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 "TextClass.h"
46 #include "WordLangTuple.h"
49 #include "frontends/alert.h"
51 #include "insets/InsetBibitem.h"
52 #include "insets/InsetLabel.h"
53 #include "insets/InsetSpecialChar.h"
55 #include "support/debug.h"
56 #include "support/docstring_list.h"
57 #include "support/ExceptionMessage.h"
58 #include "support/gettext.h"
59 #include "support/lassert.h"
60 #include "support/lstrings.h"
61 #include "support/textutils.h"
67 using namespace lyx::support;
73 /// Inset identifier (above 0x10ffff, for ucs-4)
74 char_type const META_INSET = 0x200001;
79 /////////////////////////////////////////////////////////////////////
83 /////////////////////////////////////////////////////////////////////
85 class SpellResultRange {
87 SpellResultRange(FontSpan range, SpellChecker::Result result)
88 : range_(range), result_(result)
91 FontSpan const & range() const { return range_; }
93 void range(FontSpan const & r) { range_ = r; }
95 SpellChecker::Result result() const { return result_; }
97 void result(SpellChecker::Result r) { result_ = r; }
99 bool contains(pos_type pos) const { return range_.contains(pos); }
101 bool covered(FontSpan const & r) const
103 // 1. first of new range inside current range or
104 // 2. last of new range inside current range or
105 // 3. first of current range inside new range or
106 // 4. last of current range inside new range
107 //FIXME: is this the same as !range_.intersect(r).empty() ?
108 return range_.contains(r.first) || range_.contains(r.last) ||
109 r.contains(range_.first) || r.contains(range_.last);
112 void shift(pos_type pos, int offset)
114 if (range_.first > pos) {
115 range_.first += offset;
116 range_.last += offset;
117 } else if (range_.last >= pos) {
118 range_.last += offset;
123 SpellChecker::Result result_ ;
127 /////////////////////////////////////////////////////////////////////
131 /////////////////////////////////////////////////////////////////////
133 class SpellCheckerState {
137 needs_refresh_ = true;
138 current_change_number_ = 0;
141 void setRange(FontSpan const & fp, SpellChecker::Result state)
144 RangesIterator et = ranges_.end();
145 RangesIterator it = ranges_.begin();
146 for (; it != et; ++it) {
147 if (!it->covered(fp))
148 result.push_back(SpellResultRange(it->range(), it->result()));
149 else if (state == SpellChecker::WORD_OK) {
150 // trim or split the current misspelled range
151 // store misspelled ranges only
152 FontSpan range = it->range();
153 if (fp.first > range.first) {
154 // misspelled area in front of WORD_OK
155 range.last = fp.first - 1;
156 result.push_back(SpellResultRange(range, it->result()));
159 if (fp.last < range.last) {
160 // misspelled area after WORD_OK range
161 range.first = fp.last + 1;
162 result.push_back(SpellResultRange(range, it->result()));
167 if (state != SpellChecker::WORD_OK)
168 ranges_.push_back(SpellResultRange(fp, state));
171 void increasePosAfterPos(pos_type pos)
173 correctRangesAfterPos(pos, 1);
177 void decreasePosAfterPos(pos_type pos)
179 correctRangesAfterPos(pos, -1);
183 void refreshLast(pos_type pos)
185 if (pos < refresh_.last)
189 SpellChecker::Result getState(pos_type pos) const
191 SpellChecker::Result result = SpellChecker::WORD_OK;
192 RangesIterator et = ranges_.end();
193 RangesIterator it = ranges_.begin();
194 for (; it != et; ++it) {
195 if(it->contains(pos)) {
202 FontSpan const & getRange(pos_type pos) const
204 /// empty span to indicate mismatch
205 static FontSpan empty_;
206 RangesIterator et = ranges_.end();
207 RangesIterator it = ranges_.begin();
208 for (; it != et; ++it) {
209 if(it->contains(pos)) {
216 bool needsRefresh() const
218 return needs_refresh_;
221 SpellChecker::ChangeNumber currentChangeNumber() const
223 return current_change_number_;
226 void refreshRange(pos_type & first, pos_type & last) const
228 first = refresh_.first;
229 last = refresh_.last;
232 void needsRefresh(pos_type pos)
234 if (needs_refresh_ && pos != -1) {
235 if (pos < refresh_.first)
236 refresh_.first = pos;
237 if (pos > refresh_.last)
239 } else if (pos != -1) {
240 // init request check for neighbour positions too
241 refresh_.first = pos > 0 ? pos - 1 : 0;
242 // no need for special end of paragraph check
243 refresh_.last = pos + 1;
245 needs_refresh_ = pos != -1;
248 void needsCompleteRefresh(SpellChecker::ChangeNumber change_number)
250 needs_refresh_ = true;
253 current_change_number_ = change_number;
256 typedef vector<SpellResultRange> Ranges;
257 typedef Ranges::const_iterator RangesIterator;
259 /// the area of the paragraph with pending spell check
262 /// spell state cache version number
263 SpellChecker::ChangeNumber current_change_number_;
266 void correctRangesAfterPos(pos_type pos, int offset)
268 RangesIterator et = ranges_.end();
269 Ranges::iterator it = ranges_.begin();
270 for (; it != et; ++it) {
271 it->shift(pos, offset);
277 /////////////////////////////////////////////////////////////////////
279 // Paragraph::Private
281 /////////////////////////////////////////////////////////////////////
283 class Paragraph::Private
285 // Enforce our own "copy" constructor by declaring the standard one and
286 // the assignment operator private without implementing them.
287 Private(Private const &);
288 Private & operator=(Private const &);
291 Private(Paragraph * owner, Layout const & layout);
292 /// "Copy constructor"
293 Private(Private const &, Paragraph * owner);
294 /// Copy constructor from \p beg to \p end
295 Private(Private const &, Paragraph * owner, pos_type beg, pos_type end);
298 void insertChar(pos_type pos, char_type c, Change const & change);
300 /// Output the surrogate pair formed by \p c and \p next to \p os.
301 /// \return the number of characters written.
302 int latexSurrogatePair(otexstream & os, char_type c, char_type next,
303 OutputParams const &);
305 /// Output a space in appropriate formatting (or a surrogate pair
306 /// if the next character is a combining character).
307 /// \return whether a surrogate pair was output.
308 bool simpleTeXBlanks(OutputParams const &,
311 unsigned int & column,
313 Layout const & style);
315 /// Output consecutive unicode chars, belonging to the same script as
316 /// specified by the latex macro \p ltx, to \p os starting from \p i.
317 /// \return the number of characters written.
318 int writeScriptChars(otexstream & os, docstring const & ltx,
319 Change const &, Encoding const &, pos_type & i);
321 /// This could go to ParagraphParameters if we want to.
322 int startTeXParParams(BufferParams const &, otexstream &,
323 OutputParams const &) const;
325 /// This could go to ParagraphParameters if we want to.
326 bool endTeXParParams(BufferParams const &, otexstream &,
327 OutputParams const &) const;
330 void latexInset(BufferParams const &,
335 Font const & outerfont,
337 Change & running_change,
338 Layout const & style,
340 unsigned int & column);
343 void latexSpecialChar(
345 BufferParams const & bparams,
346 OutputParams const & runparams,
347 Font const & running_font,
348 Change const & running_change,
349 Layout const & style,
352 unsigned int & column);
359 unsigned int & column);
365 unsigned int & column);
368 void validate(LaTeXFeatures & features) const;
370 /// Checks if the paragraph contains only text and no inset or font change.
371 bool onlyText(Buffer const & buf, Font const & outerfont,
372 pos_type initial) const;
374 /// a vector of speller skip positions
375 typedef vector<FontSpan> SkipPositions;
376 typedef SkipPositions::const_iterator SkipPositionsIterator;
378 void appendSkipPosition(SkipPositions & skips, pos_type const pos) const;
380 Language * getSpellLanguage(pos_type const from) const;
382 Language * locateSpellRange(pos_type & from, pos_type & to,
383 SkipPositions & skips) const;
385 bool hasSpellerChange() const
387 SpellChecker::ChangeNumber speller_change_number = 0;
388 if (theSpellChecker())
389 speller_change_number = theSpellChecker()->changeNumber();
390 return speller_change_number > speller_state_.currentChangeNumber();
393 bool ignoreWord(docstring const & word) const ;
395 void setMisspelled(pos_type from, pos_type to, SpellChecker::Result state)
397 pos_type textsize = owner_->size();
398 // check for sane arguments
399 if (to <= from || from >= textsize)
401 FontSpan fp = FontSpan(from, to - 1);
402 speller_state_.setRange(fp, state);
405 void requestSpellCheck(pos_type pos)
408 speller_state_.needsCompleteRefresh(speller_state_.currentChangeNumber());
410 speller_state_.needsRefresh(pos);
413 void readySpellCheck()
415 speller_state_.needsRefresh(-1);
418 bool needsSpellCheck() const
420 return speller_state_.needsRefresh();
423 void rangeOfSpellCheck(pos_type & first, pos_type & last) const
425 speller_state_.refreshRange(first, last);
427 last = owner_->size();
430 pos_type endpos = last;
431 owner_->locateWord(first, endpos, WHOLE_WORD);
434 owner_->locateWord(last, endpos, WHOLE_WORD);
439 int countSkips(SkipPositionsIterator & it, SkipPositionsIterator const et,
443 while (it != et && it->first < start) {
444 int skip = it->last - it->first + 1;
452 void markMisspelledWords(pos_type const & first, pos_type const & last,
453 SpellChecker::Result result,
454 docstring const & word,
455 SkipPositions const & skips);
457 InsetCode ownerCode() const
459 return inset_owner_ ? inset_owner_->lyxCode() : NO_CODE;
462 /// Which Paragraph owns us?
466 Inset const * inset_owner_;
475 ParagraphParameters params_;
477 /// for recording and looking up changes
481 InsetList insetlist_;
484 pos_type begin_of_body_;
486 typedef docstring TextContainer;
490 typedef set<docstring> Words;
491 typedef map<string, Words> LangWordsMap;
495 Layout const * layout_;
497 SpellCheckerState speller_state_;
501 Paragraph::Private::Private(Paragraph * owner, Layout const & layout)
502 : owner_(owner), inset_owner_(0), id_(-1), begin_of_body_(0), layout_(&layout)
508 // Initialization of the counter for the paragraph id's,
510 // FIXME: There should be a more intelligent way to generate and use the
511 // paragraph ids per buffer instead a global static counter for all InsetText
512 // in the running program.
513 // However, this per-session id is used in LFUN_PARAGRAPH_GOTO to
514 // switch to a different buffer, as used in the outliner for instance.
515 static int paragraph_id = -1;
517 Paragraph::Private::Private(Private const & p, Paragraph * owner)
518 : owner_(owner), inset_owner_(p.inset_owner_), fontlist_(p.fontlist_),
519 params_(p.params_), changes_(p.changes_), insetlist_(p.insetlist_),
520 begin_of_body_(p.begin_of_body_), text_(p.text_), words_(p.words_),
523 id_ = ++paragraph_id;
524 requestSpellCheck(p.text_.size());
528 Paragraph::Private::Private(Private const & p, Paragraph * owner,
529 pos_type beg, pos_type end)
530 : owner_(owner), inset_owner_(p.inset_owner_),
531 params_(p.params_), changes_(p.changes_),
532 insetlist_(p.insetlist_, beg, end),
533 begin_of_body_(p.begin_of_body_), words_(p.words_),
536 id_ = ++paragraph_id;
537 if (beg >= pos_type(p.text_.size()))
539 text_ = p.text_.substr(beg, end - beg);
541 FontList::const_iterator fcit = fontlist_.begin();
542 FontList::const_iterator fend = fontlist_.end();
543 for (; fcit != fend; ++fcit) {
544 if (fcit->pos() < beg)
546 if (fcit->pos() >= end) {
547 // Add last entry in the fontlist_.
548 fontlist_.set(text_.size() - 1, fcit->font());
551 // Add a new entry in the fontlist_.
