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(otexstream & os, char_type c, char_type next,
318 OutputParams const &);
320 /// Output a space in appropriate formatting (or a surrogate pair
321 /// if the next character is a combining character).
322 /// \return whether a surrogate pair was output.
323 bool simpleTeXBlanks(OutputParams const &,
326 unsigned int & column,
328 Layout const & style);
330 /// Output consecutive unicode chars, belonging to the same script as
331 /// specified by the latex macro \p ltx, to \p os starting from \p i.
332 /// \return the number of characters written.
333 int writeScriptChars(OutputParams const &, otexstream & os,
334 docstring const & ltx,
335 Change const &, Encoding const &,
336 std::string const, pos_type & i);
338 /// This could go to ParagraphParameters if we want to.
339 int startTeXParParams(BufferParams const &, otexstream &,
340 OutputParams const &) const;
342 /// This could go to ParagraphParameters if we want to.
343 bool endTeXParParams(BufferParams const &, otexstream &,
344 OutputParams const &) const;
347 void latexInset(BufferParams const &,
352 Font const & outerfont,
354 Change & running_change,
355 Layout const & style,
357 unsigned int & column);
360 void latexSpecialChar(
362 BufferParams const & bparams,
363 OutputParams const & runparams,
364 Font const & running_font,
365 Change const & running_change,
366 Layout const & style,
369 unsigned int & column);
376 unsigned int & column);
382 unsigned int & column);
388 unsigned int & column);
391 void validate(LaTeXFeatures & features) const;
393 /// Checks if the paragraph contains only text and no inset or font change.
394 bool onlyText(Buffer const & buf, Font const & outerfont,
395 pos_type initial) const;
397 /// a vector of speller skip positions
398 typedef vector<FontSpan> SkipPositions;
399 typedef SkipPositions::const_iterator SkipPositionsIterator;
401 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
403 Language * getSpellLanguage(pos_type const from) const;
405 Language * locateSpellRange(pos_type & from, pos_type & to,
406 SkipPositions & skips) const;
408 bool hasSpellerChange() const
410 SpellChecker::ChangeNumber speller_change_number = 0;
411 if (theSpellChecker())
412 speller_change_number = theSpellChecker()->changeNumber();
413 return speller_change_number > speller_state_.currentChangeNumber();
416 bool ignoreWord(docstring const & word) const ;
418 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
420 pos_type textsize = owner_->size();
421 // check for sane arguments
422 if (to <= from || from >= textsize)
424 FontSpan fp = FontSpan(from, to - 1);
425 speller_state_.setRange(fp, state);
428 void requestSpellCheck(pos_type pos)
431 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
433 speller_state_.needsRefresh(pos);
436 void readySpellCheck()
438 speller_state_.needsRefresh(-1);
441 bool needsSpellCheck() const
443 return speller_state_.needsRefresh();
446 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
448 speller_state_.refreshRange(first, last);
450 last = owner_->size();
453 pos_type endpos = last;
454 owner_->locateWord(first, endpos, WHOLE_WORD);
457 owner_->locateWord(last, endpos, WHOLE_WORD);
462 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
466 while (it != et && it->first < start) {
467 int skip = it->last - it->first + 1;
475 void markMisspelledWords(pos_type const & first, pos_type const & last,
476 SpellChecker::Result result,
477 docstring const & word,
478 SkipPositions const & skips);
480 InsetCode ownerCode() const
482 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
485 /// Which Paragraph owns us?
489 Inset const * inset_owner_;
498 ParagraphParameters params_;
500 /// for recording and looking up changes
504 InsetList insetlist_;
507 pos_type begin_of_body_;
509 typedef docstring TextContainer;
513 typedef set<docstring> Words;
514 typedef map<string, Words> LangWordsMap;
518 Layout const * layout_;
520 SpellCheckerState speller_state_;
524 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
525 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
532 int Paragraph::Private::make_id()
534 // The id is unique per session across buffers because it is used in
535 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
538 static atomic_uint next_id(0);
543 Paragraph::Private::Private(Private const & p, Paragraph * owner)
544 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
546 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
547 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
550 requestSpellCheck(p.text_.size());
554 Paragraph::Private::Private(Private const & p, Paragraph * owner,
555 pos_type beg, pos_type end)
556 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
557 params_(p.params_), changes_(p.changes_),
558 insetlist_(p.insetlist_, beg, end),
559 begin_of_body_(p.begin_of_body_), words_(p.words_),
562 if (beg >= pos_type(p.text_.size()))
564 text_ = p.text_.substr(beg, end - beg);
566 FontList::const_iterator fcit = fontlist_.begin();
567 FontList::const_iterator fend = fontlist_.end();
568 for (; fcit != fend; ++fcit) {
569 if (fcit->pos() < beg)
571 if (fcit->pos() >= end) {
572 // Add last entry in the fontlist_.
573 fontlist_.set(text_.size() - 1, fcit->font());
576 // Add a new entry in the fontlist_.
577 fontlist_.set(fcit->pos() - beg, fcit->font());
579 requestSpellCheck(p.text_.size());
583 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
584 bool output_active, TocBackend & backend) const
586 d->changes_.addToToc(cdit, buf, output_active, backend);
590 void Paragraph::addChangesToBuffer(Buffer const & buf) const
592 d->changes_.updateBuffer(buf);
596 bool Paragraph::isChangeUpdateRequired() const
598 return d->changes_.isUpdateRequired();
602 bool Paragraph::isDeleted(pos_type start, pos_type end) const
604 LASSERT(start >= 0 && start <= size(), return false);
605 LASSERT(end > start && end <= size() + 1, return false);
607 return d->changes_.isDeleted(start, end);
611 bool Paragraph::isChanged(pos_type start, pos_type end) const
613 LASSERT(start >= 0 && start <= size(), return false);
614 LASSERT(end > start && end <= size() + 1, return false);
616 return d->changes_.isChanged(start, end);
620 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
622 // keep the logic here in sync with the logic of eraseChars()
626 Change const & change = d->changes_.lookup(size());
627 return change.inserted() && change.currentAuthor();
631 void Paragraph::setChange(Change const & change)
633 // beware of the imaginary end-of-par character!
634 d->changes_.set(change, 0, size() + 1);
637 * Propagate the change recursively - but not in case of DELETED!
639 * Imagine that your co-author makes changes in an existing inset. He
640 * sends your document to you and you come to the conclusion that the
641 * inset should go completely. If you erase it, LyX must not delete all
642 * text within the inset. Otherwise, the change tracked insertions of
643 * your co-author get lost and there is no way to restore them later.
645 * Conclusion: An inset's content should remain untouched if you delete it
648 if (!change.deleted()) {
649 for (pos_type pos = 0; pos < size(); ++pos) {
650 if (Inset * inset = getInset(pos))
651 inset->setChange(change);
657 void Paragraph::setChange(pos_type pos, Change const & change)
659 LASSERT(pos >= 0 && pos <= size(), return);
660 d->changes_.set(change, pos);
662 // see comment in setChange(Change const &) above
663 if (!change.deleted() && pos < size())
664 if (Inset * inset = getInset(pos))
665 inset->setChange(change);
669 Change const & Paragraph::lookupChange(pos_type pos) const
671 LBUFERR(pos >= 0 && pos <= size());
672 return d->changes_.lookup(pos);
676 void Paragraph::acceptChanges(pos_type start, pos_type end)
678 LASSERT(start >= 0 && start <= size(), return);
679 LASSERT(end > start && end <= size() + 1, return);
681 for (pos_type pos = start; pos < end; ++pos) {
682 switch (lookupChange(pos).type) {
683 case Change::UNCHANGED:
684 // accept changes in nested inset
685 if (Inset * inset = getInset(pos))
686 inset->acceptChanges();
689 case Change::INSERTED:
690 d->changes_.set(Change(Change::UNCHANGED), pos);
691 // also accept changes in nested inset
692 if (Inset * inset = getInset(pos))
693 inset->acceptChanges();
696 case Change::DELETED:
697 // Suppress access to non-existent
698 // "end-of-paragraph char"
700 eraseChar(pos, false);
711 void Paragraph::rejectChanges(pos_type start, pos_type end)
713 LASSERT(start >= 0 && start <= size(), return);
714 LASSERT(end > start && end <= size() + 1, return);
716 for (pos_type pos = start; pos < end; ++pos) {
717 switch (lookupChange(pos).type) {
718 case Change::UNCHANGED:
719 // reject changes in nested inset
720 if (Inset * inset = getInset(pos))
721 inset->rejectChanges();
724 case Change::INSERTED:
725 // Suppress access to non-existent
726 // "end-of-paragraph char"
728 eraseChar(pos, false);
734 case Change::DELETED:
735 d->changes_.set(Change(Change::UNCHANGED), pos);
737 // Do NOT reject changes within a deleted inset!
738 // There may be insertions of a co-author inside of it!
746 void Paragraph::Private::insertChar(pos_type pos, char_type c,
747 Change const & change)
749 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
752 changes_.insert(change, pos);
754 // This is actually very common when parsing buffers (and
755 // maybe inserting ascii text)
756 if (pos == pos_type(text_.size())) {
757 // when appending characters, no need to update tables
759 // but we want spell checking
760 requestSpellCheck(pos);
764 text_.insert(text_.begin() + pos, c);
766 // Update the font table.
767 fontlist_.increasePosAfterPos(pos);
770 insetlist_.increasePosAfterPos(pos);
772 // Update list of misspelled positions
773 speller_state_.increasePosAfterPos(pos);
777 bool Paragraph::insertInset(pos_type pos, Inset * inset,
778 Font const & font, Change const & change)
780 LASSERT(inset, return false);
781 LASSERT(pos >= 0 && pos <= size(), return false);
783 // Paragraph::insertInset() can be used in cut/copy/paste operation where
784 // d->inset_owner_ is not set yet.
785 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
788 d->insertChar(pos, META_INSET, change);
789 LASSERT(d->text_[pos] == META_INSET, return false);
791 // Add a new entry in the insetlist_.
792 d->insetlist_.insert(inset, pos);
794 // Some insets require run of spell checker
795 requestSpellCheck(pos);
801 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
803 LASSERT(pos >= 0 && pos <= size(), return false);
805 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
808 Change change = d->changes_.lookup(pos);
810 // set the character to DELETED if
811 // a) it was previously unchanged or
812 // b) it was inserted by a co-author
814 if (!change.changed() ||
815 (change.inserted() && !change.currentAuthor())) {
816 setChange(pos, Change(Change::DELETED));
817 // request run of spell checker
818 requestSpellCheck(pos);
822 if (change.deleted())
826 // Don't physically access the imaginary end-of-paragraph character.
827 // eraseChar() can only mark it as DELETED. A physical deletion of
828 // end-of-par must be handled externally.
