncoding(prev_encoding->iconvName());
2917 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2921 bool Paragraph::emptyTag() const
2923 for (pos_type i = 0; i < size(); ++i) {
2924 if (Inset const * inset = getInset(i)) {
2925 InsetCode lyx_code = inset->lyxCode();
2926 // FIXME testing like that is wrong. What is
2928 if (lyx_code != TOC_CODE &&
2929 lyx_code != INCLUDE_CODE &&
2930 lyx_code != GRAPHICS_CODE &&
2931 lyx_code != ERT_CODE &&
2932 lyx_code != LISTINGS_CODE &&
2933 lyx_code != FLOAT_CODE &&
2934 lyx_code != TABULAR_CODE) {
2938 char_type c = d->text_[i];
2939 if (c != ' ' && c != '\t')
2947 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2950 for (pos_type i = 0; i < size(); ++i) {
2951 if (Inset const * inset = getInset(i)) {
2952 InsetCode lyx_code = inset->lyxCode();
2953 if (lyx_code == LABEL_CODE) {
2954 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2955 docstring const & id = il->getParam("name");
2956 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2964 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2968 for (i = 0; i < size(); ++i) {
2969 if (Inset const * inset = getInset(i)) {
2970 inset->docbook(os, runparams);
2972 char_type c = d->text_[i];
2975 os << sgml::escapeChar(c);
2982 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2986 for (i = 0; i < size(); ++i) {
2987 if (Inset const * inset = getInset(i)) {
2988 inset->xhtml(xs, runparams);
2990 char_type c = d->text_[i];
3000 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
3003 pos_type size = text_.size();
3004 for (pos_type i = initial; i < size; ++i) {
3005 Font font = owner_->getFont(buf.params(), i, outerfont);
3006 if (text_[i] == META_INSET)
3008 if (i != initial && font != font_old)
3017 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
3019 OutputParams const & runparams,
3020 Font const & outerfont,
3021 pos_type initial) const
3023 bool emph_flag = false;
3025 Layout const & style = *d->layout_;
3027 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3029 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3032 // parsing main loop
3033 for (pos_type i = initial; i < size(); ++i) {
3034 Font font = getFont(buf.params(), i, outerfont);
3036 // handle <emphasis> tag
3037 if (font_old.emph() != font.fontInfo().emph()) {
3038 if (font.fontInfo().emph() == FONT_ON) {
3041 } else if (i != initial) {
3042 os << "</emphasis>";
3047 if (Inset const * inset = getInset(i)) {
3048 inset->docbook(os, runparams);
3050 char_type c = d->text_[i];
3052 if (style.pass_thru)
3055 os << sgml::escapeChar(c);
3057 font_old = font.fontInfo();
3061 os << "</emphasis>";
3064 if (style.free_spacing)
3066 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3072 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
3073 vector<html::EndFontTag> & tagsToClose,
3074 bool & flag, FontState curstate, html::FontTypes type)
3076 if (curstate == FONT_ON) {
3077 tagsToOpen.push_back(html::FontTag(type));
3080 tagsToClose.push_back(html::EndFontTag(type));
3087 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3089 OutputParams const & runparams,
3090 Font const & outerfont,
3091 bool start_paragraph, bool close_paragraph,
3092 pos_type initial) const
3096 // track whether we have opened these tags
3097 bool emph_flag = false;
3098 bool bold_flag = false;
3099 bool noun_flag = false;
3100 bool ubar_flag = false;
3101 bool dbar_flag = false;
3102 bool sout_flag = false;
3103 bool xout_flag = false;
3104 bool wave_flag = false;
3106 bool shap_flag = false;
3108 bool faml_flag = false;
3110 bool size_flag = false;
3112 Layout const & style = *d->layout_;
3114 if (start_paragraph)
3115 xs.startDivision(allowEmpty());
3118 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3120 FontShape curr_fs = INHERIT_SHAPE;
3121 FontFamily curr_fam = INHERIT_FAMILY;
3122 FontSize curr_size = FONT_SIZE_INHERIT;
3124 string const default_family =
3125 buf.masterBuffer()->params().fonts_default_family;
3127 vector<html::FontTag> tagsToOpen;
3128 vector<html::EndFontTag> tagsToClose;
3130 // parsing main loop
3131 for (pos_type i = initial; i < size(); ++i) {
3132 // let's not show deleted material in the output
3136 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3139 FontState curstate = font.fontInfo().emph();
3140 if (font_old.emph() != curstate)
3141 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3144 curstate = font.fontInfo().noun();
3145 if (font_old.noun() != curstate)
3146 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3149 curstate = font.fontInfo().underbar();
3150 if (font_old.underbar() != curstate)
