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 if (docstring(1, next) == latex1) {
891 // The encoding supports the combination:
892 // output as is (combining char after base char).
893 os << latex2 << latex1;
894 return latex1.length() + latex2.length();
897 // Handle combining characters in "script" context (i.e., \textgreek and \textcyrillic)
898 docstring::size_type const brace1 = latex2.find_first_of(from_ascii("{"));
899 docstring::size_type const brace2 = latex2.find_last_of(from_ascii("}"));
900 string script = to_ascii(latex2.substr(1, brace1 - 1));
901 // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
902 // whether they are encodable or not (it only depends on the font encoding)
903 if (!runparams.isFullUnicode())
904 // This will get us a script value to deal with below
905 Encodings::isKnownScriptChar(c, script);
909 if (runparams.local_font)
910 fontenc = runparams.local_font->language()->fontenc();
912 fontenc = runparams.main_fontenc;
913 docstring scriptmacro;
915 if (script == "textgreek" || script == "textcyrillic") {
916 // We separate the script macro (\text[greek|cyr]) from the rest,
917 // since we need to include the combining char in it (#6463).
918 // This is "the rest":
921 latex2 = latex2.substr(pos, length);
922 // We only need the script macro with non-native font encodings
923 if (Encodings::needsScriptWrapper(script, fontenc)) {
924 scriptmacro = from_ascii("\\" + script + "{");
925 cb = from_ascii("}");
931 // polutonikogreek does not play nice with brackets
932 if (!runparams.local_font
933 || runparams.local_font->language()->lang() != "polutonikogreek") {
934 lb = from_ascii("{");
935 rb = from_ascii("}");
938 os << scriptmacro << latex1 << lb << latex2 << rb << cb;
939 return latex1.length() + latex2.length() + lb.length() + rb.length() + cb.length();
943 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
946 unsigned int & column,
948 Layout const & style)
950 if (style.pass_thru || runparams.pass_thru)
953 if (i + 1 < int(text_.size())) {
954 char_type next = text_[i + 1];
955 if (Encodings::isCombiningChar(next)) {
956 // This space has an accent, so we must always output it.
957 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
962 if (runparams.linelen > 0
963 && column > runparams.linelen
965 && text_[i - 1] != ' '
966 && (i + 1 < int(text_.size()))
967 // same in FreeSpacing mode
968 && !owner_->isFreeSpacing()
969 // In typewriter mode, we want to avoid
970 // ! . ? : at the end of a line
971 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
972 && (text_[i - 1] == '.'
973 || text_[i - 1] == '?'
974 || text_[i - 1] == ':'
975 || text_[i - 1] == '!'))) {
977 os.texrow().start(owner_->id(), i + 1);
979 } else if (style.free_spacing) {
988 int Paragraph::Private::writeScriptChars(OutputParams const & runparams,
990 docstring const & ltx,
991 Change const & runningChange,
992 Encoding const & encoding,
993 string const fontenc,
996 // FIXME: modifying i here is not very nice...
998 // We only arrive here when character text_[i] could not be translated
999 // into the current latex encoding (or its latex translation has been forced,)
1000 // and it belongs to a known script.
1001 // Parameter ltx contains the latex translation of text_[i] as specified
1002 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
1003 // The latex macro name "textXXX" specifies the script to which text_[i]
1004 // belongs and we use it in order to check whether characters from the
1005 // same script immediately follow, such that we can collect them in a
1006 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
1007 // for the first char but only "<spec>" for all subsequent chars.
1008 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
1009 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
1010 string script = to_ascii(ltx.substr(1, brace1 - 1));
1012 int length = brace2;
1013 bool closing_brace = true;
1014 if (!Encodings::needsScriptWrapper(script, fontenc)) {
1015 // Correct font encoding is being used, so we can avoid \text[greek|cyr].
1018 closing_brace = false;
1020 os << ltx.substr(pos, length);
1021 int size = text_.size();
1022 while (i + 1 < size) {
1023 char_type const next = text_[i + 1];
1024 // Stop here if next character belongs to another script
1025 // or there is a change in change tracking status.
1026 if (!Encodings::isKnownScriptChar(next, script) ||
1027 runningChange != owner_->lookupChange(i + 1))
1031 FontList::const_iterator cit = fontlist_.begin();
1032 FontList::const_iterator end = fontlist_.end();
1033 for (; cit != end; ++cit) {
1034 if (cit->pos() >= i && !found) {
1035 prev_font = cit->font();
1038 if (cit->pos() >= i + 1)
1041 // Stop here if there is a font attribute or encoding change.
1042 if (found && cit != end && prev_font != cit->font())
1045 // Check whether we have a combining pair
1046 char_type next_next = '\0';
1048 next_next = text_[i + 2];
1049 if (Encodings::isCombiningChar(next_next)) {
1050 length += latexSurrogatePair(os, next, next_next, runparams) - 1;
1056 docstring const latex = encoding.latexChar(next).first;
1057 docstring::size_type const b1 =
1058 latex.find_first_of(from_ascii("{"));
1059 docstring::size_type const b2 =
1060 latex.find_last_of(from_ascii("}"));
1061 int const len = b2 - b1 - 1;
1062 os << latex.substr(b1 + 1, len);
1066 if (closing_brace) {
1074 void Paragraph::Private::latexInset(BufferParams const & bparams,
1076 OutputParams & runparams,
1077 Font & running_font,
1079 Font const & outerfont,
1081 Change & running_change,
1082 Layout const & style,
1084 unsigned int & column)
1086 Inset * inset = owner_->getInset(i);
1089 if (style.pass_thru) {
1090 odocstringstream ods;
1091 inset->plaintext(ods, runparams);
1096 // FIXME: move this to InsetNewline::latex
1097 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1098 // newlines are handled differently here than
1099 // the default in simpleTeXSpecialChars().
1100 if (!style.newline_allowed) {
1104 bool needPar = false;
1105 column += running_font.latexWriteEndChanges(
1106 os, bparams, runparams,
1107 basefont, basefont, needPar);
1111 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1114 basefont = owner_->getLayoutFont(bparams, outerfont);
1115 running_font = basefont;
1117 if (runparams.moving_arg)
1121 os.texrow().start(owner_->id(), i + 1);
1125 if (owner_->isDeleted(i)) {
1126 if( ++runparams.inDeletedInset == 1)
1127 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1130 if (inset->canTrackChanges()) {
1131 column += Changes::latexMarkChange(os, bparams, running_change,
1132 Change(Change::UNCHANGED), runparams);
1133 running_change = Change(Change::UNCHANGED);
1137 odocstream::pos_type const len = os.os().tellp();
1139 if (inset->forceLTR()
1140 && !runparams.use_polyglossia
1141 && running_font.isRightToLeft()
1142 // ERT is an exception, it should be output with no
1143 // decorations at all
1144 && inset->lyxCode() != ERT_CODE) {
1145 if (running_font.language()->lang() == "farsi")
1146 os << "\\beginL" << termcmd;
1152 // FIXME: Bug: we can have an empty font change here!
1153 // if there has just been a font change, we are going to close it
1154 // right now, which means stupid latex code like \textsf{}. AFAIK,
1155 // this does not harm dvi output. A minor bug, thus (JMarc)
1157 // Some insets cannot be inside a font change command.
1158 // However, even such insets *can* be placed in \L or \R
1159 // or their equivalents (for RTL language switches), so we don't
1160 // close the language in those cases.
1161 // ArabTeX, though, cannot handle this special behavior, it seems.
1162 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1163 || running_font.language()->lang() == "arabic_arabtex";
1164 if (open_font && !inset->inheritFont()) {
1165 bool needPar = false;
1166 bool closeLanguage = arabtex
1167 || basefont.isRightToLeft() == running_font.isRightToLeft();
1168 unsigned int count = running_font.latexWriteEndChanges(os,
1169 bparams, runparams, basefont, basefont,
1170 needPar, closeLanguage);
1172 // if any font properties were closed, update the running_font,
1173 // making sure, however, to leave the language as it was
1175 // FIXME: probably a better way to keep track of the old
1176 // language, than copying the entire font?
1177 Font const copy_font(running_font);
1178 basefont = owner_->getLayoutFont(bparams, outerfont);
1179 running_font = basefont;
1181 running_font.setLanguage(copy_font.language());
1182 // leave font open if language is still open
1183 open_font = (running_font.language() == basefont.language());
1185 runparams.local_font = &basefont;
1189 size_t const previous_row_count = os.texrow().rows();
1192 runparams.lastid = id_;
1193 runparams.lastpos = i;
1194 inset->latex(os, runparams);
1195 } catch (EncodingException & e) {
1196 // add location information and throw again.
1203 if (running_font.language()->lang() == "farsi")
1204 os << "\\endL" << termcmd;
1209 if (os.texrow().rows() > previous_row_count) {
1210 os.texrow().start(owner_->id(), i + 1);
1213 column += (unsigned int)(os.os().tellp() - len);
1216 if (owner_->isDeleted(i))
1217 --runparams.inDeletedInset;
1221 void Paragraph::Private::latexSpecialChar(otexstream & os,
1222 BufferParams const & bparams,
1223 OutputParams const & runparams,
1224 Font const & running_font,
1225 Change const & running_change,
1226 Layout const & style,
1229 unsigned int & column)
1231 // With polyglossia, brackets and stuff need not be reversed
1232 // in RTL scripts (see bug #8251)
1233 char_type const c = (runparams.use_polyglossia) ?
1234 owner_->getUChar(bparams, i) : text_[i];
1236 if (style.pass_thru || runparams.pass_thru
1237 || contains(style.pass_thru_chars, c)
1238 || contains(runparams.pass_thru_chars, c)) {
1240 Encoding const * const enc = runparams.encoding;
1241 if (enc && !enc->encodable(c))
1242 throw EncodingException(c);
1248 // TIPA uses its own T3 encoding
1249 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1251 // If T1 font encoding is used, use the special
1252 // characters it provides.