552 fontlist_.set(fcit->pos() - beg, fcit->font());
554 requestSpellCheck(p.text_.size());
558 void Paragraph::addChangesToToc(DocIterator const & cdit,
559 Buffer const & buf, bool output_active) const
561 d->changes_.addToToc(cdit, buf, output_active);
565 void Paragraph::addChangesToBuffer(Buffer const & buf) const
567 d->changes_.updateBuffer(buf);
571 bool Paragraph::isChangeUpdateRequired() const
573 return d->changes_.isUpdateRequired();
577 bool Paragraph::isDeleted(pos_type start, pos_type end) const
579 LASSERT(start >= 0 && start <= size(), return false);
580 LASSERT(end > start && end <= size() + 1, return false);
582 return d->changes_.isDeleted(start, end);
586 bool Paragraph::isChanged(pos_type start, pos_type end) const
588 LASSERT(start >= 0 && start <= size(), return false);
589 LASSERT(end > start && end <= size() + 1, return false);
591 return d->changes_.isChanged(start, end);
595 bool Paragraph::isMergedOnEndOfParDeletion(bool trackChanges) const
597 // keep the logic here in sync with the logic of eraseChars()
601 Change const & change = d->changes_.lookup(size());
602 return change.inserted() && change.currentAuthor();
606 void Paragraph::setChange(Change const & change)
608 // beware of the imaginary end-of-par character!
609 d->changes_.set(change, 0, size() + 1);
612 * Propagate the change recursively - but not in case of DELETED!
614 * Imagine that your co-author makes changes in an existing inset. He
615 * sends your document to you and you come to the conclusion that the
616 * inset should go completely. If you erase it, LyX must not delete all
617 * text within the inset. Otherwise, the change tracked insertions of
618 * your co-author get lost and there is no way to restore them later.
620 * Conclusion: An inset's content should remain untouched if you delete it
623 if (!change.deleted()) {
624 for (pos_type pos = 0; pos < size(); ++pos) {
625 if (Inset * inset = getInset(pos))
626 inset->setChange(change);
632 void Paragraph::setChange(pos_type pos, Change const & change)
634 LASSERT(pos >= 0 && pos <= size(), return);
635 d->changes_.set(change, pos);
637 // see comment in setChange(Change const &) above
638 if (!change.deleted() && pos < size())
639 if (Inset * inset = getInset(pos))
640 inset->setChange(change);
644 Change const & Paragraph::lookupChange(pos_type pos) const
646 LBUFERR(pos >= 0 && pos <= size());
647 return d->changes_.lookup(pos);
651 void Paragraph::acceptChanges(pos_type start, pos_type end)
653 LASSERT(start >= 0 && start <= size(), return);
654 LASSERT(end > start && end <= size() + 1, return);
656 for (pos_type pos = start; pos < end; ++pos) {
657 switch (lookupChange(pos).type) {
658 case Change::UNCHANGED:
659 // accept changes in nested inset
660 if (Inset * inset = getInset(pos))
661 inset->acceptChanges();
664 case Change::INSERTED:
665 d->changes_.set(Change(Change::UNCHANGED), pos);
666 // also accept changes in nested inset
667 if (Inset * inset = getInset(pos))
668 inset->acceptChanges();
671 case Change::DELETED:
672 // Suppress access to non-existent
673 // "end-of-paragraph char"
675 eraseChar(pos, false);
686 void Paragraph::rejectChanges(pos_type start, pos_type end)
688 LASSERT(start >= 0 && start <= size(), return);
689 LASSERT(end > start && end <= size() + 1, return);
691 for (pos_type pos = start; pos < end; ++pos) {
692 switch (lookupChange(pos).type) {
693 case Change::UNCHANGED:
694 // reject changes in nested inset
695 if (Inset * inset = getInset(pos))
696 inset->rejectChanges();
699 case Change::INSERTED:
700 // Suppress access to non-existent
701 // "end-of-paragraph char"
703 eraseChar(pos, false);
709 case Change::DELETED:
710 d->changes_.set(Change(Change::UNCHANGED), pos);
712 // Do NOT reject changes within a deleted inset!
713 // There may be insertions of a co-author inside of it!
721 void Paragraph::Private::insertChar(pos_type pos, char_type c,
722 Change const & change)
724 LASSERT(pos >= 0 && pos <= int(text_.size()), return);
727 changes_.insert(change, pos);
729 // This is actually very common when parsing buffers (and
730 // maybe inserting ascii text)
731 if (pos == pos_type(text_.size())) {
732 // when appending characters, no need to update tables
734 // but we want spell checking
735 requestSpellCheck(pos);
739 text_.insert(text_.begin() + pos, c);
741 // Update the font table.
742 fontlist_.increasePosAfterPos(pos);
745 insetlist_.increasePosAfterPos(pos);
747 // Update list of misspelled positions
748 speller_state_.increasePosAfterPos(pos);
752 bool Paragraph::insertInset(pos_type pos, Inset * inset,
753 Font const & font, Change const & change)
755 LASSERT(inset, return false);
756 LASSERT(pos >= 0 && pos <= size(), return false);
758 // Paragraph::insertInset() can be used in cut/copy/paste operation where
759 // d->inset_owner_ is not set yet.
760 if (d->inset_owner_ && !d->inset_owner_->insetAllowed(inset->lyxCode()))
763 d->insertChar(pos, META_INSET, change);
764 LASSERT(d->text_[pos] == META_INSET, return false);
766 // Add a new entry in the insetlist_.
767 d->insetlist_.insert(inset, pos);
769 // Some insets require run of spell checker
770 requestSpellCheck(pos);
776 bool Paragraph::eraseChar(pos_type pos, bool trackChanges)
778 LASSERT(pos >= 0 && pos <= size(), return false);
780 // keep the logic here in sync with the logic of isMergedOnEndOfParDeletion()
783 Change change = d->changes_.lookup(pos);
785 // set the character to DELETED if
786 // a) it was previously unchanged or
787 // b) it was inserted by a co-author
789 if (!change.changed() ||
790 (change.inserted() && !change.currentAuthor())) {
791 setChange(pos, Change(Change::DELETED));
792 // request run of spell checker
793 requestSpellCheck(pos);
797 if (change.deleted())
801 // Don't physically access the imaginary end-of-paragraph character.
802 // eraseChar() can only mark it as DELETED. A physical deletion of
803 // end-of-par must be handled externally.
809 d->changes_.erase(pos);
811 // if it is an inset, delete the inset entry
812 if (d->text_[pos] == META_INSET)
813 d->insetlist_.erase(pos);
815 d->text_.erase(d->text_.begin() + pos);
817 // Update the fontlist_
818 d->fontlist_.erase(pos);
820 // Update the insetlist_
821 d->insetlist_.decreasePosAfterPos(pos);
823 // Update list of misspelled positions
824 d->speller_state_.decreasePosAfterPos(pos);
825 d->speller_state_.refreshLast(size());
831 int Paragraph::eraseChars(pos_type start, pos_type end, bool trackChanges)
833 LASSERT(start >= 0 && start <= size(), return 0);
834 LASSERT(end >= start && end <= size() + 1, return 0);
837 for (pos_type count = end - start; count; --count) {
838 if (!eraseChar(i, trackChanges))
845 int Paragraph::Private::latexSurrogatePair(otexstream & os, char_type c,
846 char_type next, OutputParams const & runparams)
848 // Writing next here may circumvent a possible font change between
849 // c and next. Since next is only output if it forms a surrogate pair
850 // with c we can ignore this:
851 // A font change inside a surrogate pair does not make sense and is
852 // hopefully impossible to input.
853 // FIXME: change tracking
854 // Is this correct WRT change tracking?
855 Encoding const & encoding = *(runparams.encoding);
856 docstring latex1 = encoding.latexChar(next).first;
857 if (runparams.inIPA) {
858 string const tipashortcut = Encodings::TIPAShortcut(next);
859 if (!tipashortcut.empty()) {
860 latex1 = from_ascii(tipashortcut);
863 docstring const latex2 = encoding.latexChar(c).first;
864 if (docstring(1, next) == latex1) {
865 // the encoding supports the combination
866 os << latex2 << latex1;
867 return latex1.length() + latex2.length();
868 } else if (runparams.local_font &&
869 runparams.local_font->language()->lang() == "polutonikogreek") {
870 // polutonikogreek only works without the brackets
871 os << latex1 << latex2;
872 return latex1.length() + latex2.length();
874 os << latex1 << '{' << latex2 << '}';
875 return latex1.length() + latex2.length() + 2;
879 bool Paragraph::Private::simpleTeXBlanks(OutputParams const & runparams,
882 unsigned int & column,
884 Layout const & style)
886 if (style.pass_thru || runparams.pass_thru)
889 if (i + 1 < int(text_.size())) {
890 char_type next = text_[i + 1];
891 if (Encodings::isCombiningChar(next)) {
892 // This space has an accent, so we must always output it.
893 column += latexSurrogatePair(os, ' ', next, runparams) - 1;
898 if (runparams.linelen > 0
899 && column > runparams.linelen
901 && text_[i - 1] != ' '
902 && (i + 1 < int(text_.size()))
903 // same in FreeSpacing mode
904 && !owner_->isFreeSpacing()
905 // In typewriter mode, we want to avoid
906 // ! . ? : at the end of a line
907 && !(font.fontInfo().family() == TYPEWRITER_FAMILY
908 && (text_[i - 1] == '.'
909 || text_[i - 1] == '?'
910 || text_[i - 1] == ':'
911 || text_[i - 1] == '!'))) {
913 os.texrow().start(owner_->id(), i + 1);
915 } else if (style.free_spacing) {
924 int Paragraph::Private::writeScriptChars(otexstream & os,
925 docstring const & ltx,
926 Change const & runningChange,
927 Encoding const & encoding,
930 // FIXME: modifying i here is not very nice...
932 // We only arrive here when character text_[i] could not be translated
933 // into the current latex encoding (or its latex translation has been forced,)
934 // and it belongs to a known script.
935 // TODO: We need \textcyr and \textgreek wrappers also for characters
936 // that can be encoded in the "LaTeX encoding" but not in the
937 // current *font encoding*.
938 // (See #9681 for details and test)
939 // Parameter ltx contains the latex translation of text_[i] as specified
940 // in the unicodesymbols file and is something like "\textXXX{<spec>}".
941 // The latex macro name "textXXX" specifies the script to which text_[i]
942 // belongs and we use it in order to check whether characters from the
943 // same script immediately follow, such that we can collect them in a
944 // single "\textXXX" macro. So, we have to retain "\textXXX{<spec>"
945 // for the first char but only "<spec>" for all subsequent chars.
946 docstring::size_type const brace1 = ltx.find_first_of(from_ascii("{"));
947 docstring::size_type const brace2 = ltx.find_last_of(from_ascii("}"));
948 string script = to_ascii(ltx.substr(1, brace1 - 1));
951 bool closing_brace = true;
952 if (script == "textgreek" && encoding.latexName() == "iso-8859-7") {
953 // Correct encoding is being used, so we can avoid \textgreek.
954 // TODO: wrong test: we need to check the *font encoding*
955 // (i.e. the active language and its FontEncoding tag)
956 // instead of the LaTeX *input encoding*!
957 // See #9637 for details and test-cases.
960 closing_brace = false;
962 os << ltx.substr(pos, length);
963 int size = text_.size();
964 while (i + 1 < size) {
965 char_type const next = text_[i + 1];
966 // Stop here if next character belongs to another script
967 // or there is a change in change tracking status.
968 if (!Encodings::isKnownScriptChar(next, script) ||
969 runningChange != owner_->lookupChange(i + 1))
973 FontList::const_iterator cit = fontlist_.begin();
974 FontList::const_iterator end = fontlist_.end();
975 for (; cit != end; ++cit) {
976 if (cit->pos() >= i && !found) {
977 prev_font = cit->font();
980 if (cit->pos() >= i + 1)
983 // Stop here if there is a font attribute or encoding change.
984 if (found && cit != end && prev_font != cit->font())
986 docstring const latex = encoding.latexChar(next).first;
987 docstring::size_type const b1 =
988 latex.find_first_of(from_ascii("{"));
989 docstring::size_type const b2 =
990 latex.find_last_of(from_ascii("}"));
991 int const len = b2 - b1 - 1;
992 os << latex.substr(b1 + 1, len);
1004 void Paragraph::Private::latexInset(BufferParams const & bparams,
1006 OutputParams & runparams,
1007 Font & running_font,
1009 Font const & outerfont,
1011 Change & running_change,
1012 Layout const & style,
1014 unsigned int & column)
1016 Inset * inset = owner_->getInset(i);
1019 if (style.pass_thru) {
1020 odocstringstream ods;
1021 inset->plaintext(ods, runparams);
1026 // FIXME: move this to InsetNewline::latex
1027 if (inset->lyxCode() == NEWLINE_CODE || inset->lyxCode() == SEPARATOR_CODE) {
1028 // newlines are handled differently here than
1029 // the default in simpleTeXSpecialChars().