834 d->changes_.erase(pos);
836 // if it is an inset, delete the inset entry
837 if (d->text_[pos] == META_INSET)
838 d->insetlist_.erase(pos);
840 d->text_.erase(d->text_.begin() + pos);
842 // Update the fontlist_
843 d->fontlist_.erase(pos);
845 // Update the insetlist_
846 d->insetlist_.decreasePosAfterPos(pos);
848 // Update list of misspelled positions
849 d->speller_state_.decreasePosAfterPos(pos);
850 d->speller_state_.refreshLast(size());
856 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
858 LASSERT(start >= 0 && start <= size(), return 0);
859 LASSERT(end >= start && end <= size() + 1, return 0);
862 for (pos_type count = end - start; count; --count) {
863 if (!eraseChar(i, trackChanges))
870 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
871 char_type next, OutputParams const & runparams)
873 // Writing next here may circumvent a possible font change between
874 // c and next. Since next is only output if it forms a surrogate pair
875 // with c we can ignore this:
876 // A font change inside a surrogate pair does not make sense and is
877 // hopefully impossible to input.
878 // FIXME: change tracking
879 // Is this correct WRT change tracking?
880 Encoding const & encoding = *(runparams.encoding);
881 docstring latex1 = encoding.latexChar(next).first;
882 if (runparams.inIPA) {
883 string const tipashortcut = Encodings::TIPAShortcut(next);
884 if (!tipashortcut.empty()) {
885 latex1 = from_ascii(tipashortcut);
888 docstring latex2 = encoding.latexChar(c).first;
890 docstring::size_type const brace1 = latex2.find_first_of(from_ascii("{"));
891 docstring::size_type const brace2 = latex2.find_last_of(from_ascii("}"));
892 string script = to_ascii(latex2.substr(1, brace1 - 1));
896 if (runparams.local_font)
897 fontenc = runparams.local_font->language()->fontenc();
899 fontenc = runparams.main_fontenc;
900 if ((script == "textgreek" && fontenc == "LGR")
901 ||(script == "textcyr" && (fontenc == "T2A" || fontenc == "T2B"
902 || fontenc == "T2C" || fontenc == "X2"))) {
903 // Correct font encoding is being used, so we can avoid \text[greek|cyr].
906 latex2 = latex2.substr(pos, length);
909 if (docstring(1, next) == latex1) {
910 // the encoding supports the combination
911 os << latex2 << latex1;
912 return latex1.length() + latex2.length();
913 } else if (runparams.local_font &&
914 runparams.local_font->language()->lang() == "polutonikogreek") {
915 // polutonikogreek only works without the brackets
916 os << latex1 << latex2;
917 return latex1.length() + latex2.length();
919 os << latex1 << '{' << latex2 << '}';
920 return latex1.length() + latex2.length() + 2;
924 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
927 unsigned int & column,
929 Layout const & style)
931 if (style.pass_thru || runparams.pass_thru)
934 if (i + 1 < int(text_.size())) {
935 char_type next = text_[i + 1];
936 if (Encodings::isCombiningChar(next)) {
937 // This space has an accent, so we must always output it.
938 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
943 if (runparams.linelen > 0
944 && column > runparams.linelen
946 && text_[i - 1] != ' '
947 && (i + 1 < int(text_.size()))
948 // same in FreeSpacing mode
949 && !owner_->isFreeSpacing()
950 // In typewriter mode, we want to avoid
951 // ! . ? : at the end of a line
952 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
953 && (text_[i - 1] == '.'
954 || text_[i - 1] == '?'
955 || text_[i - 1] == ':'
956 || text_[i - 1] == '!'))) {
958 os.texrow().start(owner_->id(), i + 1);
960 } else if (style.free_spacing) {
969 int Paragraph::Private::writeScriptChars(OutputParams const & runparams,
971 docstring const & ltx,
972 Change const & runningChange,
973 Encoding const & encoding,
974 string const fontenc,
977 // FIXME: modifying i here is not very nice...
979 // We only arrive here when character text_[i] could not be translated
980 // into the current latex encoding (or its latex translation has been forced,)
981 // and it belongs to a known script.
982 // Parameter ltx contains the latex translation of text_[i] as specified
983 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
984 // The latex macro name "textXXX" specifies the script to which text_[i]
985 // belongs and we use it in order to check whether characters from the
986 // same script immediately follow, such that we can collect them in a
987 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
988 // for the first char but only "<spec>" for all subsequent chars.
989 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
990 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
991 string script = to_ascii(ltx.substr(1, brace1 - 1));
994 bool closing_brace = true;
995 if ((script == "textgreek" && fontenc == "LGR")
996 ||(script == "textcyr" && (fontenc == "T2A" || fontenc == "T2B"
997 || fontenc == "T2C" || fontenc == "X2"))) {
998 // Correct font encoding is being used, so we can avoid \text[greek|cyr].
1001 closing_brace = false;
1003 os << ltx.substr(pos, length);
1004 int size = text_.size();
1005 while (i + 1 < size) {
1006 char_type const next = text_[i + 1];
1007 // Stop here if next character belongs to another script
1008 // or there is a change in change tracking status.
1009 if (!Encodings::isKnownScriptChar(next, script) ||
1010 runningChange != owner_->lookupChange(i + 1))
1014 FontList::const_iterator cit = fontlist_.begin();
1015 FontList::const_iterator end = fontlist_.end();
1016 for (; cit != end; ++cit) {
1017 if (cit->pos() >= i && !found) {
1018 prev_font = cit->font();
1021 if (cit->pos() >= i + 1)
1024 // Stop here if there is a font attribute or encoding change.
1025 if (found && cit != end && prev_font != cit->font())
1028 // Check whether we have a combining pair
1029 char_type next_next = '\0';
1031 next_next = text_[i + 2];
1032 if (Encodings::isCombiningChar(next_next)) {
1033 length += latexSurrogatePair(os, next, next_next, runparams) - 1;
1039 docstring const latex = encoding.latexChar(next).first;
1040 docstring::size_type const b1 =
1041 latex.find_first_of(from_ascii("{"));
1042 docstring::size_type const b2 =
1043 latex.find_last_of(from_ascii("}"));
1044 int const len = b2 - b1 - 1;
1045 os << latex.substr(b1 + 1, len);
1049 if (closing_brace) {
1057 void Paragraph::Private::latexInset(BufferParams const & bparams,
1059 OutputParams & runparams,
1060 Font & running_font,
1062 Font const & outerfont,
1064 Change & running_change,
1065 Layout const & style,
1067 unsigned int & column)
1069 Inset * inset = owner_->getInset(i);
1072 if (style.pass_thru) {
1073 odocstringstream ods;
1074 inset->plaintext(ods, runparams);
1079 // FIXME: move this to InsetNewline::latex
1080 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1081 // newlines are handled differently here than
1082 // the default in simpleTeXSpecialChars().
1083 if (!style.newline_allowed) {
1087 bool needPar = false;
1088 column += running_font.latexWriteEndChanges(
1089 os, bparams, runparams,
1090 basefont, basefont, needPar);
1094 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1097 basefont = owner_->getLayoutFont(bparams, outerfont);
1098 running_font = basefont;
1100 if (runparams.moving_arg)
1104 os.texrow().start(owner_->id(), i + 1);
1108 if (owner_->isDeleted(i)) {
1109 if( ++runparams.inDeletedInset == 1)
1110 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1113 if (inset->canTrackChanges()) {
1114 column += Changes::latexMarkChange(os, bparams, running_change,
1115 Change(Change::UNCHANGED), runparams);
1116 running_change = Change(Change::UNCHANGED);
1120 odocstream::pos_type const len = os.os().tellp();
1122 if (inset->forceLTR()
1123 && !runparams.use_polyglossia
1124 && running_font.isRightToLeft()
1125 // ERT is an exception, it should be output with no
1126 // decorations at all
1127 && inset->lyxCode() != ERT_CODE) {
1128 if (running_font.language()->lang() == "farsi")
1129 os << "\\beginL" << termcmd;
1135 // FIXME: Bug: we can have an empty font change here!
1136 // if there has just been a font change, we are going to close it
1137 // right now, which means stupid latex code like \textsf{}. AFAIK,
1138 // this does not harm dvi output. A minor bug, thus (JMarc)
1140 // Some insets cannot be inside a font change command.
1141 // However, even such insets *can* be placed in \L or \R
1142 // or their equivalents (for RTL language switches), so we don't
1143 // close the language in those cases.
1144 // ArabTeX, though, cannot handle this special behavior, it seems.
1145 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1146 || running_font.language()->lang() == "arabic_arabtex";
1147 if (open_font && !inset->inheritFont()) {
1148 bool needPar = false;
1149 bool closeLanguage = arabtex
1150 || basefont.isRightToLeft() == running_font.isRightToLeft();
1151 unsigned int count = running_font.latexWriteEndChanges(os,
1152 bparams, runparams, basefont, basefont,
1153 needPar, closeLanguage);
1155 // if any font properties were closed, update the running_font,
1156 // making sure, however, to leave the language as it was
1158 // FIXME: probably a better way to keep track of the old
1159 // language, than copying the entire font?
1160 Font const copy_font(running_font);
1161 basefont = owner_->getLayoutFont(bparams, outerfont);
1162 running_font = basefont;
1164 running_font.setLanguage(copy_font.language());
1165 // leave font open if language is still open
1166 open_font = (running_font.language() == basefont.language());
1168 runparams.local_font = &basefont;
1172 size_t const previous_row_count = os.texrow().rows();
1175 runparams.lastid = id_;
1176 runparams.lastpos = i;
1177 inset->latex(os, runparams);
1178 } catch (EncodingException & e) {
1179 // add location information and throw again.
1186 if (running_font.language()->lang() == "farsi")
1187 os << "\\endL" << termcmd;
1192 if (os.texrow().rows() > previous_row_count) {
1193 os.texrow().start(owner_->id(), i + 1);
1196 column += (unsigned int)(os.os().tellp() - len);
1199 if (owner_->isDeleted(i))
1200 --runparams.inDeletedInset;
1204 void Paragraph::Private::latexSpecialChar(otexstream & os,
1205 BufferParams const & bparams,
1206 OutputParams const & runparams,
1207 Font const & running_font,
1208 Change const & running_change,
1209 Layout const & style,
1212 unsigned int & column)
1214 // With polyglossia, brackets and stuff need not be reversed
1215 // in RTL scripts (see bug #8251)
1216 char_type const c = (runparams.use_polyglossia) ?
1217 owner_->getUChar(bparams, i) : text_[i];
1219 if (style.pass_thru || runparams.pass_thru
1220 || contains(style.pass_thru_chars, c)
1221 || contains(runparams.pass_thru_chars, c)) {
1223 Encoding const * const enc = runparams.encoding;
1224 if (enc && !enc->encodable(c))
1225 throw EncodingException(c);
1231 // TIPA uses its own T3 encoding
1232 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1234 // If T1 font encoding is used, use the special
1235 // characters it provides.
1236 // NOTE: Some languages reset the font encoding internally to a
1237 // non-standard font encoding. If we are using such a language,
1238 // we do not output special T1 chars.
1239 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1240 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1241 && latexSpecialT1(c, os, i, column))
1243 // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
1244 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1245 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1248 // Otherwise, we use what LaTeX provides us.
1251 os << "\\textbackslash" << termcmd;
1255 os << "\\textless" << termcmd;
1259 os << "\\textgreater" << termcmd;
1263 os << "\\textbar" << termcmd;
1268 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1269 (end_pos == -1 || i + 1 < end_pos) &&
1270 text_[i+1] == '-') {
1271 // Prevent "--" becoming an en dash and "---" an em dash.