1253 // NOTE: Some languages reset the font encoding internally to a
1254 // non-standard font encoding. If we are using such a language,
1255 // we do not output special T1 chars.
1256 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1257 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1258 && latexSpecialT1(c, os, i, column))
1260 // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
1261 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1262 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1265 // Otherwise, we use what LaTeX provides us.
1268 os << "\\textbackslash" << termcmd;
1272 os << "\\textless" << termcmd;
1276 os << "\\textgreater" << termcmd;
1280 os << "\\textbar" << termcmd;
1285 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1286 (end_pos == -1 || i + 1 < end_pos) &&
1287 text_[i+1] == '-') {
1288 // Prevent "--" becoming an en dash and "---" an em dash.
1289 // (Within \ttfamily, "---" is merged to en dash + hyphen.)
1295 os << "\\textquotedbl" << termcmd;
1300 case '%': case '#': case '{':
1308 os << "\\textasciitilde" << termcmd;
1313 os << "\\textasciicircum" << termcmd;
1320 // avoid being mistaken for optional arguments
1328 // Blanks are printed before font switching.
1329 // Sure? I am not! (try nice-latex)
1330 // I am sure it's correct. LyX might be smarter
1331 // in the future, but for now, nothing wrong is
1337 // XeTeX's dash behaviour is determined via a global setting
1338 if (bparams.use_dash_ligatures
1339 && owner_->getFontSettings(bparams, i).fontInfo().family() != TYPEWRITER_FAMILY
1341 // TODO #10961: && not in inset Flex Code
1342 // TODO #10961: && not in layout LyXCode
1343 && (!bparams.useNonTeXFonts || runparams.flavor != OutputParams::XETEX)) {
1360 Encoding const & encoding = *(runparams.encoding);
1361 char_type next = '\0';
1362 if (i + 1 < int(text_.size())) {
1363 next = text_[i + 1];
1364 if (Encodings::isCombiningChar(next)) {
1365 column += latexSurrogatePair(os, c, next, runparams) - 1;
1371 pair<docstring, bool> latex = encoding.latexChar(c);
1372 docstring nextlatex;
1373 bool nexttipas = false;
1374 string nexttipashortcut;
1375 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1376 nextlatex = encoding.latexChar(next).first;
1377 if (runparams.inIPA) {
1378 nexttipashortcut = Encodings::TIPAShortcut(next);
1379 nexttipas = !nexttipashortcut.empty();
1383 if (runparams.inIPA) {
1384 string const tipashortcut = Encodings::TIPAShortcut(c);
1385 if (!tipashortcut.empty()) {
1386 latex.first = from_ascii(tipashortcut);
1387 latex.second = false;
1392 if (running_font.language()->lang() == bparams.language->lang())
1393 fontenc = runparams.main_fontenc;
1395 fontenc = running_font.language()->fontenc();
1396 // "Script chars" need to embraced in \textcyrillic and \textgreek notwithstanding
1397 // whether they are encodable or not (it only depends on the font encoding)
1398 if (!runparams.isFullUnicode() && Encodings::isKnownScriptChar(c, script)) {
1399 docstring const wrapper = from_ascii("\\" + script + "{");
1400 docstring ltx = latex.first;
1401 if (!prefixIs(ltx, wrapper))
1402 ltx = wrapper + latex.first + from_ascii("}");
1403 column += writeScriptChars(runparams, os, ltx, running_change,
1404 encoding, fontenc, i) - 1;
1405 } else if (latex.second
1406 && ((!prefixIs(nextlatex, '\\')
1407 && !prefixIs(nextlatex, '{')
1408 && !prefixIs(nextlatex, '}'))
1410 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1412 // Prevent eating of a following
1413 // space or command corruption by
1414 // following characters
1415 if (next == ' ' || next == '\0') {
1416 column += latex.first.length() + 1;
1417 os << latex.first << "{}";
1419 column += latex.first.length();
1420 os << latex.first << " ";
1423 column += latex.first.length() - 1;
1431 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1432 pos_type i, unsigned int & column)
1438 // In T1 encoding, these characters exist
1439 // but we should avoid ligatures
1440 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1442 os << "\\textcompwordmark" << termcmd;
1449 // soul.sty breaks with \char`\"
1450 os << "\\textquotedbl" << termcmd;
1459 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1460 pos_type i, unsigned int & column)
1462 // TU encoding is currently on par with T1.
1463 return latexSpecialT1(c, os, i, column);
1467 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1468 pos_type /*i*/, unsigned int & column)
1478 os << "\\textvertline" << termcmd;
1487 void Paragraph::Private::validate(LaTeXFeatures & features) const
1489 if (layout_->inpreamble && inset_owner_) {
1490 // FIXME: Using a string stream here circumvents the encoding
1491 // switching machinery of odocstream. Therefore the
1492 // output is wrong if this paragraph contains content
1493 // that needs to switch encoding.
1494 Buffer const & buf = inset_owner_->buffer();
1495 otexstringstream os;
1496 os << layout_->preamble();
1497 size_t const length = os.length();
1498 TeXOnePar(buf, buf.text(), buf.getParFromID(owner_->id()).pit(), os,
1499 features.runparams(), string(), 0, -1, true);
1500 if (os.length() > length)
1501 features.addPreambleSnippet(os.release(), true);
1504 if (features.runparams().flavor == OutputParams::HTML
1505 && layout_->htmltitle()) {
1506 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1509 // check the params.
1510 if (!params_.spacing().isDefault())
1511 features.require("setspace");
1514 features.useLayout(layout_->name());
1517 fontlist_.validate(features);
1519 // then the indentation
1520 if (!params_.leftIndent().zero())
1521 features.require("ParagraphLeftIndent");
1524 InsetList::const_iterator icit = insetlist_.begin();
1525 InsetList::const_iterator iend = insetlist_.end();
1526 for (; icit != iend; ++icit) {
1528 features.inDeletedInset(owner_->isDeleted(icit->pos));
1529 if (icit->inset->lyxCode() == FOOT_CODE) {
1530 // FIXME: an item inset would make things much easier.
1531 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1532 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1533 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1534 && (icit->pos < begin_of_body_
1535 || (icit->pos == begin_of_body_
1536 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1537 features.saveNoteEnv("description");
1539 icit->inset->validate(features);
1540 features.inDeletedInset(false);
1541 if (layout_->needprotect &&
1542 icit->inset->lyxCode() == FOOT_CODE)
1543 features.require("NeedLyXFootnoteCode");
1547 // then the contents
1548 BufferParams const bp = features.runparams().is_child
1549 ? features.buffer().masterParams() : features.buffer().params();
1550 string bscript = "textbaltic";
1551 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1552 char_type c = text_[i];
1554 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1555 features.require("textquotedblp");
1556 else if (bp.main_font_encoding() != "T1"
1557 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1558 features.require("textquotedbl");
1559 } else if (Encodings::isKnownScriptChar(c, bscript)){
1560 string fontenc = (&owner_->getFontSettings(bp, i))->language()->fontenc();
1561 if (fontenc.empty())
1562 fontenc = features.runparams().main_fontenc;
1563 if (Encodings::needsScriptWrapper("textbaltic", fontenc))
1564 features.require("textbalticdefs");
1565 } else if (!bp.use_dash_ligatures
1566 && (c == 0x2013 || c == 0x2014)
1567 && bp.useNonTeXFonts
1568 && features.runparams().flavor == OutputParams::XETEX)
1569 // XeTeX's dash behaviour is determined via a global setting
1570 features.require("xetexdashbreakstate");
1571 BufferEncodings::validate(c, features);
1575 /////////////////////////////////////////////////////////////////////
1579 /////////////////////////////////////////////////////////////////////
1582 Layout const emptyParagraphLayout;
1585 Paragraph::Paragraph()
1586 : d(new Paragraph::Private(this, emptyParagraphLayout))
1593 Paragraph::Paragraph(Paragraph const & par)
1594 : itemdepth(par.itemdepth),
1595 d(new Paragraph::Private(*par.d, this))
1601 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1602 : itemdepth(par.itemdepth),
1603 d(new Paragraph::Private(*par.d, this, beg, end))
1609 Paragraph & Paragraph::operator=(Paragraph const & par)
1611 // needed as we will destroy the private part before copying it
1613 itemdepth = par.itemdepth;
1617 d = new Private(*par.d, this);
1624 Paragraph::~Paragraph()
1633 // this shall be called just before every "os << ..." action.
1634 void flushString(ostream & os, docstring & s)
1643 void Paragraph::write(ostream & os, BufferParams const & bparams,
1644 depth_type & dth) const
1646 // The beginning or end of a deeper (i.e. nested) area?
1647 if (dth != d->params_.depth()) {
1648 if (d->params_.depth() > dth) {
1649 while (d->params_.depth() > dth) {
1650 os << "\n\\begin_deeper";
1654 while (d->params_.depth() < dth) {
1655 os << "\n\\end_deeper";
1661 // First write the layout
1662 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1664 d->params_.write(os);
1666 Font font1(inherit_font, bparams.language);
1668 Change running_change = Change(Change::UNCHANGED);
1670 // this string is used as a buffer to avoid repetitive calls
1671 // to to_utf8(), which turn out to be expensive (JMarc)
1672 docstring write_buffer;
1675 for (pos_type i = 0; i <= size(); ++i) {
1677 Change const & change = lookupChange(i);
1678 if (change != running_change)
1679 flushString(os, write_buffer);
1680 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1681 running_change = change;
1686 // Write font changes
1687 Font font2 = getFontSettings(bparams, i);
1688 if (font2 != font1) {
1689 flushString(os, write_buffer);
1690 font2.lyxWriteChanges(font1, os);
1695 char_type const c = d->text_[i];
1698 if (Inset const * inset = getInset(i)) {
1699 flushString(os, write_buffer);
1700 if (inset->directWrite()) {
1701 // international char, let it write
1702 // code directly so it's shorter in
1708 os << "\\begin_inset ";
1710 os << "\n\\end_inset\n\n";
1713 // FIXME This can be removed again once the mystery
1714 // crash has been resolved.
