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(otexstream & os, docstring const & ltx,
334 Change const &, Encoding const &, pos_type & i);
336 /// This could go to ParagraphParameters if we want to.
337 int startTeXParParams(BufferParams const &, otexstream &,
338 OutputParams const &) const;
340 /// This could go to ParagraphParameters if we want to.
341 bool endTeXParParams(BufferParams const &, otexstream &,
342 OutputParams const &) const;
345 void latexInset(BufferParams const &,
350 Font const & outerfont,
352 Change & running_change,
353 Layout const & style,
355 unsigned int & column);
358 void latexSpecialChar(
360 BufferParams const & bparams,
361 OutputParams const & runparams,
362 Font const & running_font,
363 Change const & running_change,
364 Layout const & style,
367 unsigned int & column);
374 unsigned int & column);
380 unsigned int & column);
386 unsigned int & column);
389 void validate(LaTeXFeatures & features) const;
391 /// Checks if the paragraph contains only text and no inset or font change.
392 bool onlyText(Buffer const & buf, Font const & outerfont,
393 pos_type initial) const;
395 /// a vector of speller skip positions
396 typedef vector<FontSpan> SkipPositions;
397 typedef SkipPositions::const_iterator SkipPositionsIterator;
399 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
401 Language * getSpellLanguage(pos_type const from) const;
403 Language * locateSpellRange(pos_type & from, pos_type & to,
404 SkipPositions & skips) const;
406 bool hasSpellerChange() const
408 SpellChecker::ChangeNumber speller_change_number = 0;
409 if (theSpellChecker())
410 speller_change_number = theSpellChecker()->changeNumber();
411 return speller_change_number > speller_state_.currentChangeNumber();
414 bool ignoreWord(docstring const & word) const ;
416 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
418 pos_type textsize = owner_->size();
419 // check for sane arguments
420 if (to <= from || from >= textsize)
422 FontSpan fp = FontSpan(from, to - 1);
423 speller_state_.setRange(fp, state);
426 void requestSpellCheck(pos_type pos)
429 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
431 speller_state_.needsRefresh(pos);
434 void readySpellCheck()
436 speller_state_.needsRefresh(-1);
439 bool needsSpellCheck() const
441 return speller_state_.needsRefresh();
444 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
446 speller_state_.refreshRange(first, last);
448 last = owner_->size();
451 pos_type endpos = last;
452 owner_->locateWord(first, endpos, WHOLE_WORD);
455 owner_->locateWord(last, endpos, WHOLE_WORD);
460 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
464 while (it != et && it->first < start) {
465 int skip = it->last - it->first + 1;
473 void markMisspelledWords(pos_type const & first, pos_type const & last,
474 SpellChecker::Result result,
475 docstring const & word,
476 SkipPositions const & skips);
478 InsetCode ownerCode() const
480 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
483 /// Which Paragraph owns us?
487 Inset const * inset_owner_;
496 ParagraphParameters params_;
498 /// for recording and looking up changes
502 InsetList insetlist_;
505 pos_type begin_of_body_;
507 typedef docstring TextContainer;
511 typedef set<docstring> Words;
512 typedef map<string, Words> LangWordsMap;
516 Layout const * layout_;
518 SpellCheckerState speller_state_;
522 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
523 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
530 int Paragraph::Private::make_id()
532 // The id is unique per session across buffers because it is used in
533 // LFUN_PARAGRAPH_GOTO to switch to a different buffer, for instance in the
536 static atomic_uint next_id(0);
541 Paragraph::Private::Private(Private const & p, Paragraph * owner)
542 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
544 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
545 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
548 requestSpellCheck(p.text_.size());
552 Paragraph::Private::Private(Private const & p, Paragraph * owner,
553 pos_type beg, pos_type end)
554 : owner_(owner), inset_owner_(p.inset_owner_), id_(make_id()),
555 params_(p.params_), changes_(p.changes_),
556 insetlist_(p.insetlist_, beg, end),
557 begin_of_body_(p.begin_of_body_), words_(p.words_),
560 if (beg >= pos_type(p.text_.size()))
562 text_ = p.text_.substr(beg, end - beg);
564 FontList::const_iterator fcit = fontlist_.begin();
565 FontList::const_iterator fend = fontlist_.end();
566 for (; fcit != fend; ++fcit) {
567 if (fcit->pos() < beg)
569 if (fcit->pos() >= end) {
570 // Add last entry in the fontlist_.
571 fontlist_.set(text_.size() - 1, fcit->font());
574 // Add a new entry in the fontlist_.
575 fontlist_.set(fcit->pos() - beg, fcit->font());
577 requestSpellCheck(p.text_.size());
581 void Paragraph::addChangesToToc(DocIterator const & cdit, Buffer const & buf,
582 bool output_active, TocBackend & backend) const
584 d->changes_.addToToc(cdit, buf, output_active, backend);
588 void Paragraph::addChangesToBuffer(Buffer const & buf) const
590 d->changes_.updateBuffer(buf);
594 bool Paragraph::isChangeUpdateRequired() const
596 return d->changes_.isUpdateRequired();
600 bool Paragraph::isDeleted(pos_type start, pos_type end) const
602 LASSERT(start >= 0 && start <= size(), return false);
603 LASSERT(end > start && end <= size() + 1, return false);
605 return d->changes_.isDeleted(start, end);
609 bool Paragraph::isChanged(pos_type start, pos_type end) const
611 LASSERT(start >= 0 && start <= size(), return false);
612 LASSERT(end > start && end <= size() + 1, return false);
614 return d->changes_.isChanged(start, end);
618 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
620 // keep the logic here in sync with the logic of eraseChars()
624 Change const & change = d->changes_.lookup(size());
625 return change.inserted() && change.currentAuthor();
629 void Paragraph::setChange(Change const & change)
631 // beware of the imaginary end-of-par character!
632 d->changes_.set(change, 0, size() + 1);
635 * Propagate the change recursively - but not in case of DELETED!
637 * Imagine that your co-author makes changes in an existing inset. He
638 * sends your document to you and you come to the conclusion that the
639 * inset should go completely. If you erase it, LyX must not delete all
640 * text within the inset. Otherwise, the change tracked insertions of
641 * your co-author get lost and there is no way to restore them later.
643 * Conclusion: An inset's content should remain untouched if you delete it
646 if (!change.deleted()) {
647 for (pos_type pos = 0; pos < size(); ++pos) {
648 if (Inset * inset = getInset(pos))
649 inset->setChange(change);
655 void Paragraph::setChange(pos_type pos, Change const & change)
657 LASSERT(pos >= 0 && pos <= size(), return);
658 d->changes_.set(change, pos);
660 // see comment in setChange(Change const &) above
661 if (!change.deleted() && pos < size())
662 if (Inset * inset = getInset(pos))
663 inset->setChange(change);
667 Change const & Paragraph::lookupChange(pos_type pos) const
669 LBUFERR(pos >= 0 && pos <= size());
670 return d->changes_.lookup(pos);
674 void Paragraph::acceptChanges(pos_type start, pos_type end)
676 LASSERT(start >= 0 && start <= size(), return);
677 LASSERT(end > start && end <= size() + 1, return);
679 for (pos_type pos = start; pos < end; ++pos) {
680 switch (lookupChange(pos).type) {
681 case Change::UNCHANGED:
682 // accept changes in nested inset
683 if (Inset * inset = getInset(pos))
684 inset->acceptChanges();
687 case Change::INSERTED:
688 d->changes_.set(Change(Change::UNCHANGED), pos);
689 // also accept changes in nested inset
690 if (Inset * inset = getInset(pos))
691 inset->acceptChanges();
694 case Change::DELETED:
695 // Suppress access to non-existent
696 // "end-of-paragraph char"
698 eraseChar(pos, false);
709 void Paragraph::rejectChanges(pos_type start, pos_type end)
711 LASSERT(start >= 0 && start <= size(), return);
712 LASSERT(end > start && end <= size() + 1, return);
714 for (pos_type pos = start; pos < end; ++pos) {
715 switch (lookupChange(pos).type) {
716 case Change::UNCHANGED:
717 // reject changes in nested inset
718 if (Inset * inset = getInset(pos))
719 inset->rejectChanges();
722 case Change::INSERTED:
723 // Suppress access to non-existent
724 // "end-of-paragraph char"
726 eraseChar(pos, false);
732 case Change::DELETED:
733 d->changes_.set(Change(Change::UNCHANGED), pos);
735 // Do NOT reject changes within a deleted inset!
736 // There may be insertions of a co-author inside of it!
744 void Paragraph::Private::insertChar(pos_type pos, char_type c,
745 Change const & change)
747 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
750 changes_.insert(change, pos);
752 // This is actually very common when parsing buffers (and
753 // maybe inserting ascii text)
754 if (pos == pos_type(text_.size())) {
755 // when appending characters, no need to update tables
757 // but we want spell checking
758 requestSpellCheck(pos);
762 text_.insert(text_.begin() + pos, c);
764 // Update the font table.
765 fontlist_.increasePosAfterPos(pos);
768 insetlist_.increasePosAfterPos(pos);
770 // Update list of misspelled positions
771 speller_state_.increasePosAfterPos(pos);
775 bool Paragraph::insertInset(pos_type pos, Inset * inset,
776 Font const & font, Change const & change)
778 LASSERT(inset, return false);
779 LASSERT(pos >= 0 && pos <= size(), return false);
781 // Paragraph::insertInset() can be used in cut/copy/paste operation where
782 // d->inset_owner_ is not set yet.
783 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
786 d->insertChar(pos, META_INSET, change);
787 LASSERT(d->text_[pos] == META_INSET, return false);
789 // Add a new entry in the insetlist_.
790 d->insetlist_.insert(inset, pos);
792 // Some insets require run of spell checker
793 requestSpellCheck(pos);
799 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
801 LASSERT(pos >= 0 && pos <= size(), return false);
803 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
806 Change change = d->changes_.lookup(pos);
808 // set the character to DELETED if
809 // a) it was previously unchanged or
810 // b) it was inserted by a co-author
812 if (!change.changed() ||
813 (change.inserted() && !change.currentAuthor())) {
814 setChange(pos, Change(Change::DELETED));
815 // request run of spell checker
816 requestSpellCheck(pos);
820 if (change.deleted())
824 // Don't physically access the imaginary end-of-paragraph character.
825 // eraseChar() can only mark it as DELETED. A physical deletion of
826 // end-of-par must be handled externally.
832 d->changes_.erase(pos);
834 // if it is an inset, delete the inset entry
835 if (d->text_[pos] == META_INSET)
836 d->insetlist_.erase(pos);
838 d->text_.erase(d->text_.begin() + pos);
840 // Update the fontlist_
841 d->fontlist_.erase(pos);
843 // Update the insetlist_
844 d->insetlist_.decreasePosAfterPos(pos);
846 // Update list of misspelled positions
847 d->speller_state_.decreasePosAfterPos(pos);
848 d->speller_state_.refreshLast(size());
854 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
856 LASSERT(start >= 0 && start <= size(), return 0);
857 LASSERT(end >= start && end <= size() + 1, return 0);
860 for (pos_type count = end - start; count; --count) {
861 if (!eraseChar(i, trackChanges))
868 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
869 char_type next, OutputParams const & runparams)
871 // Writing next here may circumvent a possible font change between
872 // c and next. Since next is only output if it forms a surrogate pair
873 // with c we can ignore this:
874 // A font change inside a surrogate pair does not make sense and is
875 // hopefully impossible to input.