3151 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3154 curstate = font.fontInfo().strikeout();
3155 if (font_old.strikeout() != curstate)
3156 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3159 curstate = font.fontInfo().xout();
3160 if (font_old.xout() != curstate)
3161 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3164 curstate = font.fontInfo().uuline();
3165 if (font_old.uuline() != curstate)
3166 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3169 curstate = font.fontInfo().uwave();
3170 if (font_old.uwave() != curstate)
3171 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3174 // a little hackish, but allows us to reuse what we have.
3175 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3176 if (font_old.series() != font.fontInfo().series())
3177 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3180 curr_fs = font.fontInfo().shape();
3181 FontShape old_fs = font_old.shape();
3182 if (old_fs != curr_fs) {
3186 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3189 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3191 case SMALLCAPS_SHAPE:
3192 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3198 // the other tags are for internal use
3206 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3210 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3213 case SMALLCAPS_SHAPE:
3214 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3221 // the other tags are for internal use
3228 curr_fam = font.fontInfo().family();
3229 FontFamily old_fam = font_old.family();
3230 if (old_fam != curr_fam) {
3234 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3237 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3239 case TYPEWRITER_FAMILY:
3240 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3242 case INHERIT_FAMILY:
3245 // the other tags are for internal use
3253 // we will treat a "default" font family as roman, since we have
3254 // no other idea what to do.
3255 if (default_family != "rmdefault" && default_family != "default") {
3256 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3261 if (default_family != "sfdefault") {
3262 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3266 case TYPEWRITER_FAMILY:
3267 if (default_family != "ttdefault") {
3268 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3272 case INHERIT_FAMILY:
3275 // the other tags are for internal use
3282 curr_size = font.fontInfo().size();
3283 FontSize old_size = font_old.size();
3284 if (old_size != curr_size) {
3287 case FONT_SIZE_TINY:
3288 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3290 case FONT_SIZE_SCRIPT:
3291 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3293 case FONT_SIZE_FOOTNOTE:
3294 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3296 case FONT_SIZE_SMALL:
3297 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3299 case FONT_SIZE_LARGE:
3300 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3302 case FONT_SIZE_LARGER:
3303 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3305 case FONT_SIZE_LARGEST:
3306 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3308 case FONT_SIZE_HUGE:
3309 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3311 case FONT_SIZE_HUGER:
3312 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3314 case FONT_SIZE_INCREASE:
3315 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3317 case FONT_SIZE_DECREASE:
3318 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3320 case FONT_SIZE_INHERIT:
3321 case FONT_SIZE_NORMAL:
3324 // the other tags are for internal use
3330 switch (curr_size) {
3331 case FONT_SIZE_TINY:
3332 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3335 case FONT_SIZE_SCRIPT:
3336 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3339 case FONT_SIZE_FOOTNOTE:
3340 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3343 case FONT_SIZE_SMALL:
3344 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3347 case FONT_SIZE_LARGE:
3348 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3351 case FONT_SIZE_LARGER:
3352 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3355 case FONT_SIZE_LARGEST:
3356 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3359 case FONT_SIZE_HUGE:
3360 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3363 case FONT_SIZE_HUGER:
3364 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3367 case FONT_SIZE_INCREASE:
3368 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3371 case FONT_SIZE_DECREASE:
3372 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3375 case FONT_SIZE_NORMAL:
3376 case FONT_SIZE_INHERIT:
3379 // the other tags are for internal use
3386 // Other such tags? What about the other text ranges?