1719 flushString(os, write_buffer);
1720 os << "\n\\backslash\n";
1724 flushString(os, write_buffer);
1725 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1732 if ((column > 70 && c == ' ')
1734 flushString(os, write_buffer);
1738 // this check is to amend a bug. LyX sometimes
1739 // inserts '\0' this could cause problems.
1741 write_buffer.push_back(c);
1743 LYXERR0("NUL char in structure.");
1749 flushString(os, write_buffer);
1750 os << "\n\\end_layout\n";
1751 // FIXME This can be removed again once the mystery
1752 // crash has been resolved.
1757 void Paragraph::validate(LaTeXFeatures & features) const
1759 d->validate(features);
1760 if (needsCProtection())
1761 features.require("cprotect");
1765 void Paragraph::insert(pos_type start, docstring const & str,
1766 Font const & font, Change const & change)
1768 for (size_t i = 0, n = str.size(); i != n ; ++i)
1769 insertChar(start + i, str[i], font, change);
1773 void Paragraph::appendChar(char_type c, Font const & font,
1774 Change const & change)
1777 d->changes_.insert(change, d->text_.size());
1778 // when appending characters, no need to update tables
1779 d->text_.push_back(c);
1780 setFont(d->text_.size() - 1, font);
1781 d->requestSpellCheck(d->text_.size() - 1);
1785 void Paragraph::appendString(docstring const & s, Font const & font,
1786 Change const & change)
1788 pos_type end = s.size();
1789 size_t oldsize = d->text_.size();
1790 size_t newsize = oldsize + end;
1791 size_t capacity = d->text_.capacity();
1792 if (newsize >= capacity)
1793 d->text_.reserve(max(capacity + 100, newsize));
1795 // when appending characters, no need to update tables
1798 // FIXME: Optimize this!
1799 for (size_t i = oldsize; i != newsize; ++i) {
1801 d->changes_.insert(change, i);
1802 d->requestSpellCheck(i);
1804 d->fontlist_.set(oldsize, font);
1805 d->fontlist_.set(newsize - 1, font);
1809 void Paragraph::insertChar(pos_type pos, char_type c,
1812 d->insertChar(pos, c, Change(trackChanges ?
1813 Change::INSERTED : Change::UNCHANGED));
1817 void Paragraph::insertChar(pos_type pos, char_type c,
1818 Font const & font, bool trackChanges)
1820 d->insertChar(pos, c, Change(trackChanges ?
1821 Change::INSERTED : Change::UNCHANGED));
1826 void Paragraph::insertChar(pos_type pos, char_type c,
1827 Font const & font, Change const & change)
1829 d->insertChar(pos, c, change);
1834 void Paragraph::resetFonts(Font const & font)
1836 d->fontlist_.clear();
1837 d->fontlist_.set(0, font);
1838 d->fontlist_.set(d->text_.size() - 1, font);
1841 // Gets uninstantiated font setting at position.
1842 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1846 LYXERR0("pos: " << pos << " size: " << size());
1850 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1851 if (cit != d->fontlist_.end())
1854 if (pos == size() && !empty())
1855 return getFontSettings(bparams, pos - 1);
1857 // Optimisation: avoid a full font instantiation if there is no
1858 // language change from previous call.
1859 static Font previous_font;
1860 static Language const * previous_lang = 0;
1861 Language const * lang = getParLanguage(bparams);
1862 if (lang != previous_lang) {
1863 previous_lang = lang;
1864 previous_font = Font(inherit_font, lang);
1866 return previous_font;
1870 FontSpan Paragraph::fontSpan(pos_type pos) const
1872 LBUFERR(pos <= size());
1875 return FontSpan(pos, pos);
1878 FontList::const_iterator cit = d->fontlist_.begin();
1879 FontList::const_iterator end = d->fontlist_.end();
1880 for (; cit != end; ++cit) {
1881 if (cit->pos() >= pos) {
1882 if (pos >= beginOfBody())
1883 return FontSpan(max(start, beginOfBody()),
1886 return FontSpan(start,
1887 min(beginOfBody() - 1,
1890 start = cit->pos() + 1;
1893 // This should not happen, but if so, we take no chances.
1894 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1895 LASSERT(false, return FontSpan(pos, pos));
1899 // Gets uninstantiated font setting at position 0
1900 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1902 if (!empty() && !d->fontlist_.empty())
1903 return d->fontlist_.begin()->font();
1905 // Optimisation: avoid a full font instantiation if there is no
1906 // language change from previous call.
1907 static Font previous_font;
1908 static Language const * previous_lang = 0;
1909 if (bparams.language != previous_lang) {
1910 previous_lang = bparams.language;
1911 previous_font = Font(inherit_font, bparams.language);
1914 return previous_font;
1918 // Gets the fully instantiated font at a given position in a paragraph
1919 // This is basically the same function as Text::GetFont() in text2.cpp.
1920 // The difference is that this one is used for generating the LaTeX file,
1921 // and thus cosmetic "improvements" are disallowed: This has to deliver
1922 // the true picture of the buffer. (Asger)
1923 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1924 Font const & outerfont) const
1928 Font font = getFontSettings(bparams, pos);
1930 pos_type const body_pos = beginOfBody();
1931 FontInfo & fi = font.fontInfo();
1933 fi.realize(d->layout_->labelfont);
1935 fi.realize(d->layout_->font);
1937 fi.realize(outerfont.fontInfo());
1938 fi.realize(bparams.getFont().fontInfo());
1944 Font const Paragraph::getLabelFont
1945 (BufferParams const & bparams, Font const & outerfont) const
1947 FontInfo tmpfont = d->layout_->labelfont;
1948 tmpfont.realize(outerfont.fontInfo());
1949 tmpfont.realize(bparams.getFont().fontInfo());
1950 return Font(tmpfont, getParLanguage(bparams));
1954 Font const Paragraph::getLayoutFont
1955 (BufferParams const & bparams, Font const & outerfont) const
1957 FontInfo tmpfont = d->layout_->font;
1958 tmpfont.realize(outerfont.fontInfo());
1959 tmpfont.realize(bparams.getFont().fontInfo());
1960 return Font(tmpfont, getParLanguage(bparams));
1964 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1966 char_type c = d->text_[pos];
1967 if (!getFontSettings(bparams, pos).isRightToLeft())
1970 // FIXME: The arabic special casing is due to the difference of arabic
1971 // round brackets input introduced in r18599. Check if this should be
1972 // unified with Hebrew or at least if all bracket types should be
1973 // handled the same (file format change in either case).
1974 string const & lang = getFontSettings(bparams, pos).language()->lang();
1975 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1980 uc = arabic ? c : ')';
1983 uc = arabic ? c : '(';
2009 void Paragraph::setFont(pos_type pos, Font const & font)
2011 LASSERT(pos <= size(), return);
2013 // First, reduce font against layout/label font
2014 // Update: The setCharFont() routine in text2.cpp already
2015 // reduces font, so we don't need to do that here. (Asger)
2017 d->fontlist_.set(pos, font);
2021 void Paragraph::makeSameLayout(Paragraph const & par)
2023 d->layout_ = par.d->layout_;
2024 d->params_ = par.d->params_;
2028 bool Paragraph::stripLeadingSpaces(bool trackChanges)
2030 if (isFreeSpacing())
2036 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
2037 if (eraseChar(pos, trackChanges))
2043 return count > 0 || pos > 0;
2047 bool Paragraph::hasSameLayout(Paragraph const & par) const
2049 return par.d->layout_ == d->layout_
2050 && d->params_.sameLayout(par.d->params_);
2054 depth_type Paragraph::getDepth() const
2056 return d->params_.depth();
2060 depth_type Paragraph::getMaxDepthAfter() const
2062 if (d->layout_->isEnvironment())
2063 return d->params_.depth() + 1;
2065 return d->params_.depth();
2069 LyXAlignment Paragraph::getAlign() const
2071 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2072 return d->layout_->align;
2074 return d->params_.align();
2078 docstring const & Paragraph::labelString() const
2080 return d->params_.labelString();
2084 // the next two functions are for the manual labels
2085 docstring const Paragraph::getLabelWidthString() const
2087 if (d->layout_->margintype == MARGIN_MANUAL
2088 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2089 return d->params_.labelWidthString();
2091 return _("Senseless with this layout!");
2095 void Paragraph::setLabelWidthString(docstring const & s)
2097 d->params_.labelWidthString(s);
2101 docstring Paragraph::expandLabel(Layout const & layout,
2102 BufferParams const & bparams) const
2104 return expandParagraphLabel(layout, bparams, true);
2108 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2109 BufferParams const & bparams) const
2111 return expandParagraphLabel(layout, bparams, false);
2115 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2116 BufferParams const & bparams, bool process_appendix) const
2118 DocumentClass const & tclass = bparams.documentClass();
2119 string const & lang = getParLanguage(bparams)->code();
2120 bool const in_appendix = process_appendix && d->params_.appendix();
2121 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2123 if (fmt.empty() && !layout.counter.empty())
2124 return tclass.counters().theCounter(layout.counter, lang);
2126 // handle 'inherited level parts' in 'fmt',
2127 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2128 size_t const i = fmt.find('@', 0);
2129 if (i != docstring::npos) {
2130 size_t const j = fmt.find('@', i + 1);
2131 if (j != docstring::npos) {
2132 docstring parent(fmt, i + 1, j - i - 1);
2133 docstring label = from_ascii("??");
2134 if (tclass.hasLayout(parent))
2135 label = expandParagraphLabel(tclass[parent], bparams,
2137 fmt = docstring(fmt, 0, i) + label