1272 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1278 os << "\\textquotedbl" << termcmd;
1283 case '%': case '#': case '{':
1291 os << "\\textasciitilde" << termcmd;
1296 os << "\\textasciicircum" << termcmd;
1303 // avoid being mistaken for optional arguments
1311 // Blanks are printed before font switching.
1312 // Sure? I am not! (try nice-latex)
1313 // I am sure it's correct. LyX might be smarter
1314 // in the future, but for now, nothing wrong is
1320 // XeTeX's dash behaviour is determined via a global setting
1321 if (bparams.use_dash_ligatures
1322 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1324 // TODO #10961: && not in inset Flex Code
1325 // TODO #10961: && not in layout LyXCode
1326 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1343 Encoding const & encoding = *(runparams.encoding);
1344 char_type next = '\0';
1345 if (i + 1 < int(text_.size())) {
1346 next = text_[i + 1];
1347 if (Encodings::isCombiningChar(next)) {
1348 column += latexSurrogatePair(os, c, next, runparams) - 1;
1354 pair<docstring, bool> latex = encoding.latexChar(c);
1355 docstring nextlatex;
1356 bool nexttipas = false;
1357 string nexttipashortcut;
1358 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1359 nextlatex = encoding.latexChar(next).first;
1360 if (runparams.inIPA) {
1361 nexttipashortcut = Encodings::TIPAShortcut(next);
1362 nexttipas = !nexttipashortcut.empty();
1366 if (runparams.inIPA) {
1367 string const tipashortcut = Encodings::TIPAShortcut(c);
1368 if (!tipashortcut.empty()) {
1369 latex.first = from_ascii(tipashortcut);
1370 latex.second = false;
1375 if (running_font.language()->lang() == runparams.document_language)
1376 fontenc = runparams.main_fontenc;
1378 fontenc = running_font.language()->fontenc();
1379 if (Encodings::isKnownScriptChar(c, script)
1380 && prefixIs(latex.first, from_ascii("\\" + script)))
1381 column += writeScriptChars(runparams, os, latex.first,
1382 running_change, encoding,
1384 else if (latex.second
1385 && ((!prefixIs(nextlatex, '\\')
1386 && !prefixIs(nextlatex, '{')
1387 && !prefixIs(nextlatex, '}'))
1389 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1391 // Prevent eating of a following
1392 // space or command corruption by
1393 // following characters
1394 if (next == ' ' || next == '\0') {
1395 column += latex.first.length() + 1;
1396 os << latex.first << "{}";
1398 column += latex.first.length();
1399 os << latex.first << " ";
1402 column += latex.first.length() - 1;
1410 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1411 pos_type i, unsigned int & column)
1417 // In T1 encoding, these characters exist
1418 // but we should avoid ligatures
1419 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1421 os << "\\textcompwordmark" << termcmd;
1428 // soul.sty breaks with \char`\"
1429 os << "\\textquotedbl" << termcmd;
1438 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1439 pos_type i, unsigned int & column)
1441 // TU encoding is currently on par with T1.
1442 return latexSpecialT1(c, os, i, column);
1446 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1447 pos_type /*i*/, unsigned int & column)
1457 os << "\\textvertline" << termcmd;
1466 void Paragraph::Private::validate(LaTeXFeatures & features) const
1468 if (layout_->inpreamble && inset_owner_) {
1469 // FIXME: Using a string stream here circumvents the encoding
1470 // switching machinery of odocstream. Therefore the
1471 // output is wrong if this paragraph contains content
1472 // that needs to switch encoding.
1473 Buffer const & buf = inset_owner_->buffer();
1474 otexstringstream os;
1475 os << layout_->preamble();
1476 size_t const length = os.length();
1477 TeXOnePar(buf, buf.text(), buf.getParFromID(owner_->id()).pit(), os,
1478 features.runparams(), string(), 0, -1, true);
1479 if (os.length() > length)
1480 features.addPreambleSnippet(os.release(), true);
1483 if (features.runparams().flavor == OutputParams::HTML
1484 && layout_->htmltitle()) {
1485 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1488 // check the params.
1489 if (!params_.spacing().isDefault())
1490 features.require("setspace");
1493 features.useLayout(layout_->name());
1496 fontlist_.validate(features);
1498 // then the indentation
1499 if (!params_.leftIndent().zero())
1500 features.require("ParagraphLeftIndent");
1503 InsetList::const_iterator icit = insetlist_.begin();
1504 InsetList::const_iterator iend = insetlist_.end();
1505 for (; icit != iend; ++icit) {
1507 features.inDeletedInset(owner_->isDeleted(icit->pos));
1508 if (icit->inset->lyxCode() == FOOT_CODE) {
1509 // FIXME: an item inset would make things much easier.
1510 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1511 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1512 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1513 && (icit->pos < begin_of_body_
1514 || (icit->pos == begin_of_body_
1515 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1516 features.saveNoteEnv("description");
1518 icit->inset->validate(features);
1519 features.inDeletedInset(false);
1520 if (layout_->needprotect &&
1521 icit->inset->lyxCode() == FOOT_CODE)
1522 features.require("NeedLyXFootnoteCode");
1526 // then the contents
1527 BufferParams const bp = features.runparams().is_child
1528 ? features.buffer().masterParams() : features.buffer().params();
1529 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1530 char_type c = text_[i];
1532 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1533 features.require("textquotedblp");
1534 else if (bp.main_font_encoding() != "T1"
1535 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1536 features.require("textquotedbl");
1538 if (!bp.use_dash_ligatures
1539 && (c == 0x2013 || c == 0x2014)
1540 && bp.useNonTeXFonts
1541 && features.runparams().flavor == OutputParams::XETEX)
1542 // XeTeX's dash behaviour is determined via a global setting
1543 features.require("xetexdashbreakstate");
1544 BufferEncodings::validate(c, features);
1548 /////////////////////////////////////////////////////////////////////
1552 /////////////////////////////////////////////////////////////////////
1555 Layout const emptyParagraphLayout;
1558 Paragraph::Paragraph()
1559 : d(new Paragraph::Private(this, emptyParagraphLayout))
1566 Paragraph::Paragraph(Paragraph const & par)
1567 : itemdepth(par.itemdepth),
1568 d(new Paragraph::Private(*par.d, this))
1574 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1575 : itemdepth(par.itemdepth),
1576 d(new Paragraph::Private(*par.d, this, beg, end))
1582 Paragraph & Paragraph::operator=(Paragraph const & par)
1584 // needed as we will destroy the private part before copying it
1586 itemdepth = par.itemdepth;
1590 d = new Private(*par.d, this);
1597 Paragraph::~Paragraph()
1606 // this shall be called just before every "os << ..." action.
1607 void flushString(ostream & os, docstring & s)
1616 void Paragraph::write(ostream & os, BufferParams const & bparams,
1617 depth_type & dth) const
1619 // The beginning or end of a deeper (i.e. nested) area?
1620 if (dth != d->params_.depth()) {
1621 if (d->params_.depth() > dth) {
1622 while (d->params_.depth() > dth) {
1623 os << "\n\\begin_deeper";
1627 while (d->params_.depth() < dth) {
1628 os << "\n\\end_deeper";
1634 // First write the layout
1635 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1637 d->params_.write(os);
1639 Font font1(inherit_font, bparams.language);
1641 Change running_change = Change(Change::UNCHANGED);
1643 // this string is used as a buffer to avoid repetitive calls
1644 // to to_utf8(), which turn out to be expensive (JMarc)
1645 docstring write_buffer;
1648 for (pos_type i = 0; i <= size(); ++i) {
1650 Change const & change = lookupChange(i);
1651 if (change != running_change)
1652 flushString(os, write_buffer);
1653 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1654 running_change = change;
1659 // Write font changes
1660 Font font2 = getFontSettings(bparams, i);
1661 if (font2 != font1) {
1662 flushString(os, write_buffer);
1663 font2.lyxWriteChanges(font1, os);
1668 char_type const c = d->text_[i];
1671 if (Inset const * inset = getInset(i)) {
1672 flushString(os, write_buffer);
1673 if (inset->directWrite()) {
1674 // international char, let it write
1675 // code directly so it's shorter in
1681 os << "\\begin_inset ";
1683 os << "\n\\end_inset\n\n";
1686 // FIXME This can be removed again once the mystery
1687 // crash has been resolved.
1692 flushString(os, write_buffer);
1693 os << "\n\\backslash\n";
1697 flushString(os, write_buffer);
1698 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1705 if ((column > 70 && c == ' ')
1707 flushString(os, write_buffer);
1711 // this check is to amend a bug. LyX sometimes
1712 // inserts '\0' this could cause problems.
1714 write_buffer.push_back(c);
1716 LYXERR0("NUL char in structure.");
1722 flushString(os, write_buffer);
1723 os << "\n\\end_layout\n";
1724 // FIXME This can be removed again once the mystery
1725 // crash has been resolved.
1730 void Paragraph::validate(LaTeXFeatures & features) const
1732 d->validate(features);
1733 if (needsCProtection())
1734 features.require("cprotect");
1738 void Paragraph::insert(pos_type start, docstring const & str,
1739 Font const & font, Change const & change)
1741 for (size_t i = 0, n = str.size(); i != n ; ++i)
1742 insertChar(start + i, str[i], font, change);
1746 void Paragraph::appendChar(char_type c, Font const & font,
1747 Change const & change)
1750 d->changes_.insert(change, d->text_.size());
1751 // when appending characters, no need to update tables
1752 d->text_.push_back(c);
1753 setFont(d->text_.size() - 1, font);
1754 d->requestSpellCheck(d->text_.size() - 1);
1758 void Paragraph::appendString(docstring const & s, Font const & font,
1759 Change const & change)
1761 pos_type end = s.size();
1762 size_t oldsize = d->text_.size();
1763 size_t newsize = oldsize + end;
1764 size_t capacity = d->text_.capacity();
1765 if (newsize >= capacity)
1766 d->text_.reserve(max(capacity + 100, newsize));
1768 // when appending characters, no need to update tables
1771 // FIXME: Optimize this!
1772 for (size_t i = oldsize; i != newsize; ++i) {
1774 d->changes_.insert(change, i);
1775 d->requestSpellCheck(i);
1777 d->fontlist_.set(oldsize, font);
1778 d->fontlist_.set(newsize - 1, font);
1782 void Paragraph::insertChar(pos_type pos, char_type c,
1785 d->insertChar(pos, c, Change(trackChanges ?
1786 Change::INSERTED : Change::UNCHANGED));
1790 void Paragraph::insertChar(pos_type pos, char_type c,
1791 Font const & font, bool trackChanges)
1793 d->insertChar(pos, c, Change(trackChanges ?
1794 Change::INSERTED : Change::UNCHANGED));
1799 void Paragraph::insertChar(pos_type pos, char_type c,
1800 Font const & font, Change const & change)
1802 d->insertChar(pos, c, change);
1807 void Paragraph::resetFonts(Font const & font)
1809 d->fontlist_.clear();
1810 d->fontlist_.set(0, font);
1811 d->fontlist_.set(d->text_.size() - 1, font);
1814 // Gets uninstantiated font setting at position.