3388 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3389 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3390 for (; cit != cen; ++cit)
3393 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3394 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3395 for (; sit != sen; ++sit)
3398 tagsToClose.clear();
3401 Inset const * inset = getInset(i);
3403 if (!runparams.for_toc || inset->isInToc()) {
3404 OutputParams np = runparams;
3405 np.local_font = &font;
3406 // If the paragraph has size 1, then we are in the "special
3407 // case" where we do not output the containing paragraph info
3408 if (!inset->getLayout().htmlisblock() && size() != 1)
3409 np.html_in_par = true;
3410 retval += inset->xhtml(xs, np);
3413 char_type c = getUChar(buf.masterBuffer()->params(),
3417 font_old = font.fontInfo();
3421 // I'm worried about what happens if a branch, say, is itself
3422 // wrapped in some font stuff. I think that will not work.
3424 if (close_paragraph)
3431 bool Paragraph::isHfill(pos_type pos) const
3433 Inset const * inset = getInset(pos);
3434 return inset && inset->isHfill();
3438 bool Paragraph::isNewline(pos_type pos) const
3440 // U+2028 LINE SEPARATOR
3441 // U+2029 PARAGRAPH SEPARATOR
3442 char_type const c = d->text_[pos];
3443 if (c == 0x2028 || c == 0x2029)
3445 Inset const * inset = getInset(pos);
3446 return inset && inset->lyxCode() == NEWLINE_CODE;
3450 bool Paragraph::isEnvSeparator(pos_type pos) const
3452 Inset const * inset = getInset(pos);
3453 return inset && inset->lyxCode() == SEPARATOR_CODE;
3457 bool Paragraph::isLineSeparator(pos_type pos) const
3459 char_type const c = d->text_[pos];
3460 if (isLineSeparatorChar(c))
3462 Inset const * inset = getInset(pos);
3463 return inset && inset->isLineSeparator();
3467 bool Paragraph::isWordSeparator(pos_type pos) const
3471 if (Inset const * inset = getInset(pos))
3472 return !inset->isLetter();
3473 // if we have a hard hyphen (no en- or emdash) or apostrophe
3474 // we pass this to the spell checker
3475 // FIXME: this method is subject to change, visit
3476 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3477 // to get an impression how complex this is.
3478 if (isHardHyphenOrApostrophe(pos))
3480 char_type const c = d->text_[pos];
3481 // We want to pass the escape chars to the spellchecker
3482 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3483 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3487 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3489 pos_type const psize = size();
3492 char_type const c = d->text_[pos];
3493 if (c != '-' && c != '\'')
3495 int nextpos = pos + 1;
3496 int prevpos = pos > 0 ? pos - 1 : 0;
3497 if ((nextpos == psize || isSpace(nextpos))
3498 && (pos == 0 || isSpace(prevpos)))
3504 bool Paragraph::needsCProtection(bool const fragile) const
3506 // first check the layout of the paragraph, but only in insets
3507 InsetText const * textinset = inInset().asInsetText();
3508 bool const maintext = textinset
3509 ? textinset->text().isMainText()
3512 if (!maintext && layout().needcprotect) {
3513 // Environments need cprotection regardless the content
3514 if (layout().latextype == LATEX_ENVIRONMENT)
3517 // Commands need cprotection if they contain specific chars
3518 int const nchars_escape = 9;
3519 static char_type const chars_escape[nchars_escape] = {
3520 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3522 docstring const pars = asString();
3523 for (int k = 0; k < nchars_escape; k++) {
3524 if (contains(pars, chars_escape[k]))
3529 // now check whether we have insets that need cprotection
3530 pos_type size = d->text_.size();
3531 for (pos_type i = 0; i < size; ++i)
3532 if (isInset(i) && getInset(i)->needsCProtection(maintext, fragile))
3539 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3541 return d->speller_state_.getRange(pos);
3545 bool Paragraph::isChar(pos_type pos) const
3547 if (Inset const * inset = getInset(pos))
3548 return inset->isChar();
3549 char_type const c = d->text_[pos];