2138 + docstring(fmt, j + 1, docstring::npos);
2142 return tclass.counters().counterLabel(fmt, lang);
2146 void Paragraph::applyLayout(Layout const & new_layout)
2148 d->layout_ = &new_layout;
2149 LyXAlignment const oldAlign = d->params_.align();
2151 if (!(oldAlign & d->layout_->alignpossible)) {
2152 frontend::Alert::warning(_("Alignment not permitted"),
2153 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2154 d->params_.align(LYX_ALIGN_LAYOUT);
2159 pos_type Paragraph::beginOfBody() const
2161 return d->begin_of_body_;
2165 void Paragraph::setBeginOfBody()
2167 if (d->layout_->labeltype != LABEL_MANUAL) {
2168 d->begin_of_body_ = 0;
2172 // Unroll the first two cycles of the loop
2173 // and remember the previous character to
2174 // remove unnecessary getChar() calls
2176 pos_type end = size();
2177 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2180 char_type previous_char = d->text_[i];
2181 if (!(isNewline(i) || isEnvSeparator(i))) {
2183 while (i < end && previous_char != ' ') {
2184 char_type temp = d->text_[i];
2185 if (isNewline(i) || isEnvSeparator(i))
2188 previous_char = temp;
2194 d->begin_of_body_ = i;
2198 bool Paragraph::allowParagraphCustomization() const
2200 return inInset().allowParagraphCustomization();
2204 bool Paragraph::usePlainLayout() const
2206 return inInset().usePlainLayout();
2210 bool Paragraph::isPassThru() const
2212 return inInset().isPassThru() || d->layout_->pass_thru;
2217 // paragraphs inside floats need different alignment tags to avoid
2220 bool noTrivlistCentering(InsetCode code)
2222 return code == FLOAT_CODE
2223 || code == WRAP_CODE
2224 || code == CELL_CODE;
2228 string correction(string const & orig)
2230 if (orig == "flushleft")
2231 return "raggedright";
2232 if (orig == "flushright")
2233 return "raggedleft";
2234 if (orig == "center")
2240 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2241 InsetCode code, bool const lastpar, int & col)
2243 string macro = suffix + "{";
2244 if (noTrivlistCentering(code)) {
2246 // the last paragraph in non-trivlist-aligned
2247 // context is special (to avoid unwanted whitespace)
2248 if (suffix == "\\begin") {
2249 macro = "\\" + correction(env) + "{}";
2250 os << from_ascii(macro);
2251 col += macro.size();
2256 macro += correction(env);
2260 if (suffix == "\\par\\end") {
2264 os << from_ascii(macro);
2265 col += macro.size();
2266 if (suffix == "\\begin") {
2276 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2277 otexstream & os, OutputParams const & runparams) const
2282 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2283 (layout_->toggle_indent != ITOGGLE_NEVER) :
2284 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2286 if (canindent && params_.noindent() && !layout_->pass_thru) {
2287 os << "\\noindent ";
2291 LyXAlignment const curAlign = params_.align();
2293 if (curAlign == layout_->align)
2297 case LYX_ALIGN_NONE:
2298 case LYX_ALIGN_BLOCK:
2299 case LYX_ALIGN_LAYOUT:
2300 case LYX_ALIGN_SPECIAL:
2301 case LYX_ALIGN_DECIMAL:
2303 case LYX_ALIGN_LEFT:
2304 case LYX_ALIGN_RIGHT:
2305 case LYX_ALIGN_CENTER:
2306 if (runparams.moving_arg) {
2313 string const begin_tag = "\\begin";
2314 InsetCode code = ownerCode();
2315 bool const lastpar = runparams.isLastPar;
2318 case LYX_ALIGN_NONE:
2319 case LYX_ALIGN_BLOCK:
2320 case LYX_ALIGN_LAYOUT:
2321 case LYX_ALIGN_SPECIAL:
2322 case LYX_ALIGN_DECIMAL:
2324 case LYX_ALIGN_LEFT: {
2325 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2326 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2328 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2330 } case LYX_ALIGN_RIGHT: {
2331 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2332 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2334 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2336 } case LYX_ALIGN_CENTER: {
2337 corrected_env(os, begin_tag, "center", code, lastpar, column);
2346 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2347 otexstream & os, OutputParams const & runparams) const
2349 LyXAlignment const curAlign = params_.align();
2351 if (curAlign == layout_->align)
2355 case LYX_ALIGN_NONE:
2356 case LYX_ALIGN_BLOCK:
2357 case LYX_ALIGN_LAYOUT:
2358 case LYX_ALIGN_SPECIAL:
2359 case LYX_ALIGN_DECIMAL:
2361 case LYX_ALIGN_LEFT:
2362 case LYX_ALIGN_RIGHT:
2363 case LYX_ALIGN_CENTER:
2364 if (runparams.moving_arg)
2369 bool output = false;
2371 string const end_tag = "\\par\\end";
2372 InsetCode code = ownerCode();
2373 bool const lastpar = runparams.isLastPar;
2376 case LYX_ALIGN_NONE:
2377 case LYX_ALIGN_BLOCK:
2378 case LYX_ALIGN_LAYOUT:
2379 case LYX_ALIGN_SPECIAL:
2380 case LYX_ALIGN_DECIMAL:
2382 case LYX_ALIGN_LEFT: {
2383 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2384 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2386 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2388 } case LYX_ALIGN_RIGHT: {
2389 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2390 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2392 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2394 } case LYX_ALIGN_CENTER: {
2395 corrected_env(os, end_tag, "center", code, lastpar, col);
2400 return output || lastpar;
2404 // This one spits out the text of the paragraph
2405 void Paragraph::latex(BufferParams const & bparams,
2406 Font const & outerfont,
2408 OutputParams const & runparams,
2409 int start_pos, int end_pos, bool force) const
2411 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2413 // FIXME This check should not be needed. Perhaps issue an
2414 // error if it triggers.
2415 Layout const & style = inInset().forcePlainLayout() ?
2416 bparams.documentClass().plainLayout() : *d->layout_;
2418 if (!force && style.inpreamble)
2421 bool const allowcust = allowParagraphCustomization();
2423 // Current base font for all inherited font changes, without any
2424 // change caused by an individual character, except for the language:
2425 // It is set to the language of the first character.
2426 // As long as we are in the label, this font is the base font of the
2427 // label. Before the first body character it is set to the base font
2431 // Maybe we have to create a optional argument.
2432 pos_type body_pos = beginOfBody();
2433 unsigned int column = 0;
2436 // the optional argument is kept in curly brackets in
2437 // case it contains a ']'
2438 // This is not strictly needed, but if this is changed it
2439 // would be a file format change, and tex2lyx would need
2440 // to be adjusted, since it unconditionally removes the
2441 // braces when it parses \item.
2444 basefont = getLabelFont(bparams, outerfont);
2446 basefont = getLayoutFont(bparams, outerfont);
2449 // Which font is currently active?
2450 Font running_font(basefont);
2451 // Do we have an open font change?
2452 bool open_font = false;
2454 Change runningChange = Change(Change::UNCHANGED);
2456 Encoding const * const prev_encoding = runparams.encoding;
2458 os.texrow().start(id(), 0);
2460 // if the paragraph is empty, the loop will not be entered at all
2462 // For InTitle commands, we have already opened a group
2463 // in output_latex::TeXOnePar.
2464 if (style.isCommand() && !style.intitle) {
2468 if (!style.leftdelim().empty()) {
2469 os << style.leftdelim();
2470 column += style.leftdelim().size();
2473 column += d->startTeXParParams(bparams, os, runparams);
2476 // Whether a \par can be issued for insets typeset inline with text.
2477 // Yes if greater than 0. This has to be static.
2478 THREAD_LOCAL_STATIC int parInline = 0;
2480 for (pos_type i = 0; i < size(); ++i) {
2481 // First char in paragraph or after label?
2482 if (i == body_pos) {
2485 bool needPar = false;
2486 column += running_font.latexWriteEndChanges(
2487 os, bparams, runparams,
2488 basefont, basefont, needPar);
2491 basefont = getLayoutFont(bparams, outerfont);
2492 running_font = basefont;
2494 column += Changes::latexMarkChange(os, bparams,
2495 runningChange, Change(Change::UNCHANGED),
2497 runningChange = Change(Change::UNCHANGED);
2502 // For InTitle commands, we have already opened a group
2503 // in output_latex::TeXOnePar.
2504 if (style.isCommand() && !style.intitle) {
2509 if (!style.leftdelim().empty()) {
2510 os << style.leftdelim();
2511 column += style.leftdelim().size();
2515 column += d->startTeXParParams(bparams, os,
2519 runparams.wasDisplayMath = runparams.inDisplayMath;
2520 runparams.inDisplayMath = false;
2521 bool deleted_display_math = false;
2522 Change const & change = runparams.inDeletedInset
2523 ? runparams.changeOfDeletedInset : lookupChange(i);
2525 // Check whether a display math inset follows
2526 if (d->text_[i] == META_INSET
2527 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2528 InsetMath const * im = getInset(i)->asInsetMath();
2529 if (im && im->asHullInset()
2530 && im->asHullInset()->outerDisplay()) {
2531 runparams.inDisplayMath = true;
2532 // runparams.inDeletedInset will be set by
2533 // latexInset later, but we need this info
2534 // before it is called. On the other hand, we
2535 // cannot set it here because it is a counter.
2536 deleted_display_math = isDeleted(i);
2538 if (bparams.output_changes && deleted_display_math
2539 && runningChange == change
2540 && change.type == Change::DELETED
2541 && !os.afterParbreak()) {
2542 // A display math in the same paragraph follows.
2543 // We have to close and then reopen \lyxdeleted,
2544 // otherwise the math will be shifted up.