1815 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1819 LYXERR0("pos: " << pos << " size: " << size());
1823 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1824 if (cit != d->fontlist_.end())
1827 if (pos == size() && !empty())
1828 return getFontSettings(bparams, pos - 1);
1830 // Optimisation: avoid a full font instantiation if there is no
1831 // language change from previous call.
1832 static Font previous_font;
1833 static Language const * previous_lang = 0;
1834 Language const * lang = getParLanguage(bparams);
1835 if (lang != previous_lang) {
1836 previous_lang = lang;
1837 previous_font = Font(inherit_font, lang);
1839 return previous_font;
1843 FontSpan Paragraph::fontSpan(pos_type pos) const
1845 LBUFERR(pos <= size());
1848 return FontSpan(pos, pos);
1851 FontList::const_iterator cit = d->fontlist_.begin();
1852 FontList::const_iterator end = d->fontlist_.end();
1853 for (; cit != end; ++cit) {
1854 if (cit->pos() >= pos) {
1855 if (pos >= beginOfBody())
1856 return FontSpan(max(start, beginOfBody()),
1859 return FontSpan(start,
1860 min(beginOfBody() - 1,
1863 start = cit->pos() + 1;
1866 // This should not happen, but if so, we take no chances.
1867 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1868 LASSERT(false, return FontSpan(pos, pos));
1872 // Gets uninstantiated font setting at position 0
1873 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1875 if (!empty() && !d->fontlist_.empty())
1876 return d->fontlist_.begin()->font();
1878 // Optimisation: avoid a full font instantiation if there is no
1879 // language change from previous call.
1880 static Font previous_font;
1881 static Language const * previous_lang = 0;
1882 if (bparams.language != previous_lang) {
1883 previous_lang = bparams.language;
1884 previous_font = Font(inherit_font, bparams.language);
1887 return previous_font;
1891 // Gets the fully instantiated font at a given position in a paragraph
1892 // This is basically the same function as Text::GetFont() in text2.cpp.
1893 // The difference is that this one is used for generating the LaTeX file,
1894 // and thus cosmetic "improvements" are disallowed: This has to deliver
1895 // the true picture of the buffer. (Asger)
1896 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1897 Font const & outerfont) const
1901 Font font = getFontSettings(bparams, pos);
1903 pos_type const body_pos = beginOfBody();
1904 FontInfo & fi = font.fontInfo();
1906 fi.realize(d->layout_->labelfont);
1908 fi.realize(d->layout_->font);
1910 fi.realize(outerfont.fontInfo());
1911 fi.realize(bparams.getFont().fontInfo());
1917 Font const Paragraph::getLabelFont
1918 (BufferParams const & bparams, Font const & outerfont) const
1920 FontInfo tmpfont = d->layout_->labelfont;
1921 tmpfont.realize(outerfont.fontInfo());
1922 tmpfont.realize(bparams.getFont().fontInfo());
1923 return Font(tmpfont, getParLanguage(bparams));
1927 Font const Paragraph::getLayoutFont
1928 (BufferParams const & bparams, Font const & outerfont) const
1930 FontInfo tmpfont = d->layout_->font;
1931 tmpfont.realize(outerfont.fontInfo());
1932 tmpfont.realize(bparams.getFont().fontInfo());
1933 return Font(tmpfont, getParLanguage(bparams));
1937 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1939 char_type c = d->text_[pos];
1940 if (!getFontSettings(bparams, pos).isRightToLeft())
1943 // FIXME: The arabic special casing is due to the difference of arabic
1944 // round brackets input introduced in r18599. Check if this should be
1945 // unified with Hebrew or at least if all bracket types should be
1946 // handled the same (file format change in either case).
1947 string const & lang = getFontSettings(bparams, pos).language()->lang();
1948 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1953 uc = arabic ? c : ')';
1956 uc = arabic ? c : '(';
1982 void Paragraph::setFont(pos_type pos, Font const & font)
1984 LASSERT(pos <= size(), return);
1986 // First, reduce font against layout/label font
1987 // Update: The setCharFont() routine in text2.cpp already
1988 // reduces font, so we don't need to do that here. (Asger)
1990 d->fontlist_.set(pos, font);
1994 void Paragraph::makeSameLayout(Paragraph const & par)
1996 d->layout_ = par.d->layout_;
1997 d->params_ = par.d->params_;
2001 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2003 if (isFreeSpacing())
2009 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2010 if (eraseChar(pos, trackChanges))
2016 return count > 0 || pos > 0;
2020 bool Paragraph::hasSameLayout(Paragraph const & par) const
2022 return par.d->layout_ == d->layout_
2023 && d->params_.sameLayout(par.d->params_);
2027 depth_type Paragraph::getDepth() const
2029 return d->params_.depth();
2033 depth_type Paragraph::getMaxDepthAfter() const
2035 if (d->layout_->isEnvironment())
2036 return d->params_.depth() + 1;
2038 return d->params_.depth();
2042 LyXAlignment Paragraph::getAlign() const
2044 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2045 return d->layout_->align;
2047 return d->params_.align();
2051 docstring const & Paragraph::labelString() const
2053 return d->params_.labelString();
2057 // the next two functions are for the manual labels
2058 docstring const Paragraph::getLabelWidthString() const
2060 if (d->layout_->margintype == MARGIN_MANUAL
2061 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2062 return d->params_.labelWidthString();
2064 return _("Senseless with this layout!");
2068 void Paragraph::setLabelWidthString(docstring const & s)
2070 d->params_.labelWidthString(s);
2074 docstring Paragraph::expandLabel(Layout const & layout,
2075 BufferParams const & bparams) const
2077 return expandParagraphLabel(layout, bparams, true);
2081 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2082 BufferParams const & bparams) const
2084 return expandParagraphLabel(layout, bparams, false);
2088 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2089 BufferParams const & bparams, bool process_appendix) const
2091 DocumentClass const & tclass = bparams.documentClass();
2092 string const & lang = getParLanguage(bparams)->code();
2093 bool const in_appendix = process_appendix && d->params_.appendix();
2094 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2096 if (fmt.empty() && !layout.counter.empty())
2097 return tclass.counters().theCounter(layout.counter, lang);
2099 // handle 'inherited level parts' in 'fmt',
2100 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2101 size_t const i = fmt.find('@', 0);
2102 if (i != docstring::npos) {
2103 size_t const j = fmt.find('@', i + 1);
2104 if (j != docstring::npos) {
2105 docstring parent(fmt, i + 1, j - i - 1);
2106 docstring label = from_ascii("??");
2107 if (tclass.hasLayout(parent))
2108 label = expandParagraphLabel(tclass[parent], bparams,
2110 fmt = docstring(fmt, 0, i) + label
2111 + docstring(fmt, j + 1, docstring::npos);
2115 return tclass.counters().counterLabel(fmt, lang);
2119 void Paragraph::applyLayout(Layout const & new_layout)
2121 d->layout_ = &new_layout;
2122 LyXAlignment const oldAlign = d->params_.align();
2124 if (!(oldAlign & d->layout_->alignpossible)) {
2125 frontend::Alert::warning(_("Alignment not permitted"),
2126 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2127 d->params_.align(LYX_ALIGN_LAYOUT);
2132 pos_type Paragraph::beginOfBody() const
2134 return d->begin_of_body_;
2138 void Paragraph::setBeginOfBody()
2140 if (d->layout_->labeltype != LABEL_MANUAL) {
2141 d->begin_of_body_ = 0;
2145 // Unroll the first two cycles of the loop
2146 // and remember the previous character to
2147 // remove unnecessary getChar() calls
2149 pos_type end = size();
2150 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2153 char_type previous_char = d->text_[i];
2154 if (!(isNewline(i) || isEnvSeparator(i))) {
2156 while (i < end && previous_char != ' ') {
2157 char_type temp = d->text_[i];
2158 if (isNewline(i) || isEnvSeparator(i))
2161 previous_char = temp;
2167 d->begin_of_body_ = i;
2171 bool Paragraph::allowParagraphCustomization() const
2173 return inInset().allowParagraphCustomization();
2177 bool Paragraph::usePlainLayout() const
2179 return inInset().usePlainLayout();
2183 bool Paragraph::isPassThru() const
2185 return inInset().isPassThru() || d->layout_->pass_thru;
2190 // paragraphs inside floats need different alignment tags to avoid
2193 bool noTrivlistCentering(InsetCode code)
2195 return code == FLOAT_CODE
2196 || code == WRAP_CODE
2197 || code == CELL_CODE;
2201 string correction(string const & orig)
2203 if (orig == "flushleft")
2204 return "raggedright";
2205 if (orig == "flushright")
2206 return "raggedleft";
2207 if (orig == "center")
2213 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2214 InsetCode code, bool const lastpar, int & col)
2216 string macro = suffix + "{";
2217 if (noTrivlistCentering(code)) {
2219 // the last paragraph in non-trivlist-aligned
2220 // context is special (to avoid unwanted whitespace)
2221 if (suffix == "\\begin") {
2222 macro = "\\" + correction(env) + "{}";
2223 os << from_ascii(macro);
2224 col += macro.size();
2229 macro += correction(env);
2233 if (suffix == "\\par\\end") {
2237 os << from_ascii(macro);
2238 col += macro.size();
2239 if (suffix == "\\begin") {
2249 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2250 otexstream & os, OutputParams const & runparams) const
2255 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2256 (layout_->toggle_indent != ITOGGLE_NEVER) :
2257 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2259 if (canindent && params_.noindent() && !layout_->pass_thru) {
2260 os << "\\noindent ";
2264 LyXAlignment const curAlign = params_.align();
2266 if (curAlign == layout_->align)
2270 case LYX_ALIGN_NONE:
2271 case LYX_ALIGN_BLOCK:
2272 case LYX_ALIGN_LAYOUT:
2273 case LYX_ALIGN_SPECIAL:
2274 case LYX_ALIGN_DECIMAL:
2276 case LYX_ALIGN_LEFT:
2277 case LYX_ALIGN_RIGHT:
2278 case LYX_ALIGN_CENTER:
2279 if (runparams.moving_arg) {
2286 string const begin_tag = "\\begin";
2287 InsetCode code = ownerCode();
2288 bool const lastpar = runparams.isLastPar;
2291 case LYX_ALIGN_NONE:
2292 case LYX_ALIGN_BLOCK:
2293 case LYX_ALIGN_LAYOUT:
2294 case LYX_ALIGN_SPECIAL:
2295 case LYX_ALIGN_DECIMAL:
2297 case LYX_ALIGN_LEFT: {
2298 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2299 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2301 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2303 } case LYX_ALIGN_RIGHT: {
2304 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2305 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2307 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2309 } case LYX_ALIGN_CENTER: {
2310 corrected_env(os, begin_tag, "center", code, lastpar, column);
2319 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2320 otexstream & os, OutputParams const & runparams) const
2322 LyXAlignment const curAlign = params_.align();
2324 if (curAlign == layout_->align)
2328 case LYX_ALIGN_NONE:
2329 case LYX_ALIGN_BLOCK:
2330 case LYX_ALIGN_LAYOUT:
2331 case LYX_ALIGN_SPECIAL:
2332 case LYX_ALIGN_DECIMAL:
2334 case LYX_ALIGN_LEFT:
2335 case LYX_ALIGN_RIGHT:
2336 case LYX_ALIGN_CENTER:
2337 if (runparams.moving_arg)
2342 bool output = false;
2344 string const end_tag = "\\par\\end";
2345 InsetCode code = ownerCode();
2346 bool const lastpar = runparams.isLastPar;
2349 case LYX_ALIGN_NONE:
2350 case LYX_ALIGN_BLOCK:
2351 case LYX_ALIGN_LAYOUT:
2352 case LYX_ALIGN_SPECIAL:
2353 case LYX_ALIGN_DECIMAL:
2355 case LYX_ALIGN_LEFT: {
2356 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2357 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2359 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2361 } case LYX_ALIGN_RIGHT: {
2362 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2363 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2365 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2367 } case LYX_ALIGN_CENTER: {
2368 corrected_env(os, end_tag, "center", code, lastpar, col);
2373 return output || lastpar;
2377 // This one spits out the text of the paragraph
2378 void Paragraph::latex(BufferParams const & bparams,
2379 Font const & outerfont,
2381 OutputParams const & runparams,
2382 int start_pos, int end_pos, bool force) const
2384 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2386 // FIXME This check should not be needed. Perhaps issue an
2387 // error if it triggers.