3550 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3554 bool Paragraph::isSpace(pos_type pos) const
3556 if (Inset const * inset = getInset(pos))
3557 return inset->isSpace();
3558 char_type const c = d->text_[pos];
3559 return lyx::isSpace(c);
3564 Paragraph::getParLanguage(BufferParams const & bparams) const
3567 return getFirstFontSettings(bparams).language();
3568 // FIXME: we should check the prev par as well (Lgb)
3569 return bparams.language;
3573 bool Paragraph::isRTL(BufferParams const & bparams) const
3575 return getParLanguage(bparams)->rightToLeft()
3576 && !inInset().getLayout().forceLTR();
3580 void Paragraph::changeLanguage(BufferParams const & bparams,
3581 Language const * from, Language const * to)
3583 // change language including dummy font change at the end
3584 for (pos_type i = 0; i <= size(); ++i) {
3585 Font font = getFontSettings(bparams, i);
3586 if (font.language() == from) {
3587 font.setLanguage(to);
3589 d->requestSpellCheck(i);
3595 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3597 Language const * doc_language = bparams.language;
3598 for (auto const & f : d->fontlist_)
3599 if (f.font().language() != ignore_language &&
3600 f.font().language() != latex_language &&
3601 f.font().language() != doc_language)
3607 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3609 for (auto const & f : d->fontlist_) {
3610 Language const * lang = f.font().language();
3611 if (lang != ignore_language &&
3612 lang != latex_language)
3618 docstring Paragraph::asString(int options) const
3620 return asString(0, size(), options);
3624 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3626 odocstringstream os;
3629 && options & AS_STR_LABEL
3630 && !d->params_.labelString().empty())
3631 os << d->params_.labelString() << ' ';
3633 for (pos_type i = beg; i < end; ++i) {
3634 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3636 char_type const c = d->text_[i];
3637 if (isPrintable(c) || c == '\t'
3638 || (c == '\n' && (options & AS_STR_NEWLINES)))
3640 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3641 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3642 LASSERT(runparams != 0, return docstring());
3643 getInset(i)->plaintext(os, *runparams);
3645 getInset(i)->toString(os);
3654 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3655 bool const shorten, bool const label) const
3657 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3658 if (label && !labelString().empty())
3659 os += labelString() + ' ';
3660 if (!layout().isTocCaption())
3662 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3665 char_type const c = d->text_[i];
3668 else if (c == META_INSET)
3669 getInset(i)->forOutliner(os, tmplen, false);
3672 Text::shortenForOutliner(os, maxlen);
3676 void Paragraph::setInsetOwner(Inset const * inset)
3678 d->inset_owner_ = inset;
3682 int Paragraph::id() const
3688 void Paragraph::setId(int id)
3694 Layout const & Paragraph::layout() const
3700 void Paragraph::setLayout(Layout const & layout)
3702 d->layout_ = &layout;
3706 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3708 setLayout(tc.defaultLayout());
3712 void Paragraph::setPlainLayout(DocumentClass const & tc)
3714 setLayout(tc.plainLayout());
3718 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3720 if (usePlainLayout())
3721 setPlainLayout(tclass);
3723 setDefaultLayout(tclass);
3727 Inset const & Paragraph::inInset() const
3729 LBUFERR(d->inset_owner_);
3730 return *d->inset_owner_;
3734 ParagraphParameters & Paragraph::params()
3740 ParagraphParameters const & Paragraph::params() const
3746 bool Paragraph::isFreeSpacing() const
3748 if (d->layout_->free_spacing)
3750 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3754 bool Paragraph::allowEmpty() const
3756 if (d->layout_->keepempty)
3758 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3762 bool Paragraph::brokenBiblio() const
3764 // there is a problem if there is no bibitem at position 0 or
3765 // if there is another bibitem in the paragraph.