1030 if (!style.newline_allowed) {
1034 column += running_font.latexWriteEndChanges(
1035 os, bparams, runparams,
1036 basefont, basefont);
1040 if (running_font.fontInfo().family() == TYPEWRITER_FAMILY)
1043 basefont = owner_->getLayoutFont(bparams, outerfont);
1044 running_font = basefont;
1046 if (runparams.moving_arg)
1050 os.texrow().start(owner_->id(), i + 1);
1054 if (owner_->isDeleted(i)) {
1055 if( ++runparams.inDeletedInset == 1)
1056 runparams.changeOfDeletedInset = owner_->lookupChange(i);
1059 if (inset->canTrackChanges()) {
1060 column += Changes::latexMarkChange(os, bparams, running_change,
1061 Change(Change::UNCHANGED), runparams);
1062 running_change = Change(Change::UNCHANGED);
1066 odocstream::pos_type const len = os.os().tellp();
1068 if (inset->forceLTR()
1069 && !runparams.use_polyglossia
1070 && running_font.isRightToLeft()
1071 // ERT is an exception, it should be output with no
1072 // decorations at all
1073 && inset->lyxCode() != ERT_CODE) {
1074 if (running_font.language()->lang() == "farsi")
1081 // FIXME: Bug: we can have an empty font change here!
1082 // if there has just been a font change, we are going to close it
1083 // right now, which means stupid latex code like \textsf{}. AFAIK,
1084 // this does not harm dvi output. A minor bug, thus (JMarc)
1086 // Some insets cannot be inside a font change command.
1087 // However, even such insets *can* be placed in \L or \R
1088 // or their equivalents (for RTL language switches), so we don't
1089 // close the language in those cases.
1090 // ArabTeX, though, cannot handle this special behavior, it seems.
1091 bool arabtex = basefont.language()->lang() == "arabic_arabtex"
1092 || running_font.language()->lang() == "arabic_arabtex";
1093 if (open_font && !inset->inheritFont()) {
1094 bool closeLanguage = arabtex
1095 || basefont.isRightToLeft() == running_font.isRightToLeft();
1096 unsigned int count = running_font.latexWriteEndChanges(os,
1097 bparams, runparams, basefont, basefont, closeLanguage);
1099 // if any font properties were closed, update the running_font,
1100 // making sure, however, to leave the language as it was
1102 // FIXME: probably a better way to keep track of the old
1103 // language, than copying the entire font?
1104 Font const copy_font(running_font);
1105 basefont = owner_->getLayoutFont(bparams, outerfont);
1106 running_font = basefont;
1108 running_font.setLanguage(copy_font.language());
1109 // leave font open if language is still open
1110 open_font = (running_font.language() == basefont.language());
1112 runparams.local_font = &basefont;
1116 int prev_rows = os.texrow().rows();
1119 runparams.lastid = id_;
1120 runparams.lastpos = i;
1121 inset->latex(os, runparams);
1122 } catch (EncodingException & e) {
1123 // add location information and throw again.
1130 if (running_font.language()->lang() == "farsi")
1136 if (os.texrow().rows() > prev_rows) {
1137 os.texrow().start(owner_->id(), i + 1);
1140 column += (unsigned int)(os.os().tellp() - len);
1143 if (owner_->isDeleted(i))
1144 --runparams.inDeletedInset;
1148 void Paragraph::Private::latexSpecialChar(otexstream & os,
1149 BufferParams const & bparams,
1150 OutputParams const & runparams,
1151 Font const & running_font,
1152 Change const & running_change,
1153 Layout const & style,
1156 unsigned int & column)
1158 // With polyglossia, brackets and stuff need not be reversed
1159 // in RTL scripts (see bug #8251)
1160 char_type const c = (runparams.use_polyglossia) ?
1161 owner_->getUChar(bparams, i) : text_[i];
1163 if (style.pass_thru || runparams.pass_thru
1164 || contains(style.pass_thru_chars, c)
1165 || contains(runparams.pass_thru_chars, c)) {
1167 Encoding const * const enc = runparams.encoding;
1168 if (enc && !enc->encodable(c))
1169 throw EncodingException(c);
1175 // TIPA uses its own T3 encoding
1176 if (runparams.inIPA && latexSpecialT3(c, os, i, column))
1178 // If T1 font encoding is used, use the special
1179 // characters it provides.
1180 // NOTE: Some languages reset the font encoding internally to a
1181 // non-standard font encoding. If we are using such a language,
1182 // we do not output special T1 chars.
1183 if (!runparams.inIPA && !running_font.language()->internalFontEncoding()
1184 && bparams.font_encoding() == "T1" && latexSpecialT1(c, os, i, column))
1187 // Otherwise, we use what LaTeX provides us.
1190 os << "\\textbackslash{}";
1194 os << "\\textless{}";
1198 os << "\\textgreater{}";
1202 os << "\\textbar{}";
1207 if (i + 1 < static_cast<pos_type>(text_.size()) &&
1208 (end_pos == -1 || i + 1 < end_pos) &&
1209 text_[i+1] == '-') {
1210 // Prevent "--" becoming an endash and "---" becoming
1212 // Within \ttfamily, "--" is merged to "-" (no endash)
1213 // so we avoid this rather irritating ligature as well
1219 os << "\\char`\\\"{}";
1224 case '%': case '#': case '{':
1232 os << "\\textasciitilde{}";
1237 os << "\\textasciicircum{}";
1244 // avoid being mistaken for optional arguments
1252 // Blanks are printed before font switching.
1253 // Sure? I am not! (try nice-latex)
1254 // I am sure it's correct. LyX might be smarter
1255 // in the future, but for now, nothing wrong is
1263 Encoding const & encoding = *(runparams.encoding);
1264 char_type next = '\0';
1265 if (i + 1 < int(text_.size())) {
1266 next = text_[i + 1];
1267 if (Encodings::isCombiningChar(next)) {
1268 column += latexSurrogatePair(os, c, next, runparams) - 1;
1274 pair<docstring, bool> latex = encoding.latexChar(c);
1275 docstring nextlatex;
1276 bool nexttipas = false;
1277 string nexttipashortcut;
1278 if (next != '\0' && next != META_INSET && encoding.encodable(next)) {
1279 nextlatex = encoding.latexChar(next).first;
1280 if (runparams.inIPA) {
1281 nexttipashortcut = Encodings::TIPAShortcut(next);
1282 nexttipas = !nexttipashortcut.empty();
1286 if (runparams.inIPA) {
1287 string const tipashortcut = Encodings::TIPAShortcut(c);
1288 if (!tipashortcut.empty()) {
1289 latex.first = from_ascii(tipashortcut);
1290 latex.second = false;
1294 if (Encodings::isKnownScriptChar(c, script)
1295 && prefixIs(latex.first, from_ascii("\\" + script)))
1296 column += writeScriptChars(os, latex.first,
1297 running_change, encoding, i) - 1;
1298 else if (latex.second
1299 && ((!prefixIs(nextlatex, '\\')
1300 && !prefixIs(nextlatex, '{')
1301 && !prefixIs(nextlatex, '}'))
1303 && !prefixIs(from_ascii(nexttipashortcut), '\\')))
1305 // Prevent eating of a following
1306 // space or command corruption by
1307 // following characters
1308 if (next == ' ' || next == '\0') {
1309 column += latex.first.length() + 1;
1310 os << latex.first << "{}";
1312 column += latex.first.length();
1313 os << latex.first << " ";
1316 column += latex.first.length() - 1;
1324 bool Paragraph::Private::latexSpecialT1(char_type const c, otexstream & os,
1325 pos_type i, unsigned int & column)
1331 // In T1 encoding, these characters exist
1332 // but we should avoid ligatures
1333 if (i + 1 >= int(text_.size()) || text_[i + 1] != c)
1335 os << "\\textcompwordmark{}";
1342 // soul.sty breaks with \char`\"
1343 os << "\\textquotedbl{}";
1352 bool Paragraph::Private::latexSpecialT3(char_type const c, otexstream & os,
1353 pos_type /*i*/, unsigned int & column)
1363 os << "\\textvertline{}";
1372 void Paragraph::Private::validate(LaTeXFeatures & features) const
1374 if (layout_->inpreamble && inset_owner_) {
1375 bool const is_command = layout_->latextype == LATEX_COMMAND;
1376 Buffer const & buf = inset_owner_->buffer();
1377 BufferParams const & bp = features.runparams().is_child
1378 ? buf.masterParams() : buf.params();
1381 // Using a string stream here circumvents the encoding
1382 // switching machinery of odocstream. Therefore the
1383 // output is wrong if this paragraph contains content
1384 // that needs to switch encoding.
1385 odocstringstream ods;
1386 otexstream os(ods, texrow);
1388 os << '\\' << from_ascii(layout_->latexname());
1389 // we have to provide all the optional arguments here, even though
1390 // the last one is the only one we care about.
1391 // Separate handling of optional argument inset.
1392 if (!layout_->latexargs().empty()) {
1393 OutputParams rp = features.runparams();
1394 rp.local_font = &owner_->getFirstFontSettings(bp);
1395 latexArgInsets(*owner_, os, rp, layout_->latexargs());
1397 os << from_ascii(layout_->latexparam());
1399 docstring::size_type const length = ods.str().length();
1400 // this will output "{" at the beginning, but not at the end
1401 owner_->latex(bp, f, os, features.runparams(), 0, -1, true);
1402 if (ods.str().length() > length) {
1405 if (!layout_->postcommandargs().empty()) {
1406 OutputParams rp = features.runparams();
1407 rp.local_font = &owner_->getFirstFontSettings(bp);
1408 latexArgInsets(*owner_, os, rp, layout_->postcommandargs(), "post:");
1411 string const snippet = to_utf8(ods.str());
1412 features.addPreambleSnippet(snippet, true);
1416 if (features.runparams().flavor == OutputParams::HTML
1417 && layout_->htmltitle()) {
1418 features.setHTMLTitle(owner_->asString(AS_STR_INSETS | AS_STR_SKIPDELETE));
1421 // check the params.
1422 if (!params_.spacing().isDefault())
1423 features.require("setspace");
1426 features.useLayout(layout_->name());
1429 fontlist_.validate(features);
1431 // then the indentation
1432 if (!params_.leftIndent().zero())
1433 features.require("ParagraphLeftIndent");
1436 InsetList::const_iterator icit = insetlist_.begin();
1437 InsetList::const_iterator iend = insetlist_.end();
1438 for (; icit != iend; ++icit) {
1440 icit->inset->validate(features);
1441 if (layout_->needprotect &&
1442 icit->inset->lyxCode() == FOOT_CODE)
1443 features.require("NeedLyXFootnoteCode");
1447 // then the contents
1448 for (pos_type i = 0; i < int(text_.size()) ; ++i) {
1449 BufferEncodings::validate(text_[i], features);
1453 /////////////////////////////////////////////////////////////////////
1457 /////////////////////////////////////////////////////////////////////
1460 Layout const emptyParagraphLayout;
1463 Paragraph::Paragraph()
1464 : d(new Paragraph::Private(this, emptyParagraphLayout))
1471 Paragraph::Paragraph(Paragraph const & par)
1472 : itemdepth(par.itemdepth),
1473 d(new Paragraph::Private(*par.d, this))
1479 Paragraph::Paragraph(Paragraph const & par, pos_type beg, pos_type end)
1480 : itemdepth(par.itemdepth),
1481 d(new Paragraph::Private(*par.d, this, beg, end))
1487 Paragraph & Paragraph::operator=(Paragraph const & par)
1489 // needed as we will destroy the private part before copying it
1491 itemdepth = par.itemdepth;
1495 d = new Private(*par.d, this);
1502 Paragraph::~Paragraph()
1511 // this shall be called just before every "os << ..." action.
1512 void flushString(ostream & os, docstring & s)
1521 void Paragraph::write(ostream & os, BufferParams const & bparams,
1522 depth_type & dth) const
1524 // The beginning or end of a deeper (i.e. nested) area?