2388 Layout const & style = inInset().forcePlainLayout() ?
2389 bparams.documentClass().plainLayout() : *d->layout_;
2391 if (!force && style.inpreamble)
2394 bool const allowcust = allowParagraphCustomization();
2396 // Current base font for all inherited font changes, without any
2397 // change caused by an individual character, except for the language:
2398 // It is set to the language of the first character.
2399 // As long as we are in the label, this font is the base font of the
2400 // label. Before the first body character it is set to the base font
2404 // Maybe we have to create a optional argument.
2405 pos_type body_pos = beginOfBody();
2406 unsigned int column = 0;
2409 // the optional argument is kept in curly brackets in
2410 // case it contains a ']'
2411 // This is not strictly needed, but if this is changed it
2412 // would be a file format change, and tex2lyx would need
2413 // to be adjusted, since it unconditionally removes the
2414 // braces when it parses \item.
2417 basefont = getLabelFont(bparams, outerfont);
2419 basefont = getLayoutFont(bparams, outerfont);
2422 // Which font is currently active?
2423 Font running_font(basefont);
2424 // Do we have an open font change?
2425 bool open_font = false;
2427 Change runningChange = Change(Change::UNCHANGED);
2429 Encoding const * const prev_encoding = runparams.encoding;
2431 os.texrow().start(id(), 0);
2433 // if the paragraph is empty, the loop will not be entered at all
2435 // For InTitle commands, we have already opened a group
2436 // in output_latex::TeXOnePar.
2437 if (style.isCommand() && !style.intitle) {
2441 if (!style.leftdelim().empty()) {
2442 os << style.leftdelim();
2443 column += style.leftdelim().size();
2446 column += d->startTeXParParams(bparams, os, runparams);
2449 // Whether a \par can be issued for insets typeset inline with text.
2450 // Yes if greater than 0. This has to be static.
2451 THREAD_LOCAL_STATIC int parInline = 0;
2453 for (pos_type i = 0; i < size(); ++i) {
2454 // First char in paragraph or after label?
2455 if (i == body_pos) {
2458 bool needPar = false;
2459 column += running_font.latexWriteEndChanges(
2460 os, bparams, runparams,
2461 basefont, basefont, needPar);
2464 basefont = getLayoutFont(bparams, outerfont);
2465 running_font = basefont;
2467 column += Changes::latexMarkChange(os, bparams,
2468 runningChange, Change(Change::UNCHANGED),
2470 runningChange = Change(Change::UNCHANGED);
2475 // For InTitle commands, we have already opened a group
2476 // in output_latex::TeXOnePar.
2477 if (style.isCommand() && !style.intitle) {
2482 if (!style.leftdelim().empty()) {
2483 os << style.leftdelim();
2484 column += style.leftdelim().size();
2488 column += d->startTeXParParams(bparams, os,
2492 runparams.wasDisplayMath = runparams.inDisplayMath;
2493 runparams.inDisplayMath = false;
2494 bool deleted_display_math = false;
2495 Change const & change = runparams.inDeletedInset
2496 ? runparams.changeOfDeletedInset : lookupChange(i);
2498 // Check whether a display math inset follows
2499 if (d->text_[i] == META_INSET
2500 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2501 InsetMath const * im = getInset(i)->asInsetMath();
2502 if (im && im->asHullInset()
2503 && im->asHullInset()->outerDisplay()) {
2504 runparams.inDisplayMath = true;
2505 // runparams.inDeletedInset will be set by
2506 // latexInset later, but we need this info
2507 // before it is called. On the other hand, we
2508 // cannot set it here because it is a counter.
2509 deleted_display_math = isDeleted(i);
2511 if (bparams.output_changes && deleted_display_math
2512 && runningChange == change
2513 && change.type == Change::DELETED
2514 && !os.afterParbreak()) {
2515 // A display math in the same paragraph follows.
2516 // We have to close and then reopen \lyxdeleted,
2517 // otherwise the math will be shifted up.
2518 OutputParams rp = runparams;
2520 bool needPar = false;
2521 column += running_font.latexWriteEndChanges(
2522 os, bparams, rp, basefont,
2526 basefont = getLayoutFont(bparams, outerfont);
2527 running_font = basefont;
2528 column += Changes::latexMarkChange(os, bparams,
2529 Change(Change::INSERTED), change, rp);
2533 if (bparams.output_changes && runningChange != change) {
2535 bool needPar = false;
2536 column += running_font.latexWriteEndChanges(
2537 os, bparams, runparams,
2538 basefont, basefont, needPar);
2541 basefont = getLayoutFont(bparams, outerfont);
2542 running_font = basefont;
2543 column += Changes::latexMarkChange(os, bparams, runningChange,
2545 runningChange = change;
2548 // do not output text which is marked deleted
2549 // if change tracking output is disabled
2550 if (!bparams.output_changes && change.deleted()) {
2556 // Fully instantiated font
2557 Font const current_font = getFont(bparams, i, outerfont);
2559 Font const last_font = running_font;
2561 // Do we need to close the previous font?
2563 (current_font != running_font ||
2564 current_font.language() != running_font.language()))
2566 bool needPar = false;
2567 column += running_font.latexWriteEndChanges(
2568 os, bparams, runparams, basefont,
2569 (i == body_pos-1) ? basefont : current_font,
2571 running_font = basefont;
2575 string const running_lang = runparams.use_polyglossia ?
2576 running_font.language()->polyglossia() : running_font.language()->babel();
2577 // close babel's font environment before opening CJK.
2578 string const lang_end_command = runparams.use_polyglossia ?
2579 "\\end{$$lang}" : lyxrc.language_command_end;
2580 bool const using_begin_end = runparams.use_polyglossia ||
2581 !lang_end_command.empty();
2582 if (!running_lang.empty() &&
2583 (!using_begin_end || running_lang == openLanguageName()) &&
2584 current_font.language()->encoding()->package() == Encoding::CJK) {
2585 string end_tag = subst(lang_end_command,
2588 os << from_ascii(end_tag);
2589 column += end_tag.length();
2590 if (using_begin_end)
2594 // Switch file encoding if necessary (and allowed)
2595 if (!runparams.pass_thru && !style.pass_thru &&
2596 runparams.encoding->package() != Encoding::none &&
2597 current_font.language()->encoding()->package() != Encoding::none) {
2598 pair<bool, int> const enc_switch =
2599 switchEncoding(os.os(), bparams, runparams,
2600 *(current_font.language()->encoding()));
2601 if (enc_switch.first) {
2602 column += enc_switch.second;
2603 runparams.encoding = current_font.language()->encoding();
2607 char_type const c = d->text_[i];
2609 // A display math inset inside an ulem command will be output
2610 // as a box of width \linewidth, so we have to either disable
2611 // indentation if the inset starts a paragraph, or start a new
2612 // line to accommodate such box. This has to be done before
2613 // writing any font changing commands.
2614 if (runparams.inDisplayMath && !deleted_display_math
2615 && runparams.inulemcmd) {
2616 if (os.afterParbreak())
2622 // Do we need to change font?
2623 if ((current_font != running_font ||
2624 current_font.language() != running_font.language()) &&
2627 odocstringstream ods;
2628 column += current_font.latexWriteStartChanges(ods, bparams,
2629 runparams, basefont,
2631 // Check again for display math in ulem commands as a
2632 // font change may also occur just before a math inset.
2633 if (runparams.inDisplayMath && !deleted_display_math
2634 && runparams.inulemcmd) {
2635 if (os.afterParbreak())
2640 running_font = current_font;
2642 docstring fontchange = ods.str();
2643 // check whether the fontchange ends with a \\textcolor
2644 // modifier and the text starts with a space (bug 4473)
2645 docstring const last_modifier = rsplit(fontchange, '\\');
2646 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2647 os << fontchange << from_ascii("{}");
2648 // check if the fontchange ends with a trailing blank
2649 // (like "\small " (see bug 3382)
2650 else if (suffixIs(fontchange, ' ') && c == ' ')
2651 os << fontchange.substr(0, fontchange.size() - 1)
2652 << from_ascii("{}");
2657 // FIXME: think about end_pos implementation...
2658 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2659 // FIXME: integrate this case in latexSpecialChar
2660 // Do not print the separation of the optional argument
2661 // if style.pass_thru is false. This works because
2662 // latexSpecialChar ignores spaces if
2663 // style.pass_thru is false.
2664 if (i != body_pos - 1) {
2665 if (d->simpleTeXBlanks(runparams, os,
2666 i, column, current_font, style)) {
2667 // A surrogate pair was output. We
2668 // must not call latexSpecialChar
2669 // in this iteration, since it would output
2670 // the combining character again.
2677 OutputParams rp = runparams;
2678 rp.free_spacing = style.free_spacing;
2679 rp.local_font = ¤t_font;
2680 rp.intitle = style.intitle;
2682 // Two major modes: LaTeX or plain
2683 // Handle here those cases common to both modes
2684 // and then split to handle the two modes separately.
2685 if (c == META_INSET) {
2686 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2687 // Greyedout notes and, in general, all insets
2688 // with InsetLayout::isDisplay() == false,
2689 // are typeset inline with the text. So, we
2690 // can add a \par to the last paragraph of
2691 // such insets only if nothing else follows.
2692 bool incremented = false;
2693 Inset const * inset = getInset(i);
2694 InsetText const * textinset = inset
2695 ? inset->asInsetText()
2697 if (i + 1 == size() && textinset
2698 && !inset->getLayout().isDisplay()) {
2699 ParagraphList const & pars =
2700 textinset->text().paragraphs();
2701 pit_type const pit = pars.size() - 1;
2702 Font const lastfont =
2703 pit < 0 || pars[pit].empty()
2704 ? pars[pit].getLayoutFont(
2707 : pars[pit].getFont(bparams,
2708 pars[pit].size() - 1,
2710 if (lastfont.fontInfo().size() !=
2711 basefont.fontInfo().size()) {
2716 d->latexInset(bparams, os, rp, running_font,
2717 basefont, outerfont, open_font,
2718 runningChange, style, i, column);
2722 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2724 d->latexSpecialChar(os, bparams, rp,
2725 running_font, runningChange,
2726 style, i, end_pos, column);
2727 } catch (EncodingException & e) {
2728 if (runparams.dryrun) {
2729 os << "<" << _("LyX Warning: ")
2730 << _("uncodable character") << " '";
2734 // add location information and throw again.