3766 return d->layout_->labeltype == LABEL_BIBLIO
3767 && (d->insetlist_.find(BIBITEM_CODE) != 0
3768 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3772 int Paragraph::fixBiblio(Buffer const & buffer)
3774 // FIXME: What about the case where paragraph is not BIBLIO
3775 // but there is an InsetBibitem?
3776 // FIXME: when there was already an inset at 0, the return value is 1,
3777 // which does not tell whether another inset has been remove; the
3778 // cursor cannot be correctly updated.
3780 if (d->layout_->labeltype != LABEL_BIBLIO)
3783 bool const track_changes = buffer.params().track_changes;
3784 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3785 bool const hasbibitem0 = bibitem_pos == 0;
3788 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3789 // There was an InsetBibitem at pos 0, and no other one => OK
3790 if (bibitem_pos == -1)
3792 // there is a bibitem at the 0 position, but since
3793 // there is a second one, we copy the second on the
3794 // first. We're assuming there are at most two of
3795 // these, which there should be.
3796 // FIXME: why does it make sense to do that rather
3797 // than keep the first? (JMarc)
3798 Inset * inset = releaseInset(bibitem_pos);
3799 d->insetlist_.begin()->inset = inset;
3800 return -bibitem_pos;
3803 // We need to create an inset at the beginning
3805 if (bibitem_pos > 0) {
3806 // there was one somewhere in the paragraph, let's move it
3807 inset = d->insetlist_.release(bibitem_pos);
3808 eraseChar(bibitem_pos, track_changes);
3811 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3812 InsetCommandParams(BIBITEM_CODE));
3814 Font font(inherit_font, buffer.params().language);
3815 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3816 : Change::UNCHANGED));
3822 void Paragraph::checkAuthors(AuthorList const & authorList)
3824 d->changes_.checkAuthors(authorList);
3828 bool Paragraph::isChanged(pos_type pos) const
3830 return lookupChange(pos).changed();
3834 bool Paragraph::isInserted(pos_type pos) const
3836 return lookupChange(pos).inserted();
3840 bool Paragraph::isDeleted(pos_type pos) const
3842 return lookupChange(pos).deleted();
3846 InsetList const & Paragraph::insetList() const
3848 return d->insetlist_;
3852 void Paragraph::setInsetBuffers(Buffer & b)
3854 d->insetlist_.setBuffer(b);
3858 void Paragraph::resetBuffer()
3860 d->insetlist_.resetBuffer();
3864 Inset * Paragraph::releaseInset(pos_type pos)
3866 Inset * inset = d->insetlist_.release(pos);
3867 /// does not honour change tracking!
3868 eraseChar(pos, false);
3873 Inset * Paragraph::getInset(pos_type pos)
3875 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3876 ? d->insetlist_.get(pos) : 0;
3880 Inset const * Paragraph::getInset(pos_type pos) const
3882 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3883 ? d->insetlist_.get(pos) : 0;
3887 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3888 pos_type & right, TextCase action)
3890 // process sequences of modified characters; in change
3891 // tracking mode, this approach results in much better
3892 // usability than changing case on a char-by-char basis
3893 // We also need to track the current font, since font
3894 // changes within sequences can occur.
3895 vector<pair<char_type, Font> > changes;
3897 bool const trackChanges = bparams.track_changes;
3899 bool capitalize = true;
3901 for (; pos < right; ++pos) {
3902 char_type oldChar = d->text_[pos];
3903 char_type newChar = oldChar;
3905 // ignore insets and don't play with deleted text!