876 // FIXME: change tracking
877 // Is this correct WRT change tracking?
878 Encoding const & encoding = *(runparams.encoding);
879 docstring latex1 = encoding.latexChar(next).first;
880 if (runparams.inIPA) {
881 string const tipashortcut = Encodings::TIPAShortcut(next);
882 if (!tipashortcut.empty()) {
883 latex1 = from_ascii(tipashortcut);
886 docstring const latex2 = encoding.latexChar(c).first;
887 if (docstring(1, next) == latex1) {
888 // the encoding supports the combination
889 os << latex2 << latex1;
890 return latex1.length() + latex2.length();
891 } else if (runparams.local_font &&
892 runparams.local_font->language()->lang() == "polutonikogreek") {
893 // polutonikogreek only works without the brackets
894 os << latex1 << latex2;
895 return latex1.length() + latex2.length();
897 os << latex1 << '{' << latex2 << '}';
898 return latex1.length() + latex2.length() + 2;
902 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
905 unsigned int & column,
907 Layout const & style)
909 if (style.pass_thru || runparams.pass_thru)
912 if (i + 1 < int(text_.size())) {
913 char_type next = text_[i + 1];
914 if (Encodings::isCombiningChar(next)) {
915 // This space has an accent, so we must always output it.
916 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
921 if (runparams.linelen > 0
922 && column > runparams.linelen
924 && text_[i - 1] != ' '
925 && (i + 1 < int(text_.size()))
926 // same in FreeSpacing mode
927 && !owner_->isFreeSpacing()
928 // In typewriter mode, we want to avoid
929 // ! . ? : at the end of a line
930 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
931 && (text_[i - 1] == '.'
932 || text_[i - 1] == '?'
933 || text_[i - 1] == ':'
934 || text_[i - 1] == '!'))) {
936 os.texrow().start(owner_->id(), i + 1);
938 } else if (style.free_spacing) {
947 int Paragraph::Private::writeScriptChars(otexstream & os,
948 docstring const & ltx,
949 Change const & runningChange,
950 Encoding const & encoding,
953 // FIXME: modifying i here is not very nice...
955 // We only arrive here when character text_[i] could not be translated
956 // into the current latex encoding (or its latex translation has been forced,)
957 // and it belongs to a known script.
958 // TODO: We need \textcyr and \textgreek wrappers also for characters
959 // that can be encoded in the "LaTeX encoding" but not in the
960 // current *font encoding*.
961 // (See #9681 for details and test)
962 // Parameter ltx contains the latex translation of text_[i] as specified
963 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
964 // The latex macro name "textXXX" specifies the script to which text_[i]
965 // belongs and we use it in order to check whether characters from the
966 // same script immediately follow, such that we can collect them in a
967 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
968 // for the first char but only "<spec>" for all subsequent chars.
969 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
970 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
971 string script = to_ascii(ltx.substr(1, brace1 - 1));
974 bool closing_brace = true;
975 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
976 // Correct encoding is being used, so we can avoid \textgreek.
977 // TODO: wrong test: we need to check the *font encoding*
978 // (i.e. the active language and its FontEncoding tag)
979 // instead of the LaTeX *input encoding*!
980 // See #9637 for details and test-cases.
983 closing_brace = false;
985 os << ltx.substr(pos, length);
986 int size = text_.size();
987 while (i + 1 < size) {
988 char_type const next = text_[i + 1];
989 // Stop here if next character belongs to another script
990 // or there is a change in change tracking status.
991 if (!Encodings::isKnownScriptChar(next, script) ||
992 runningChange != owner_->lookupChange(i + 1))
996 FontList::const_iterator cit = fontlist_.begin();
997 FontList::const_iterator end = fontlist_.end();
998 for (; cit != end; ++cit) {
999 if (cit->pos() >= i && !found) {
1000 prev_font = cit->font();
1003 if (cit->pos() >= i + 1)
1006 // Stop here if there is a font attribute or encoding change.
1007 if (found && cit != end && prev_font != cit->font())
1009 docstring const latex = encoding.latexChar(next).first;
1010 docstring::size_type const b1 =
1011 latex.find_first_of(from_ascii("{"));
1012 docstring::size_type const b2 =
1013 latex.find_last_of(from_ascii("}"));
1014 int const len = b2 - b1 - 1;
1015 os << latex.substr(b1 + 1, len);
1019 if (closing_brace) {
1027 void Paragraph::Private::latexInset(BufferParams const & bparams,
1029 OutputParams & runparams,
1030 Font & running_font,
1032 Font const & outerfont,
1034 Change & running_change,
1035 Layout const & style,
1037 unsigned int & column)
1039 Inset * inset = owner_->getInset(i);
1042 if (style.pass_thru) {
1043 odocstringstream ods;
1044 inset->plaintext(ods, runparams);
1049 // FIXME: move this to InsetNewline::latex
1050 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1051 // newlines are handled differently here than
1052 // the default in simpleTeXSpecialChars().
1053 if (!style.newline_allowed) {
1057 bool needPar = false;
1058 column += running_font.latexWriteEndChanges(
1059 os, bparams, runparams,
1060 basefont, basefont, needPar);
1064 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1067 basefont = owner_->getLayoutFont(bparams, outerfont);
1068 running_font = basefont;
1070 if (runparams.moving_arg)
1074 os.texrow().start(owner_->id(), i + 1);
1078 if (owner_->isDeleted(i)) {
1079 if( ++runparams.inDeletedInset == 1)
1080 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1083 if (inset->canTrackChanges()) {
1084 column += Changes::latexMarkChange(os, bparams, running_change,
1085 Change(Change::UNCHANGED), runparams);
1086 running_change = Change(Change::UNCHANGED);
1090 odocstream::pos_type const len = os.os().tellp();
1092 if (inset->forceLTR()
1093 && !runparams.use_polyglossia
1094 && running_font.isRightToLeft()
1095 // ERT is an exception, it should be output with no
1096 // decorations at all
1097 && inset->lyxCode() != ERT_CODE) {
1098 if (running_font.language()->lang() == "farsi")
1099 os << "\\beginL" << termcmd;
1105 // FIXME: Bug: we can have an empty font change here!
1106 // if there has just been a font change, we are going to close it
1107 // right now, which means stupid latex code like \textsf{}. AFAIK,
1108 // this does not harm dvi output. A minor bug, thus (JMarc)
1110 // Some insets cannot be inside a font change command.
1111 // However, even such insets *can* be placed in \L or \R
1112 // or their equivalents (for RTL language switches), so we don't
1113 // close the language in those cases.
1114 // ArabTeX, though, cannot handle this special behavior, it seems.
1115 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1116 || running_font.language()->lang() == "arabic_arabtex";
1117 if (open_font && !inset->inheritFont()) {
1118 bool needPar = false;
1119 bool closeLanguage = arabtex
1120 || basefont.isRightToLeft() == running_font.isRightToLeft();
1121 unsigned int count = running_font.latexWriteEndChanges(os,
1122 bparams, runparams, basefont, basefont,
1123 needPar, closeLanguage);
1125 // if any font properties were closed, update the running_font,
1126 // making sure, however, to leave the language as it was
1128 // FIXME: probably a better way to keep track of the old
1129 // language, than copying the entire font?
1130 Font const copy_font(running_font);
1131 basefont = owner_->getLayoutFont(bparams, outerfont);
1132 running_font = basefont;
1134 running_font.setLanguage(copy_font.language());
1135 // leave font open if language is still open
1136 open_font = (running_font.language() == basefont.language());
1138 runparams.local_font = &basefont;
1142 size_t const previous_row_count = os.texrow().rows();
1145 runparams.lastid = id_;
1146 runparams.lastpos = i;
1147 inset->latex(os, runparams);
1148 } catch (EncodingException & e) {
1149 // add location information and throw again.
1156 if (running_font.language()->lang() == "farsi")
1157 os << "\\endL" << termcmd;
1162 if (os.texrow().rows() > previous_row_count) {
1163 os.texrow().start(owner_->id(), i + 1);
1166 column += (unsigned int)(os.os().tellp() - len);
1169 if (owner_->isDeleted(i))
1170 --runparams.inDeletedInset;
1174 void Paragraph::Private::latexSpecialChar(otexstream & os,
1175 BufferParams const & bparams,
1176 OutputParams const & runparams,
1177 Font const & running_font,
1178 Change const & running_change,
1179 Layout const & style,
1182 unsigned int & column)
1184 // With polyglossia, brackets and stuff need not be reversed
1185 // in RTL scripts (see bug #8251)
1186 char_type const c = (runparams.use_polyglossia) ?
1187 owner_->getUChar(bparams, i) : text_[i];
1189 if (style.pass_thru || runparams.pass_thru
1190 || contains(style.pass_thru_chars, c)
1191 || contains(runparams.pass_thru_chars, c)) {
1193 Encoding const * const enc = runparams.encoding;
1194 if (enc && !enc->encodable(c))
1195 throw EncodingException(c);
1201 // TIPA uses its own T3 encoding
1202 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1204 // If T1 font encoding is used, use the special
1205 // characters it provides.
1206 // NOTE: Some languages reset the font encoding internally to a
1207 // non-standard font encoding. If we are using such a language,
1208 // we do not output special T1 chars.
1209 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1210 && !runparams.isFullUnicode() && bparams.main_font_encoding() == "T1"
1211 && latexSpecialT1(c, os, i, column))
1213 // NOTE: XeTeX and LuaTeX use EU1/2 (pre 2017) or TU (as of 2017) encoding
1214 else if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1215 && runparams.isFullUnicode() && latexSpecialTU(c, os, i, column))
1218 // Otherwise, we use what LaTeX provides us.
1221 os << "\\textbackslash" << termcmd;
1225 os << "\\textless" << termcmd;
1229 os << "\\textgreater" << termcmd;
1233 os << "\\textbar" << termcmd;
1238 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1239 (end_pos == -1 || i + 1 < end_pos) &&
1240 text_[i+1] == '-') {
1241 // Prevent "--" becoming an endash and "---" becoming
1243 // Within \ttfamily, "--" is merged to "-" (no endash)
1244 // so we avoid this rather irritating ligature as well
1250 os << "\\textquotedbl" << termcmd;
1255 case '%': case '#': case '{':
1263 os << "\\textasciitilde" << termcmd;
1268 os << "\\textasciicircum" << termcmd;
1275 // avoid being mistaken for optional arguments
1283 // Blanks are printed before font switching.