2545 OutputParams rp = runparams;
2547 bool needPar = false;
2548 column += running_font.latexWriteEndChanges(
2549 os, bparams, rp, basefont,
2553 basefont = getLayoutFont(bparams, outerfont);
2554 running_font = basefont;
2555 column += Changes::latexMarkChange(os, bparams,
2556 Change(Change::INSERTED), change, rp);
2560 if (bparams.output_changes && runningChange != change) {
2562 bool needPar = false;
2563 column += running_font.latexWriteEndChanges(
2564 os, bparams, runparams,
2565 basefont, basefont, needPar);
2568 basefont = getLayoutFont(bparams, outerfont);
2569 running_font = basefont;
2570 column += Changes::latexMarkChange(os, bparams, runningChange,
2572 runningChange = change;
2575 // do not output text which is marked deleted
2576 // if change tracking output is disabled
2577 if (!bparams.output_changes && change.deleted()) {
2583 // Fully instantiated font
2584 Font const current_font = getFont(bparams, i, outerfont);
2586 Font const last_font = running_font;
2588 // Do we need to close the previous font?
2590 (current_font != running_font ||
2591 current_font.language() != running_font.language()))
2593 bool needPar = false;
2594 column += running_font.latexWriteEndChanges(
2595 os, bparams, runparams, basefont,
2596 (i == body_pos-1) ? basefont : current_font,
2598 running_font = basefont;
2602 string const running_lang = runparams.use_polyglossia ?
2603 running_font.language()->polyglossia() : running_font.language()->babel();
2604 // close babel's font environment before opening CJK.
2605 string const lang_end_command = runparams.use_polyglossia ?
2606 "\\end{$$lang}" : lyxrc.language_command_end;
2607 bool const using_begin_end = runparams.use_polyglossia ||
2608 !lang_end_command.empty();
2609 if (!running_lang.empty() &&
2610 (!using_begin_end || running_lang == openLanguageName()) &&
2611 current_font.language()->encoding()->package() == Encoding::CJK) {
2612 string end_tag = subst(lang_end_command,
2615 os << from_ascii(end_tag);
2616 column += end_tag.length();
2617 if (using_begin_end)
2621 // Switch file encoding if necessary (and allowed)
2622 if (!runparams.pass_thru && !style.pass_thru &&
2623 runparams.encoding->package() != Encoding::none &&
2624 current_font.language()->encoding()->package() != Encoding::none) {
2625 pair<bool, int> const enc_switch =
2626 switchEncoding(os.os(), bparams, runparams,
2627 *(current_font.language()->encoding()));
2628 if (enc_switch.first) {
2629 column += enc_switch.second;
2630 runparams.encoding = current_font.language()->encoding();
2634 char_type const c = d->text_[i];
2636 // A display math inset inside an ulem command will be output
2637 // as a box of width \linewidth, so we have to either disable
2638 // indentation if the inset starts a paragraph, or start a new
2639 // line to accommodate such box. This has to be done before
2640 // writing any font changing commands.
2641 if (runparams.inDisplayMath && !deleted_display_math
2642 && runparams.inulemcmd) {
2643 if (os.afterParbreak())
2649 // Do we need to change font?
2650 if ((current_font != running_font ||
2651 current_font.language() != running_font.language()) &&
2654 odocstringstream ods;
2655 column += current_font.latexWriteStartChanges(ods, bparams,
2656 runparams, basefont,
2658 // Check again for display math in ulem commands as a
2659 // font change may also occur just before a math inset.
2660 if (runparams.inDisplayMath && !deleted_display_math
2661 && runparams.inulemcmd) {
2662 if (os.afterParbreak())
2667 running_font = current_font;
2669 docstring fontchange = ods.str();
2670 // check whether the fontchange ends with a \\textcolor
2671 // modifier and the text starts with a space (bug 4473)
2672 docstring const last_modifier = rsplit(fontchange, '\\');
2673 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2674 os << fontchange << from_ascii("{}");
2675 // check if the fontchange ends with a trailing blank
2676 // (like "\small " (see bug 3382)
2677 else if (suffixIs(fontchange, ' ') && c == ' ')
2678 os << fontchange.substr(0, fontchange.size() - 1)
2679 << from_ascii("{}");
2684 // FIXME: think about end_pos implementation...
2685 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2686 // FIXME: integrate this case in latexSpecialChar
2687 // Do not print the separation of the optional argument
2688 // if style.pass_thru is false. This works because
2689 // latexSpecialChar ignores spaces if
2690 // style.pass_thru is false.
2691 if (i != body_pos - 1) {
2692 if (d->simpleTeXBlanks(runparams, os,
2693 i, column, current_font, style)) {
2694 // A surrogate pair was output. We
2695 // must not call latexSpecialChar
2696 // in this iteration, since it would output
2697 // the combining character again.
2704 OutputParams rp = runparams;
2705 rp.free_spacing = style.free_spacing;
2706 rp.local_font = ¤t_font;
2707 rp.intitle = style.intitle;
2709 // Two major modes: LaTeX or plain
2710 // Handle here those cases common to both modes
2711 // and then split to handle the two modes separately.
2712 if (c == META_INSET) {
2713 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2714 // Greyedout notes and, in general, all insets
2715 // with InsetLayout::isDisplay() == false,
2716 // are typeset inline with the text. So, we
2717 // can add a \par to the last paragraph of
2718 // such insets only if nothing else follows.
2719 bool incremented = false;
2720 Inset const * inset = getInset(i);
2721 InsetText const * textinset = inset
2722 ? inset->asInsetText()
2724 if (i + 1 == size() && textinset
2725 && !inset->getLayout().isDisplay()) {
2726 ParagraphList const & pars =
2727 textinset->text().paragraphs();
2728 pit_type const pit = pars.size() - 1;
2729 Font const lastfont =
2730 pit < 0 || pars[pit].empty()
2731 ? pars[pit].getLayoutFont(
2734 : pars[pit].getFont(bparams,
2735 pars[pit].size() - 1,
2737 if (lastfont.fontInfo().size() !=
2738 basefont.fontInfo().size()) {
2743 d->latexInset(bparams, os, rp, running_font,
2744 basefont, outerfont, open_font,
2745 runningChange, style, i, column);
2749 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2751 d->latexSpecialChar(os, bparams, rp,
2752 running_font, runningChange,
2753 style, i, end_pos, column);
2754 } catch (EncodingException & e) {
2755 if (runparams.dryrun) {
2756 os << "<" << _("LyX Warning: ")
2757 << _("uncodable character") << " '";
2761 // add location information and throw again.
2769 // Set the encoding to that returned from latexSpecialChar (see
2770 // comment for encoding member in OutputParams.h)
2771 runparams.encoding = rp.encoding;
2773 // Also carry on the info on a closed ulem command for insets
2774 // such as Note that do not produce any output, so that no
2775 // command is ever executed but its opening was recorded.
2776 runparams.inulemcmd = rp.inulemcmd;
2779 // If we have an open font definition, we have to close it
2781 // Make sure that \\par is done with the font of the last
2782 // character if this has another size as the default.
2783 // This is necessary because LaTeX (and LyX on the screen)
2784 // calculates the space between the baselines according
2785 // to this font. (Matthias)
2787 // We must not change the font for the last paragraph
2788 // of non-multipar insets, tabular cells or commands,
2789 // since this produces unwanted whitespace.
2791 Font const font = empty()
2792 ? getLayoutFont(bparams, outerfont)
2793 : getFont(bparams, size() - 1, outerfont);
2795 InsetText const * textinset = inInset().asInsetText();
2797 bool const maintext = textinset
2798 ? textinset->text().isMainText()
2801 size_t const numpars = textinset
2802 ? textinset->text().paragraphs().size()
2805 bool needPar = false;
2807 if (style.resfont.size() != font.fontInfo().size()
2808 && (!runparams.isLastPar || maintext
2809 || (numpars > 1 && d->ownerCode() != CELL_CODE
2810 && (inInset().getLayout().isDisplay()
2812 && !style.isCommand()) {
2815 #ifdef FIXED_LANGUAGE_END_DETECTION
2817 running_font.latexWriteEndChanges(os, bparams,
2818 runparams, basefont,
2819 next_->getFont(bparams, 0, outerfont),
2822 running_font.latexWriteEndChanges(os, bparams,
2823 runparams, basefont, basefont, needPar);
2826 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2827 //FIXME: there as we start another \selectlanguage with the next paragraph if
2828 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2829 running_font.latexWriteEndChanges(os, bparams, runparams,
2830 basefont, basefont, needPar);
2833 // The \par could not be inserted at the same nesting
2834 // level of the font size change, so do it now.
2835 os << "{\\" << font.latexSize() << "\\par}";
2839 column += Changes::latexMarkChange(os, bparams, runningChange,
2840 Change(Change::UNCHANGED), runparams);
2842 // Needed if there is an optional argument but no contents.