1525 if (dth != d->params_.depth()) {
1526 if (d->params_.depth() > dth) {
1527 while (d->params_.depth() > dth) {
1528 os << "\n\\begin_deeper";
1532 while (d->params_.depth() < dth) {
1533 os << "\n\\end_deeper";
1539 // First write the layout
1540 os << "\n\\begin_layout " << to_utf8(d->layout_->name()) << '\n';
1542 d->params_.write(os);
1544 Font font1(inherit_font, bparams.language);
1546 Change running_change = Change(Change::UNCHANGED);
1548 // this string is used as a buffer to avoid repetitive calls
1549 // to to_utf8(), which turn out to be expensive (JMarc)
1550 docstring write_buffer;
1553 for (pos_type i = 0; i <= size(); ++i) {
1555 Change const & change = lookupChange(i);
1556 if (change != running_change)
1557 flushString(os, write_buffer);
1558 Changes::lyxMarkChange(os, bparams, column, running_change, change);
1559 running_change = change;
1564 // Write font changes
1565 Font font2 = getFontSettings(bparams, i);
1566 if (font2 != font1) {
1567 flushString(os, write_buffer);
1568 font2.lyxWriteChanges(font1, os);
1573 char_type const c = d->text_[i];
1576 if (Inset const * inset = getInset(i)) {
1577 flushString(os, write_buffer);
1578 if (inset->directWrite()) {
1579 // international char, let it write
1580 // code directly so it's shorter in
1586 os << "\\begin_inset ";
1588 os << "\n\\end_inset\n\n";
1591 // FIXME This can be removed again once the mystery
1592 // crash has been resolved.
1597 flushString(os, write_buffer);
1598 os << "\n\\backslash\n";
1602 flushString(os, write_buffer);
1603 if (i + 1 < size() && d->text_[i + 1] == ' ') {
1610 if ((column > 70 && c == ' ')
1612 flushString(os, write_buffer);
1616 // this check is to amend a bug. LyX sometimes
1617 // inserts '\0' this could cause problems.
1619 write_buffer.push_back(c);
1621 LYXERR0("NUL char in structure.");
1627 flushString(os, write_buffer);
1628 os << "\n\\end_layout\n";
1629 // FIXME This can be removed again once the mystery
1630 // crash has been resolved.
1635 void Paragraph::validate(LaTeXFeatures & features) const
1637 d->validate(features);
1641 void Paragraph::insert(pos_type start, docstring const & str,
1642 Font const & font, Change const & change)
1644 for (size_t i = 0, n = str.size(); i != n ; ++i)
1645 insertChar(start + i, str[i], font, change);
1649 void Paragraph::appendChar(char_type c, Font const & font,
1650 Change const & change)
1653 d->changes_.insert(change, d->text_.size());
1654 // when appending characters, no need to update tables
1655 d->text_.push_back(c);
1656 setFont(d->text_.size() - 1, font);
1657 d->requestSpellCheck(d->text_.size() - 1);
1661 void Paragraph::appendString(docstring const & s, Font const & font,
1662 Change const & change)
1664 pos_type end = s.size();
1665 size_t oldsize = d->text_.size();
1666 size_t newsize = oldsize + end;
1667 size_t capacity = d->text_.capacity();
1668 if (newsize >= capacity)
1669 d->text_.reserve(max(capacity + 100, newsize));
1671 // when appending characters, no need to update tables
1674 // FIXME: Optimize this!
1675 for (size_t i = oldsize; i != newsize; ++i) {
1677 d->changes_.insert(change, i);
1678 d->requestSpellCheck(i);
1680 d->fontlist_.set(oldsize, font);
1681 d->fontlist_.set(newsize - 1, font);
1685 void Paragraph::insertChar(pos_type pos, char_type c,
1688 d->insertChar(pos, c, Change(trackChanges ?
1689 Change::INSERTED : Change::UNCHANGED));
1693 void Paragraph::insertChar(pos_type pos, char_type c,
1694 Font const & font, bool trackChanges)
1696 d->insertChar(pos, c, Change(trackChanges ?
1697 Change::INSERTED : Change::UNCHANGED));
1702 void Paragraph::insertChar(pos_type pos, char_type c,
1703 Font const & font, Change const & change)
1705 d->insertChar(pos, c, change);
1710 void Paragraph::resetFonts(Font const & font)
1712 d->fontlist_.clear();
1713 d->fontlist_.set(0, font);
1714 d->fontlist_.set(d->text_.size() - 1, font);
1717 // Gets uninstantiated font setting at position.
1718 Font const & Paragraph::getFontSettings(BufferParams const & bparams,
1722 LYXERR0("pos: " << pos << " size: " << size());
1726 FontList::const_iterator cit = d->fontlist_.fontIterator(pos);
1727 if (cit != d->fontlist_.end())
1730 if (pos == size() && !empty())
1731 return getFontSettings(bparams, pos - 1);
1733 // Optimisation: avoid a full font instantiation if there is no
1734 // language change from previous call.
1735 static Font previous_font;
1736 static Language const * previous_lang = 0;
1737 Language const * lang = getParLanguage(bparams);
1738 if (lang != previous_lang) {
1739 previous_lang = lang;
1740 previous_font = Font(inherit_font, lang);
1742 return previous_font;
1746 FontSpan Paragraph::fontSpan(pos_type pos) const
1748 LBUFERR(pos <= size());
1751 return FontSpan(pos, pos);
1754 FontList::const_iterator cit = d->fontlist_.begin();
1755 FontList::const_iterator end = d->fontlist_.end();
1756 for (; cit != end; ++cit) {
1757 if (cit->pos() >= pos) {
1758 if (pos >= beginOfBody())
1759 return FontSpan(max(start, beginOfBody()),
1762 return FontSpan(start,
1763 min(beginOfBody() - 1,
1766 start = cit->pos() + 1;
1769 // This should not happen, but if so, we take no chances.
1770 LYXERR0("Paragraph::fontSpan: position not found in fontinfo table!");
1771 LASSERT(false, return FontSpan(pos, pos));
1775 // Gets uninstantiated font setting at position 0
1776 Font const & Paragraph::getFirstFontSettings(BufferParams const & bparams) const
1778 if (!empty() && !d->fontlist_.empty())
1779 return d->fontlist_.begin()->font();
1781 // Optimisation: avoid a full font instantiation if there is no
1782 // language change from previous call.
1783 static Font previous_font;
1784 static Language const * previous_lang = 0;
1785 if (bparams.language != previous_lang) {
1786 previous_lang = bparams.language;
1787 previous_font = Font(inherit_font, bparams.language);
1790 return previous_font;
1794 // Gets the fully instantiated font at a given position in a paragraph
1795 // This is basically the same function as Text::GetFont() in text2.cpp.
1796 // The difference is that this one is used for generating the LaTeX file,
1797 // and thus cosmetic "improvements" are disallowed: This has to deliver
1798 // the true picture of the buffer. (Asger)
1799 Font const Paragraph::getFont(BufferParams const & bparams, pos_type pos,
1800 Font const & outerfont) const
1804 Font font = getFontSettings(bparams, pos);
1806 pos_type const body_pos = beginOfBody();
1807 FontInfo & fi = font.fontInfo();
1809 fi.realize(d->layout_->labelfont);
1811 fi.realize(d->layout_->font);
1813 fi.realize(outerfont.fontInfo());
1814 fi.realize(bparams.getFont().fontInfo());
1820 Font const Paragraph::getLabelFont
1821 (BufferParams const & bparams, Font const & outerfont) const
1823 FontInfo tmpfont = d->layout_->labelfont;
1824 tmpfont.realize(outerfont.fontInfo());
1825 tmpfont.realize(bparams.getFont().fontInfo());
1826 return Font(tmpfont, getParLanguage(bparams));
1830 Font const Paragraph::getLayoutFont
1831 (BufferParams const & bparams, Font const & outerfont) const
1833 FontInfo tmpfont = d->layout_->font;
1834 tmpfont.realize(outerfont.fontInfo());
1835 tmpfont.realize(bparams.getFont().fontInfo());
1836 return Font(tmpfont, getParLanguage(bparams));
1840 char_type Paragraph::getUChar(BufferParams const & bparams, pos_type pos) const
1842 char_type c = d->text_[pos];
1843 if (!getFontSettings(bparams, pos).isRightToLeft())
1846 // FIXME: The arabic special casing is due to the difference of arabic
1847 // round brackets input introduced in r18599. Check if this should be
1848 // unified with Hebrew or at least if all bracket types should be
1849 // handled the same (file format change in either case).
1850 string const & lang = getFontSettings(bparams, pos).language()->lang();
1851 bool const arabic = lang == "arabic_arabtex" || lang == "arabic_arabi"
1856 uc = arabic ? c : ')';
1859 uc = arabic ? c : '(';
1885 void Paragraph::setFont(pos_type pos, Font const & font)
1887 LASSERT(pos <= size(), return);
1889 // First, reduce font against layout/label font
1890 // Update: The setCharFont() routine in text2.cpp already
1891 // reduces font, so we don't need to do that here. (Asger)
1893 d->fontlist_.set(pos, font);
1897 void Paragraph::makeSameLayout(Paragraph const & par)
1899 d->layout_ = par.d->layout_;
1900 d->params_ = par.d->params_;
1904 bool Paragraph::stripLeadingSpaces(bool trackChanges)
1906 if (isFreeSpacing())
1912 while (pos < size() && (isNewline(pos) || isLineSeparator(pos))) {
1913 if (eraseChar(pos, trackChanges))
1919 return count > 0 || pos > 0;
1923 bool Paragraph::hasSameLayout(Paragraph const & par) const
1925 return par.d->layout_ == d->layout_
1926 && d->params_.sameLayout(par.d->params_);
1930 depth_type Paragraph::getDepth() const
1932 return d->params_.depth();
1936 depth_type Paragraph::getMaxDepthAfter() const
1938 if (d->layout_->isEnvironment())
1939 return d->params_.depth() + 1;
1941 return d->params_.depth();
1945 char Paragraph::getAlign() const
1947 if (d->params_.align() == LYX_ALIGN_LAYOUT)
1948 return d->layout_->align;
1950 return d->params_.align();
1954 docstring const & Paragraph::labelString() const
1956 return d->params_.labelString();
1960 // the next two functions are for the manual labels
1961 docstring const Paragraph::getLabelWidthString() const
1963 if (d->layout_->margintype == MARGIN_MANUAL
1964 || d->layout_->latextype == LATEX_BIB_ENVIRONMENT)
1965 return d->params_.labelWidthString();
1967 return _("Senseless with this layout!");
1971 void Paragraph::setLabelWidthString(docstring const & s)
1973 d->params_.labelWidthString(s);
1977 docstring Paragraph::expandLabel(Layout const & layout,
1978 BufferParams const & bparams) const
1980 return expandParagraphLabel(layout, bparams, true);
1984 docstring Paragraph::expandDocBookLabel(Layout const & layout,
1985 BufferParams const & bparams) const
1987 return expandParagraphLabel(layout, bparams, false);
1991 docstring Paragraph::expandParagraphLabel(Layout const & layout,
1992 BufferParams const & bparams, bool process_appendix) const
1994 DocumentClass const & tclass = bparams.documentClass();
1995 string const & lang = getParLanguage(bparams)->code();
1996 bool const in_appendix = process_appendix && d->params_.appendix();
1997 docstring fmt = translateIfPossible(layout.labelstring(in_appendix), lang);
1999 if (fmt.empty() && !layout.counter.empty())
2000 return tclass.counters().theCounter(layout.counter, lang);
2002 // handle 'inherited level parts' in 'fmt',
2003 // i.e. the stuff between '@' in '@Section@.\arabic{subsection}'
2004 size_t const i = fmt.find('@', 0);
2005 if (i != docstring::npos) {
2006 size_t const j = fmt.find('@', i + 1);
2007 if (j != docstring::npos) {
2008 docstring parent(fmt, i + 1, j - i - 1);
2009 docstring label = from_ascii("??");
2010 if (tclass.hasLayout(parent))
2011 label = expandParagraphLabel(tclass[parent], bparams,
2013 fmt = docstring(fmt, 0, i) + label
2014 + docstring(fmt, j + 1, docstring::npos);
2018 return tclass.counters().counterLabel(fmt, lang);
2022 void Paragraph::applyLayout(Layout const & new_layout)
2024 d->layout_ = &new_layout;
2025 LyXAlignment const oldAlign = d->params_.align();
2027 if (!(oldAlign & d->layout_->alignpossible)) {
2028 frontend::Alert::warning(_("Alignment not permitted"),
2029 _("The new layout does not permit the alignment previously used.\nSetting to default."));
2030 d->params_.align(LYX_ALIGN_LAYOUT);
2035 pos_type Paragraph::beginOfBody() const
2037 return d->begin_of_body_;
2041 void Paragraph::setBeginOfBody()
2043 if (d->layout_->labeltype != LABEL_MANUAL) {
2044 d->begin_of_body_ = 0;
2048 // Unroll the first two cycles of the loop
2049 // and remember the previous character to
2050 // remove unnecessary getChar() calls
2052 pos_type end = size();
2053 if (i < end && !(isNewline(i) || isEnvSeparator(i))) {
2056 char_type previous_char = d->text_[i];
2057 if (!(isNewline(i) || isEnvSeparator(i))) {
2059 while (i < end && previous_char != ' ') {
2060 char_type temp = d->text_[i];
2061 if (isNewline(i) || isEnvSeparator(i))
2064 previous_char = temp;
2070 d->begin_of_body_ = i;
2074 bool Paragraph::allowParagraphCustomization() const
2076 return inInset().allowParagraphCustomization();
2080 bool Paragraph::usePlainLayout() const
2082 return inInset().usePlainLayout();
2086 bool Paragraph::isPassThru() const
2088 return inInset().isPassThru() || d->layout_->pass_thru;
2093 // paragraphs inside floats need different alignment tags to avoid
2096 bool noTrivlistCentering(InsetCode code)
2098 return code == FLOAT_CODE
2099 || code == WRAP_CODE
2100 || code == CELL_CODE;
2104 string correction(string const & orig)
2106 if (orig == "flushleft")
2107 return "raggedright";
2108 if (orig == "flushright")
2109 return "raggedleft";
2110 if (orig == "center")
2116 bool corrected_env(otexstream & os, string const & suffix, string const & env,
2117 InsetCode code, bool const lastpar, int & col)
2119 string macro = suffix + "{";
2120 if (noTrivlistCentering(code)) {
2122 // the last paragraph in non-trivlist-aligned
2123 // context is special (to avoid unwanted whitespace)
2124 if (suffix == "\\begin") {
2125 macro = "\\" + correction(env) + "{}";
2126 os << from_ascii(macro);
2127 col += macro.size();
2132 macro += correction(env);
2136 if (suffix == "\\par\\end") {
2140 os << from_ascii(macro);
2141 col += macro.size();
2142 if (suffix == "\\begin") {
2152 int Paragraph::Private::startTeXParParams(BufferParams const & bparams,
2153 otexstream & os, OutputParams const & runparams) const
2158 (bparams.paragraph_separation == BufferParams::ParagraphIndentSeparation) ?