1284 // Sure? I am not! (try nice-latex)
1285 // I am sure it's correct. LyX might be smarter
1286 // in the future, but for now, nothing wrong is
1292 if (bparams.use_dash_ligatures && !bparams.useNonTeXFonts) {
1309 Encoding const & encoding = *(runparams.encoding);
1310 char_type next = '\0';
1311 if (i + 1 < int(text_.size())) {
1312 next = text_[i + 1];
1313 if (Encodings::isCombiningChar(next)) {
1314 column += latexSurrogatePair(os, c, next, runparams) - 1;
1320 pair<docstring, bool> latex = encoding.latexChar(c);
1321 docstring nextlatex;
1322 bool nexttipas = false;
1323 string nexttipashortcut;
1324 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1325 nextlatex = encoding.latexChar(next).first;
1326 if (runparams.inIPA) {
1327 nexttipashortcut = Encodings::TIPAShortcut(next);
1328 nexttipas = !nexttipashortcut.empty();
1332 if (runparams.inIPA) {
1333 string const tipashortcut = Encodings::TIPAShortcut(c);
1334 if (!tipashortcut.empty()) {
1335 latex.first = from_ascii(tipashortcut);
1336 latex.second = false;
1340 if (Encodings::isKnownScriptChar(c, script)
1341 && prefixIs(latex.first, from_ascii("\\" + script)))
1342 column += writeScriptChars(os, latex.first,
1343 running_change, encoding, i) - 1;
1344 else if (latex.second
1345 && ((!prefixIs(nextlatex, '\\')
1346 && !prefixIs(nextlatex, '{')
1347 && !prefixIs(nextlatex, '}'))
1349 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1351 // Prevent eating of a following
1352 // space or command corruption by
1353 // following characters
1354 if (next == ' ' || next == '\0') {
1355 column += latex.first.length() + 1;
1356 os << latex.first << "{}";
1358 column += latex.first.length();
1359 os << latex.first << " ";
1362 column += latex.first.length() - 1;
1370 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1371 pos_type i, unsigned int & column)
1377 // In T1 encoding, these characters exist
1378 // but we should avoid ligatures
1379 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1381 os << "\\textcompwordmark" << termcmd;
1388 // soul.sty breaks with \char`\"
1389 os << "\\textquotedbl" << termcmd;
1398 bool Paragraph::Private::latexSpecialTU(char_type const c, otexstream & os,
1399 pos_type i, unsigned int & column)
1401 // TU encoding is currently on par with T1.
1402 return latexSpecialT1(c, os, i, column);
1406 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1407 pos_type /*i*/, unsigned int & column)
1417 os << "\\textvertline" << termcmd;
1426 void Paragraph::Private::validate(LaTeXFeatures & features) const
1428 if (layout_->inpreamble && inset_owner_) {
1429 bool const is_command = layout_->latextype == LATEX_COMMAND;
1430 Buffer const & buf = inset_owner_->buffer();
1431 BufferParams const & bp = features.runparams().is_child
1432 ? buf.masterParams() : buf.params();
1434 // Using a string stream here circumvents the encoding
1435 // switching machinery of odocstream. Therefore the
1436 // output is wrong if this paragraph contains content
1437 // that needs to switch encoding.
1438 otexstringstream os;
1439 os << layout_->preamble();
1441 os << '\\' << from_ascii(layout_->latexname());
1442 // we have to provide all the optional arguments here, even though
1443 // the last one is the only one we care about.
1444 // Separate handling of optional argument inset.
1445 if (!layout_->latexargs().empty()) {
1446 OutputParams rp = features.runparams();
1447 rp.local_font = &owner_->getFirstFontSettings(bp);
1448 latexArgInsets(*owner_, os, rp, layout_->latexargs());
1450 os << from_ascii(layout_->latexparam());
1452 size_t const length = os.length();
1453 // this will output "{" at the beginning, but not at the end
1454 owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
1455 if (os.length() > length) {
1458 if (!layout_->postcommandargs().empty()) {
1459 OutputParams rp = features.runparams();
1460 rp.local_font = &owner_->getFirstFontSettings(bp);
1461 latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
1464 features.addPreambleSnippet(os.release(), true);
1468 if (features.runparams().flavor == OutputParams::HTML
1469 && layout_->htmltitle()) {
1470 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1473 // check the params.
1474 if (!params_.spacing().isDefault())
1475 features.require("setspace");
1478 features.useLayout(layout_->name());
1481 fontlist_.validate(features);
1483 // then the indentation
1484 if (!params_.leftIndent().zero())
1485 features.require("ParagraphLeftIndent");
1488 InsetList::const_iterator icit = insetlist_.begin();
1489 InsetList::const_iterator iend = insetlist_.end();
1490 for (; icit != iend; ++icit) {
1492 features.inDeletedInset(owner_->isDeleted(icit->pos));
1493 if (icit->inset->lyxCode() == FOOT_CODE) {
1494 // FIXME: an item inset would make things much easier.
1495 if ((layout_->latextype == LATEX_LIST_ENVIRONMENT
1496 || (layout_->latextype == LATEX_ITEM_ENVIRONMENT
1497 && layout_->margintype == MARGIN_FIRST_DYNAMIC))
1498 && (icit->pos < begin_of_body_
1499 || (icit->pos == begin_of_body_
1500 && (icit->pos == 0 || text_[icit->pos - 1] != ' '))))
1501 features.saveNoteEnv("description");
1503 icit->inset->validate(features);
1504 features.inDeletedInset(false);
1505 if (layout_->needprotect &&
1506 icit->inset->lyxCode() == FOOT_CODE)
1507 features.require("NeedLyXFootnoteCode");
1511 // then the contents
1512 BufferParams const bp = features.buffer().masterParams();
1513 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1514 char_type c = text_[i];
1516 if (features.runparams().isFullUnicode() && bp.useNonTeXFonts)
1517 features.require("textquotedblp");
1518 else if (bp.main_font_encoding() != "T1"
1519 || ((&owner_->getFontSettings(bp, i))->language()->internalFontEncoding()))
1520 features.require("textquotedbl");
1522 BufferEncodings::validate(c, features);
1526 /////////////////////////////////////////////////////////////////////
1530 /////////////////////////////////////////////////////////////////////
1533 Layout const emptyParagraphLayout;
1536 Paragraph::Paragraph()
1537 : d(new Paragraph::Private(this, emptyParagraphLayout))
1544 Paragraph::Paragraph(Paragraph const & par)
1545 : itemdepth(par.itemdepth),
1546 d(new Paragraph::Private(*par.d, this))
1552 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1553 : itemdepth(par.itemdepth),
1554 d(new Paragraph::Private(*par.d, this, beg, end))
1560 Paragraph & Paragraph::operator=(Paragraph const & par)
1562 // needed as we will destroy the private part before copying it
1564 itemdepth = par.itemdepth;
1568 d = new Private(*par.d, this);
1575 Paragraph::~Paragraph()
1584 // this shall be called just before every "os << ..." action.
1585 void flushString(ostream & os, docstring & s)
1594 void Paragraph::write(ostream & os, BufferParams const & bparams,
1595 depth_type & dth) const
1597 // The beginning or end of a deeper (i.e. nested) area?
1598 if (dth != d->params_.depth()) {
1599 if (d->params_.depth() > dth) {
1600 while (d->params_.depth() > dth) {
1601 os << "\n\\begin_deeper";
1605 while (d->params_.depth() < dth) {
1606 os << "\n\\end_deeper";
1612 // First write the layout
1613 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1615 d->params_.write(os);
1617 Font font1(inherit_font, bparams.language);
1619 Change running_change = Change(Change::UNCHANGED);
1621 // this string is used as a buffer to avoid repetitive calls
1622 // to to_utf8(), which turn out to be expensive (JMarc)
1623 docstring write_buffer;
1626 for (pos_type i = 0; i <= size(); ++i) {
1628 Change const & change = lookupChange(i);
1629 if (change != running_change)
1630 flushString(os, write_buffer);
1631 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1632 running_change = change;
1637 // Write font changes
1638 Font font2 = getFontSettings(bparams, i);
1639 if (font2 != font1) {
1640 flushString(os, write_buffer);
1641 font2.lyxWriteChanges(font1, os);
1646 char_type const c = d->text_[i];
1649 if (Inset const * inset = getInset(i)) {
1650 flushString(os, write_buffer);
1651 if (inset->directWrite()) {
1652 // international char, let it write
1653 // code directly so it's shorter in
1659 os << "\\begin_inset ";
1661 os << "\n\\end_inset\n\n";
1664 // FIXME This can be removed again once the mystery
1665 // crash has been resolved.
1670 flushString(os, write_buffer);
1671 os << "\n\\backslash\n";
1675 flushString(os, write_buffer);
1676 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1683 if ((column > 70 && c == ' ')
1685 flushString(os, write_buffer);
1689 // this check is to amend a bug. LyX sometimes
1690 // inserts '\0' this could cause problems.
1692 write_buffer.push_back(c);
1694 LYXERR0("NUL char in structure.");
1700 flushString(os, write_buffer);
1701 os << "\n\\end_layout\n";
1702 // FIXME This can be removed again once the mystery
1703 // crash has been resolved.
1708 void Paragraph::validate(LaTeXFeatures & features) const
1710 d->validate(features);
1714 void Paragraph::insert(pos_type start, docstring const & str,
1715 Font const & font, Change const & change)
1717 for (size_t i = 0, n = str.size(); i != n ; ++i)
1718 insertChar(start + i, str[i], font, change);
1722 void Paragraph::appendChar(char_type c, Font const & font,
1723 Change const & change)
1726 d->changes_.insert(change, d->text_.size());
1727 // when appending characters, no need to update tables
1728 d->text_.push_back(c);
1729 setFont(d->text_.size() - 1, font);
1730 d->requestSpellCheck(d->text_.size() - 1);
1734 void Paragraph::appendString(docstring const & s, Font const & font,
1735 Change const & change)
1737 pos_type end = s.size();
1738 size_t oldsize = d->text_.size();
1739 size_t newsize = oldsize + end;
1740 size_t capacity = d->text_.capacity();
1741 if (newsize >= capacity)
1742 d->text_.reserve(max(capacity + 100, newsize));
1744 // when appending characters, no need to update tables
1747 // FIXME: Optimize this!
1748 for (size_t i = oldsize; i != newsize; ++i) {
1750 d->changes_.insert(change, i);
1751 d->requestSpellCheck(i);
1753 d->fontlist_.set(oldsize, font);
1754 d->fontlist_.set(newsize - 1, font);
1758 void Paragraph::insertChar(pos_type pos, char_type c,
1761 d->insertChar(pos, c, Change(trackChanges ?
1762 Change::INSERTED : Change::UNCHANGED));
1766 void Paragraph::insertChar(pos_type pos, char_type c,
1767 Font const & font, bool trackChanges)
1769 d->insertChar(pos, c, Change(trackChanges ?
1770 Change::INSERTED : Change::UNCHANGED));
1775 void Paragraph::insertChar(pos_type pos, char_type c,
1776 Font const & font, Change const & change)
1778 d->insertChar(pos, c, change);
1783 void Paragraph::resetFonts(Font const & font)
1785 d->fontlist_.clear();
1786 d->fontlist_.set(0, font);
1787 d->fontlist_.set(d->text_.size() - 1, font);
1790 // Gets uninstantiated font setting at position.