2843 if (body_pos > 0 && body_pos == size()) {
2847 if (!style.rightdelim().empty()) {
2848 os << style.rightdelim();
2849 column += style.rightdelim().size();
2852 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2853 && runparams.encoding != prev_encoding) {
2854 runparams.encoding = prev_encoding;
2855 os << setEncoding(prev_encoding->iconvName());
2858 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2862 bool Paragraph::emptyTag() const
2864 for (pos_type i = 0; i < size(); ++i) {
2865 if (Inset const * inset = getInset(i)) {
2866 InsetCode lyx_code = inset->lyxCode();
2867 // FIXME testing like that is wrong. What is
2869 if (lyx_code != TOC_CODE &&
2870 lyx_code != INCLUDE_CODE &&
2871 lyx_code != GRAPHICS_CODE &&
2872 lyx_code != ERT_CODE &&
2873 lyx_code != LISTINGS_CODE &&
2874 lyx_code != FLOAT_CODE &&
2875 lyx_code != TABULAR_CODE) {
2879 char_type c = d->text_[i];
2880 if (c != ' ' && c != '\t')
2888 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2891 for (pos_type i = 0; i < size(); ++i) {
2892 if (Inset const * inset = getInset(i)) {
2893 InsetCode lyx_code = inset->lyxCode();
2894 if (lyx_code == LABEL_CODE) {
2895 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2896 docstring const & id = il->getParam("name");
2897 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2905 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2909 for (i = 0; i < size(); ++i) {
2910 if (Inset const * inset = getInset(i)) {
2911 inset->docbook(os, runparams);
2913 char_type c = d->text_[i];
2916 os << sgml::escapeChar(c);
2923 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2927 for (i = 0; i < size(); ++i) {
2928 if (Inset const * inset = getInset(i)) {
2929 inset->xhtml(xs, runparams);
2931 char_type c = d->text_[i];
2941 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2944 pos_type size = text_.size();
2945 for (pos_type i = initial; i < size; ++i) {
2946 Font font = owner_->getFont(buf.params(), i, outerfont);
2947 if (text_[i] == META_INSET)
2949 if (i != initial && font != font_old)
2958 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2960 OutputParams const & runparams,
2961 Font const & outerfont,
2962 pos_type initial) const
2964 bool emph_flag = false;
2966 Layout const & style = *d->layout_;
2968 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2970 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2973 // parsing main loop
2974 for (pos_type i = initial; i < size(); ++i) {
2975 Font font = getFont(buf.params(), i, outerfont);
2977 // handle <emphasis> tag
2978 if (font_old.emph() != font.fontInfo().emph()) {
2979 if (font.fontInfo().emph() == FONT_ON) {
2982 } else if (i != initial) {
2983 os << "</emphasis>";
2988 if (Inset const * inset = getInset(i)) {
2989 inset->docbook(os, runparams);
2991 char_type c = d->text_[i];
2993 if (style.pass_thru)
2996 os << sgml::escapeChar(c);
2998 font_old = font.fontInfo();
3002 os << "</emphasis>";
3005 if (style.free_spacing)
3007 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
3013 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
3014 vector<html::EndFontTag> & tagsToClose,
3015 bool & flag, FontState curstate, html::FontTypes type)
3017 if (curstate == FONT_ON) {
3018 tagsToOpen.push_back(html::FontTag(type));
3021 tagsToClose.push_back(html::EndFontTag(type));
3028 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
3030 OutputParams const & runparams,
3031 Font const & outerfont,
3032 bool start_paragraph, bool close_paragraph,
3033 pos_type initial) const
3037 // track whether we have opened these tags
3038 bool emph_flag = false;
3039 bool bold_flag = false;
3040 bool noun_flag = false;
3041 bool ubar_flag = false;
3042 bool dbar_flag = false;
3043 bool sout_flag = false;
3044 bool xout_flag = false;
3045 bool wave_flag = false;
3047 bool shap_flag = false;
3049 bool faml_flag = false;
3051 bool size_flag = false;
3053 Layout const & style = *d->layout_;
3055 if (start_paragraph)
3056 xs.startDivision(allowEmpty());
3059 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3061 FontShape curr_fs = INHERIT_SHAPE;
3062 FontFamily curr_fam = INHERIT_FAMILY;
3063 FontSize curr_size = FONT_SIZE_INHERIT;
3065 string const default_family =
3066 buf.masterBuffer()->params().fonts_default_family;
3068 vector<html::FontTag> tagsToOpen;
3069 vector<html::EndFontTag> tagsToClose;
3071 // parsing main loop
3072 for (pos_type i = initial; i < size(); ++i) {
3073 // let's not show deleted material in the output
3077 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3080 FontState curstate = font.fontInfo().emph();
3081 if (font_old.emph() != curstate)
3082 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3085 curstate = font.fontInfo().noun();
3086 if (font_old.noun() != curstate)
3087 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3090 curstate = font.fontInfo().underbar();
3091 if (font_old.underbar() != curstate)
3092 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3095 curstate = font.fontInfo().strikeout();
3096 if (font_old.strikeout() != curstate)
3097 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3100 curstate = font.fontInfo().xout();
3101 if (font_old.xout() != curstate)
3102 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3105 curstate = font.fontInfo().uuline();
3106 if (font_old.uuline() != curstate)
3107 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3110 curstate = font.fontInfo().uwave();
3111 if (font_old.uwave() != curstate)
3112 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3115 // a little hackish, but allows us to reuse what we have.
3116 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3117 if (font_old.series() != font.fontInfo().series())
3118 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3121 curr_fs = font.fontInfo().shape();
3122 FontShape old_fs = font_old.shape();
3123 if (old_fs != curr_fs) {
3127 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3130 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3132 case SMALLCAPS_SHAPE:
3133 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3139 // the other tags are for internal use
3147 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3151 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3154 case SMALLCAPS_SHAPE:
3155 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3162 // the other tags are for internal use
3169 curr_fam = font.fontInfo().family();
3170 FontFamily old_fam = font_old.family();
3171 if (old_fam != curr_fam) {
3175 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3178 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3180 case TYPEWRITER_FAMILY:
3181 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3183 case INHERIT_FAMILY:
3186 // the other tags are for internal use
3194 // we will treat a "default" font family as roman, since we have
3195 // no other idea what to do.
3196 if (default_family != "rmdefault" && default_family != "default") {
3197 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3202 if (default_family != "sfdefault") {
3203 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3207 case TYPEWRITER_FAMILY:
3208 if (default_family != "ttdefault") {
3209 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3213 case INHERIT_FAMILY:
3216 // the other tags are for internal use
3223 curr_size = font.fontInfo().size();
3224 FontSize old_size = font_old.size();
3225 if (old_size != curr_size) {
3228 case FONT_SIZE_TINY:
3229 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3231 case FONT_SIZE_SCRIPT:
3232 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3234 case FONT_SIZE_FOOTNOTE:
3235 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3237 case FONT_SIZE_SMALL:
3238 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3240 case FONT_SIZE_LARGE:
3241 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3243 case FONT_SIZE_LARGER:
3244 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3246 case FONT_SIZE_LARGEST:
3247 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3249 case FONT_SIZE_HUGE:
3250 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3252 case FONT_SIZE_HUGER:
3253 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3255 case FONT_SIZE_INCREASE:
3256 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3258 case FONT_SIZE_DECREASE:
3259 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3261 case FONT_SIZE_INHERIT:
3262 case FONT_SIZE_NORMAL:
3265 // the other tags are for internal use
3271 switch (curr_size) {
3272 case FONT_SIZE_TINY:
3273 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3276 case FONT_SIZE_SCRIPT:
3277 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3280 case FONT_SIZE_FOOTNOTE:
3281 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3284 case FONT_SIZE_SMALL:
3285 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3288 case FONT_SIZE_LARGE:
3289 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3292 case FONT_SIZE_LARGER:
3293 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3296 case FONT_SIZE_LARGEST:
3297 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3300 case FONT_SIZE_HUGE:
3301 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3304 case FONT_SIZE_HUGER:
3305 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3308 case FONT_SIZE_INCREASE:
3309 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3312 case FONT_SIZE_DECREASE:
3313 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3316 case FONT_SIZE_NORMAL:
3317 case FONT_SIZE_INHERIT:
3320 // the other tags are for internal use
3327 // Other such tags? What about the other text ranges?
3329 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3330 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3331 for (; cit != cen; ++cit)
3334 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3335 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3336 for (; sit != sen; ++sit)
3339 tagsToClose.clear();
3342 Inset const * inset = getInset(i);
3344 if (!runparams.for_toc || inset->isInToc()) {
3345 OutputParams np = runparams;
3346 np.local_font = &font;
3347 // If the paragraph has size 1, then we are in the "special
3348 // case" where we do not output the containing paragraph info
3349 if (!inset->getLayout().htmlisblock() && size() != 1)
3350 np.html_in_par = true;
3351 retval += inset->xhtml(xs, np);
3354 char_type c = getUChar(buf.masterBuffer()->params(), i);
3357 font_old = font.fontInfo();
3361 // I'm worried about what happens if a branch, say, is itself
3362 // wrapped in some font stuff. I think that will not work.
3364 if (close_paragraph)
3371 bool Paragraph::isHfill(pos_type pos) const
3373 Inset const * inset = getInset(pos);
3374 return inset && inset->isHfill();
3378 bool Paragraph::isNewline(pos_type pos) const
3380 // U+2028 LINE SEPARATOR
3381 // U+2029 PARAGRAPH SEPARATOR
3382 char_type const c = d->text_[pos];
3383 if (c == 0x2028 || c == 0x2029)
3385 Inset const * inset = getInset(pos);
3386 return inset && inset->lyxCode() == NEWLINE_CODE;
3390 bool Paragraph::isEnvSeparator(pos_type pos) const
3392 Inset const * inset = getInset(pos);
3393 return inset && inset->lyxCode() == SEPARATOR_CODE;
3397 bool Paragraph::isLineSeparator(pos_type pos) const
3399 char_type const c = d->text_[pos];
3400 if (isLineSeparatorChar(c))
3402 Inset const * inset = getInset(pos);
3403 return inset && inset->isLineSeparator();
3407 bool Paragraph::isWordSeparator(pos_type pos) const
3411 if (Inset const * inset = getInset(pos))
3412 return !inset->isLetter();
3413 // if we have a hard hyphen (no en- or emdash) or apostrophe
3414 // we pass this to the spell checker
3415 // FIXME: this method is subject to change, visit
3416 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3417 // to get an impression how complex this is.