2159 (layout_->toggle_indent != ITOGGLE_NEVER) :
2160 (layout_->toggle_indent == ITOGGLE_ALWAYS);
2162 if (canindent && params_.noindent() && !layout_->pass_thru) {
2163 os << "\\noindent ";
2167 LyXAlignment const curAlign = params_.align();
2169 if (curAlign == layout_->align)
2173 case LYX_ALIGN_NONE:
2174 case LYX_ALIGN_BLOCK:
2175 case LYX_ALIGN_LAYOUT:
2176 case LYX_ALIGN_SPECIAL:
2177 case LYX_ALIGN_DECIMAL:
2179 case LYX_ALIGN_LEFT:
2180 case LYX_ALIGN_RIGHT:
2181 case LYX_ALIGN_CENTER:
2182 if (runparams.moving_arg) {
2189 string const begin_tag = "\\begin";
2190 InsetCode code = ownerCode();
2191 bool const lastpar = runparams.isLastPar;
2194 case LYX_ALIGN_NONE:
2195 case LYX_ALIGN_BLOCK:
2196 case LYX_ALIGN_LAYOUT:
2197 case LYX_ALIGN_SPECIAL:
2198 case LYX_ALIGN_DECIMAL:
2200 case LYX_ALIGN_LEFT: {
2201 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2202 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2204 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2206 } case LYX_ALIGN_RIGHT: {
2207 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2208 corrected_env(os, begin_tag, "flushright", code, lastpar, column);
2210 corrected_env(os, begin_tag, "flushleft", code, lastpar, column);
2212 } case LYX_ALIGN_CENTER: {
2213 corrected_env(os, begin_tag, "center", code, lastpar, column);
2222 bool Paragraph::Private::endTeXParParams(BufferParams const & bparams,
2223 otexstream & os, OutputParams const & runparams) const
2225 LyXAlignment const curAlign = params_.align();
2227 if (curAlign == layout_->align)
2231 case LYX_ALIGN_NONE:
2232 case LYX_ALIGN_BLOCK:
2233 case LYX_ALIGN_LAYOUT:
2234 case LYX_ALIGN_SPECIAL:
2235 case LYX_ALIGN_DECIMAL:
2237 case LYX_ALIGN_LEFT:
2238 case LYX_ALIGN_RIGHT:
2239 case LYX_ALIGN_CENTER:
2240 if (runparams.moving_arg)
2245 bool output = false;
2247 string const end_tag = "\\par\\end";
2248 InsetCode code = ownerCode();
2249 bool const lastpar = runparams.isLastPar;
2252 case LYX_ALIGN_NONE:
2253 case LYX_ALIGN_BLOCK:
2254 case LYX_ALIGN_LAYOUT:
2255 case LYX_ALIGN_SPECIAL:
2256 case LYX_ALIGN_DECIMAL:
2258 case LYX_ALIGN_LEFT: {
2259 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2260 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2262 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2264 } case LYX_ALIGN_RIGHT: {
2265 if (owner_->getParLanguage(bparams)->babel() != "hebrew")
2266 output = corrected_env(os, end_tag, "flushright", code, lastpar, col);
2268 output = corrected_env(os, end_tag, "flushleft", code, lastpar, col);
2270 } case LYX_ALIGN_CENTER: {
2271 corrected_env(os, end_tag, "center", code, lastpar, col);
2276 return output || lastpar;
2280 // This one spits out the text of the paragraph
2281 void Paragraph::latex(BufferParams const & bparams,
2282 Font const & outerfont,
2284 OutputParams const & runparams,
2285 int start_pos, int end_pos, bool force) const
2287 LYXERR(Debug::LATEX, "Paragraph::latex... " << this);
2289 // FIXME This check should not be needed. Perhaps issue an
2290 // error if it triggers.
2291 Layout const & style = inInset().forcePlainLayout() ?
2292 bparams.documentClass().plainLayout() : *d->layout_;
2294 if (!force && style.inpreamble)
2297 bool const allowcust = allowParagraphCustomization();
2299 // Current base font for all inherited font changes, without any
2300 // change caused by an individual character, except for the language:
2301 // It is set to the language of the first character.
2302 // As long as we are in the label, this font is the base font of the
2303 // label. Before the first body character it is set to the base font
2307 // Maybe we have to create a optional argument.
2308 pos_type body_pos = beginOfBody();
2309 unsigned int column = 0;
2312 // the optional argument is kept in curly brackets in
2313 // case it contains a ']'
2314 // This is not strictly needed, but if this is changed it
2315 // would be a file format change, and tex2lyx would need
2316 // to be adjusted, since it unconditionally removes the
2317 // braces when it parses \item.
2320 basefont = getLabelFont(bparams, outerfont);
2322 basefont = getLayoutFont(bparams, outerfont);
2325 // Which font is currently active?
2326 Font running_font(basefont);
2327 // Do we have an open font change?
2328 bool open_font = false;
2330 Change runningChange = Change(Change::UNCHANGED);
2332 Encoding const * const prev_encoding = runparams.encoding;
2334 os.texrow().start(id(), 0);
2336 // if the paragraph is empty, the loop will not be entered at all
2338 if (style.isCommand()) {
2342 if (!style.leftdelim().empty()) {
2343 os << style.leftdelim();
2344 column += style.leftdelim().size();
2347 column += d->startTeXParParams(bparams, os, runparams);
2350 for (pos_type i = 0; i < size(); ++i) {
2351 // First char in paragraph or after label?
2352 if (i == body_pos) {
2355 column += running_font.latexWriteEndChanges(
2356 os, bparams, runparams,
2357 basefont, basefont);
2360 basefont = getLayoutFont(bparams, outerfont);
2361 running_font = basefont;
2363 column += Changes::latexMarkChange(os, bparams,
2364 runningChange, Change(Change::UNCHANGED),
2366 runningChange = Change(Change::UNCHANGED);
2371 if (style.isCommand()) {
2376 if (!style.leftdelim().empty()) {
2377 os << style.leftdelim();
2378 column += style.leftdelim().size();
2382 column += d->startTeXParParams(bparams, os,
2386 Change const & change = runparams.inDeletedInset
2387 ? runparams.changeOfDeletedInset : lookupChange(i);
2389 if (bparams.output_changes && runningChange != change) {
2391 column += running_font.latexWriteEndChanges(
2392 os, bparams, runparams, basefont, basefont);
2395 basefont = getLayoutFont(bparams, outerfont);
2396 running_font = basefont;
2398 column += Changes::latexMarkChange(os, bparams, runningChange,
2400 runningChange = change;
2403 // do not output text which is marked deleted
2404 // if change tracking output is disabled
2405 if (!bparams.output_changes && change.deleted()) {
2411 // Fully instantiated font
2412 Font const font = getFont(bparams, i, outerfont);
2414 Font const last_font = running_font;
2416 // Do we need to close the previous font?
2418 (font != running_font ||
2419 font.language() != running_font.language()))
2421 column += running_font.latexWriteEndChanges(
2422 os, bparams, runparams, basefont,
2423 (i == body_pos-1) ? basefont : font);
2424 running_font = basefont;
2428 string const running_lang = runparams.use_polyglossia ?
2429 running_font.language()->polyglossia() : running_font.language()->babel();
2430 // close babel's font environment before opening CJK.
2431 string const lang_end_command = runparams.use_polyglossia ?
2432 "\\end{$$lang}" : lyxrc.language_command_end;
2433 if (!running_lang.empty() &&
2434 font.language()->encoding()->package() == Encoding::CJK) {
2435 string end_tag = subst(lang_end_command,
2438 os << from_ascii(end_tag);
2439 column += end_tag.length();
2442 // Switch file encoding if necessary (and allowed)
2443 if (!runparams.pass_thru && !style.pass_thru &&
2444 runparams.encoding->package() != Encoding::none &&
2445 font.language()->encoding()->package() != Encoding::none) {
2446 pair<bool, int> const enc_switch =
2447 switchEncoding(os.os(), bparams, runparams,
2448 *(font.language()->encoding()));
2449 if (enc_switch.first) {
2450 column += enc_switch.second;
2451 runparams.encoding = font.language()->encoding();
2455 char_type const c = d->text_[i];
2457 // Do we need to change font?
2458 if ((font != running_font ||
2459 font.language() != running_font.language()) &&
2462 odocstringstream ods;
2463 column += font.latexWriteStartChanges(ods, bparams,
2464 runparams, basefont,
2466 running_font = font;
2468 docstring fontchange = ods.str();
2469 // check whether the fontchange ends with a \\textcolor
2470 // modifier and the text starts with a space (bug 4473)
2471 docstring const last_modifier = rsplit(fontchange, '\\');
2472 if (prefixIs(last_modifier, from_ascii("textcolor")) && c == ' ')
2473 os << fontchange << from_ascii("{}");
2474 // check if the fontchange ends with a trailing blank
2475 // (like "\small " (see bug 3382)
2476 else if (suffixIs(fontchange, ' ') && c == ' ')
2477 os << fontchange.substr(0, fontchange.size() - 1)
2478 << from_ascii("{}");
2483 // FIXME: think about end_pos implementation...
2484 if (c == ' ' && i >= start_pos && (end_pos == -1 || i < end_pos)) {
2485 // FIXME: integrate this case in latexSpecialChar
2486 // Do not print the separation of the optional argument
2487 // if style.pass_thru is false. This works because
2488 // latexSpecialChar ignores spaces if
2489 // style.pass_thru is false.
2490 if (i != body_pos - 1) {
2491 if (d->simpleTeXBlanks(runparams, os,
2492 i, column, font, style)) {
2493 // A surrogate pair was output. We
2494 // must not call latexSpecialChar
2495 // in this iteration, since it would output
2496 // the combining character again.
2503 OutputParams rp = runparams;
2504 rp.free_spacing = style.free_spacing;
2505 rp.local_font = &font;
2506 rp.intitle = style.intitle;
2508 // Two major modes: LaTeX or plain
2509 // Handle here those cases common to both modes
2510 // and then split to handle the two modes separately.
2511 if (c == META_INSET) {
2512 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2513 d->latexInset(bparams, os, rp, running_font,
2514 basefont, outerfont, open_font,
2515 runningChange, style, i, column);
2518 if (i >= start_pos && (end_pos == -1 || i < end_pos)) {
2520 d->latexSpecialChar(os, bparams, rp, running_font, runningChange,
2521 style, i, end_pos, column);
2522 } catch (EncodingException & e) {
2523 if (runparams.dryrun) {
2524 os << "<" << _("LyX Warning: ")
2525 << _("uncodable character") << " '";
2529 // add location information and throw again.