1791 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1795 LYXERR0("pos: " << pos << " size: " << size());
1799 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1800 if (cit != d->fontlist_.end())
1803 if (pos == size() && !empty())
1804 return getFontSettings(bparams, pos - 1);
1806 // Optimisation: avoid a full font instantiation if there is no
1807 // language change from previous call.
1808 static Font previous_font;
1809 static Language const * previous_lang = 0;
1810 Language const * lang = getParLanguage(bparams);
1811 if (lang != previous_lang) {
1812 previous_lang = lang;
1813 previous_font = Font(inherit_font, lang);
1815 return previous_font;
1819 FontSpan Paragraph::fontSpan(pos_type pos) const
1821 LBUFERR(pos <= size());
1824 return FontSpan(pos, pos);
1827 FontList::const_iterator cit = d->fontlist_.begin();
1828 FontList::const_iterator end = d->fontlist_.end();
1829 for (; cit != end; ++cit) {
1830 if (cit->pos() >= pos) {
1831 if (pos >= beginOfBody())
1832 return FontSpan(max(start, beginOfBody()),
1835 return FontSpan(start,
1836 min(beginOfBody() - 1,
1839 start = cit->pos() + 1;
1842 // This should not happen, but if so, we take no chances.
1843 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1844 LASSERT(false, return FontSpan(pos, pos));
1848 // Gets uninstantiated font setting at position 0
1849 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1851 if (!empty() && !d->fontlist_.empty())
1852 return d->fontlist_.begin()->font();
1854 // Optimisation: avoid a full font instantiation if there is no
1855 // language change from previous call.
1856 static Font previous_font;
1857 static Language const * previous_lang = 0;
1858 if (bparams.language != previous_lang) {
1859 previous_lang = bparams.language;
1860 previous_font = Font(inherit_font, bparams.language);
1863 return previous_font;
1867 // Gets the fully instantiated font at a given position in a paragraph
1868 // This is basically the same function as Text::GetFont() in text2.cpp.
1869 // The difference is that this one is used for generating the LaTeX file,
1870 // and thus cosmetic "improvements" are disallowed: This has to deliver
1871 // the true picture of the buffer. (Asger)
1872 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1873 Font const & outerfont) const
1877 Font font = getFontSettings(bparams, pos);
1879 pos_type const body_pos = beginOfBody();
1880 FontInfo & fi = font.fontInfo();
1882 fi.realize(d->layout_->labelfont);
1884 fi.realize(d->layout_->font);
1886 fi.realize(outerfont.fontInfo());
1887 fi.realize(bparams.getFont().fontInfo());
1893 Font const Paragraph::getLabelFont
1894 (BufferParams const & bparams, Font const & outerfont) const
1896 FontInfo tmpfont = d->layout_->labelfont;
1897 tmpfont.realize(outerfont.fontInfo());
1898 tmpfont.realize(bparams.getFont().fontInfo());
1899 return Font(tmpfont, getParLanguage(bparams));
1903 Font const Paragraph::getLayoutFont
1904 (BufferParams const & bparams, Font const & outerfont) const
1906 FontInfo tmpfont = d->layout_->font;
1907 tmpfont.realize(outerfont.fontInfo());
1908 tmpfont.realize(bparams.getFont().fontInfo());
1909 return Font(tmpfont, getParLanguage(bparams));
1913 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1915 char_type c = d->text_[pos];
1916 if (!getFontSettings(bparams, pos).isRightToLeft())
1919 // FIXME: The arabic special casing is due to the difference of arabic
1920 // round brackets input introduced in r18599. Check if this should be
1921 // unified with Hebrew or at least if all bracket types should be
1922 // handled the same (file format change in either case).
1923 string const & lang = getFontSettings(bparams, pos).language()->lang();
1924 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1929 uc = arabic ? c : ')';
1932 uc = arabic ? c : '(';
1958 void Paragraph::setFont(pos_type pos, Font const & font)
1960 LASSERT(pos <= size(), return);
1962 // First, reduce font against layout/label font
1963 // Update: The setCharFont() routine in text2.cpp already
1964 // reduces font, so we don't need to do that here. (Asger)
1966 d->fontlist_.set(pos, font);
1970 void Paragraph::makeSameLayout(Paragraph const & par)
1972 d->layout_ = par.d->layout_;
1973 d->params_ = par.d->params_;
1977 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1979 if (isFreeSpacing())
1985 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1986 if (eraseChar(pos, trackChanges))
1992 return count > 0 || pos > 0;
1996 bool Paragraph::hasSameLayout(Paragraph const & par) const
1998 return par.d->layout_ == d->layout_
1999 && d->params_.sameLayout(par.d->params_);
2003 depth_type Paragraph::getDepth() const
2005 return d->params_.depth();
2009 depth_type Paragraph::getMaxDepthAfter() const
2011 if (d->layout_->isEnvironment())
2012 return d->params_.depth() + 1;
2014 return d->params_.depth();
2018 LyXAlignment Paragraph::getAlign() const
2020 if (d->params_.align() == LYX_ALIGN_LAYOUT)
2021 return d->layout_->align;
2023 return d->params_.align();
2027 docstring const & Paragraph::labelString() const
2029 return d->params_.labelString();
2033 // the next two functions are for the manual labels
2034 docstring const Paragraph::getLabelWidthString() const
2036 if (d->layout_->margintype == MARGIN_MANUAL
2037 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
2038 return d->params_.labelWidthString();
2040 return _("Senseless with this layout!");
2044 void Paragraph::setLabelWidthString(docstring const & s)
2046 d->params_.labelWidthString(s);
2050 docstring Paragraph::expandLabel(Layout const & layout,
2051 BufferParams const & bparams) const
2053 return expandParagraphLabel(layout, bparams, true);
2057 docstring Paragraph::expandDocBookLabel(Layout const & layout,
2058 BufferParams const & bparams) const
2060 return expandParagraphLabel(layout, bparams, false);
2064 docstring Paragraph::expandParagraphLabel(Layout const & layout,
2065 BufferParams const & bparams, bool process_appendix) const
2067 DocumentClass const & tclass = bparams.documentClass();
2068 string const & lang = getParLanguage(bparams)->code();
2069 bool const in_appendix = process_appendix && d->params_.appendix();
2070 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
2072 if (fmt.empty() && !layout.counter.empty())
2073 return tclass.counters().theCounter(layout.counter, lang);
2075 // handle 'inherited level parts' in 'fmt',
2076 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2077 size_t const i = fmt.find('@', 0);
2078 if (i != docstring::npos) {
2079 size_t const j = fmt.find('@', i + 1);
2080 if (j != docstring::npos) {
2081 docstring parent(fmt, i + 1, j - i - 1);
2082 docstring label = from_ascii("??");
2083 if (tclass.hasLayout(parent))
2084 label = expandParagraphLabel(tclass[parent], bparams,
2086 fmt = docstring(fmt, 0, i) + label
2087 + docstring(fmt, j + 1, docstring::npos);
2091 return tclass.counters().counterLabel(fmt, lang);
2095 void Paragraph::applyLayout(Layout const & new_layout)
2097 d->layout_ = &new_layout;
2098 LyXAlignment const oldAlign = d->params_.align();
2100 if (!(oldAlign & d->layout_->alignpossible)) {
2101 frontend::Alert::warning(_("Alignment not permitted"),
2102 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2103 d->params_.align(LYX_ALIGN_LAYOUT);
2108 pos_type Paragraph::beginOfBody() const
2110 return d->begin_of_body_;
2114 void Paragraph::setBeginOfBody()
2116 if (d->layout_->labeltype != LABEL_MANUAL) {
2117 d->begin_of_body_ = 0;
2121 // Unroll the first two cycles of the loop
2122 // and remember the previous character to
2123 // remove unnecessary getChar() calls
2125 pos_type end = size();
2126 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2129 char_type previous_char = d->text_[i];
2130 if (!(isNewline(i) || isEnvSeparator(i))) {
2132 while (i < end && previous_char != ' ') {
2133 char_type temp = d->text_[i];
2134 if (isNewline(i) || isEnvSeparator(i))
2137 previous_char = temp;
2143 d->begin_of_body_ = i;
2147 bool Paragraph::allowParagraphCustomization() const
2149 return inInset().allowParagraphCustomization();
2153 bool Paragraph::usePlainLayout() const
2155 return inInset().usePlainLayout();
2159 bool Paragraph::isPassThru() const
2161 return inInset().isPassThru() || d->layout_->pass_thru;
2166 // paragraphs inside floats need different alignment tags to avoid
2169 bool noTrivlistCentering(InsetCode code)
2171 return code == FLOAT_CODE
2172 || code == WRAP_CODE
2173 || code == CELL_CODE;
2177 string correction(string const & orig)
2179 if (orig == "flushleft")
2180 return "raggedright";
2181 if (orig == "flushright")
2182 return "raggedleft";
2183 if (orig == "center")
2189 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2190 InsetCode code, bool const lastpar, int & col)
2192 string macro = suffix + "{";
2193 if (noTrivlistCentering(code)) {
2195 // the last paragraph in non-trivlist-aligned
2196 // context is special (to avoid unwanted whitespace)
2197 if (suffix == "\\begin") {
2198 macro = "\\" + correction(env) + "{}";
2199 os << from_ascii(macro);
2200 col += macro.size();
2205 macro += correction(env);
2209 if (suffix == "\\par\\end") {
2213 os << from_ascii(macro);
2214 col += macro.size();
2215 if (suffix == "\\begin") {
2225 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2226 otexstream & os, OutputParams const & runparams) const
2231 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2232 (layout_->toggle_indent != ITOGGLE_NEVER) :
2233 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2235 if (canindent && params_.noindent() && !layout_->pass_thru) {
2236 os << "\\noindent ";
2240 LyXAlignment const curAlign = params_.align();
2242 if (curAlign == layout_->align)
2246 case LYX_ALIGN_NONE:
2247 case LYX_ALIGN_BLOCK:
2248 case LYX_ALIGN_LAYOUT:
2249 case LYX_ALIGN_SPECIAL:
2250 case LYX_ALIGN_DECIMAL:
2252 case LYX_ALIGN_LEFT:
2253 case LYX_ALIGN_RIGHT:
2254 case LYX_ALIGN_CENTER:
2255 if (runparams.moving_arg) {
2262 string const begin_tag = "\\begin";
2263 InsetCode code = ownerCode();
2264 bool const lastpar = runparams.isLastPar;
2267 case LYX_ALIGN_NONE:
2268 case LYX_ALIGN_BLOCK:
2269 case LYX_ALIGN_LAYOUT:
2270 case LYX_ALIGN_SPECIAL:
2271 case LYX_ALIGN_DECIMAL:
2273 case LYX_ALIGN_LEFT: {
2274 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2275 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2277 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2279 } case LYX_ALIGN_RIGHT: {
2280 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2281 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2283 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2285 } case LYX_ALIGN_CENTER: {
2286 corrected_env(os, begin_tag, "center", code, lastpar, column);
2295 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2296 otexstream & os, OutputParams const & runparams) const
2298 LyXAlignment const curAlign = params_.align();
2300 if (curAlign == layout_->align)
2304 case LYX_ALIGN_NONE:
2305 case LYX_ALIGN_BLOCK:
2306 case LYX_ALIGN_LAYOUT:
2307 case LYX_ALIGN_SPECIAL:
2308 case LYX_ALIGN_DECIMAL:
2310 case LYX_ALIGN_LEFT:
2311 case LYX_ALIGN_RIGHT:
2312 case LYX_ALIGN_CENTER:
2313 if (runparams.moving_arg)
2318 bool output = false;
2320 string const end_tag = "\\par\\end";
2321 InsetCode code = ownerCode();
2322 bool const lastpar = runparams.isLastPar;
2325 case LYX_ALIGN_NONE:
2326 case LYX_ALIGN_BLOCK:
2327 case LYX_ALIGN_LAYOUT:
2328 case LYX_ALIGN_SPECIAL:
2329 case LYX_ALIGN_DECIMAL:
2331 case LYX_ALIGN_LEFT: {
2332 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2333 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2335 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2337 } case LYX_ALIGN_RIGHT: {
2338 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2339 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2341 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2343 } case LYX_ALIGN_CENTER: {
2344 corrected_env(os, end_tag, "center", code, lastpar, col);
2349 return output || lastpar;
2353 // This one spits out the text of the paragraph
2354 void Paragraph::latex(BufferParams const & bparams,
2355 Font const & outerfont,
2357 OutputParams const & runparams,
2358 int start_pos, int end_pos, bool force) const
2360 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2362 // FIXME This check should not be needed. Perhaps issue an
2363 // error if it triggers.