3418 if (isHardHyphenOrApostrophe(pos))
3420 char_type const c = d->text_[pos];
3421 // We want to pass the escape chars to the spellchecker
3422 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3423 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3427 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3429 pos_type const psize = size();
3432 char_type const c = d->text_[pos];
3433 if (c != '-' && c != '\'')
3435 int nextpos = pos + 1;
3436 int prevpos = pos > 0 ? pos - 1 : 0;
3437 if ((nextpos == psize || isSpace(nextpos))
3438 && (pos == 0 || isSpace(prevpos)))
3444 bool Paragraph::needsCProtection() const
3446 // first check the layout of the paragraph
3447 if (layout().needcprotect) {
3448 // Environments need cprotection regardless the content
3449 if (layout().latextype == LATEX_ENVIRONMENT)
3452 // Commands need cprotection if they contain specific chars
3453 int const nchars_escape = 9;
3454 static char_type const chars_escape[nchars_escape] = {
3455 '&', '_', '$', '%', '#', '^', '{', '}', '\\'};
3457 docstring const pars = asString();
3458 for (int k = 0; k < nchars_escape; k++) {
3459 if (contains(pars, chars_escape[k]))
3464 // now check whether we have insets that need cprotection
3465 pos_type size = d->text_.size();
3466 for (pos_type i = 0; i < size; ++i)
3467 if (isInset(i) && getInset(i)->needsCProtection())
3474 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3476 return d->speller_state_.getRange(pos);
3480 bool Paragraph::isChar(pos_type pos) const
3482 if (Inset const * inset = getInset(pos))
3483 return inset->isChar();
3484 char_type const c = d->text_[pos];
3485 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3489 bool Paragraph::isSpace(pos_type pos) const
3491 if (Inset const * inset = getInset(pos))
3492 return inset->isSpace();
3493 char_type const c = d->text_[pos];
3494 return lyx::isSpace(c);
3499 Paragraph::getParLanguage(BufferParams const & bparams) const
3502 return getFirstFontSettings(bparams).language();
3503 // FIXME: we should check the prev par as well (Lgb)
3504 return bparams.language;
3508 bool Paragraph::isRTL(BufferParams const & bparams) const
3510 return getParLanguage(bparams)->rightToLeft()
3511 && !inInset().getLayout().forceLTR();
3515 void Paragraph::changeLanguage(BufferParams const & bparams,
3516 Language const * from, Language const * to)
3518 // change language including dummy font change at the end
3519 for (pos_type i = 0; i <= size(); ++i) {
3520 Font font = getFontSettings(bparams, i);
3521 if (font.language() == from) {
3522 font.setLanguage(to);
3524 d->requestSpellCheck(i);
3530 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3532 Language const * doc_language = bparams.language;
3533 for (auto const & f : d->fontlist_)
3534 if (f.font().language() != ignore_language &&
3535 f.font().language() != latex_language &&
3536 f.font().language() != doc_language)
3542 void Paragraph::getLanguages(std::set<Language const *> & langs) const
3544 for (auto const & f : d->fontlist_) {
3545 Language const * lang = f.font().language();
3546 if (lang != ignore_language &&
3547 lang != latex_language)
3553 docstring Paragraph::asString(int options) const
3555 return asString(0, size(), options);
3559 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3561 odocstringstream os;
3564 && options & AS_STR_LABEL
3565 && !d->params_.labelString().empty())
3566 os << d->params_.labelString() << ' ';
3568 for (pos_type i = beg; i < end; ++i) {
3569 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3571 char_type const c = d->text_[i];
3572 if (isPrintable(c) || c == '\t'
3573 || (c == '\n' && (options & AS_STR_NEWLINES)))
3575 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3576 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3577 LASSERT(runparams != 0, return docstring());
3578 getInset(i)->plaintext(os, *runparams);
3580 getInset(i)->toString(os);
3589 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3590 bool const shorten, bool const label) const
3592 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3593 if (label && !labelString().empty())
3594 os += labelString() + ' ';
3595 if (!layout().isTocCaption())
3597 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3600 char_type const c = d->text_[i];
3603 else if (c == META_INSET)
3604 getInset(i)->forOutliner(os, tmplen, false);
3607 Text::shortenForOutliner(os, maxlen);
3611 void Paragraph::setInsetOwner(Inset const * inset)
3613 d->inset_owner_ = inset;
3617 int Paragraph::id() const
3623 void Paragraph::setId(int id)
3629 Layout const & Paragraph::layout() const
3635 void Paragraph::setLayout(Layout const & layout)
3637 d->layout_ = &layout;
3641 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3643 setLayout(tc.defaultLayout());
3647 void Paragraph::setPlainLayout(DocumentClass const & tc)
3649 setLayout(tc.plainLayout());
3653 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3655 if (usePlainLayout())
3656 setPlainLayout(tclass);
3658 setDefaultLayout(tclass);
3662 Inset const & Paragraph::inInset() const
3664 LBUFERR(d->inset_owner_);
3665 return *d->inset_owner_;
3669 ParagraphParameters & Paragraph::params()
3675 ParagraphParameters const & Paragraph::params() const
3681 bool Paragraph::isFreeSpacing() const
3683 if (d->layout_->free_spacing)
3685 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3689 bool Paragraph::allowEmpty() const
3691 if (d->layout_->keepempty)
3693 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3697 bool Paragraph::brokenBiblio() const
3699 // there is a problem if there is no bibitem at position 0 or
3700 // if there is another bibitem in the paragraph.
3701 return d->layout_->labeltype == LABEL_BIBLIO
3702 && (d->insetlist_.find(BIBITEM_CODE) != 0
3703 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3707 int Paragraph::fixBiblio(Buffer const & buffer)
3709 // FIXME: What about the case where paragraph is not BIBLIO
3710 // but there is an InsetBibitem?
3711 // FIXME: when there was already an inset at 0, the return value is 1,
3712 // which does not tell whether another inset has been remove; the
3713 // cursor cannot be correctly updated.
3715 if (d->layout_->labeltype != LABEL_BIBLIO)
3718 bool const track_changes = buffer.params().track_changes;
3719 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3720 bool const hasbibitem0 = bibitem_pos == 0;
3723 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3724 // There was an InsetBibitem at pos 0, and no other one => OK
3725 if (bibitem_pos == -1)
3727 // there is a bibitem at the 0 position, but since
3728 // there is a second one, we copy the second on the
3729 // first. We're assuming there are at most two of
3730 // these, which there should be.
3731 // FIXME: why does it make sense to do that rather
3732 // than keep the first? (JMarc)
3733 Inset * inset = releaseInset(bibitem_pos);
3734 d->insetlist_.begin()->inset = inset;
3735 return -bibitem_pos;
3738 // We need to create an inset at the beginning
3740 if (bibitem_pos > 0) {
3741 // there was one somewhere in the paragraph, let's move it
3742 inset = d->insetlist_.release(bibitem_pos);
3743 eraseChar(bibitem_pos, track_changes);
3746 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3747 InsetCommandParams(BIBITEM_CODE));
3749 Font font(inherit_font, buffer.params().language);
3750 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3751 : Change::UNCHANGED));
3757 void Paragraph::checkAuthors(AuthorList const & authorList)
3759 d->changes_.checkAuthors(authorList);
3763 bool Paragraph::isChanged(pos_type pos) const
3765 return lookupChange(pos).changed();
3769 bool Paragraph::isInserted(pos_type pos) const
3771 return lookupChange(pos).inserted();
3775 bool Paragraph::isDeleted(pos_type pos) const
3777 return lookupChange(pos).deleted();
3781 InsetList const & Paragraph::insetList() const
3783 return d->insetlist_;
3787 void Paragraph::setInsetBuffers(Buffer & b)
3789 d->insetlist_.setBuffer(b);
3793 void Paragraph::resetBuffer()
3795 d->insetlist_.resetBuffer();
3799 Inset * Paragraph::releaseInset(pos_type pos)
3801 Inset * inset = d->insetlist_.release(pos);
3802 /// does not honour change tracking!
3803 eraseChar(pos, false);
3808 Inset * Paragraph::getInset(pos_type pos)
3810 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3811 ? d->insetlist_.get(pos) : 0;
3815 Inset const * Paragraph::getInset(pos_type pos) const
3817 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3818 ? d->insetlist_.get(pos) : 0;
3822 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3823 pos_type & right, TextCase action)
3825 // process sequences of modified characters; in change
3826 // tracking mode, this approach results in much better
3827 // usability than changing case on a char-by-char basis
3828 // We also need to track the current font, since font
3829 // changes within sequences can occur.
3830 vector<pair<char_type, Font> > changes;
3832 bool const trackChanges = bparams.track_changes;
3834 bool capitalize = true;
3836 for (; pos < right; ++pos) {
3837 char_type oldChar = d->text_[pos];
3838 char_type newChar = oldChar;
3840 // ignore insets and don't play with deleted text!
3841 if (oldChar != META_INSET && !isDeleted(pos)) {
3843 case text_lowercase:
3844 newChar = lowercase(oldChar);
3846 case text_capitalization:
3848 newChar = uppercase(oldChar);
3852 case text_uppercase:
3853 newChar = uppercase(oldChar);
3858 if (isWordSeparator(pos) || isDeleted(pos)) {
3859 // permit capitalization again
3863 if (oldChar != newChar) {
3864 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3865 if (pos != right - 1)
3867 // step behind the changing area
3871 int erasePos = pos - changes.size();
3872 for (size_t i = 0; i < changes.size(); i++) {
3873 insertChar(pos, changes[i].first,
3876 if (!eraseChar(erasePos, trackChanges)) {
3879 ++right; // expand selection
3887 int Paragraph::find(docstring const & str, bool cs, bool mw,
3888 pos_type start_pos, bool del) const
3890 pos_type pos = start_pos;
3891 int const strsize = str.length();
3893 pos_type const parsize = d->text_.size();
3894 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3895 // Ignore "invisible" letters such as ligature breaks
3896 // and hyphenation chars while searching
3897 while (pos < parsize - 1 && isInset(pos)) {
3898 odocstringstream os;
3899 getInset(pos)->toString(os);
3900 if (!getInset(pos)->isLetter() || !os.str().empty())
3904 if (cs && str[i] != d->text_[pos])
3906 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3908 if (!del && isDeleted(pos))
3915 // if necessary, check whether string matches word
3917 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3920 && !isWordSeparator(pos))
3924 return pos - start_pos;
3928 char_type Paragraph::getChar(pos_type pos) const
3930 return d->text_[pos];
3934 pos_type Paragraph::size() const
3936 return d->text_.size();
3940 bool Paragraph::empty() const
3942 return d->text_.empty();
3946 bool Paragraph::isInset(pos_type pos) const
3948 return d->text_[pos] == META_INSET;
3952 bool Paragraph::isSeparator(pos_type pos) const
3954 //FIXME: Are we sure this can be the only separator?