3906 if (oldChar != META_INSET && !isDeleted(pos)) {
3908 case text_lowercase:
3909 newChar = lowercase(oldChar);
3911 case text_capitalization:
3913 newChar = uppercase(oldChar);
3917 case text_uppercase:
3918 newChar = uppercase(oldChar);
3923 if (isWordSeparator(pos) || isDeleted(pos)) {
3924 // permit capitalization again
3928 if (oldChar != newChar) {
3929 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3930 if (pos != right - 1)
3932 // step behind the changing area
3936 int erasePos = pos - changes.size();
3937 for (size_t i = 0; i < changes.size(); i++) {
3938 insertChar(pos, changes[i].first,
3941 if (!eraseChar(erasePos, trackChanges)) {
3944 ++right; // expand selection
3952 int Paragraph::find(docstring const & str, bool cs, bool mw,
3953 pos_type start_pos, bool del) const
3955 pos_type pos = start_pos;
3956 int const strsize = str.length();
3958 pos_type const parsize = d->text_.size();
3959 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3960 // Ignore "invisible" letters such as ligature breaks
3961 // and hyphenation chars while searching
3962 while (pos < parsize - 1 && isInset(pos)) {
3963 odocstringstream os;
3964 getInset(pos)->toString(os);
3965 if (!getInset(pos)->isLetter() || !os.str().empty())
3969 if (cs && str[i] != d->text_[pos])
3971 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3973 if (!del && isDeleted(pos))
3980 // if necessary, check whether string matches word
3982 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3985 && !isWordSeparator(pos))
3989 return pos - start_pos;
3993 char_type Paragraph::getChar(pos_type pos) const
3995 return d->text_[pos];
3999 pos_type Paragraph::size() const
4001 return d->text_.size();
4005 bool Paragraph::empty() const
4007 return d->text_.empty();
4011 bool Paragraph::isInset(pos_type pos) const
4013 return d->text_[pos] == META_INSET;
4017 bool Paragraph::isSeparator(pos_type pos) const
4019 //FIXME: Are we sure this can be the only separator?
4020 return d->text_[pos] == ' ';
4024 void Paragraph::deregisterWords()
4026 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4027 Private::LangWordsMap::const_iterator ite = d->words_.end();
4028 for (; itl != ite; ++itl) {
4029 WordList & wl = theWordList(itl->first);
4030 Private::Words::const_iterator it = (itl->second).begin();
4031 Private::Words::const_iterator et = (itl->second).end();
4032 for (; it != et; ++it)
4039 void Paragraph::locateWord(pos_type & from, pos_type & to,
4040 word_location const loc) const
4043 case WHOLE_WORD_STRICT:
4044 if (from == 0 || from == size()
4045 || isWordSeparator(from)
4046 || isWordSeparator(from - 1)) {
4053 // If we are already at the beginning of a word, do nothing
4054 if (!from || isWordSeparator(from - 1))
4059 // always move the cursor to the beginning of previous word
4060 while (from && !isWordSeparator(from - 1))
4064 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4067 // no need to move the 'from' cursor
4071 while (to < size() && !isWordSeparator(to))
4076 void Paragraph::collectWords()
4078 for (pos_type pos = 0; pos < size(); ++pos) {
4079 if (isWordSeparator(pos))
4081 pos_type from = pos;
4082 locateWord(from, pos, WHOLE_WORD);
4083 // Work around MSVC warning: The statement
4084 // if (pos < from + lyxrc.completion_minlength)
4085 // triggers a signed vs. unsigned warning.
4086 // I don't know why this happens, it could be a MSVC bug, or
4087 // related to LLP64 (windows) vs. LP64 (unix) programming
4088 // model, or the C++ standard might be ambigous in the section
4089 // defining the "usual arithmetic conversions". However, using
4090 // a temporary variable is safe and works on all compilers.