2364 Layout const & style = inInset().forcePlainLayout() ?
2365 bparams.documentClass().plainLayout() : *d->layout_;
2367 if (!force && style.inpreamble)
2370 bool const allowcust = allowParagraphCustomization();
2372 // Current base font for all inherited font changes, without any
2373 // change caused by an individual character, except for the language:
2374 // It is set to the language of the first character.
2375 // As long as we are in the label, this font is the base font of the
2376 // label. Before the first body character it is set to the base font
2380 // Maybe we have to create a optional argument.
2381 pos_type body_pos = beginOfBody();
2382 unsigned int column = 0;
2385 // the optional argument is kept in curly brackets in
2386 // case it contains a ']'
2387 // This is not strictly needed, but if this is changed it
2388 // would be a file format change, and tex2lyx would need
2389 // to be adjusted, since it unconditionally removes the
2390 // braces when it parses \item.
2393 basefont = getLabelFont(bparams, outerfont);
2395 basefont = getLayoutFont(bparams, outerfont);
2398 // Which font is currently active?
2399 Font running_font(basefont);
2400 // Do we have an open font change?
2401 bool open_font = false;
2403 Change runningChange = Change(Change::UNCHANGED);
2405 Encoding const * const prev_encoding = runparams.encoding;
2407 os.texrow().start(id(), 0);
2409 // if the paragraph is empty, the loop will not be entered at all
2411 if (style.isCommand()) {
2415 if (!style.leftdelim().empty()) {
2416 os << style.leftdelim();
2417 column += style.leftdelim().size();
2420 column += d->startTeXParParams(bparams, os, runparams);
2423 // Whether a \par can be issued for insets typeset inline with text.
2424 // Yes if greater than 0. This has to be static.
2425 THREAD_LOCAL_STATIC int parInline = 0;
2427 for (pos_type i = 0; i < size(); ++i) {
2428 // First char in paragraph or after label?
2429 if (i == body_pos) {
2432 bool needPar = false;
2433 column += running_font.latexWriteEndChanges(
2434 os, bparams, runparams,
2435 basefont, basefont, needPar);
2438 basefont = getLayoutFont(bparams, outerfont);
2439 running_font = basefont;
2441 column += Changes::latexMarkChange(os, bparams,
2442 runningChange, Change(Change::UNCHANGED),
2444 runningChange = Change(Change::UNCHANGED);
2449 if (style.isCommand()) {
2454 if (!style.leftdelim().empty()) {
2455 os << style.leftdelim();
2456 column += style.leftdelim().size();
2460 column += d->startTeXParParams(bparams, os,
2464 runparams.wasDisplayMath = runparams.inDisplayMath;
2465 runparams.inDisplayMath = false;
2466 bool deleted_display_math = false;
2467 Change const & change = runparams.inDeletedInset
2468 ? runparams.changeOfDeletedInset : lookupChange(i);
2470 // Check whether a display math inset follows
2471 if (d->text_[i] == META_INSET
2472 && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2473 InsetMath const * im = getInset(i)->asInsetMath();
2474 if (im && im->asHullInset()
2475 && im->asHullInset()->outerDisplay()) {
2476 runparams.inDisplayMath = true;
2477 // runparams.inDeletedInset will be set by
2478 // latexInset later, but we need this info
2479 // before it is called. On the other hand, we
2480 // cannot set it here because it is a counter.
2481 deleted_display_math = isDeleted(i);
2483 if (bparams.output_changes && deleted_display_math
2484 && runningChange == change
2485 && change.type == Change::DELETED
2486 && !os.afterParbreak()) {
2487 // A display math in the same paragraph follows.
2488 // We have to close and then reopen \lyxdeleted,
2489 // otherwise the math will be shifted up.
2490 OutputParams rp = runparams;
2492 bool needPar = false;
2493 column += running_font.latexWriteEndChanges(
2494 os, bparams, rp, basefont,
2498 basefont = getLayoutFont(bparams, outerfont);
2499 running_font = basefont;
2500 column += Changes::latexMarkChange(os, bparams,
2501 Change(Change::INSERTED), change, rp);
2505 if (bparams.output_changes && runningChange != change) {
2507 bool needPar = false;
2508 column += running_font.latexWriteEndChanges(
2509 os, bparams, runparams,
2510 basefont, basefont, needPar);
2513 basefont = getLayoutFont(bparams, outerfont);
2514 running_font = basefont;
2515 column += Changes::latexMarkChange(os, bparams, runningChange,
2517 runningChange = change;
2520 // do not output text which is marked deleted
2521 // if change tracking output is disabled
2522 if (!bparams.output_changes && change.deleted()) {
2528 // Fully instantiated font
2529 Font const current_font = getFont(bparams, i, outerfont);
2531 Font const last_font = running_font;
2533 // Do we need to close the previous font?
2535 (current_font != running_font ||
2536 current_font.language() != running_font.language()))
2538 bool needPar = false;
2539 column += running_font.latexWriteEndChanges(
2540 os, bparams, runparams, basefont,
2541 (i == body_pos-1) ? basefont : current_font,
2543 running_font = basefont;
2547 string const running_lang = runparams.use_polyglossia ?
2548 running_font.language()->polyglossia() : running_font.language()->babel();
2549 // close babel's font environment before opening CJK.
2550 string const lang_end_command = runparams.use_polyglossia ?
2551 "\\end{$$lang}" : lyxrc.language_command_end;
2552 bool const using_begin_end = runparams.use_polyglossia ||
2553 !lang_end_command.empty();
2554 if (!running_lang.empty() &&
2555 (!using_begin_end || running_lang == openLanguageName()) &&
2556 current_font.language()->encoding()->package() == Encoding::CJK) {
2557 string end_tag = subst(lang_end_command,
2560 os << from_ascii(end_tag);
2561 column += end_tag.length();
2562 if (using_begin_end)
2566 // Switch file encoding if necessary (and allowed)
2567 if (!runparams.pass_thru && !style.pass_thru &&
2568 runparams.encoding->package() != Encoding::none &&
2569 current_font.language()->encoding()->package() != Encoding::none) {
2570 pair<bool, int> const enc_switch =
2571 switchEncoding(os.os(), bparams, runparams,
2572 *(current_font.language()->encoding()));
2573 if (enc_switch.first) {
2574 column += enc_switch.second;
2575 runparams.encoding = current_font.language()->encoding();
2579 char_type const c = d->text_[i];
2581 // A display math inset inside an ulem command will be output
2582 // as a box of width \linewidth, so we have to either disable
2583 // indentation if the inset starts a paragraph, or start a new
2584 // line to accommodate such box. This has to be done before
2585 // writing any font changing commands.
2586 if (runparams.inDisplayMath && !deleted_display_math
2587 && runparams.inulemcmd) {
2588 if (os.afterParbreak())
2594 // Do we need to change font?
2595 if ((current_font != running_font ||
2596 current_font.language() != running_font.language()) &&
2599 odocstringstream ods;
2600 column += current_font.latexWriteStartChanges(ods, bparams,
2601 runparams, basefont,
2603 // Check again for display math in ulem commands as a
2604 // font change may also occur just before a math inset.
2605 if (runparams.inDisplayMath && !deleted_display_math
2606 && runparams.inulemcmd) {
2607 if (os.afterParbreak())
2612 running_font = current_font;
2614 docstring fontchange = ods.str();
2615 // check whether the fontchange ends with a \\textcolor
2616 // modifier and the text starts with a space (bug 4473)
2617 docstring const last_modifier = rsplit(fontchange, '\\');
2618 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2619 os << fontchange << from_ascii("{}");
2620 // check if the fontchange ends with a trailing blank
2621 // (like "\small " (see bug 3382)
2622 else if (suffixIs(fontchange, ' ') && c == ' ')
2623 os << fontchange.substr(0, fontchange.size() - 1)
2624 << from_ascii("{}");
2629 // FIXME: think about end_pos implementation...
2630 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2631 // FIXME: integrate this case in latexSpecialChar
2632 // Do not print the separation of the optional argument
2633 // if style.pass_thru is false. This works because
2634 // latexSpecialChar ignores spaces if
2635 // style.pass_thru is false.
2636 if (i != body_pos - 1) {
2637 if (d->simpleTeXBlanks(runparams, os,
2638 i, column, current_font, style)) {
2639 // A surrogate pair was output. We
2640 // must not call latexSpecialChar
2641 // in this iteration, since it would output
2642 // the combining character again.
2649 OutputParams rp = runparams;
2650 rp.free_spacing = style.free_spacing;
2651 rp.local_font = ¤t_font;
2652 rp.intitle = style.intitle;
2654 // Two major modes: LaTeX or plain
2655 // Handle here those cases common to both modes
2656 // and then split to handle the two modes separately.
2657 if (c == META_INSET) {
2658 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2659 // Greyedout notes and, in general, all insets
2660 // with InsetLayout::isDisplay() == false,
2661 // are typeset inline with the text. So, we
2662 // can add a \par to the last paragraph of
2663 // such insets only if nothing else follows.
2664 bool incremented = false;
2665 Inset const * inset = getInset(i);
2666 InsetText const * textinset = inset
2667 ? inset->asInsetText()
2669 if (i + 1 == size() && textinset
2670 && !inset->getLayout().isDisplay()) {
2671 ParagraphList const & pars =
2672 textinset->text().paragraphs();
2673 pit_type const pit = pars.size() - 1;
2674 Font const last_font =
2675 pit < 0 || pars[pit].empty()
2676 ? pars[pit].getLayoutFont(
2679 : pars[pit].getFont(bparams,
2680 pars[pit].size() - 1,
2682 if (last_font.fontInfo().size() !=
2683 basefont.fontInfo().size()) {
2688 d->latexInset(bparams, os, rp, running_font,
2689 basefont, outerfont, open_font,
2690 runningChange, style, i, column);
2694 } else if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2696 d->latexSpecialChar(os, bparams, rp,
2697 running_font, runningChange,
2698 style, i, end_pos, column);
2699 } catch (EncodingException & e) {
2700 if (runparams.dryrun) {
2701 os << "<" << _("LyX Warning: ")
2702 << _("uncodable character") << " '";
2706 // add location information and throw again.