2742 // Set the encoding to that returned from latexSpecialChar (see
2743 // comment for encoding member in OutputParams.h)
2744 runparams.encoding = rp.encoding;
2746 // Also carry on the info on a closed ulem command for insets
2747 // such as Note that do not produce any output, so that no
2748 // command is ever executed but its opening was recorded.
2749 runparams.inulemcmd = rp.inulemcmd;
2752 // If we have an open font definition, we have to close it
2754 // Make sure that \\par is done with the font of the last
2755 // character if this has another size as the default.
2756 // This is necessary because LaTeX (and LyX on the screen)
2757 // calculates the space between the baselines according
2758 // to this font. (Matthias)
2760 // We must not change the font for the last paragraph
2761 // of non-multipar insets, tabular cells or commands,
2762 // since this produces unwanted whitespace.
2764 Font const font = empty()
2765 ? getLayoutFont(bparams, outerfont)
2766 : getFont(bparams, size() - 1, outerfont);
2768 InsetText const * textinset = inInset().asInsetText();
2770 bool const maintext = textinset
2771 ? textinset->text().isMainText()
2774 size_t const numpars = textinset
2775 ? textinset->text().paragraphs().size()
2778 bool needPar = false;
2780 if (style.resfont.size() != font.fontInfo().size()
2781 && (!runparams.isLastPar || maintext
2782 || (numpars > 1 && d->ownerCode() != CELL_CODE
2783 && (inInset().getLayout().isDisplay()
2785 && !style.isCommand()) {
2788 #ifdef FIXED_LANGUAGE_END_DETECTION
2790 running_font.latexWriteEndChanges(os, bparams,
2791 runparams, basefont,
2792 next_->getFont(bparams, 0, outerfont),
2795 running_font.latexWriteEndChanges(os, bparams,
2796 runparams, basefont, basefont, needPar);
2799 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2800 //FIXME: there as we start another \selectlanguage with the next paragraph if
2801 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2802 running_font.latexWriteEndChanges(os, bparams, runparams,
2803 basefont, basefont, needPar);
2806 // The \par could not be inserted at the same nesting
2807 // level of the font size change, so do it now.
2808 os << "{\\" << font.latexSize() << "\\par}";
2812 column += Changes::latexMarkChange(os, bparams, runningChange,
2813 Change(Change::UNCHANGED), runparams);
2815 // Needed if there is an optional argument but no contents.
2816 if (body_pos > 0 && body_pos == size()) {
2820 if (!style.rightdelim().empty()) {
2821 os << style.rightdelim();
2822 column += style.rightdelim().size();
2825 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2826 && runparams.encoding != prev_encoding) {
2827 runparams.encoding = prev_encoding;
2828 os << setEncoding(prev_encoding->iconvName());
2831 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2835 bool Paragraph::emptyTag() const
2837 for (pos_type i = 0; i < size(); ++i) {
2838 if (Inset const * inset = getInset(i)) {
2839 InsetCode lyx_code = inset->lyxCode();
2840 // FIXME testing like that is wrong. What is
2842 if (lyx_code != TOC_CODE &&
2843 lyx_code != INCLUDE_CODE &&
2844 lyx_code != GRAPHICS_CODE &&
2845 lyx_code != ERT_CODE &&
2846 lyx_code != LISTINGS_CODE &&
2847 lyx_code != FLOAT_CODE &&
2848 lyx_code != TABULAR_CODE) {
2852 char_type c = d->text_[i];
2853 if (c != ' ' && c != '\t')
2861 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2864 for (pos_type i = 0; i < size(); ++i) {
2865 if (Inset const * inset = getInset(i)) {
2866 InsetCode lyx_code = inset->lyxCode();
2867 if (lyx_code == LABEL_CODE) {
2868 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2869 docstring const & id = il->getParam("name");
2870 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2878 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2882 for (i = 0; i < size(); ++i) {
2883 if (Inset const * inset = getInset(i)) {
2884 inset->docbook(os, runparams);
2886 char_type c = d->text_[i];
2889 os << sgml::escapeChar(c);
2896 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2900 for (i = 0; i < size(); ++i) {
2901 if (Inset const * inset = getInset(i)) {
2902 inset->xhtml(xs, runparams);
2904 char_type c = d->text_[i];
2914 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2917 pos_type size = text_.size();
2918 for (pos_type i = initial; i < size; ++i) {
2919 Font font = owner_->getFont(buf.params(), i, outerfont);
2920 if (text_[i] == META_INSET)
2922 if (i != initial && font != font_old)
2931 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2933 OutputParams const & runparams,
2934 Font const & outerfont,
2935 pos_type initial) const
2937 bool emph_flag = false;
2939 Layout const & style = *d->layout_;
2941 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2943 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2946 // parsing main loop
2947 for (pos_type i = initial; i < size(); ++i) {
2948 Font font = getFont(buf.params(), i, outerfont);
2950 // handle <emphasis> tag
2951 if (font_old.emph() != font.fontInfo().emph()) {
2952 if (font.fontInfo().emph() == FONT_ON) {
2955 } else if (i != initial) {
2956 os << "</emphasis>";
2961 if (Inset const * inset = getInset(i)) {
2962 inset->docbook(os, runparams);
2964 char_type c = d->text_[i];
2966 if (style.pass_thru)
2969 os << sgml::escapeChar(c);
2971 font_old = font.fontInfo();
2975 os << "</emphasis>";
2978 if (style.free_spacing)
2980 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2986 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2987 vector<html::EndFontTag> & tagsToClose,
2988 bool & flag, FontState curstate, html::FontTypes type)
2990 if (curstate == FONT_ON) {
2991 tagsToOpen.push_back(html::FontTag(type));
2994 tagsToClose.push_back(html::EndFontTag(type));
3001 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3003 OutputParams const & runparams,
3004 Font const & outerfont,
3005 bool start_paragraph, bool close_paragraph,
3006 pos_type initial) const
3010 // track whether we have opened these tags
3011 bool emph_flag = false;
3012 bool bold_flag = false;
3013 bool noun_flag = false;
3014 bool ubar_flag = false;
3015 bool dbar_flag = false;
3016 bool sout_flag = false;
3017 bool xout_flag = false;
3018 bool wave_flag = false;
3020 bool shap_flag = false;
3022 bool faml_flag = false;
3024 bool size_flag = false;
3026 Layout const & style = *d->layout_;
3028 if (start_paragraph)
3029 xs.startDivision(allowEmpty());
3032 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3034 FontShape curr_fs = INHERIT_SHAPE;
3035 FontFamily curr_fam = INHERIT_FAMILY;
3036 FontSize curr_size = FONT_SIZE_INHERIT;
3038 string const default_family =
3039 buf.masterBuffer()->params().fonts_default_family;
3041 vector<html::FontTag> tagsToOpen;
3042 vector<html::EndFontTag> tagsToClose;
3044 // parsing main loop
3045 for (pos_type i = initial; i < size(); ++i) {
3046 // let's not show deleted material in the output
3050 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3053 FontState curstate = font.fontInfo().emph();
3054 if (font_old.emph() != curstate)
3055 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3058 curstate = font.fontInfo().noun();
3059 if (font_old.noun() != curstate)
3060 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3063 curstate = font.fontInfo().underbar();
3064 if (font_old.underbar() != curstate)
3065 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3068 curstate = font.fontInfo().strikeout();
3069 if (font_old.strikeout() != curstate)
3070 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3073 curstate = font.fontInfo().xout();
3074 if (font_old.xout() != curstate)
3075 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3078 curstate = font.fontInfo().uuline();
3079 if (font_old.uuline() != curstate)
3080 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3083 curstate = font.fontInfo().uwave();
3084 if (font_old.uwave() != curstate)
3085 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3088 // a little hackish, but allows us to reuse what we have.
3089 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3090 if (font_old.series() != font.fontInfo().series())
3091 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3094 curr_fs = font.fontInfo().shape();
3095 FontShape old_fs = font_old.shape();
3096 if (old_fs != curr_fs) {
3100 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3103 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3105 case SMALLCAPS_SHAPE:
3106 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3112 // the other tags are for internal use
3120 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3124 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3127 case SMALLCAPS_SHAPE:
3128 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3135 // the other tags are for internal use
3142 curr_fam = font.fontInfo().family();
3143 FontFamily old_fam = font_old.family();
3144 if (old_fam != curr_fam) {
3148 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3151 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3153 case TYPEWRITER_FAMILY:
3154 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3156 case INHERIT_FAMILY:
3159 // the other tags are for internal use
3167 // we will treat a "default" font family as roman, since we have
3168 // no other idea what to do.
3169 if (default_family != "rmdefault" && default_family != "default") {
3170 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3175 if (default_family != "sfdefault") {
3176 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3180 case TYPEWRITER_FAMILY:
3181 if (default_family != "ttdefault") {
3182 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3186 case INHERIT_FAMILY:
3189 // the other tags are for internal use
3196 curr_size = font.fontInfo().size();
3197 FontSize old_size = font_old.size();
3198 if (old_size != curr_size) {
3201 case FONT_SIZE_TINY:
3202 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3204 case FONT_SIZE_SCRIPT:
3205 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3207 case FONT_SIZE_FOOTNOTE:
3208 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3210 case FONT_SIZE_SMALL:
3211 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3213 case FONT_SIZE_LARGE:
3214 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3216 case FONT_SIZE_LARGER:
3217 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3219 case FONT_SIZE_LARGEST:
3220 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3222 case FONT_SIZE_HUGE:
3223 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3225 case FONT_SIZE_HUGER:
3226 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3228 case FONT_SIZE_INCREASE:
3229 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3231 case FONT_SIZE_DECREASE:
3232 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3234 case FONT_SIZE_INHERIT:
3235 case FONT_SIZE_NORMAL:
3238 // the other tags are for internal use
3244 switch (curr_size) {
3245 case FONT_SIZE_TINY:
3246 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3249 case FONT_SIZE_SCRIPT:
3250 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3253 case FONT_SIZE_FOOTNOTE:
3254 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3257 case FONT_SIZE_SMALL:
3258 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3261 case FONT_SIZE_LARGE:
3262 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3265 case FONT_SIZE_LARGER:
3266 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3269 case FONT_SIZE_LARGEST:
3270 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3273 case FONT_SIZE_HUGE:
3274 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3277 case FONT_SIZE_HUGER:
3278 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3281 case FONT_SIZE_INCREASE:
3282 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3285 case FONT_SIZE_DECREASE:
3286 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3289 case FONT_SIZE_NORMAL:
3290 case FONT_SIZE_INHERIT:
3293 // the other tags are for internal use
3300 // Other such tags? What about the other text ranges?