2538 // Set the encoding to that returned from latexSpecialChar (see
2539 // comment for encoding member in OutputParams.h)
2540 runparams.encoding = rp.encoding;
2543 // If we have an open font definition, we have to close it
2545 #ifdef FIXED_LANGUAGE_END_DETECTION
2547 running_font.latexWriteEndChanges(os, bparams,
2548 runparams, basefont,
2549 next_->getFont(bparams, 0, outerfont));
2551 running_font.latexWriteEndChanges(os, bparams,
2552 runparams, basefont, basefont);
2555 //FIXME: For now we ALWAYS have to close the foreign font settings if they are
2556 //FIXME: there as we start another \selectlanguage with the next paragraph if
2557 //FIXME: we are in need of this. This should be fixed sometime (Jug)
2558 running_font.latexWriteEndChanges(os, bparams, runparams,
2559 basefont, basefont);
2563 column += Changes::latexMarkChange(os, bparams, runningChange,
2564 Change(Change::UNCHANGED), runparams);
2566 // Needed if there is an optional argument but no contents.
2567 if (body_pos > 0 && body_pos == size()) {
2571 if (!style.rightdelim().empty()) {
2572 os << style.rightdelim();
2573 column += style.rightdelim().size();
2576 if (allowcust && d->endTeXParParams(bparams, os, runparams)
2577 && runparams.encoding != prev_encoding) {
2578 runparams.encoding = prev_encoding;
2579 os << setEncoding(prev_encoding->iconvName());
2582 LYXERR(Debug::LATEX, "Paragraph::latex... done " << this);
2586 bool Paragraph::emptyTag() const
2588 for (pos_type i = 0; i < size(); ++i) {
2589 if (Inset const * inset = getInset(i)) {
2590 InsetCode lyx_code = inset->lyxCode();
2591 // FIXME testing like that is wrong. What is
2593 if (lyx_code != TOC_CODE &&
2594 lyx_code != INCLUDE_CODE &&
2595 lyx_code != GRAPHICS_CODE &&
2596 lyx_code != ERT_CODE &&
2597 lyx_code != LISTINGS_CODE &&
2598 lyx_code != FLOAT_CODE &&
2599 lyx_code != TABULAR_CODE) {
2603 char_type c = d->text_[i];
2604 if (c != ' ' && c != '\t')
2612 string Paragraph::getID(Buffer const & buf, OutputParams const & runparams)
2615 for (pos_type i = 0; i < size(); ++i) {
2616 if (Inset const * inset = getInset(i)) {
2617 InsetCode lyx_code = inset->lyxCode();
2618 if (lyx_code == LABEL_CODE) {
2619 InsetLabel const * const il = static_cast<InsetLabel const *>(inset);
2620 docstring const & id = il->getParam("name");
2621 return "id='" + to_utf8(sgml::cleanID(buf, runparams, id)) + "'";
2629 pos_type Paragraph::firstWordDocBook(odocstream & os, OutputParams const & runparams)
2633 for (i = 0; i < size(); ++i) {
2634 if (Inset const * inset = getInset(i)) {
2635 inset->docbook(os, runparams);
2637 char_type c = d->text_[i];
2640 os << sgml::escapeChar(c);
2647 pos_type Paragraph::firstWordLyXHTML(XHTMLStream & xs, OutputParams const & runparams)
2651 for (i = 0; i < size(); ++i) {
2652 if (Inset const * inset = getInset(i)) {
2653 inset->xhtml(xs, runparams);
2655 char_type c = d->text_[i];
2665 bool Paragraph::Private::onlyText(Buffer const & buf, Font const & outerfont, pos_type initial) const
2668 pos_type size = text_.size();
2669 for (pos_type i = initial; i < size; ++i) {
2670 Font font = owner_->getFont(buf.params(), i, outerfont);
2671 if (text_[i] == META_INSET)
2673 if (i != initial && font != font_old)
2682 void Paragraph::simpleDocBookOnePar(Buffer const & buf,
2684 OutputParams const & runparams,
2685 Font const & outerfont,
2686 pos_type initial) const
2688 bool emph_flag = false;
2690 Layout const & style = *d->layout_;
2692 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2694 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2697 // parsing main loop
2698 for (pos_type i = initial; i < size(); ++i) {
2699 Font font = getFont(buf.params(), i, outerfont);
2701 // handle <emphasis> tag
2702 if (font_old.emph() != font.fontInfo().emph()) {
2703 if (font.fontInfo().emph() == FONT_ON) {
2706 } else if (i != initial) {
2707 os << "</emphasis>";
2712 if (Inset const * inset = getInset(i)) {
2713 inset->docbook(os, runparams);
2715 char_type c = d->text_[i];
2717 if (style.pass_thru)
2720 os << sgml::escapeChar(c);
2722 font_old = font.fontInfo();
2726 os << "</emphasis>";
2729 if (style.free_spacing)
2731 if (style.pass_thru && !d->onlyText(buf, outerfont, initial))
2737 void doFontSwitch(vector<html::FontTag> & tagsToOpen,
2738 vector<html::EndFontTag> & tagsToClose,
2739 bool & flag, FontState curstate, html::FontTypes type)
2741 if (curstate == FONT_ON) {
2742 tagsToOpen.push_back(html::FontTag(type));
2745 tagsToClose.push_back(html::EndFontTag(type));
2752 docstring Paragraph::simpleLyXHTMLOnePar(Buffer const & buf,
2754 OutputParams const & runparams,
2755 Font const & outerfont,
2756 pos_type initial) const
2760 // track whether we have opened these tags
2761 bool emph_flag = false;
2762 bool bold_flag = false;
2763 bool noun_flag = false;
2764 bool ubar_flag = false;
2765 bool dbar_flag = false;
2766 bool sout_flag = false;
2767 bool wave_flag = false;
2769 bool shap_flag = false;
2771 bool faml_flag = false;
2773 bool size_flag = false;
2775 Layout const & style = *d->layout_;
2777 xs.startParagraph(allowEmpty());
2780 style.labeltype == LABEL_MANUAL ? style.labelfont : style.font;
2782 FontShape curr_fs = INHERIT_SHAPE;
2783 FontFamily curr_fam = INHERIT_FAMILY;
2784 FontSize curr_size = FONT_SIZE_INHERIT;
2786 string const default_family =
2787 buf.masterBuffer()->params().fonts_default_family;
2789 vector<html::FontTag> tagsToOpen;
2790 vector<html::EndFontTag> tagsToClose;
2792 // parsing main loop
2793 for (pos_type i = initial; i < size(); ++i) {
2794 // let's not show deleted material in the output
2798 Font const font = getFont(buf.masterBuffer()->params(), i, outerfont);
2801 FontState curstate = font.fontInfo().emph();
2802 if (font_old.emph() != curstate)
2803 doFontSwitch(tagsToOpen, tagsToClose, emph_flag, curstate, html::FT_EMPH);
2806 curstate = font.fontInfo().noun();
2807 if (font_old.noun() != curstate)
2808 doFontSwitch(tagsToOpen, tagsToClose, noun_flag, curstate, html::FT_NOUN);
2811 curstate = font.fontInfo().underbar();
2812 if (font_old.underbar() != curstate)
2813 doFontSwitch(tagsToOpen, tagsToClose, ubar_flag, curstate, html::FT_UBAR);
2816 curstate = font.fontInfo().strikeout();
2817 if (font_old.strikeout() != curstate)
2818 doFontSwitch(tagsToOpen, tagsToClose, sout_flag, curstate, html::FT_SOUT);
2821 curstate = font.fontInfo().uuline();
2822 if (font_old.uuline() != curstate)
2823 doFontSwitch(tagsToOpen, tagsToClose, dbar_flag, curstate, html::FT_DBAR);
2826 curstate = font.fontInfo().uwave();
2827 if (font_old.uwave() != curstate)
2828 doFontSwitch(tagsToOpen, tagsToClose, wave_flag, curstate, html::FT_WAVE);
2831 // a little hackish, but allows us to reuse what we have.
2832 curstate = (font.fontInfo().series() == BOLD_SERIES ? FONT_ON : FONT_OFF);
2833 if (font_old.series() != font.fontInfo().series())
2834 doFontSwitch(tagsToOpen, tagsToClose, bold_flag, curstate, html::FT_BOLD);
2837 curr_fs = font.fontInfo().shape();
2838 FontShape old_fs = font_old.shape();
2839 if (old_fs != curr_fs) {
2843 tagsToClose.push_back(html::EndFontTag(html::FT_ITALIC));
2846 tagsToClose.push_back(html::EndFontTag(html::FT_SLANTED));
2848 case SMALLCAPS_SHAPE:
2849 tagsToClose.push_back(html::EndFontTag(html::FT_SMALLCAPS));
2855 // the other tags are for internal use
2863 tagsToOpen.push_back(html::FontTag(html::FT_ITALIC));
2867 tagsToOpen.push_back(html::FontTag(html::FT_SLANTED));
2870 case SMALLCAPS_SHAPE:
2871 tagsToOpen.push_back(html::FontTag(html::FT_SMALLCAPS));
2878 // the other tags are for internal use
2885 curr_fam = font.fontInfo().family();
2886 FontFamily old_fam = font_old.family();
2887 if (old_fam != curr_fam) {
2891 tagsToClose.push_back(html::EndFontTag(html::FT_ROMAN));
2894 tagsToClose.push_back(html::EndFontTag(html::FT_SANS));
2896 case TYPEWRITER_FAMILY:
2897 tagsToClose.push_back(html::EndFontTag(html::FT_TYPE));
2899 case INHERIT_FAMILY:
2902 // the other tags are for internal use
2910 // we will treat a "default" font family as roman, since we have
2911 // no other idea what to do.
2912 if (default_family != "rmdefault" && default_family != "default") {
2913 tagsToOpen.push_back(html::FontTag(html::FT_ROMAN));
2918 if (default_family != "sfdefault") {
2919 tagsToOpen.push_back(html::FontTag(html::FT_SANS));
2923 case TYPEWRITER_FAMILY:
2924 if (default_family != "ttdefault") {
2925 tagsToOpen.push_back(html::FontTag(html::FT_TYPE));
2929 case INHERIT_FAMILY:
2932 // the other tags are for internal use
2939 curr_size = font.fontInfo().size();
2940 FontSize old_size = font_old.size();
2941 if (old_size != curr_size) {
2944 case FONT_SIZE_TINY:
2945 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_TINY));
2947 case FONT_SIZE_SCRIPT:
2948 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SCRIPT));
2950 case FONT_SIZE_FOOTNOTE:
2951 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_FOOTNOTE));
2953 case FONT_SIZE_SMALL:
2954 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_SMALL));
2956 case FONT_SIZE_LARGE:
2957 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGE));
2959 case FONT_SIZE_LARGER:
2960 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGER));
2962 case FONT_SIZE_LARGEST:
2963 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_LARGEST));
2965 case FONT_SIZE_HUGE:
2966 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGE));
2968 case FONT_SIZE_HUGER:
2969 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_HUGER));
2971 case FONT_SIZE_INCREASE:
2972 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_INCREASE));
2974 case FONT_SIZE_DECREASE:
2975 tagsToClose.push_back(html::EndFontTag(html::FT_SIZE_DECREASE));
2977 case FONT_SIZE_INHERIT:
2978 case FONT_SIZE_NORMAL:
2981 // the other tags are for internal use
2987 switch (curr_size) {
2988 case FONT_SIZE_TINY:
2989 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_TINY));
2992 case FONT_SIZE_SCRIPT:
2993 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SCRIPT));
2996 case FONT_SIZE_FOOTNOTE:
2997 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_FOOTNOTE));
3000 case FONT_SIZE_SMALL:
3001 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_SMALL));
3004 case FONT_SIZE_LARGE:
3005 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGE));
3008 case FONT_SIZE_LARGER:
3009 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGER));
3012 case FONT_SIZE_LARGEST:
3013 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_LARGEST));
3016 case FONT_SIZE_HUGE:
3017 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGE));
3020 case FONT_SIZE_HUGER:
3021 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_HUGER));
3024 case FONT_SIZE_INCREASE:
3025 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_INCREASE));
3028 case FONT_SIZE_DECREASE:
3029 tagsToOpen.push_back(html::FontTag(html::FT_SIZE_DECREASE));
3032 case FONT_SIZE_NORMAL:
3033 case FONT_SIZE_INHERIT:
3036 // the other tags are for internal use
3043 // Other such tags? What about the other text ranges?