2714 // Set the encoding to that returned from latexSpecialChar (see
2715 // comment for encoding member in OutputParams.h)
2716 runparams.encoding = rp.encoding;
2718 // Also carry on the info on a closed ulem command for insets
2719 // such as Note that do not produce any output, so that no
2720 // command is ever executed but its opening was recorded.
2721 runparams.inulemcmd = rp.inulemcmd;
2724 // If we have an open font definition, we have to close it
2726 // Make sure that \\par is done with the font of the last
2727 // character if this has another size as the default.
2728 // This is necessary because LaTeX (and LyX on the screen)
2729 // calculates the space between the baselines according
2730 // to this font. (Matthias)
2732 // We must not change the font for the last paragraph
2733 // of non-multipar insets, tabular cells or commands,
2734 // since this produces unwanted whitespace.
2736 Font const font = empty()
2737 ? getLayoutFont(bparams, outerfont)
2738 : getFont(bparams, size() - 1, outerfont);
2740 InsetText const * textinset = inInset().asInsetText();
2742 bool const maintext = textinset
2743 ? textinset->text().isMainText()
2746 size_t const numpars = textinset
2747 ? textinset->text().paragraphs().size()
2750 bool needPar = false;
2752 if (style.resfont.size() != font.fontInfo().size()
2753 && (!runparams.isLastPar || maintext
2754 || (numpars > 1 && d->ownerCode() != CELL_CODE
2755 && (inInset().getLayout().isDisplay()
2757 && !style.isCommand()) {
2760 #ifdef FIXED_LANGUAGE_END_DETECTION
2762 running_font.latexWriteEndChanges(os, bparams,
2763 runparams, basefont,
2764 next_->getFont(bparams, 0, outerfont),
2767 running_font.latexWriteEndChanges(os, bparams,
2768 runparams, basefont, basefont, needPar);
2771 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2772 //FIXME: there as we start another \selectlanguage with the next paragraph if
2773 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2774 running_font.latexWriteEndChanges(os, bparams, runparams,
2775 basefont, basefont, needPar);
2778 // The \par could not be inserted at the same nesting
2779 // level of the font size change, so do it now.
2780 os << "{\\" << font.latexSize() << "\\par}";
2784 column += Changes::latexMarkChange(os, bparams, runningChange,
2785 Change(Change::UNCHANGED), runparams);
2787 // Needed if there is an optional argument but no contents.
2788 if (body_pos > 0 && body_pos == size()) {
2792 if (!style.rightdelim().empty()) {
2793 os << style.rightdelim();
2794 column += style.rightdelim().size();
2797 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2798 && runparams.encoding != prev_encoding) {
2799 runparams.encoding = prev_encoding;
2800 os << setEncoding(prev_encoding->iconvName());
2803 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2807 bool Paragraph::emptyTag() const
2809 for (pos_type i = 0; i < size(); ++i) {
2810 if (Inset const * inset = getInset(i)) {
2811 InsetCode lyx_code = inset->lyxCode();
2812 // FIXME testing like that is wrong. What is
2814 if (lyx_code != TOC_CODE &&
2815 lyx_code != INCLUDE_CODE &&
2816 lyx_code != GRAPHICS_CODE &&
2817 lyx_code != ERT_CODE &&
2818 lyx_code != LISTINGS_CODE &&
2819 lyx_code != FLOAT_CODE &&
2820 lyx_code != TABULAR_CODE) {
2824 char_type c = d->text_[i];
2825 if (c != ' ' && c != '\t')
2833 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2836 for (pos_type i = 0; i < size(); ++i) {
2837 if (Inset const * inset = getInset(i)) {
2838 InsetCode lyx_code = inset->lyxCode();
2839 if (lyx_code == LABEL_CODE) {
2840 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2841 docstring const & id = il->getParam("name");
2842 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2850 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2854 for (i = 0; i < size(); ++i) {
2855 if (Inset const * inset = getInset(i)) {
2856 inset->docbook(os, runparams);
2858 char_type c = d->text_[i];
2861 os << sgml::escapeChar(c);
2868 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2872 for (i = 0; i < size(); ++i) {
2873 if (Inset const * inset = getInset(i)) {
2874 inset->xhtml(xs, runparams);
2876 char_type c = d->text_[i];
2886 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2889 pos_type size = text_.size();
2890 for (pos_type i = initial; i < size; ++i) {
2891 Font font = owner_->getFont(buf.params(), i, outerfont);
2892 if (text_[i] == META_INSET)
2894 if (i != initial && font != font_old)
2903 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2905 OutputParams const & runparams,
2906 Font const & outerfont,
2907 pos_type initial) const
2909 bool emph_flag = false;
2911 Layout const & style = *d->layout_;
2913 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2915 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2918 // parsing main loop
2919 for (pos_type i = initial; i < size(); ++i) {
2920 Font font = getFont(buf.params(), i, outerfont);
2922 // handle <emphasis> tag
2923 if (font_old.emph() != font.fontInfo().emph()) {
2924 if (font.fontInfo().emph() == FONT_ON) {
2927 } else if (i != initial) {
2928 os << "</emphasis>";
2933 if (Inset const * inset = getInset(i)) {
2934 inset->docbook(os, runparams);
2936 char_type c = d->text_[i];
2938 if (style.pass_thru)
2941 os << sgml::escapeChar(c);
2943 font_old = font.fontInfo();
2947 os << "</emphasis>";
2950 if (style.free_spacing)
2952 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2958 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2959 vector<html::EndFontTag> & tagsToClose,
2960 bool & flag, FontState curstate, html::FontTypes type)
2962 if (curstate == FONT_ON) {
2963 tagsToOpen.push_back(html::FontTag(type));
2966 tagsToClose.push_back(html::EndFontTag(type));
2973 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2975 OutputParams const & runparams,
2976 Font const & outerfont,
2977 bool start_paragraph, bool close_paragraph,
2978 pos_type initial) const
2982 // track whether we have opened these tags
2983 bool emph_flag = false;
2984 bool bold_flag = false;
2985 bool noun_flag = false;
2986 bool ubar_flag = false;
2987 bool dbar_flag = false;
2988 bool sout_flag = false;
2989 bool xout_flag = false;
2990 bool wave_flag = false;
2992 bool shap_flag = false;
2994 bool faml_flag = false;
2996 bool size_flag = false;
2998 Layout const & style = *d->layout_;
3000 if (start_paragraph)
3001 xs.startDivision(allowEmpty());
3004 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
3006 FontShape curr_fs = INHERIT_SHAPE;
3007 FontFamily curr_fam = INHERIT_FAMILY;
3008 FontSize curr_size = FONT_SIZE_INHERIT;
3010 string const default_family =
3011 buf.masterBuffer()->params().fonts_default_family;
3013 vector<html::FontTag> tagsToOpen;
3014 vector<html::EndFontTag> tagsToClose;
3016 // parsing main loop
3017 for (pos_type i = initial; i < size(); ++i) {
3018 // let's not show deleted material in the output
3022 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
3025 FontState curstate = font.fontInfo().emph();
3026 if (font_old.emph() != curstate)
3027 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
3030 curstate = font.fontInfo().noun();
3031 if (font_old.noun() != curstate)
3032 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
3035 curstate = font.fontInfo().underbar();
3036 if (font_old.underbar() != curstate)
3037 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
3040 curstate = font.fontInfo().strikeout();
3041 if (font_old.strikeout() != curstate)
3042 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
3045 curstate = font.fontInfo().xout();
3046 if (font_old.xout() != curstate)
3047 doFontSwitch(tagsToOpen, tagsToClose, xout_flag, curstate, html::FT_XOUT);
3050 curstate = font.fontInfo().uuline();
3051 if (font_old.uuline() != curstate)
3052 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
3055 curstate = font.fontInfo().uwave();
3056 if (font_old.uwave() != curstate)
3057 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
3060 // a little hackish, but allows us to reuse what we have.
3061 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
3062 if (font_old.series() != font.fontInfo().series())
3063 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
3066 curr_fs = font.fontInfo().shape();
3067 FontShape old_fs = font_old.shape();
3068 if (old_fs != curr_fs) {
3072 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
3075 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
3077 case SMALLCAPS_SHAPE:
3078 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
3084 // the other tags are for internal use
3092 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
3096 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
3099 case SMALLCAPS_SHAPE:
3100 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
3107 // the other tags are for internal use
3114 curr_fam = font.fontInfo().family();
3115 FontFamily old_fam = font_old.family();
3116 if (old_fam != curr_fam) {
3120 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
3123 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
3125 case TYPEWRITER_FAMILY:
3126 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
3128 case INHERIT_FAMILY:
3131 // the other tags are for internal use
3139 // we will treat a "default" font family as roman, since we have
3140 // no other idea what to do.
3141 if (default_family != "rmdefault" && default_family != "default") {
3142 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
3147 if (default_family != "sfdefault") {
3148 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
3152 case TYPEWRITER_FAMILY:
3153 if (default_family != "ttdefault") {
3154 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
3158 case INHERIT_FAMILY:
3161 // the other tags are for internal use
3168 curr_size = font.fontInfo().size();
3169 FontSize old_size = font_old.size();
3170 if (old_size != curr_size) {
3173 case FONT_SIZE_TINY:
3174 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
3176 case FONT_SIZE_SCRIPT:
3177 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
3179 case FONT_SIZE_FOOTNOTE:
3180 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
3182 case FONT_SIZE_SMALL:
3183 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
3185 case FONT_SIZE_LARGE:
3186 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
3188 case FONT_SIZE_LARGER:
3189 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
3191 case FONT_SIZE_LARGEST:
3192 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
3194 case FONT_SIZE_HUGE:
3195 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
3197 case FONT_SIZE_HUGER:
3198 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
3200 case FONT_SIZE_INCREASE:
3201 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
3203 case FONT_SIZE_DECREASE:
3204 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
3206 case FONT_SIZE_INHERIT:
3207 case FONT_SIZE_NORMAL:
3210 // the other tags are for internal use
3216 switch (curr_size) {
3217 case FONT_SIZE_TINY:
3218 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
3221 case FONT_SIZE_SCRIPT:
3222 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
3225 case FONT_SIZE_FOOTNOTE:
3226 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3229 case FONT_SIZE_SMALL:
3230 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3233 case FONT_SIZE_LARGE:
3234 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3237 case FONT_SIZE_LARGER:
3238 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3241 case FONT_SIZE_LARGEST:
3242 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3245 case FONT_SIZE_HUGE:
3246 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3249 case FONT_SIZE_HUGER:
3250 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3253 case FONT_SIZE_INCREASE:
3254 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3257 case FONT_SIZE_DECREASE:
3258 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3261 case FONT_SIZE_NORMAL:
3262 case FONT_SIZE_INHERIT:
3265 // the other tags are for internal use
3272 // Other such tags? What about the other text ranges?