3302 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3303 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3304 for (; cit != cen; ++cit)
3307 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3308 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3309 for (; sit != sen; ++sit)
3312 tagsToClose.clear();
3315 Inset const * inset = getInset(i);
3317 if (!runparams.for_toc || inset->isInToc()) {
3318 OutputParams np = runparams;
3319 np.local_font = &font;
3320 // If the paragraph has size 1, then we are in the "special
3321 // case" where we do not output the containing paragraph info
3322 if (!inset->getLayout().htmlisblock() && size() != 1)
3323 np.html_in_par = true;
3324 retval += inset->xhtml(xs, np);
3327 char_type c = getUChar(buf.masterBuffer()->params(), i);
3330 font_old = font.fontInfo();
3334 // I'm worried about what happens if a branch, say, is itself
3335 // wrapped in some font stuff. I think that will not work.
3337 if (close_paragraph)
3344 bool Paragraph::isHfill(pos_type pos) const
3346 Inset const * inset = getInset(pos);
3347 return inset && inset->isHfill();
3351 bool Paragraph::isNewline(pos_type pos) const
3353 // U+2028 LINE SEPARATOR
3354 // U+2029 PARAGRAPH SEPARATOR
3355 char_type const c = d->text_[pos];
3356 if (c == 0x2028 || c == 0x2029)
3358 Inset const * inset = getInset(pos);
3359 return inset && inset->lyxCode() == NEWLINE_CODE;
3363 bool Paragraph::isEnvSeparator(pos_type pos) const
3365 Inset const * inset = getInset(pos);
3366 return inset && inset->lyxCode() == SEPARATOR_CODE;
3370 bool Paragraph::isLineSeparator(pos_type pos) const
3372 char_type const c = d->text_[pos];
3373 if (isLineSeparatorChar(c))
3375 Inset const * inset = getInset(pos);
3376 return inset && inset->isLineSeparator();
3380 bool Paragraph::isWordSeparator(pos_type pos) const
3384 if (Inset const * inset = getInset(pos))
3385 return !inset->isLetter();
3386 // if we have a hard hyphen (no en- or emdash) or apostrophe
3387 // we pass this to the spell checker
3388 // FIXME: this method is subject to change, visit
3389 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3390 // to get an impression how complex this is.
3391 if (isHardHyphenOrApostrophe(pos))
3393 char_type const c = d->text_[pos];
3394 // We want to pass the escape chars to the spellchecker
3395 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3396 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3400 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3402 pos_type const psize = size();
3405 char_type const c = d->text_[pos];
3406 if (c != '-' && c != '\'')
3408 int nextpos = pos + 1;
3409 int prevpos = pos > 0 ? pos - 1 : 0;
3410 if ((nextpos == psize || isSpace(nextpos))
3411 && (pos == 0 || isSpace(prevpos)))
3417 bool Paragraph::needsCProtection() const
3419 // first check the layout of the paragraph
3420 if (layout().needcprotect) {
3421 // Environments need cprotection regardless the content
3422 if (layout().latextype == LATEX_ENVIRONMENT)
3425 // Commands need cprotection if they contain specific chars
3426 int const nchars_escape = 9;
3427 static char_type const chars_escape[nchars_escape] = {
3428 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3430 docstring const pars = asString();
3431 for (int k = 0; k < nchars_escape; k++) {
3432 if (contains(pars, chars_escape[k]))
3437 // now check whether we have insets that need cprotection
3438 pos_type size = d->text_.size();
3439 for (pos_type i = 0; i < size; ++i)
3440 if (isInset(i) && getInset(i)->needsCProtection())
3447 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3449 return d->speller_state_.getRange(pos);
3453 bool Paragraph::isChar(pos_type pos) const
3455 if (Inset const * inset = getInset(pos))
3456 return inset->isChar();
3457 char_type const c = d->text_[pos];
3458 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3462 bool Paragraph::isSpace(pos_type pos) const
3464 if (Inset const * inset = getInset(pos))
3465 return inset->isSpace();
3466 char_type const c = d->text_[pos];
3467 return lyx::isSpace(c);
3472 Paragraph::getParLanguage(BufferParams const & bparams) const
3475 return getFirstFontSettings(bparams).language();
3476 // FIXME: we should check the prev par as well (Lgb)
3477 return bparams.language;
3481 bool Paragraph::isRTL(BufferParams const & bparams) const
3483 return getParLanguage(bparams)->rightToLeft()
3484 && !inInset().getLayout().forceLTR();
3488 void Paragraph::changeLanguage(BufferParams const & bparams,
3489 Language const * from, Language const * to)
3491 // change language including dummy font change at the end
3492 for (pos_type i = 0; i <= size(); ++i) {
3493 Font font = getFontSettings(bparams, i);
3494 if (font.language() == from) {
3495 font.setLanguage(to);
3497 d->requestSpellCheck(i);
3503 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3505 Language const * doc_language = bparams.language;
3506 for (auto const & f : d->fontlist_)
3507 if (f.font().language() != ignore_language &&
3508 f.font().language() != latex_language &&
3509 f.font().language() != doc_language)
3515 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3517 for (auto const & f : d->fontlist_) {
3518 Language const * lang = f.font().language();
3519 if (lang != ignore_language &&
3520 lang != latex_language)
3526 docstring Paragraph::asString(int options) const
3528 return asString(0, size(), options);
3532 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3534 odocstringstream os;
3537 && options & AS_STR_LABEL
3538 && !d->params_.labelString().empty())
3539 os << d->params_.labelString() << ' ';
3541 for (pos_type i = beg; i < end; ++i) {
3542 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3544 char_type const c = d->text_[i];
3545 if (isPrintable(c) || c == '\t'
3546 || (c == '\n' && (options & AS_STR_NEWLINES)))
3548 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3549 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3550 LASSERT(runparams != 0, return docstring());
3551 getInset(i)->plaintext(os, *runparams);
3553 getInset(i)->toString(os);
3562 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3563 bool const shorten, bool const label) const
3565 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3566 if (label && !labelString().empty())
3567 os += labelString() + ' ';
3568 if (!layout().isTocCaption())
3570 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3573 char_type const c = d->text_[i];
3576 else if (c == META_INSET)
3577 getInset(i)->forOutliner(os, tmplen, false);
3580 Text::shortenForOutliner(os, maxlen);
3584 void Paragraph::setInsetOwner(Inset const * inset)
3586 d->inset_owner_ = inset;
3590 int Paragraph::id() const
3596 void Paragraph::setId(int id)
3602 Layout const & Paragraph::layout() const
3608 void Paragraph::setLayout(Layout const & layout)
3610 d->layout_ = &layout;
3614 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3616 setLayout(tc.defaultLayout());
3620 void Paragraph::setPlainLayout(DocumentClass const & tc)
3622 setLayout(tc.plainLayout());
3626 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3628 if (usePlainLayout())
3629 setPlainLayout(tclass);
3631 setDefaultLayout(tclass);
3635 Inset const & Paragraph::inInset() const
3637 LBUFERR(d->inset_owner_);
3638 return *d->inset_owner_;
3642 ParagraphParameters & Paragraph::params()
3648 ParagraphParameters const & Paragraph::params() const
3654 bool Paragraph::isFreeSpacing() const
3656 if (d->layout_->free_spacing)
3658 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3662 bool Paragraph::allowEmpty() const
3664 if (d->layout_->keepempty)
3666 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3670 bool Paragraph::brokenBiblio() const
3672 // there is a problem if there is no bibitem at position 0 or
3673 // if there is another bibitem in the paragraph.
3674 return d->layout_->labeltype == LABEL_BIBLIO
3675 && (d->insetlist_.find(BIBITEM_CODE) != 0
3676 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3680 int Paragraph::fixBiblio(Buffer const & buffer)
3682 // FIXME: What about the case where paragraph is not BIBLIO
3683 // but there is an InsetBibitem?
3684 // FIXME: when there was already an inset at 0, the return value is 1,
3685 // which does not tell whether another inset has been remove; the
3686 // cursor cannot be correctly updated.
3688 if (d->layout_->labeltype != LABEL_BIBLIO)
3691 bool const track_changes = buffer.params().track_changes;
3692 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3693 bool const hasbibitem0 = bibitem_pos == 0;
3696 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3697 // There was an InsetBibitem at pos 0, and no other one => OK
3698 if (bibitem_pos == -1)
3700 // there is a bibitem at the 0 position, but since
3701 // there is a second one, we copy the second on the
3702 // first. We're assuming there are at most two of
3703 // these, which there should be.
3704 // FIXME: why does it make sense to do that rather
3705 // than keep the first? (JMarc)
3706 Inset * inset = releaseInset(bibitem_pos);
3707 d->insetlist_.begin()->inset = inset;
3708 return -bibitem_pos;
3711 // We need to create an inset at the beginning
3713 if (bibitem_pos > 0) {
3714 // there was one somewhere in the paragraph, let's move it
3715 inset = d->insetlist_.release(bibitem_pos);
3716 eraseChar(bibitem_pos, track_changes);
3719 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3720 InsetCommandParams(BIBITEM_CODE));
3722 Font font(inherit_font, buffer.params().language);
3723 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3724 : Change::UNCHANGED));
3730 void Paragraph::checkAuthors(AuthorList const & authorList)
3732 d->changes_.checkAuthors(authorList);
3736 bool Paragraph::isChanged(pos_type pos) const
3738 return lookupChange(pos).changed();
3742 bool Paragraph::isInserted(pos_type pos) const
3744 return lookupChange(pos).inserted();
3748 bool Paragraph::isDeleted(pos_type pos) const
3750 return lookupChange(pos).deleted();
3754 InsetList const & Paragraph::insetList() const
3756 return d->insetlist_;
3760 void Paragraph::setInsetBuffers(Buffer & b)
3762 d->insetlist_.setBuffer(b);
3766 void Paragraph::resetBuffer()
3768 d->insetlist_.resetBuffer();
3772 Inset * Paragraph::releaseInset(pos_type pos)
3774 Inset * inset = d->insetlist_.release(pos);
3775 /// does not honour change tracking!
3776 eraseChar(pos, false);
3781 Inset * Paragraph::getInset(pos_type pos)
3783 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3784 ? d->insetlist_.get(pos) : 0;
3788 Inset const * Paragraph::getInset(pos_type pos) const
3790 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3791 ? d->insetlist_.get(pos) : 0;
3795 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3796 pos_type & right, TextCase action)
3798 // process sequences of modified characters; in change
3799 // tracking mode, this approach results in much better
3800 // usability than changing case on a char-by-char basis
3801 // We also need to track the current font, since font
3802 // changes within sequences can occur.
3803 vector<pair<char_type, Font> > changes;
3805 bool const trackChanges = bparams.track_changes;
3807 bool capitalize = true;
3809 for (; pos < right; ++pos) {
3810 char_type oldChar = d->text_[pos];
3811 char_type newChar = oldChar;
3813 // ignore insets and don't play with deleted text!