3955 return d->text_[pos] == ' ';
3959 void Paragraph::deregisterWords()
3961 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3962 Private::LangWordsMap::const_iterator ite = d->words_.end();
3963 for (; itl != ite; ++itl) {
3964 WordList & wl = theWordList(itl->first);
3965 Private::Words::const_iterator it = (itl->second).begin();
3966 Private::Words::const_iterator et = (itl->second).end();
3967 for (; it != et; ++it)
3974 void Paragraph::locateWord(pos_type & from, pos_type & to,
3975 word_location const loc) const
3978 case WHOLE_WORD_STRICT:
3979 if (from == 0 || from == size()
3980 || isWordSeparator(from)
3981 || isWordSeparator(from - 1)) {
3988 // If we are already at the beginning of a word, do nothing
3989 if (!from || isWordSeparator(from - 1))
3994 // always move the cursor to the beginning of previous word
3995 while (from && !isWordSeparator(from - 1))
3999 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
4002 // no need to move the 'from' cursor
4006 while (to < size() && !isWordSeparator(to))
4011 void Paragraph::collectWords()
4013 for (pos_type pos = 0; pos < size(); ++pos) {
4014 if (isWordSeparator(pos))
4016 pos_type from = pos;
4017 locateWord(from, pos, WHOLE_WORD);
4018 // Work around MSVC warning: The statement
4019 // if (pos < from + lyxrc.completion_minlength)
4020 // triggers a signed vs. unsigned warning.
4021 // I don't know why this happens, it could be a MSVC bug, or
4022 // related to LLP64 (windows) vs. LP64 (unix) programming
4023 // model, or the C++ standard might be ambigous in the section
4024 // defining the "usual arithmetic conversions". However, using
4025 // a temporary variable is safe and works on all compilers.
4026 pos_type const endpos = from + lyxrc.completion_minlength;
4029 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
4030 if (cit == d->fontlist_.end())
4032 Language const * lang = cit->font().language();
4033 docstring const word = asString(from, pos, AS_STR_NONE);
4034 d->words_[lang->lang()].insert(word);
4039 void Paragraph::registerWords()
4041 Private::LangWordsMap::const_iterator itl = d->words_.begin();
4042 Private::LangWordsMap::const_iterator ite = d->words_.end();
4043 for (; itl != ite; ++itl) {
4044 WordList & wl = theWordList(itl->first);
4045 Private::Words::const_iterator it = (itl->second).begin();
4046 Private::Words::const_iterator et = (itl->second).end();
4047 for (; it != et; ++it)
4053 void Paragraph::updateWords()
4061 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
4063 SkipPositionsIterator begin = skips.begin();
4064 SkipPositions::iterator end = skips.end();
4065 if (pos > 0 && begin < end) {
4067 if (end->last == pos - 1) {
4072 skips.insert(end, FontSpan(pos, pos));
4076 Language * Paragraph::Private::locateSpellRange(
4077 pos_type & from, pos_type & to,
4078 SkipPositions & skips) const
4080 // skip leading white space
4081 while (from < to && owner_->isWordSeparator(from))
4083 // don't check empty range
4086 // get current language
4087 Language * lang = getSpellLanguage(from);
4088 pos_type last = from;
4089 bool samelang = true;
4090 bool sameinset = true;
4091 while (last < to && samelang && sameinset) {
4092 // hop to end of word
4093 while (last < to && !owner_->isWordSeparator(last)) {
4094 if (owner_->getInset(last)) {
4095 appendSkipPosition(skips, last);
4096 } else if (owner_->isDeleted(last)) {
4097 appendSkipPosition(skips, last);
4101 // hop to next word while checking for insets
4102 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4103 if (Inset const * inset = owner_->getInset(last))
4104 sameinset = inset->isChar() && inset->isLetter();
4105 if (sameinset && owner_->isDeleted(last)) {
4106 appendSkipPosition(skips, last);
4111 if (sameinset && last < to) {
4112 // now check for language change
4113 samelang = lang == getSpellLanguage(last);
4116 // if language change detected backstep is needed
4124 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4127 const_cast<Language *>(owner_->getFontSettings(
4128 inset_owner_->buffer().params(), from).language());
4129 if (lang == inset_owner_->buffer().params().language
4130 && !lyxrc.spellchecker_alt_lang.empty()) {
4132 string const lang_variety =
4133 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4134 lang->setCode(lang_code);
4135 lang->setVariety(lang_variety);
4141 void Paragraph::requestSpellCheck(pos_type pos)
4143 d->requestSpellCheck(pos);
4147 bool Paragraph::needsSpellCheck() const
4149 SpellChecker::ChangeNumber speller_change_number = 0;
4150 if (theSpellChecker())
4151 speller_change_number = theSpellChecker()->changeNumber();
4152 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4153 d->speller_state_.needsCompleteRefresh(speller_change_number);
4155 return d->needsSpellCheck();
4159 bool Paragraph::Private::ignoreWord(docstring const & word) const
4161 // Ignore words with digits
4162 // FIXME: make this customizable
4163 // (note that some checkers ignore words with digits by default)
4164 docstring::const_iterator cit = word.begin();
4165 docstring::const_iterator const end = word.end();
4166 for (; cit != end; ++cit) {
4167 if (isNumber((*cit)))
4174 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4175 WordLangTuple & wl, docstring_list & suggestions,
4176 bool do_suggestion, bool check_learned) const
4178 SpellChecker::Result result = SpellChecker::WORD_OK;
4179 SpellChecker * speller = theSpellChecker();
4183 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4186 locateWord(from, to, WHOLE_WORD);
4187 if (from == to || from >= size())
4190 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4191 Language * lang = d->getSpellLanguage(from);
4193 wl = WordLangTuple(word, lang);
4198 if (needsSpellCheck() || check_learned) {
4200 if (!d->ignoreWord(word)) {
4201 bool const trailing_dot = to < size() && d->text_[to] == '.';
4202 result = speller->check(wl);
4203 if (SpellChecker::misspelled(result) && trailing_dot) {
4204 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4205 result = speller->check(wl);
4206 if (!SpellChecker::misspelled(result)) {
4207 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4209 from << ".." << to << "]");
4211 // spell check with dot appended failed too
4212 // restore original word/lang value
4213 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4214 wl = WordLangTuple(word, lang);
4218 if (!SpellChecker::misspelled(result)) {
4219 // area up to the begin of the next word is not misspelled
4220 while (end < size() && isWordSeparator(end))
4223 d->setMisspelled(from, end, result);
4225 result = d->speller_state_.getState(from);
4229 suggestions.clear();
4231 if (SpellChecker::misspelled(result)) {
4232 LYXERR(Debug::GUI, "misspelled word: \"" <<
4234 from << ".." << to << "]");
4236 speller->suggest(wl, suggestions);
4242 void Paragraph::anonymize()
4244 // This is a very crude anonymization for now
4245 for (char_type & c : d->text_)
4246 if (isLetterChar(c) || isNumber(c))
4251 void Paragraph::Private::markMisspelledWords(
4252 pos_type const & first, pos_type const & last,
4253 SpellChecker::Result result,
4254 docstring const & word,
4255 SkipPositions const & skips)
4257 if (!SpellChecker::misspelled(result)) {
4258 setMisspelled(first, last, SpellChecker::WORD_OK);
4262 SpellChecker * speller = theSpellChecker();
4263 // locate and enumerate the error positions
4264 int nerrors = speller->numMisspelledWords();
4266 SkipPositionsIterator it = skips.begin();
4267 SkipPositionsIterator et = skips.end();
4268 for (int index = 0; index < nerrors; ++index) {
4271 speller->misspelledWord(index, wstart, wlen);
4272 /// should not happen if speller supports range checks
4273 if (!wlen) continue;
4274 docstring const misspelled = word.substr(wstart, wlen);
4275 wstart += first + numskipped;
4276 if (snext < wstart) {
4277 /// mark the range of correct spelling
4278 numskipped += countSkips(it, et, wstart);
4279 setMisspelled(snext,
4280 wstart - 1, SpellChecker::WORD_OK);
4282 snext = wstart + wlen;
4283 numskipped += countSkips(it, et, snext);
4284 /// mark the range of misspelling
4285 setMisspelled(wstart, snext, result);
4286 LYXERR(Debug::GUI, "misspelled word: \"" <<
4287 misspelled << "\" [" <<
4288 wstart << ".." << (snext-1) << "]");
4291 if (snext <= last) {
4292 /// mark the range of correct spelling at end
4293 setMisspelled(snext, last, SpellChecker::WORD_OK);
4298 void Paragraph::spellCheck() const
4300 SpellChecker * speller = theSpellChecker();
4301 if (!speller || empty() ||!needsSpellCheck())
4305 d->rangeOfSpellCheck(start, endpos);
4306 if (speller->canCheckParagraph()) {
4307 // loop until we leave the range
4308 for (pos_type first = start; first < endpos; ) {
4309 pos_type last = endpos;
4310 Private::SkipPositions skips;
4311 Language * lang = d->locateSpellRange(first, last, skips);
4312 if (first >= endpos)
4314 // start the spell checker on the unit of meaning
4315 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4316 WordLangTuple wl = WordLangTuple(word, lang);
4317 SpellChecker::Result result = word.size() ?
4318 speller->check(wl) : SpellChecker::WORD_OK;
4319 d->markMisspelledWords(first, last, result, word, skips);
4323 static docstring_list suggestions;
4324 pos_type to = endpos;
4325 while (start < endpos) {
4327 spellCheck(start, to, wl, suggestions, false);
4331 d->readySpellCheck();
4335 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4337 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4338 if (result || pos <= 0 || pos > size())
4340 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4341 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4346 string Paragraph::magicLabel() const
4349 ss << "magicparlabel-" << id();