3274 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3275 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3276 for (; cit != cen; ++cit)
3279 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3280 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3281 for (; sit != sen; ++sit)
3284 tagsToClose.clear();
3287 Inset const * inset = getInset(i);
3289 if (!runparams.for_toc || inset->isInToc()) {
3290 OutputParams np = runparams;
3291 np.local_font = &font;
3292 // If the paragraph has size 1, then we are in the "special
3293 // case" where we do not output the containing paragraph info
3294 if (!inset->getLayout().htmlisblock() && size() != 1)
3295 np.html_in_par = true;
3296 retval += inset->xhtml(xs, np);
3299 char_type c = getUChar(buf.masterBuffer()->params(), i);
3302 font_old = font.fontInfo();
3306 // I'm worried about what happens if a branch, say, is itself
3307 // wrapped in some font stuff. I think that will not work.
3309 if (close_paragraph)
3316 bool Paragraph::isHfill(pos_type pos) const
3318 Inset const * inset = getInset(pos);
3319 return inset && inset->isHfill();
3323 bool Paragraph::isNewline(pos_type pos) const
3325 // U+2028 LINE SEPARATOR
3326 // U+2029 PARAGRAPH SEPARATOR
3327 char_type const c = d->text_[pos];
3328 if (c == 0x2028 || c == 0x2029)
3330 Inset const * inset = getInset(pos);
3331 return inset && inset->lyxCode() == NEWLINE_CODE;
3335 bool Paragraph::isEnvSeparator(pos_type pos) const
3337 Inset const * inset = getInset(pos);
3338 return inset && inset->lyxCode() == SEPARATOR_CODE;
3342 bool Paragraph::isLineSeparator(pos_type pos) const
3344 char_type const c = d->text_[pos];
3345 if (isLineSeparatorChar(c))
3347 Inset const * inset = getInset(pos);
3348 return inset && inset->isLineSeparator();
3352 bool Paragraph::isWordSeparator(pos_type pos) const
3356 if (Inset const * inset = getInset(pos))
3357 return !inset->isLetter();
3358 // if we have a hard hyphen (no en- or emdash) or apostrophe
3359 // we pass this to the spell checker
3360 // FIXME: this method is subject to change, visit
3361 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3362 // to get an impression how complex this is.
3363 if (isHardHyphenOrApostrophe(pos))
3365 char_type const c = d->text_[pos];
3366 // We want to pass the escape chars to the spellchecker
3367 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3368 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3372 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3374 pos_type const psize = size();
3377 char_type const c = d->text_[pos];
3378 if (c != '-' && c != '\'')
3380 int nextpos = pos + 1;
3381 int prevpos = pos > 0 ? pos - 1 : 0;
3382 if ((nextpos == psize || isSpace(nextpos))
3383 && (pos == 0 || isSpace(prevpos)))
3389 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3391 return d->speller_state_.getRange(pos);
3395 bool Paragraph::isChar(pos_type pos) const
3397 if (Inset const * inset = getInset(pos))
3398 return inset->isChar();
3399 char_type const c = d->text_[pos];
3400 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3404 bool Paragraph::isSpace(pos_type pos) const
3406 if (Inset const * inset = getInset(pos))
3407 return inset->isSpace();
3408 char_type const c = d->text_[pos];
3409 return lyx::isSpace(c);
3414 Paragraph::getParLanguage(BufferParams const & bparams) const
3417 return getFirstFontSettings(bparams).language();
3418 // FIXME: we should check the prev par as well (Lgb)
3419 return bparams.language;
3423 bool Paragraph::isRTL(BufferParams const & bparams) const
3425 return getParLanguage(bparams)->rightToLeft()
3426 && !inInset().getLayout().forceLTR();
3430 void Paragraph::changeLanguage(BufferParams const & bparams,
3431 Language const * from, Language const * to)
3433 // change language including dummy font change at the end
3434 for (pos_type i = 0; i <= size(); ++i) {
3435 Font font = getFontSettings(bparams, i);
3436 if (font.language() == from) {
3437 font.setLanguage(to);
3439 d->requestSpellCheck(i);
3445 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3447 Language const * doc_language = bparams.language;
3448 FontList::const_iterator cit = d->fontlist_.begin();
3449 FontList::const_iterator end = d->fontlist_.end();
3451 for (; cit != end; ++cit)
3452 if (cit->font().language() != ignore_language &&
3453 cit->font().language() != latex_language &&
3454 cit->font().language() != doc_language)
3460 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3462 FontList::const_iterator cit = d->fontlist_.begin();
3463 FontList::const_iterator end = d->fontlist_.end();
3465 for (; cit != end; ++cit) {
3466 Language const * lang = cit->font().language();
3467 if (lang != ignore_language &&
3468 lang != latex_language)
3469 languages.insert(lang);
3474 docstring Paragraph::asString(int options) const
3476 return asString(0, size(), options);
3480 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3482 odocstringstream os;
3485 && options & AS_STR_LABEL
3486 && !d->params_.labelString().empty())
3487 os << d->params_.labelString() << ' ';
3489 for (pos_type i = beg; i < end; ++i) {
3490 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3492 char_type const c = d->text_[i];
3493 if (isPrintable(c) || c == '\t'
3494 || (c == '\n' && (options & AS_STR_NEWLINES)))
3496 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3497 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3498 LASSERT(runparams != 0, return docstring());
3499 getInset(i)->plaintext(os, *runparams);
3501 getInset(i)->toString(os);
3510 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3511 bool const shorten, bool const label) const
3513 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3514 if (label && !labelString().empty())
3515 os += labelString() + ' ';
3516 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3519 char_type const c = d->text_[i];
3522 else if (c == META_INSET)
3523 getInset(i)->forOutliner(os, tmplen, false);
3526 Text::shortenForOutliner(os, maxlen);
3530 void Paragraph::setInsetOwner(Inset const * inset)
3532 d->inset_owner_ = inset;
3536 int Paragraph::id() const
3542 void Paragraph::setId(int id)
3548 Layout const & Paragraph::layout() const
3554 void Paragraph::setLayout(Layout const & layout)
3556 d->layout_ = &layout;
3560 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3562 setLayout(tc.defaultLayout());
3566 void Paragraph::setPlainLayout(DocumentClass const & tc)
3568 setLayout(tc.plainLayout());
3572 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3574 if (usePlainLayout())
3575 setPlainLayout(tclass);
3577 setDefaultLayout(tclass);
3581 Inset const & Paragraph::inInset() const
3583 LBUFERR(d->inset_owner_);
3584 return *d->inset_owner_;
3588 ParagraphParameters & Paragraph::params()
3594 ParagraphParameters const & Paragraph::params() const
3600 bool Paragraph::isFreeSpacing() const
3602 if (d->layout_->free_spacing)
3604 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3608 bool Paragraph::allowEmpty() const
3610 if (d->layout_->keepempty)
3612 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3616 bool Paragraph::brokenBiblio() const
3618 // there is a problem if there is no bibitem at position 0 or
3619 // if there is another bibitem in the paragraph.
3620 return d->layout_->labeltype == LABEL_BIBLIO
3621 && (d->insetlist_.find(BIBITEM_CODE) != 0
3622 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3626 int Paragraph::fixBiblio(Buffer const & buffer)
3628 // FIXME: What about the case where paragraph is not BIBLIO
3629 // but there is an InsetBibitem?
3630 // FIXME: when there was already an inset at 0, the return value is 1,
3631 // which does not tell whether another inset has been remove; the
3632 // cursor cannot be correctly updated.
3634 if (d->layout_->labeltype != LABEL_BIBLIO)
3637 bool const track_changes = buffer.params().track_changes;
3638 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3639 bool const hasbibitem0 = bibitem_pos == 0;
3642 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3643 // There was an InsetBibitem at pos 0, and no other one => OK
3644 if (bibitem_pos == -1)
3646 // there is a bibitem at the 0 position, but since
3647 // there is a second one, we copy the second on the
3648 // first. We're assuming there are at most two of
3649 // these, which there should be.
3650 // FIXME: why does it make sense to do that rather
3651 // than keep the first? (JMarc)
3652 Inset * inset = releaseInset(bibitem_pos);
3653 d->insetlist_.begin()->inset = inset;
3654 return -bibitem_pos;
3657 // We need to create an inset at the beginning
3659 if (bibitem_pos > 0) {
3660 // there was one somewhere in the paragraph, let's move it
3661 inset = d->insetlist_.release(bibitem_pos);
3662 eraseChar(bibitem_pos, track_changes);
3665 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3666 InsetCommandParams(BIBITEM_CODE));
3668 Font font(inherit_font, buffer.params().language);
3669 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3670 : Change::UNCHANGED));
3676 void Paragraph::checkAuthors(AuthorList const & authorList)
3678 d->changes_.checkAuthors(authorList);
3682 bool Paragraph::isChanged(pos_type pos) const
3684 return lookupChange(pos).changed();
3688 bool Paragraph::isInserted(pos_type pos) const
3690 return lookupChange(pos).inserted();
3694 bool Paragraph::isDeleted(pos_type pos) const
3696 return lookupChange(pos).deleted();
3700 InsetList const & Paragraph::insetList() const
3702 return d->insetlist_;
3706 void Paragraph::setBuffer(Buffer & b)
3708 d->insetlist_.setBuffer(b);
3712 void Paragraph::resetBuffer()
3714 d->insetlist_.resetBuffer();
3718 Inset * Paragraph::releaseInset(pos_type pos)
3720 Inset * inset = d->insetlist_.release(pos);
3721 /// does not honour change tracking!
3722 eraseChar(pos, false);
3727 Inset * Paragraph::getInset(pos_type pos)
3729 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3730 ? d->insetlist_.get(pos) : 0;
3734 Inset const * Paragraph::getInset(pos_type pos) const
3736 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3737 ? d->insetlist_.get(pos) : 0;
3741 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3742 pos_type & right, TextCase action)
3744 // process sequences of modified characters; in change
3745 // tracking mode, this approach results in much better
3746 // usability than changing case on a char-by-char basis
3747 // We also need to track the current font, since font
3748 // changes within sequences can occur.
3749 vector<pair<char_type, Font> > changes;
3751 bool const trackChanges = bparams.track_changes;
3753 bool capitalize = true;
3755 for (; pos < right; ++pos) {
3756 char_type oldChar = d->text_[pos];
3757 char_type newChar = oldChar;
3759 // ignore insets and don't play with deleted text!