3814 if (oldChar != META_INSET && !isDeleted(pos)) {
3816 case text_lowercase:
3817 newChar = lowercase(oldChar);
3819 case text_capitalization:
3821 newChar = uppercase(oldChar);
3825 case text_uppercase:
3826 newChar = uppercase(oldChar);
3831 if (isWordSeparator(pos) || isDeleted(pos)) {
3832 // permit capitalization again
3836 if (oldChar != newChar) {
3837 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3838 if (pos != right - 1)
3840 // step behind the changing area
3844 int erasePos = pos - changes.size();
3845 for (size_t i = 0; i < changes.size(); i++) {
3846 insertChar(pos, changes[i].first,
3849 if (!eraseChar(erasePos, trackChanges)) {
3852 ++right; // expand selection
3860 int Paragraph::find(docstring const & str, bool cs, bool mw,
3861 pos_type start_pos, bool del) const
3863 pos_type pos = start_pos;
3864 int const strsize = str.length();
3866 pos_type const parsize = d->text_.size();
3867 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3868 // Ignore "invisible" letters such as ligature breaks
3869 // and hyphenation chars while searching
3870 while (pos < parsize - 1 && isInset(pos)) {
3871 odocstringstream os;
3872 getInset(pos)->toString(os);
3873 if (!getInset(pos)->isLetter() || !os.str().empty())
3877 if (cs && str[i] != d->text_[pos])
3879 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3881 if (!del && isDeleted(pos))
3888 // if necessary, check whether string matches word
3890 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3893 && !isWordSeparator(pos))
3897 return pos - start_pos;
3901 char_type Paragraph::getChar(pos_type pos) const
3903 return d->text_[pos];
3907 pos_type Paragraph::size() const
3909 return d->text_.size();
3913 bool Paragraph::empty() const
3915 return d->text_.empty();
3919 bool Paragraph::isInset(pos_type pos) const
3921 return d->text_[pos] == META_INSET;
3925 bool Paragraph::isSeparator(pos_type pos) const
3927 //FIXME: Are we sure this can be the only separator?
3928 return d->text_[pos] == ' ';
3932 void Paragraph::deregisterWords()
3934 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3935 Private::LangWordsMap::const_iterator ite = d->words_.end();
3936 for (; itl != ite; ++itl) {
3937 WordList & wl = theWordList(itl->first);
3938 Private::Words::const_iterator it = (itl->second).begin();
3939 Private::Words::const_iterator et = (itl->second).end();
3940 for (; it != et; ++it)
3947 void Paragraph::locateWord(pos_type & from, pos_type & to,
3948 word_location const loc) const
3951 case WHOLE_WORD_STRICT:
3952 if (from == 0 || from == size()
3953 || isWordSeparator(from)
3954 || isWordSeparator(from - 1)) {
3961 // If we are already at the beginning of a word, do nothing
3962 if (!from || isWordSeparator(from - 1))
3967 // always move the cursor to the beginning of previous word
3968 while (from && !isWordSeparator(from - 1))
3972 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3975 // no need to move the 'from' cursor
3979 while (to < size() && !isWordSeparator(to))
3984 void Paragraph::collectWords()
3986 for (pos_type pos = 0; pos < size(); ++pos) {
3987 if (isWordSeparator(pos))
3989 pos_type from = pos;
3990 locateWord(from, pos, WHOLE_WORD);
3991 // Work around MSVC warning: The statement
3992 // if (pos < from + lyxrc.completion_minlength)
3993 // triggers a signed vs. unsigned warning.
3994 // I don't know why this happens, it could be a MSVC bug, or
3995 // related to LLP64 (windows) vs. LP64 (unix) programming
3996 // model, or the C++ standard might be ambigous in the section
3997 // defining the "usual arithmetic conversions". However, using
3998 // a temporary variable is safe and works on all compilers.
3999 pos_type const endpos = from + lyxrc.completion_minlength;
4002 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4003 if (cit == d->fontlist_.end())
4005 Language const * lang = cit->font().language();
4006 docstring const word = asString(from, pos, AS_STR_NONE);
4007 d->words_[lang->lang()].insert(word);
4012 void Paragraph::registerWords()
4014 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4015 Private::LangWordsMap::const_iterator ite = d->words_.end();
4016 for (; itl != ite; ++itl) {
4017 WordList & wl = theWordList(itl->first);
4018 Private::Words::const_iterator it = (itl->second).begin();
4019 Private::Words::const_iterator et = (itl->second).end();
4020 for (; it != et; ++it)
4026 void Paragraph::updateWords()
4034 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4036 SkipPositionsIterator begin = skips.begin();
4037 SkipPositions::iterator end = skips.end();
4038 if (pos > 0 && begin < end) {
4040 if (end->last == pos - 1) {
4045 skips.insert(end, FontSpan(pos, pos));
4049 Language * Paragraph::Private::locateSpellRange(
4050 pos_type & from, pos_type & to,
4051 SkipPositions & skips) const
4053 // skip leading white space
4054 while (from < to && owner_->isWordSeparator(from))
4056 // don't check empty range
4059 // get current language
4060 Language * lang = getSpellLanguage(from);
4061 pos_type last = from;
4062 bool samelang = true;
4063 bool sameinset = true;
4064 while (last < to && samelang && sameinset) {
4065 // hop to end of word
4066 while (last < to && !owner_->isWordSeparator(last)) {
4067 if (owner_->getInset(last)) {
4068 appendSkipPosition(skips, last);
4069 } else if (owner_->isDeleted(last)) {
4070 appendSkipPosition(skips, last);
4074 // hop to next word while checking for insets
4075 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4076 if (Inset const * inset = owner_->getInset(last))
4077 sameinset = inset->isChar() && inset->isLetter();
4078 if (sameinset && owner_->isDeleted(last)) {
4079 appendSkipPosition(skips, last);
4084 if (sameinset && last < to) {
4085 // now check for language change
4086 samelang = lang == getSpellLanguage(last);
4089 // if language change detected backstep is needed
4097 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4100 const_cast<Language *>(owner_->getFontSettings(
4101 inset_owner_->buffer().params(), from).language());
4102 if (lang == inset_owner_->buffer().params().language
4103 && !lyxrc.spellchecker_alt_lang.empty()) {
4105 string const lang_variety =
4106 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4107 lang->setCode(lang_code);
4108 lang->setVariety(lang_variety);
4114 void Paragraph::requestSpellCheck(pos_type pos)
4116 d->requestSpellCheck(pos);
4120 bool Paragraph::needsSpellCheck() const
4122 SpellChecker::ChangeNumber speller_change_number = 0;
4123 if (theSpellChecker())
4124 speller_change_number = theSpellChecker()->changeNumber();
4125 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4126 d->speller_state_.needsCompleteRefresh(speller_change_number);
4128 return d->needsSpellCheck();
4132 bool Paragraph::Private::ignoreWord(docstring const & word) const
4134 // Ignore words with digits
4135 // FIXME: make this customizable
4136 // (note that some checkers ignore words with digits by default)
4137 docstring::const_iterator cit = word.begin();
4138 docstring::const_iterator const end = word.end();
4139 for (; cit != end; ++cit) {
4140 if (isNumber((*cit)))
4147 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4148 WordLangTuple & wl, docstring_list & suggestions,
4149 bool do_suggestion, bool check_learned) const
4151 SpellChecker::Result result = SpellChecker::WORD_OK;
4152 SpellChecker * speller = theSpellChecker();
4156 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4159 locateWord(from, to, WHOLE_WORD);
4160 if (from == to || from >= size())
4163 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4164 Language * lang = d->getSpellLanguage(from);
4166 wl = WordLangTuple(word, lang);
4171 if (needsSpellCheck() || check_learned) {
4173 if (!d->ignoreWord(word)) {
4174 bool const trailing_dot = to < size() && d->text_[to] == '.';
4175 result = speller->check(wl);
4176 if (SpellChecker::misspelled(result) && trailing_dot) {
4177 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4178 result = speller->check(wl);
4179 if (!SpellChecker::misspelled(result)) {
4180 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4182 from << ".." << to << "]");
4184 // spell check with dot appended failed too
4185 // restore original word/lang value
4186 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4187 wl = WordLangTuple(word, lang);
4191 if (!SpellChecker::misspelled(result)) {
4192 // area up to the begin of the next word is not misspelled
4193 while (end < size() && isWordSeparator(end))
4196 d->setMisspelled(from, end, result);
4198 result = d->speller_state_.getState(from);
4202 suggestions.clear();
4204 if (SpellChecker::misspelled(result)) {
4205 LYXERR(Debug::GUI, "misspelled word: \"" <<
4207 from << ".." << to << "]");
4209 speller->suggest(wl, suggestions);
4215 void Paragraph::anonymize()
4217 // This is a very crude anonymization for now
4218 for (char_type & c : d->text_)
4219 if (isLetterChar(c) || isNumber(c))
4224 void Paragraph::Private::markMisspelledWords(
4225 pos_type const & first, pos_type const & last,
4226 SpellChecker::Result result,
4227 docstring const & word,
4228 SkipPositions const & skips)
4230 if (!SpellChecker::misspelled(result)) {
4231 setMisspelled(first, last, SpellChecker::WORD_OK);
4235 SpellChecker * speller = theSpellChecker();
4236 // locate and enumerate the error positions
4237 int nerrors = speller->numMisspelledWords();
4239 SkipPositionsIterator it = skips.begin();
4240 SkipPositionsIterator et = skips.end();
4241 for (int index = 0; index < nerrors; ++index) {
4244 speller->misspelledWord(index, wstart, wlen);
4245 /// should not happen if speller supports range checks
4246 if (!wlen) continue;
4247 docstring const misspelled = word.substr(wstart, wlen);
4248 wstart += first + numskipped;
4249 if (snext < wstart) {
4250 /// mark the range of correct spelling
4251 numskipped += countSkips(it, et, wstart);
4252 setMisspelled(snext,
4253 wstart - 1, SpellChecker::WORD_OK);
4255 snext = wstart + wlen;
4256 numskipped += countSkips(it, et, snext);
4257 /// mark the range of misspelling
4258 setMisspelled(wstart, snext, result);
4259 LYXERR(Debug::GUI, "misspelled word: \"" <<
4260 misspelled << "\" [" <<
4261 wstart << ".." << (snext-1) << "]");
4264 if (snext <= last) {
4265 /// mark the range of correct spelling at end
4266 setMisspelled(snext, last, SpellChecker::WORD_OK);
4271 void Paragraph::spellCheck() const
4273 SpellChecker * speller = theSpellChecker();
4274 if (!speller || empty() ||!needsSpellCheck())
4278 d->rangeOfSpellCheck(start, endpos);
4279 if (speller->canCheckParagraph()) {
4280 // loop until we leave the range
4281 for (pos_type first = start; first < endpos; ) {
4282 pos_type last = endpos;
4283 Private::SkipPositions skips;
4284 Language * lang = d->locateSpellRange(first, last, skips);
4285 if (first >= endpos)
4287 // start the spell checker on the unit of meaning
4288 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4289 WordLangTuple wl = WordLangTuple(word, lang);
4290 SpellChecker::Result result = word.size() ?
4291 speller->check(wl) : SpellChecker::WORD_OK;
4292 d->markMisspelledWords(first, last, result, word, skips);
4296 static docstring_list suggestions;
4297 pos_type to = endpos;
4298 while (start < endpos) {
4300 spellCheck(start, to, wl, suggestions, false);
4304 d->readySpellCheck();
4308 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4310 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4311 if (result || pos <= 0 || pos > size())
4313 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4314 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4319 string Paragraph::magicLabel() const
4322 ss << "magicparlabel-" << id();