4091 pos_type const endpos = from + lyxrc.completion_minlength;
4094 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4095 if (cit == d->fontlist_.end())
4097 Language const * lang = cit->font().language();
4098 docstring const word = asString(from, pos, AS_STR_NONE);
4099 d->words_[lang->lang()].insert(word);
4104 void Paragraph::registerWords()
4106 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4107 Private::LangWordsMap::const_iterator ite = d->words_.end();
4108 for (; itl != ite; ++itl) {
4109 WordList & wl = theWordList(itl->first);
4110 Private::Words::const_iterator it = (itl->second).begin();
4111 Private::Words::const_iterator et = (itl->second).end();
4112 for (; it != et; ++it)
4118 void Paragraph::updateWords()
4126 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4128 SkipPositionsIterator begin = skips.begin();
4129 SkipPositions::iterator end = skips.end();
4130 if (pos > 0 && begin < end) {
4132 if (end->last == pos - 1) {
4137 skips.insert(end, FontSpan(pos, pos));
4141 Language * Paragraph::Private::locateSpellRange(
4142 pos_type & from, pos_type & to,
4143 SkipPositions & skips) const
4145 // skip leading white space
4146 while (from < to && owner_->isWordSeparator(from))
4148 // don't check empty range
4151 // get current language
4152 Language * lang = getSpellLanguage(from);
4153 pos_type last = from;
4154 bool samelang = true;
4155 bool sameinset = true;
4156 while (last < to && samelang && sameinset) {
4157 // hop to end of word
4158 while (last < to && !owner_->isWordSeparator(last)) {
4159 if (owner_->getInset(last)) {
4160 appendSkipPosition(skips, last);
4161 } else if (owner_->isDeleted(last)) {
4162 appendSkipPosition(skips, last);
4166 // hop to next word while checking for insets
4167 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4168 if (Inset const * inset = owner_->getInset(last))
4169 sameinset = inset->isChar() && inset->isLetter();
4170 if (sameinset && owner_->isDeleted(last)) {
4171 appendSkipPosition(skips, last);
4176 if (sameinset && last < to) {
4177 // now check for language change
4178 samelang = lang == getSpellLanguage(last);
4181 // if language change detected backstep is needed
4189 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4192 const_cast<Language *>(owner_->getFontSettings(
4193 inset_owner_->buffer().params(), from).language());
4194 if (lang == inset_owner_->buffer().params().language
4195 && !lyxrc.spellchecker_alt_lang.empty()) {
4197 string const lang_variety =
4198 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4199 lang->setCode(lang_code);
4200 lang->setVariety(lang_variety);
4206 void Paragraph::requestSpellCheck(pos_type pos)
4208 d->requestSpellCheck(pos);
4212 bool Paragraph::needsSpellCheck() const
4214 SpellChecker::ChangeNumber speller_change_number = 0;
4215 if (theSpellChecker())
4216 speller_change_number = theSpellChecker()->changeNumber();
4217 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4218 d->speller_state_.needsCompleteRefresh(speller_change_number);
4220 return d->needsSpellCheck();
4224 bool Paragraph::Private::ignoreWord(docstring const & word) const
4226 // Ignore words with digits
4227 // FIXME: make this customizable
4228 // (note that some checkers ignore words with digits by default)
4229 docstring::const_iterator cit = word.begin();
4230 docstring::const_iterator const end = word.end();
4231 for (; cit != end; ++cit) {
4232 if (isNumber((*cit)))
4239 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4240 WordLangTuple & wl, docstring_list & suggestions,
4241 bool do_suggestion, bool check_learned) const
4243 SpellChecker::Result result = SpellChecker::WORD_OK;
4244 SpellChecker * speller = theSpellChecker();
4248 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4251 locateWord(from, to, WHOLE_WORD);
4252 if (from == to || from >= size())
4255 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4256 Language * lang = d->getSpellLanguage(from);
4258 if (getFontSettings(d->inset_owner_->buffer().params(), from).fontInfo().nospellcheck() == FONT_ON)