3760 if (oldChar != META_INSET && !isDeleted(pos)) {
3762 case text_lowercase:
3763 newChar = lowercase(oldChar);
3765 case text_capitalization:
3767 newChar = uppercase(oldChar);
3771 case text_uppercase:
3772 newChar = uppercase(oldChar);
3777 if (isWordSeparator(pos) || isDeleted(pos)) {
3778 // permit capitalization again
3782 if (oldChar != newChar) {
3783 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3784 if (pos != right - 1)
3786 // step behind the changing area
3790 int erasePos = pos - changes.size();
3791 for (size_t i = 0; i < changes.size(); i++) {
3792 insertChar(pos, changes[i].first,
3795 if (!eraseChar(erasePos, trackChanges)) {
3798 ++right; // expand selection
3806 int Paragraph::find(docstring const & str, bool cs, bool mw,
3807 pos_type start_pos, bool del) const
3809 pos_type pos = start_pos;
3810 int const strsize = str.length();
3812 pos_type const parsize = d->text_.size();
3813 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3814 // Ignore "invisible" letters such as ligature breaks
3815 // and hyphenation chars while searching
3816 while (pos < parsize - 1 && isInset(pos)) {
3817 odocstringstream os;
3818 getInset(pos)->toString(os);
3819 if (!getInset(pos)->isLetter() || !os.str().empty())
3823 if (cs && str[i] != d->text_[pos])
3825 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3827 if (!del && isDeleted(pos))
3834 // if necessary, check whether string matches word
3836 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3839 && !isWordSeparator(pos))
3843 return pos - start_pos;
3847 char_type Paragraph::getChar(pos_type pos) const
3849 return d->text_[pos];
3853 pos_type Paragraph::size() const
3855 return d->text_.size();
3859 bool Paragraph::empty() const
3861 return d->text_.empty();
3865 bool Paragraph::isInset(pos_type pos) const
3867 return d->text_[pos] == META_INSET;
3871 bool Paragraph::isSeparator(pos_type pos) const
3873 //FIXME: Are we sure this can be the only separator?
3874 return d->text_[pos] == ' ';
3878 void Paragraph::deregisterWords()
3880 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3881 Private::LangWordsMap::const_iterator ite = d->words_.end();
3882 for (; itl != ite; ++itl) {
3883 WordList & wl = theWordList(itl->first);
3884 Private::Words::const_iterator it = (itl->second).begin();
3885 Private::Words::const_iterator et = (itl->second).end();
3886 for (; it != et; ++it)
3893 void Paragraph::locateWord(pos_type & from, pos_type & to,
3894 word_location const loc) const
3897 case WHOLE_WORD_STRICT:
3898 if (from == 0 || from == size()
3899 || isWordSeparator(from)
3900 || isWordSeparator(from - 1)) {
3907 // If we are already at the beginning of a word, do nothing
3908 if (!from || isWordSeparator(from - 1))
3913 // always move the cursor to the beginning of previous word
3914 while (from && !isWordSeparator(from - 1))
3918 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3921 // no need to move the 'from' cursor
3925 while (to < size() && !isWordSeparator(to))
3930 void Paragraph::collectWords()
3932 for (pos_type pos = 0; pos < size(); ++pos) {
3933 if (isWordSeparator(pos))
3935 pos_type from = pos;
3936 locateWord(from, pos, WHOLE_WORD);
3937 // Work around MSVC warning: The statement
3938 // if (pos < from + lyxrc.completion_minlength)
3939 // triggers a signed vs. unsigned warning.
3940 // I don't know why this happens, it could be a MSVC bug, or
3941 // related to LLP64 (windows) vs. LP64 (unix) programming
3942 // model, or the C++ standard might be ambigous in the section
3943 // defining the "usual arithmetic conversions". However, using
3944 // a temporary variable is safe and works on all compilers.
3945 pos_type const endpos = from + lyxrc.completion_minlength;
3948 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3949 if (cit == d->fontlist_.end())
3951 Language const * lang = cit->font().language();
3952 docstring const word = asString(from, pos, AS_STR_NONE);
3953 d->words_[lang->lang()].insert(word);
3958 void Paragraph::registerWords()
3960 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3961 Private::LangWordsMap::const_iterator ite = d->words_.end();
3962 for (; itl != ite; ++itl) {
3963 WordList & wl = theWordList(itl->first);
3964 Private::Words::const_iterator it = (itl->second).begin();
3965 Private::Words::const_iterator et = (itl->second).end();
3966 for (; it != et; ++it)
3972 void Paragraph::updateWords()
3980 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3982 SkipPositionsIterator begin = skips.begin();
3983 SkipPositions::iterator end = skips.end();
3984 if (pos > 0 && begin < end) {
3986 if (end->last == pos - 1) {
3991 skips.insert(end, FontSpan(pos, pos));
3995 Language * Paragraph::Private::locateSpellRange(
3996 pos_type & from, pos_type & to,
3997 SkipPositions & skips) const
3999 // skip leading white space
4000 while (from < to && owner_->isWordSeparator(from))
4002 // don't check empty range
4005 // get current language
4006 Language * lang = getSpellLanguage(from);
4007 pos_type last = from;
4008 bool samelang = true;
4009 bool sameinset = true;
4010 while (last < to && samelang && sameinset) {
4011 // hop to end of word
4012 while (last < to && !owner_->isWordSeparator(last)) {
4013 if (owner_->getInset(last)) {
4014 appendSkipPosition(skips, last);
4015 } else if (owner_->isDeleted(last)) {
4016 appendSkipPosition(skips, last);
4020 // hop to next word while checking for insets
4021 while (sameinset && last < to && owner_->isWordSeparator(last)) {
4022 if (Inset const * inset = owner_->getInset(last))
4023 sameinset = inset->isChar() && inset->isLetter();
4024 if (sameinset && owner_->isDeleted(last)) {
4025 appendSkipPosition(skips, last);
4030 if (sameinset && last < to) {
4031 // now check for language change
4032 samelang = lang == getSpellLanguage(last);
4035 // if language change detected backstep is needed
4043 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
4046 const_cast<Language *>(owner_->getFontSettings(
4047 inset_owner_->buffer().params(), from).language());
4048 if (lang == inset_owner_->buffer().params().language
4049 && !lyxrc.spellchecker_alt_lang.empty()) {
4051 string const lang_variety =
4052 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
4053 lang->setCode(lang_code);
4054 lang->setVariety(lang_variety);
4060 void Paragraph::requestSpellCheck(pos_type pos)
4062 d->requestSpellCheck(pos);
4066 bool Paragraph::needsSpellCheck() const
4068 SpellChecker::ChangeNumber speller_change_number = 0;
4069 if (theSpellChecker())
4070 speller_change_number = theSpellChecker()->changeNumber();
4071 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
4072 d->speller_state_.needsCompleteRefresh(speller_change_number);
4074 return d->needsSpellCheck();
4078 bool Paragraph::Private::ignoreWord(docstring const & word) const
4080 // Ignore words with digits
4081 // FIXME: make this customizable
4082 // (note that some checkers ignore words with digits by default)
4083 docstring::const_iterator cit = word.begin();
4084 docstring::const_iterator const end = word.end();
4085 for (; cit != end; ++cit) {
4086 if (isNumber((*cit)))
4093 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
4094 WordLangTuple & wl, docstring_list & suggestions,
4095 bool do_suggestion, bool check_learned) const
4097 SpellChecker::Result result = SpellChecker::WORD_OK;
4098 SpellChecker * speller = theSpellChecker();
4102 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
4105 locateWord(from, to, WHOLE_WORD);
4106 if (from == to || from >= size())
4109 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4110 Language * lang = d->getSpellLanguage(from);
4112 wl = WordLangTuple(word, lang);
4117 if (needsSpellCheck() || check_learned) {
4119 if (!d->ignoreWord(word)) {
4120 bool const trailing_dot = to < size() && d->text_[to] == '.';
4121 result = speller->check(wl);
4122 if (SpellChecker::misspelled(result) && trailing_dot) {
4123 wl = WordLangTuple(word.append(from_ascii(".")), lang);
4124 result = speller->check(wl);
4125 if (!SpellChecker::misspelled(result)) {
4126 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
4128 from << ".." << to << "]");
4130 // spell check with dot appended failed too
4131 // restore original word/lang value
4132 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
4133 wl = WordLangTuple(word, lang);
4137 if (!SpellChecker::misspelled(result)) {
4138 // area up to the begin of the next word is not misspelled
4139 while (end < size() && isWordSeparator(end))
4142 d->setMisspelled(from, end, result);
4144 result = d->speller_state_.getState(from);
4148 suggestions.clear();
4150 if (SpellChecker::misspelled(result)) {
4151 LYXERR(Debug::GUI, "misspelled word: \"" <<
4153 from << ".." << to << "]");
4155 speller->suggest(wl, suggestions);
4161 void Paragraph::Private::markMisspelledWords(
4162 pos_type const & first, pos_type const & last,
4163 SpellChecker::Result result,
4164 docstring const & word,
4165 SkipPositions const & skips)
4167 if (!SpellChecker::misspelled(result)) {
4168 setMisspelled(first, last, SpellChecker::WORD_OK);
4172 SpellChecker * speller = theSpellChecker();
4173 // locate and enumerate the error positions
4174 int nerrors = speller->numMisspelledWords();
4176 SkipPositionsIterator it = skips.begin();
4177 SkipPositionsIterator et = skips.end();
4178 for (int index = 0; index < nerrors; ++index) {
4181 speller->misspelledWord(index, wstart, wlen);
4182 /// should not happen if speller supports range checks
4183 if (!wlen) continue;
4184 docstring const misspelled = word.substr(wstart, wlen);
4185 wstart += first + numskipped;
4186 if (snext < wstart) {
4187 /// mark the range of correct spelling
4188 numskipped += countSkips(it, et, wstart);
4189 setMisspelled(snext,
4190 wstart - 1, SpellChecker::WORD_OK);
4192 snext = wstart + wlen;
4193 numskipped += countSkips(it, et, snext);
4194 /// mark the range of misspelling
4195 setMisspelled(wstart, snext, result);
4196 LYXERR(Debug::GUI, "misspelled word: \"" <<
4197 misspelled << "\" [" <<
4198 wstart << ".." << (snext-1) << "]");
4201 if (snext <= last) {
4202 /// mark the range of correct spelling at end
4203 setMisspelled(snext, last, SpellChecker::WORD_OK);
4208 void Paragraph::spellCheck() const
4210 SpellChecker * speller = theSpellChecker();
4211 if (!speller || empty() ||!needsSpellCheck())
4215 d->rangeOfSpellCheck(start, endpos);
4216 if (speller->canCheckParagraph()) {
4217 // loop until we leave the range
4218 for (pos_type first = start; first < endpos; ) {
4219 pos_type last = endpos;
4220 Private::SkipPositions skips;
4221 Language * lang = d->locateSpellRange(first, last, skips);
4222 if (first >= endpos)
4224 // start the spell checker on the unit of meaning
4225 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
4226 WordLangTuple wl = WordLangTuple(word, lang);
4227 SpellChecker::Result result = word.size() ?
4228 speller->check(wl) : SpellChecker::WORD_OK;
4229 d->markMisspelledWords(first, last, result, word, skips);
4233 static docstring_list suggestions;
4234 pos_type to = endpos;
4235 while (start < endpos) {
4237 spellCheck(start, to, wl, suggestions, false);
4241 d->readySpellCheck();
4245 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4247 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4248 if (result || pos <= 0 || pos > size())
4250 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4251 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4256 string Paragraph::magicLabel() const
4259 ss << "magicparlabel-" << id();