3045 vector<html::EndFontTag>::const_iterator cit = tagsToClose.begin();
3046 vector<html::EndFontTag>::const_iterator cen = tagsToClose.end();
3047 for (; cit != cen; ++cit)
3050 vector<html::FontTag>::const_iterator sit = tagsToOpen.begin();
3051 vector<html::FontTag>::const_iterator sen = tagsToOpen.end();
3052 for (; sit != sen; ++sit)
3055 tagsToClose.clear();
3058 Inset const * inset = getInset(i);
3060 if (!runparams.for_toc || inset->isInToc()) {
3061 OutputParams np = runparams;
3062 np.local_font = &font;
3063 if (!inset->getLayout().htmlisblock())
3064 np.html_in_par = true;
3065 retval += inset->xhtml(xs, np);
3068 char_type c = getUChar(buf.masterBuffer()->params(), i);
3071 font_old = font.fontInfo();
3080 bool Paragraph::isHfill(pos_type pos) const
3082 Inset const * inset = getInset(pos);
3083 return inset && inset->isHfill();
3087 bool Paragraph::isNewline(pos_type pos) const
3089 Inset const * inset = getInset(pos);
3090 return inset && inset->lyxCode() == NEWLINE_CODE;
3094 bool Paragraph::isEnvSeparator(pos_type pos) const
3096 Inset const * inset = getInset(pos);
3097 return inset && inset->lyxCode() == SEPARATOR_CODE;
3101 bool Paragraph::isLineSeparator(pos_type pos) const
3103 char_type const c = d->text_[pos];
3104 if (isLineSeparatorChar(c))
3106 Inset const * inset = getInset(pos);
3107 return inset && inset->isLineSeparator();
3111 bool Paragraph::isWordSeparator(pos_type pos) const
3115 if (Inset const * inset = getInset(pos))
3116 return !inset->isLetter();
3117 // if we have a hard hyphen (no en- or emdash) or apostrophe
3118 // we pass this to the spell checker
3119 // FIXME: this method is subject to change, visit
3120 // https://bugzilla.mozilla.org/show_bug.cgi?id=355178
3121 // to get an impression how complex this is.
3122 if (isHardHyphenOrApostrophe(pos))
3124 char_type const c = d->text_[pos];
3125 // We want to pass the escape chars to the spellchecker
3126 docstring const escape_chars = from_utf8(lyxrc.spellchecker_esc_chars);
3127 return !isLetterChar(c) && !isDigitASCII(c) && !contains(escape_chars, c);
3131 bool Paragraph::isHardHyphenOrApostrophe(pos_type pos) const
3133 pos_type const psize = size();
3136 char_type const c = d->text_[pos];
3137 if (c != '-' && c != '\'')
3139 int nextpos = pos + 1;
3140 int prevpos = pos > 0 ? pos - 1 : 0;
3141 if ((nextpos == psize || isSpace(nextpos))
3142 && (pos == 0 || isSpace(prevpos)))
3148 FontSpan const & Paragraph::getSpellRange(pos_type pos) const
3150 return d->speller_state_.getRange(pos);
3154 bool Paragraph::isChar(pos_type pos) const
3156 if (Inset const * inset = getInset(pos))
3157 return inset->isChar();
3158 char_type const c = d->text_[pos];
3159 return !isLetterChar(c) && !isDigitASCII(c) && !lyx::isSpace(c);
3163 bool Paragraph::isSpace(pos_type pos) const
3165 if (Inset const * inset = getInset(pos))
3166 return inset->isSpace();
3167 char_type const c = d->text_[pos];
3168 return lyx::isSpace(c);
3173 Paragraph::getParLanguage(BufferParams const & bparams) const
3176 return getFirstFontSettings(bparams).language();
3177 // FIXME: we should check the prev par as well (Lgb)
3178 return bparams.language;
3182 bool Paragraph::isRTL(BufferParams const & bparams) const
3184 return getParLanguage(bparams)->rightToLeft()
3185 && !inInset().getLayout().forceLTR();
3189 void Paragraph::changeLanguage(BufferParams const & bparams,
3190 Language const * from, Language const * to)
3192 // change language including dummy font change at the end
3193 for (pos_type i = 0; i <= size(); ++i) {
3194 Font font = getFontSettings(bparams, i);
3195 if (font.language() == from) {
3196 font.setLanguage(to);
3198 d->requestSpellCheck(i);
3204 bool Paragraph::isMultiLingual(BufferParams const & bparams) const
3206 Language const * doc_language = bparams.language;
3207 FontList::const_iterator cit = d->fontlist_.begin();
3208 FontList::const_iterator end = d->fontlist_.end();
3210 for (; cit != end; ++cit)
3211 if (cit->font().language() != ignore_language &&
3212 cit->font().language() != latex_language &&
3213 cit->font().language() != doc_language)
3219 void Paragraph::getLanguages(std::set<Language const *> & languages) const
3221 FontList::const_iterator cit = d->fontlist_.begin();
3222 FontList::const_iterator end = d->fontlist_.end();
3224 for (; cit != end; ++cit) {
3225 Language const * lang = cit->font().language();
3226 if (lang != ignore_language &&
3227 lang != latex_language)
3228 languages.insert(lang);
3233 docstring Paragraph::asString(int options) const
3235 return asString(0, size(), options);
3239 docstring Paragraph::asString(pos_type beg, pos_type end, int options, const OutputParams *runparams) const
3241 odocstringstream os;
3244 && options & AS_STR_LABEL
3245 && !d->params_.labelString().empty())
3246 os << d->params_.labelString() << ' ';
3248 for (pos_type i = beg; i < end; ++i) {
3249 if ((options & AS_STR_SKIPDELETE) && isDeleted(i))
3251 char_type const c = d->text_[i];
3252 if (isPrintable(c) || c == '\t'
3253 || (c == '\n' && (options & AS_STR_NEWLINES)))
3255 else if (c == META_INSET && (options & AS_STR_INSETS)) {
3256 if (c == META_INSET && (options & AS_STR_PLAINTEXT)) {
3257 LASSERT(runparams != 0, return docstring());
3258 getInset(i)->plaintext(os, *runparams);
3260 getInset(i)->toString(os);
3269 void Paragraph::forOutliner(docstring & os, size_t const maxlen,
3270 bool const shorten) const
3272 size_t tmplen = shorten ? maxlen + 1 : maxlen;
3273 if (!d->params_.labelString().empty())
3274 os += d->params_.labelString() + ' ';
3275 for (pos_type i = 0; i < size() && os.length() < tmplen; ++i) {
3278 char_type const c = d->text_[i];
3281 else if (c == META_INSET)
3282 getInset(i)->forOutliner(os, tmplen, false);
3285 Text::shortenForOutliner(os, maxlen);
3289 void Paragraph::setInsetOwner(Inset const * inset)
3291 d->inset_owner_ = inset;
3295 int Paragraph::id() const
3301 void Paragraph::setId(int id)
3307 Layout const & Paragraph::layout() const
3313 void Paragraph::setLayout(Layout const & layout)
3315 d->layout_ = &layout;
3319 void Paragraph::setDefaultLayout(DocumentClass const & tc)
3321 setLayout(tc.defaultLayout());
3325 void Paragraph::setPlainLayout(DocumentClass const & tc)
3327 setLayout(tc.plainLayout());
3331 void Paragraph::setPlainOrDefaultLayout(DocumentClass const & tclass)
3333 if (usePlainLayout())
3334 setPlainLayout(tclass);
3336 setDefaultLayout(tclass);
3340 Inset const & Paragraph::inInset() const
3342 LBUFERR(d->inset_owner_);
3343 return *d->inset_owner_;
3347 ParagraphParameters & Paragraph::params()
3353 ParagraphParameters const & Paragraph::params() const
3359 bool Paragraph::isFreeSpacing() const
3361 if (d->layout_->free_spacing)
3363 return d->inset_owner_ && d->inset_owner_->isFreeSpacing();
3367 bool Paragraph::allowEmpty() const
3369 if (d->layout_->keepempty)
3371 return d->inset_owner_ && d->inset_owner_->allowEmpty();
3375 bool Paragraph::brokenBiblio() const
3377 // there is a problem if there is no bibitem at position 0 or
3378 // if there is another bibitem in the paragraph.
3379 return d->layout_->labeltype == LABEL_BIBLIO
3380 && (d->insetlist_.find(BIBITEM_CODE) != 0
3381 || d->insetlist_.find(BIBITEM_CODE, 1) > 0);
3385 int Paragraph::fixBiblio(Buffer const & buffer)
3387 // FIXME: What about the case where paragraph is not BIBLIO
3388 // but there is an InsetBibitem?
3389 // FIXME: when there was already an inset at 0, the return value is 1,
3390 // which does not tell whether another inset has been remove; the
3391 // cursor cannot be correctly updated.
3393 if (d->layout_->labeltype != LABEL_BIBLIO)
3396 bool const track_changes = buffer.params().track_changes;
3397 int bibitem_pos = d->insetlist_.find(BIBITEM_CODE);
3398 bool const hasbibitem0 = bibitem_pos == 0;
3401 bibitem_pos = d->insetlist_.find(BIBITEM_CODE, 1);
3402 // There was an InsetBibitem at pos 0, and no other one => OK
3403 if (bibitem_pos == -1)
3405 // there is a bibitem at the 0 position, but since
3406 // there is a second one, we copy the second on the
3407 // first. We're assuming there are at most two of
3408 // these, which there should be.
3409 // FIXME: why does it make sense to do that rather
3410 // than keep the first? (JMarc)
3411 Inset * inset = releaseInset(bibitem_pos);
3412 d->insetlist_.begin()->inset = inset;
3413 return -bibitem_pos;
3416 // We need to create an inset at the beginning
3418 if (bibitem_pos > 0) {
3419 // there was one somewhere in the paragraph, let's move it
3420 inset = d->insetlist_.release(bibitem_pos);
3421 eraseChar(bibitem_pos, track_changes);
3424 inset = new InsetBibitem(const_cast<Buffer *>(&buffer),
3425 InsetCommandParams(BIBITEM_CODE));
3427 Font font(inherit_font, buffer.params().language);
3428 insertInset(0, inset, font, Change(track_changes ? Change::INSERTED
3429 : Change::UNCHANGED));
3435 void Paragraph::checkAuthors(AuthorList const & authorList)
3437 d->changes_.checkAuthors(authorList);
3441 bool Paragraph::isChanged(pos_type pos) const
3443 return lookupChange(pos).changed();
3447 bool Paragraph::isInserted(pos_type pos) const
3449 return lookupChange(pos).inserted();
3453 bool Paragraph::isDeleted(pos_type pos) const
3455 return lookupChange(pos).deleted();
3459 InsetList const & Paragraph::insetList() const
3461 return d->insetlist_;
3465 void Paragraph::setBuffer(Buffer & b)
3467 d->insetlist_.setBuffer(b);
3471 Inset * Paragraph::releaseInset(pos_type pos)
3473 Inset * inset = d->insetlist_.release(pos);
3474 /// does not honour change tracking!
3475 eraseChar(pos, false);
3480 Inset * Paragraph::getInset(pos_type pos)
3482 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3483 ? d->insetlist_.get(pos) : 0;
3487 Inset const * Paragraph::getInset(pos_type pos) const
3489 return (pos < pos_type(d->text_.size()) && d->text_[pos] == META_INSET)
3490 ? d->insetlist_.get(pos) : 0;
3494 void Paragraph::changeCase(BufferParams const & bparams, pos_type pos,
3495 pos_type & right, TextCase action)
3497 // process sequences of modified characters; in change
3498 // tracking mode, this approach results in much better
3499 // usability than changing case on a char-by-char basis
3500 // We also need to track the current font, since font
3501 // changes within sequences can occur.
3502 vector<pair<char_type, Font> > changes;
3504 bool const trackChanges = bparams.track_changes;
3506 bool capitalize = true;
3508 for (; pos < right; ++pos) {
3509 char_type oldChar = d->text_[pos];
3510 char_type newChar = oldChar;
3512 // ignore insets and don't play with deleted text!