4261 wl = WordLangTuple(word, lang);
4266 if (needsSpellCheck() || check_learned) {
4268 if (!d->ignoreWord(word)) {
4269 bool const trailing_dot = to < size() && d->text_[to] == '.';
4270 result = speller->check(wl);
4271 if (SpellChecker::misspelled(result) && trailing_dot) {
4272 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4273 result = speller->check(wl);
4274 if (!SpellChecker::misspelled(result)) {
4275 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4277 from << ".." << to << "]");
4279 // spell check with dot appended failed too
4280 // restore original word/lang value
4281 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4282 wl = WordLangTuple(word, lang);
4286 if (!SpellChecker::misspelled(result)) {
4287 // area up to the begin of the next word is not misspelled
4288 while (end < size() && isWordSeparator(end))
4291 d->setMisspelled(from, end, result);
4293 result = d->speller_state_.getState(from);
4297 suggestions.clear();
4299 if (SpellChecker::misspelled(result)) {
4300 LYXERR(Debug::GUI, "misspelled word: \"" <<
4302 from << ".." << to << "]");
4304 speller->suggest(wl, suggestions);
4310 void Paragraph::anonymize()
4312 // This is a very crude anonymization for now
4313 for (char_type & c : d->text_)
4314 if (isLetterChar(c) || isNumber(c))
4319 void Paragraph::Private::markMisspelledWords(
4320 pos_type const & first, pos_type const & last,
4321 SpellChecker::Result result,
4322 docstring const & word,
4323 SkipPositions const & skips)
4325 if (!SpellChecker::misspelled(result)) {
4326 setMisspelled(first, last, SpellChecker::WORD_OK);
4330 SpellChecker * speller = theSpellChecker();
4331 // locate and enumerate the error positions
4332 int nerrors = speller->numMisspelledWords();
4334 SkipPositionsIterator it = skips.begin();
4335 SkipPositionsIterator et = skips.end();
4336 for (int index = 0; index < nerrors; ++index) {
4339 speller->misspelledWord(index, wstart, wlen);
4340 /// should not happen if speller supports range checks
4341 if (!wlen) continue;
4342 docstring const misspelled = word.substr(wstart, wlen);
4343 wstart += first + numskipped;
4344 if (snext < wstart) {
4345 /// mark the range of correct spelling
4346 numskipped += countSkips(it, et, wstart);
4347 setMisspelled(snext,
4348 wstart - 1, SpellChecker::WORD_OK);
4350 snext = wstart + wlen;
4351 numskipped += countSkips(it, et, snext);
4352 /// mark the range of misspelling
4353 setMisspelled(wstart, snext, result);
4354 LYXERR(Debug::GUI, "misspelled word: \"" <<
4355 misspelled << "\" [" <<
4356 wstart << ".." << (snext-1) << "]");
4359 if (snext <= last) {
4360 /// mark the range of correct spelling at end
4361 setMisspelled(snext, last, SpellChecker::WORD_OK);
4366 void Paragraph::spellCheck() const
4368 SpellChecker * speller = theSpellChecker();
4369 if (!speller || empty() ||!needsSpellCheck())
4373 d->rangeOfSpellCheck(start, endpos);
4374 if (speller->canCheckParagraph()) {
4375 // loop until we leave the range
4376 for (pos_type first = start; first < endpos; ) {
4377 pos_type last = endpos;
4378 Private::SkipPositions skips;
4379 Language * lang = d->locateSpellRange(first, last, skips);
4380 if (first >= endpos)
4382 // start the spell checker on the unit of meaning
4383 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4384 WordLangTuple wl = WordLangTuple(word, lang);
4385 SpellChecker::Result result = word.size() ?
4386 speller->check(wl) : SpellChecker::WORD_OK;
4387 d->markMisspelledWords(first, last, result, word, skips);
4391 static docstring_list suggestions;
4392 pos_type to = endpos;
4393 while (start < endpos) {
4395 spellCheck(start, to, wl, suggestions, false);
4399 d->readySpellCheck();
4403 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4405 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4406 if (result || pos <= 0 || pos > size())
4408 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4409 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4414 string Paragraph::magicLabel() const
4417 ss << "magicparlabel-" << id();