3513 if (oldChar != META_INSET && !isDeleted(pos)) {
3515 case text_lowercase:
3516 newChar = lowercase(oldChar);
3518 case text_capitalization:
3520 newChar = uppercase(oldChar);
3524 case text_uppercase:
3525 newChar = uppercase(oldChar);
3530 if (isWordSeparator(pos) || isDeleted(pos)) {
3531 // permit capitalization again
3535 if (oldChar != newChar) {
3536 changes.push_back(make_pair(newChar, getFontSettings(bparams, pos)));
3537 if (pos != right - 1)
3539 // step behind the changing area
3543 int erasePos = pos - changes.size();
3544 for (size_t i = 0; i < changes.size(); i++) {
3545 insertChar(pos, changes[i].first,
3548 if (!eraseChar(erasePos, trackChanges)) {
3551 ++right; // expand selection
3559 int Paragraph::find(docstring const & str, bool cs, bool mw,
3560 pos_type start_pos, bool del) const
3562 pos_type pos = start_pos;
3563 int const strsize = str.length();
3565 pos_type const parsize = d->text_.size();
3566 for (i = 0; i < strsize && pos < parsize; ++i, ++pos) {
3567 // Ignore "invisible" letters such as ligature breaks
3568 // and hyphenation chars while searching
3569 while (pos < parsize - 1 && isInset(pos)) {
3570 odocstringstream os;
3571 getInset(pos)->toString(os);
3572 if (!getInset(pos)->isLetter() || !os.str().empty())
3576 if (cs && str[i] != d->text_[pos])
3578 if (!cs && uppercase(str[i]) != uppercase(d->text_[pos]))
3580 if (!del && isDeleted(pos))
3587 // if necessary, check whether string matches word
3589 if (start_pos > 0 && !isWordSeparator(start_pos - 1))
3592 && !isWordSeparator(pos))
3596 return pos - start_pos;
3600 char_type Paragraph::getChar(pos_type pos) const
3602 return d->text_[pos];
3606 pos_type Paragraph::size() const
3608 return d->text_.size();
3612 bool Paragraph::empty() const
3614 return d->text_.empty();
3618 bool Paragraph::isInset(pos_type pos) const
3620 return d->text_[pos] == META_INSET;
3624 bool Paragraph::isSeparator(pos_type pos) const
3626 //FIXME: Are we sure this can be the only separator?
3627 return d->text_[pos] == ' ';
3631 void Paragraph::deregisterWords()
3633 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3634 Private::LangWordsMap::const_iterator ite = d->words_.end();
3635 for (; itl != ite; ++itl) {
3636 WordList * wl = theWordList(itl->first);
3637 Private::Words::const_iterator it = (itl->second).begin();
3638 Private::Words::const_iterator et = (itl->second).end();
3639 for (; it != et; ++it)
3646 void Paragraph::locateWord(pos_type & from, pos_type & to,
3647 word_location const loc) const
3650 case WHOLE_WORD_STRICT:
3651 if (from == 0 || from == size()
3652 || isWordSeparator(from)
3653 || isWordSeparator(from - 1)) {
3657 // no break here, we go to the next
3660 // If we are already at the beginning of a word, do nothing
3661 if (!from || isWordSeparator(from - 1))
3663 // no break here, we go to the next
3666 // always move the cursor to the beginning of previous word
3667 while (from && !isWordSeparator(from - 1))
3671 LYXERR0("Paragraph::locateWord: NEXT_WORD not implemented yet");
3674 // no need to move the 'from' cursor
3678 while (to < size() && !isWordSeparator(to))
3683 void Paragraph::collectWords()
3685 for (pos_type pos = 0; pos < size(); ++pos) {
3686 if (isWordSeparator(pos))
3688 pos_type from = pos;
3689 locateWord(from, pos, WHOLE_WORD);
3690 // Work around MSVC warning: The statement
3691 // if (pos < from + lyxrc.completion_minlength)
3692 // triggers a signed vs. unsigned warning.
3693 // I don't know why this happens, it could be a MSVC bug, or
3694 // related to LLP64 (windows) vs. LP64 (unix) programming
3695 // model, or the C++ standard might be ambigous in the section
3696 // defining the "usual arithmetic conversions". However, using
3697 // a temporary variable is safe and works on all compilers.
3698 pos_type const endpos = from + lyxrc.completion_minlength;
3701 FontList::const_iterator cit = d->fontlist_.fontIterator(from);
3702 if (cit == d->fontlist_.end())
3704 Language const * lang = cit->font().language();
3705 docstring const word = asString(from, pos, AS_STR_NONE);
3706 d->words_[lang->lang()].insert(word);
3711 void Paragraph::registerWords()
3713 Private::LangWordsMap::const_iterator itl = d->words_.begin();
3714 Private::LangWordsMap::const_iterator ite = d->words_.end();
3715 for (; itl != ite; ++itl) {
3716 WordList * wl = theWordList(itl->first);
3717 Private::Words::const_iterator it = (itl->second).begin();
3718 Private::Words::const_iterator et = (itl->second).end();
3719 for (; it != et; ++it)
3725 void Paragraph::updateWords()
3733 void Paragraph::Private::appendSkipPosition(SkipPositions & skips, pos_type const pos) const
3735 SkipPositionsIterator begin = skips.begin();
3736 SkipPositions::iterator end = skips.end();
3737 if (pos > 0 && begin < end) {
3739 if (end->last == pos - 1) {
3744 skips.insert(end, FontSpan(pos, pos));
3748 Language * Paragraph::Private::locateSpellRange(
3749 pos_type & from, pos_type & to,
3750 SkipPositions & skips) const
3752 // skip leading white space
3753 while (from < to && owner_->isWordSeparator(from))
3755 // don't check empty range
3758 // get current language
3759 Language * lang = getSpellLanguage(from);
3760 pos_type last = from;
3761 bool samelang = true;
3762 bool sameinset = true;
3763 while (last < to && samelang && sameinset) {
3764 // hop to end of word
3765 while (last < to && !owner_->isWordSeparator(last)) {
3766 if (owner_->getInset(last)) {
3767 appendSkipPosition(skips, last);
3768 } else if (owner_->isDeleted(last)) {
3769 appendSkipPosition(skips, last);
3773 // hop to next word while checking for insets
3774 while (sameinset && last < to && owner_->isWordSeparator(last)) {
3775 if (Inset const * inset = owner_->getInset(last))
3776 sameinset = inset->isChar() && inset->isLetter();
3777 if (sameinset && owner_->isDeleted(last)) {
3778 appendSkipPosition(skips, last);
3783 if (sameinset && last < to) {
3784 // now check for language change
3785 samelang = lang == getSpellLanguage(last);
3788 // if language change detected backstep is needed
3796 Language * Paragraph::Private::getSpellLanguage(pos_type const from) const
3799 const_cast<Language *>(owner_->getFontSettings(
3800 inset_owner_->buffer().params(), from).language());
3801 if (lang == inset_owner_->buffer().params().language
3802 && !lyxrc.spellchecker_alt_lang.empty()) {
3804 string const lang_variety =
3805 split(lyxrc.spellchecker_alt_lang, lang_code, '-');
3806 lang->setCode(lang_code);
3807 lang->setVariety(lang_variety);
3813 void Paragraph::requestSpellCheck(pos_type pos)
3815 d->requestSpellCheck(pos);
3819 bool Paragraph::needsSpellCheck() const
3821 SpellChecker::ChangeNumber speller_change_number = 0;
3822 if (theSpellChecker())
3823 speller_change_number = theSpellChecker()->changeNumber();
3824 if (speller_change_number > d->speller_state_.currentChangeNumber()) {
3825 d->speller_state_.needsCompleteRefresh(speller_change_number);
3827 return d->needsSpellCheck();
3831 bool Paragraph::Private::ignoreWord(docstring const & word) const
3833 // Ignore words with digits
3834 // FIXME: make this customizable
3835 // (note that some checkers ignore words with digits by default)
3836 docstring::const_iterator cit = word.begin();
3837 docstring::const_iterator const end = word.end();
3838 for (; cit != end; ++cit) {
3839 if (isNumber((*cit)))
3846 SpellChecker::Result Paragraph::spellCheck(pos_type & from, pos_type & to,
3847 WordLangTuple & wl, docstring_list & suggestions,
3848 bool do_suggestion, bool check_learned) const
3850 SpellChecker::Result result = SpellChecker::WORD_OK;
3851 SpellChecker * speller = theSpellChecker();
3855 if (!d->layout_->spellcheck || !inInset().allowSpellCheck())
3858 locateWord(from, to, WHOLE_WORD);
3859 if (from == to || from >= size())
3862 docstring word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3863 Language * lang = d->getSpellLanguage(from);
3865 wl = WordLangTuple(word, lang);
3870 if (needsSpellCheck() || check_learned) {
3872 if (!d->ignoreWord(word)) {
3873 bool const trailing_dot = to < size() && d->text_[to] == '.';
3874 result = speller->check(wl);
3875 if (SpellChecker::misspelled(result) && trailing_dot) {
3876 wl = WordLangTuple(word.append(from_ascii(".")), lang);
3877 result = speller->check(wl);
3878 if (!SpellChecker::misspelled(result)) {
3879 LYXERR(Debug::GUI, "misspelled word is correct with dot: \"" <<
3881 from << ".." << to << "]");
3883 // spell check with dot appended failed too
3884 // restore original word/lang value
3885 word = asString(from, to, AS_STR_INSETS | AS_STR_SKIPDELETE);
3886 wl = WordLangTuple(word, lang);
3890 if (!SpellChecker::misspelled(result)) {
3891 // area up to the begin of the next word is not misspelled
3892 while (end < size() && isWordSeparator(end))
3895 d->setMisspelled(from, end, result);
3897 result = d->speller_state_.getState(from);
3901 suggestions.clear();
3903 if (SpellChecker::misspelled(result)) {
3904 LYXERR(Debug::GUI, "misspelled word: \"" <<
3906 from << ".." << to << "]");
3908 speller->suggest(wl, suggestions);
3914 void Paragraph::Private::markMisspelledWords(
3915 pos_type const & first, pos_type const & last,
3916 SpellChecker::Result result,
3917 docstring const & word,
3918 SkipPositions const & skips)
3920 if (!SpellChecker::misspelled(result)) {
3921 setMisspelled(first, last, SpellChecker::WORD_OK);
3925 SpellChecker * speller = theSpellChecker();
3926 // locate and enumerate the error positions
3927 int nerrors = speller->numMisspelledWords();
3929 SkipPositionsIterator it = skips.begin();
3930 SkipPositionsIterator et = skips.end();
3931 for (int index = 0; index < nerrors; ++index) {
3934 speller->misspelledWord(index, wstart, wlen);
3935 /// should not happen if speller supports range checks
3936 if (!wlen) continue;
3937 docstring const misspelled = word.substr(wstart, wlen);
3938 wstart += first + numskipped;
3939 if (snext < wstart) {
3940 /// mark the range of correct spelling
3941 numskipped += countSkips(it, et, wstart);
3942 setMisspelled(snext,
3943 wstart - 1, SpellChecker::WORD_OK);
3945 snext = wstart + wlen;
3946 numskipped += countSkips(it, et, snext);
3947 /// mark the range of misspelling
3948 setMisspelled(wstart, snext, result);
3949 LYXERR(Debug::GUI, "misspelled word: \"" <<
3950 misspelled << "\" [" <<
3951 wstart << ".." << (snext-1) << "]");
3954 if (snext <= last) {
3955 /// mark the range of correct spelling at end
3956 setMisspelled(snext, last, SpellChecker::WORD_OK);
3961 void Paragraph::spellCheck() const
3963 SpellChecker * speller = theSpellChecker();
3964 if (!speller || empty() ||!needsSpellCheck())
3968 d->rangeOfSpellCheck(start, endpos);
3969 if (speller->canCheckParagraph()) {
3970 // loop until we leave the range
3971 for (pos_type first = start; first < endpos; ) {
3972 pos_type last = endpos;
3973 Private::SkipPositions skips;
3974 Language * lang = d->locateSpellRange(first, last, skips);
3975 if (first >= endpos)
3977 // start the spell checker on the unit of meaning
3978 docstring word = asString(first, last, AS_STR_INSETS + AS_STR_SKIPDELETE);
3979 WordLangTuple wl = WordLangTuple(word, lang);
3980 SpellChecker::Result result = word.size() ?
3981 speller->check(wl) : SpellChecker::WORD_OK;
3982 d->markMisspelledWords(first, last, result, word, skips);
3986 static docstring_list suggestions;
3987 pos_type to = endpos;
3988 while (start < endpos) {
3990 spellCheck(start, to, wl, suggestions, false);
3994 d->readySpellCheck();
3998 bool Paragraph::isMisspelled(pos_type pos, bool check_boundary) const
4000 bool result = SpellChecker::misspelled(d->speller_state_.getState(pos));
4001 if (result || pos <= 0 || pos > size())
4003 if (check_boundary && (pos == size() || isWordSeparator(pos)))
4004 result = SpellChecker::misspelled(d->speller_state_.getState(pos - 1));
4009 string Paragraph::magicLabel() const
4012 ss << "magicparlabel-" << id();